"John A. Weeks III" wrote:

In article , Dick West
wrote:

I would first check those 4 transistors. Remember, they must be used in pairs
matched
for hFe, else DC offset results.

Did the original poster ever remove the MOSFET's from the unit. If
so, were they put back in exactly the same place? If not, it is
possible that the matched sets were mismatched in the process.
That would be enough to cause the problem that Dick is talking about.

-john-

John, if I recall parts of this thread correctly, the OP said that he also found
one of the DC power supply wires to the circuit card was not attached. Obviously,
this would cause a huge set of problems such as he described.

Now, we do know that the two N-channel MOSFETs on a channel must be matched on Vge
within 10%, which is why each MOSFET is marked with the Hafler grading number.
Ditto for the P-channel MOSFETs. These devices are not the same 4 to which I
referred. I referred to the 4 input differential amp circuit on the circuit card
which must be matched, not the 4 MOSFETs attached to the heatsink.

Let's wait to hear more from the OP before making anymore guesses about what is
wrong. Besides, without the use of a Variac his efforts are probably doomed to
failure.

Dick

John A. Weeks III wrote:

Did the original poster ever remove the MOSFET's from the unit. If
so, were they put back in exactly the same place? If not, it is
possible that the matched sets were mismatched in the process.
That would be enough to cause the problem that Dick is talking about.

Here I am, the original poster! :-)

I removed the MOSFETS at one point, and they all went back into the same
slots. Exactly.

And as for Dick's suggestions, let me recap and intersperse a few
questions.

1) Q9, the transistor in the bias circuit, was definitely fried. It was
measuring a hfE between 200 and 1700, depending on the time of day,
when we tested it out of circuit. (specs for it should be about 60).
And yes, the reading changed drastically from hour to hour.
2) Replacing it and fixing a broken solder joint seemed to fix things,
but after something less than three hours of playing, it died again.
This time the speaker fuse was shot as well, and when I replaced it,
I was back to the same scenario I had before replacing Q9. The
symptoms a
-20VDC at the speaker outputs.
No sound to the speaker. (But I'm not about to hook them up again!)
3) And for the record, for that brief while that the amp was working,
the DC output at the speaker terminals was tiny--definitely less
than 80mV. (with no input signal, of course)

So my plan is to check Q9 again, and I'm expecting to find it dead.
After that, I'll go after the Q1/Q2 pair, and see if they've drifted
apart. Next will be the opposite pair, Q5 and Q6 I believe (I don't
have the schematic in front of me). After that, it's going to be a
process of testing every resistor, diode, and cap on the board.

Then--MAYBE--I'll plug it in again.

Colin

John A. Weeks III wrote:

Did the original poster ever remove the MOSFET's from the unit. If
so, were they put back in exactly the same place? If not, it is
possible that the matched sets were mismatched in the process.
That would be enough to cause the problem that Dick is talking about.

Here I am, the original poster! :-)

I removed the MOSFETS at one point, and they all went back into the same
slots. Exactly.

And as for Dick's suggestions, let me recap and intersperse a few
questions.

1) Q9, the transistor in the bias circuit, was definitely fried. It was
measuring a hfE between 200 and 1700, depending on the time of day,
when we tested it out of circuit. (specs for it should be about 60).
And yes, the reading changed drastically from hour to hour.
2) Replacing it and fixing a broken solder joint seemed to fix things,
but after something less than three hours of playing, it died again.
This time the speaker fuse was shot as well, and when I replaced it,
I was back to the same scenario I had before replacing Q9. The
symptoms a
-20VDC at the speaker outputs.
No sound to the speaker. (But I'm not about to hook them up again!)
3) And for the record, for that brief while that the amp was working,
the DC output at the speaker terminals was tiny--definitely less
than 80mV. (with no input signal, of course)

So my plan is to check Q9 again, and I'm expecting to find it dead.
After that, I'll go after the Q1/Q2 pair, and see if they've drifted
apart. Next will be the opposite pair, Q5 and Q6 I believe (I don't
have the schematic in front of me). After that, it's going to be a
process of testing every resistor, diode, and cap on the board.

Then--MAYBE--I'll plug it in again.

Colin

John A. Weeks III wrote:

Did the original poster ever remove the MOSFET's from the unit. If
so, were they put back in exactly the same place? If not, it is
possible that the matched sets were mismatched in the process.
That would be enough to cause the problem that Dick is talking about.

Here I am, the original poster! :-)

I removed the MOSFETS at one point, and they all went back into the same
slots. Exactly.

And as for Dick's suggestions, let me recap and intersperse a few
questions.

1) Q9, the transistor in the bias circuit, was definitely fried. It was
measuring a hfE between 200 and 1700, depending on the time of day,
when we tested it out of circuit. (specs for it should be about 60).
And yes, the reading changed drastically from hour to hour.
2) Replacing it and fixing a broken solder joint seemed to fix things,
but after something less than three hours of playing, it died again.
This time the speaker fuse was shot as well, and when I replaced it,
I was back to the same scenario I had before replacing Q9. The
symptoms a
-20VDC at the speaker outputs.
No sound to the speaker. (But I'm not about to hook them up again!)
3) And for the record, for that brief while that the amp was working,
the DC output at the speaker terminals was tiny--definitely less
than 80mV. (with no input signal, of course)

So my plan is to check Q9 again, and I'm expecting to find it dead.
After that, I'll go after the Q1/Q2 pair, and see if they've drifted
apart. Next will be the opposite pair, Q5 and Q6 I believe (I don't
have the schematic in front of me). After that, it's going to be a
process of testing every resistor, diode, and cap on the board.

Then--MAYBE--I'll plug it in again.

Colin

John A. Weeks III wrote:

Did the original poster ever remove the MOSFET's from the unit. If
so, were they put back in exactly the same place? If not, it is
possible that the matched sets were mismatched in the process.
That would be enough to cause the problem that Dick is talking about.

Here I am, the original poster! :-)

I removed the MOSFETS at one point, and they all went back into the same
slots. Exactly.

And as for Dick's suggestions, let me recap and intersperse a few
questions.

1) Q9, the transistor in the bias circuit, was definitely fried. It was
measuring a hfE between 200 and 1700, depending on the time of day,
when we tested it out of circuit. (specs for it should be about 60).
And yes, the reading changed drastically from hour to hour.
2) Replacing it and fixing a broken solder joint seemed to fix things,
but after something less than three hours of playing, it died again.
This time the speaker fuse was shot as well, and when I replaced it,
I was back to the same scenario I had before replacing Q9. The
symptoms a
-20VDC at the speaker outputs.
No sound to the speaker. (But I'm not about to hook them up again!)
3) And for the record, for that brief while that the amp was working,
the DC output at the speaker terminals was tiny--definitely less
than 80mV. (with no input signal, of course)

So my plan is to check Q9 again, and I'm expecting to find it dead.
After that, I'll go after the Q1/Q2 pair, and see if they've drifted
apart. Next will be the opposite pair, Q5 and Q6 I believe (I don't
have the schematic in front of me). After that, it's going to be a
process of testing every resistor, diode, and cap on the board.

Then--MAYBE--I'll plug it in again.

Colin

"Colin B." wrote:

John A. Weeks III wrote:

Did the original poster ever remove the MOSFET's from the unit. If
so, were they put back in exactly the same place? If not, it is
possible that the matched sets were mismatched in the process.
That would be enough to cause the problem that Dick is talking about.

Here I am, the original poster! :-)

I removed the MOSFETS at one point, and they all went back into the same
slots. Exactly.

And as for Dick's suggestions, let me recap and intersperse a few
questions.

1) Q9, the transistor in the bias circuit, was definitely fried. It was
measuring a hfE between 200 and 1700, depending on the time of day,
when we tested it out of circuit. (specs for it should be about 60).
And yes, the reading changed drastically from hour to hour.
2) Replacing it and fixing a broken solder joint seemed to fix things,
but after something less than three hours of playing, it died again.
This time the speaker fuse was shot as well, and when I replaced it,
I was back to the same scenario I had before replacing Q9. The
symptoms a
-20VDC at the speaker outputs.
No sound to the speaker. (But I'm not about to hook them up again!)
3) And for the record, for that brief while that the amp was working,
the DC output at the speaker terminals was tiny--definitely less
than 80mV. (with no input signal, of course)

So my plan is to check Q9 again, and I'm expecting to find it dead.
After that, I'll go after the Q1/Q2 pair, and see if they've drifted
apart. Next will be the opposite pair, Q5 and Q6 I believe (I don't
have the schematic in front of me). After that, it's going to be a
process of testing every resistor, diode, and cap on the board.

Then--MAYBE--I'll plug it in again.

Colin

Colin, this is Dick, the "HaflerLover." Thanks for summarizing your
progress.

Here are some suggestions and opinions:

1. I doubt Q9 is the main offender. Its hfe should be around 150 but a
modestly lower or higher hfe will result mostly in the inability of P1 to
adjust the amount of bias current. You seem to be getting dramatic changes
in voltages as a function of time (and probably temperature) too large to be
related just to bias adjustment.

2. Dramatic changes in parameters that seem related to temperature could be
caused by a break (a lesion) in a trace, a solder joint, or even an internal
connection in an active device (transistor or diode) or a resistor. Things
expand and contract due to temperature and only a microscopic circuit break
is required to produce significant changes in output. But, without better
control in input AC voltage it is difficult to track down this type of
problem. You need a Variac. However a change in DC offset from 80 mV to 20V
is dramatic and makes me think you have an either/or thing going on, such as
a temperature related break in the circuit or a component.

Here is a repair strategy to consider:

1. Get a Variac or wire up 2-3 60 watt light bulbs in series with the input
AC. By switching a bypass to each bulb in turn you can slowly ramp up the
input AC surge and voltage -- giving you some control over events in the
circuit.

2. Use two meters, a Variac, a can of "cool-it" and a heat gun (hair dryer).
Monitor bias current and DC offset voltage with the two meters as you slowly
ramp up the incoming AC voltage. Put a 10 ohm 10 watt resistor across the
speaker outputs in the channel under test. This will help you more reliably
test for DC offset when you connect meter probes. And, make all voltage
measurements related to circuit ground -- the connection between the two
power supply filter caps.

Selectively heat and cool various transistors and monitor the effect of
these events by watching the changes in bias and, especially, the DC offset.
This will help you make some assumptions about which part of the circuit is
bad.

Also, use a plastic tool (toothbrush handle) to tap or gently strike various
components and circuit traces to see if doing so changes the readings on the
meters. This process may help you find any small circuit breaks that might
exist. On one occasion I took the time to reflow all solder joints on the
back of one circuit card that was causing a problem, and doing this fixed
the circuit.

3. What color is your circuit card? Is it phenolic board or the gray-green
fiberglass? What is the serial number of your amp? It is on a 2" square
white sticker on the bottom or back side of your amp. There were 3 revisions
to the DH-200 and I can tell you which you have based on the serial number.
Earlier amps with phenolic cards often had microscopic dry solder joints
around the driver and pre-driver transistors (the 4 with the cooling fins).
This was because the solder bath temp and time this card could tolerate was
too low to produce good joints all of the time. Later versions of the amp
used a small plastic washer at the bottom of each of these 4 transistors and
the attachment of the traces on the rear of the card was improved (better
glue), which resulted in higher tolerances in solder bath time and
temperatures which resulted in improved solder connections without heat
damage to components.

As a minimum I would re-flow the solder joints around the BCE of these four
transistors, the ones with the small metal can tops.

4. And, of course, you should check for the obvious. Broken wires or bad
solder joints in other parts of the amp, especially power supply connections
and the feedback to the amp from the speaker protection fuse. However
because your problem is only with one channel I doubt anything is wrong with
the power supply unless it is a power supply connection to the card. You
could monitor the voltages at the + and - supply rails on the card as you
tap its components and its connections back to the PS. They should be ~65
+/- VDC.

5. You should lay out both heat sinks to enable access to both circuit
cards. Then compare voltages at suspicious parts by noting voltages on the
good circuit and the same spot on the bad card. But, of course, this
prolonged comparison of voltages works best when a variac is used so you can
keep the bad card at energized at a low voltage that precludes blowing up or
frying the rest of its channel.

Well, that's enough for now and I am sure enough to draw some flames from
others. Tell me the serial number and I will make other suggestions.

Dick

"Colin B." wrote:

John A. Weeks III wrote:

Did the original poster ever remove the MOSFET's from the unit. If
so, were they put back in exactly the same place? If not, it is
possible that the matched sets were mismatched in the process.
That would be enough to cause the problem that Dick is talking about.

Here I am, the original poster! :-)

I removed the MOSFETS at one point, and they all went back into the same
slots. Exactly.

And as for Dick's suggestions, let me recap and intersperse a few
questions.

1) Q9, the transistor in the bias circuit, was definitely fried. It was
measuring a hfE between 200 and 1700, depending on the time of day,
when we tested it out of circuit. (specs for it should be about 60).
And yes, the reading changed drastically from hour to hour.
2) Replacing it and fixing a broken solder joint seemed to fix things,
but after something less than three hours of playing, it died again.
This time the speaker fuse was shot as well, and when I replaced it,
I was back to the same scenario I had before replacing Q9. The
symptoms a
-20VDC at the speaker outputs.
No sound to the speaker. (But I'm not about to hook them up again!)
3) And for the record, for that brief while that the amp was working,
the DC output at the speaker terminals was tiny--definitely less
than 80mV. (with no input signal, of course)

So my plan is to check Q9 again, and I'm expecting to find it dead.
After that, I'll go after the Q1/Q2 pair, and see if they've drifted
apart. Next will be the opposite pair, Q5 and Q6 I believe (I don't
have the schematic in front of me). After that, it's going to be a
process of testing every resistor, diode, and cap on the board.

Then--MAYBE--I'll plug it in again.

Colin

Colin, this is Dick, the "HaflerLover." Thanks for summarizing your
progress.

Here are some suggestions and opinions:

1. I doubt Q9 is the main offender. Its hfe should be around 150 but a
modestly lower or higher hfe will result mostly in the inability of P1 to
adjust the amount of bias current. You seem to be getting dramatic changes
in voltages as a function of time (and probably temperature) too large to be
related just to bias adjustment.

2. Dramatic changes in parameters that seem related to temperature could be
caused by a break (a lesion) in a trace, a solder joint, or even an internal
connection in an active device (transistor or diode) or a resistor. Things
expand and contract due to temperature and only a microscopic circuit break
is required to produce significant changes in output. But, without better
control in input AC voltage it is difficult to track down this type of
problem. You need a Variac. However a change in DC offset from 80 mV to 20V
is dramatic and makes me think you have an either/or thing going on, such as
a temperature related break in the circuit or a component.

Here is a repair strategy to consider:

1. Get a Variac or wire up 2-3 60 watt light bulbs in series with the input
AC. By switching a bypass to each bulb in turn you can slowly ramp up the
input AC surge and voltage -- giving you some control over events in the
circuit.

2. Use two meters, a Variac, a can of "cool-it" and a heat gun (hair dryer).
Monitor bias current and DC offset voltage with the two meters as you slowly
ramp up the incoming AC voltage. Put a 10 ohm 10 watt resistor across the
speaker outputs in the channel under test. This will help you more reliably
test for DC offset when you connect meter probes. And, make all voltage
measurements related to circuit ground -- the connection between the two
power supply filter caps.

Selectively heat and cool various transistors and monitor the effect of
these events by watching the changes in bias and, especially, the DC offset.
This will help you make some assumptions about which part of the circuit is
bad.

Also, use a plastic tool (toothbrush handle) to tap or gently strike various
components and circuit traces to see if doing so changes the readings on the
meters. This process may help you find any small circuit breaks that might
exist. On one occasion I took the time to reflow all solder joints on the
back of one circuit card that was causing a problem, and doing this fixed
the circuit.

3. What color is your circuit card? Is it phenolic board or the gray-green
fiberglass? What is the serial number of your amp? It is on a 2" square
white sticker on the bottom or back side of your amp. There were 3 revisions
to the DH-200 and I can tell you which you have based on the serial number.
Earlier amps with phenolic cards often had microscopic dry solder joints
around the driver and pre-driver transistors (the 4 with the cooling fins).
This was because the solder bath temp and time this card could tolerate was
too low to produce good joints all of the time. Later versions of the amp
used a small plastic washer at the bottom of each of these 4 transistors and
the attachment of the traces on the rear of the card was improved (better
glue), which resulted in higher tolerances in solder bath time and
temperatures which resulted in improved solder connections without heat
damage to components.

As a minimum I would re-flow the solder joints around the BCE of these four
transistors, the ones with the small metal can tops.

4. And, of course, you should check for the obvious. Broken wires or bad
solder joints in other parts of the amp, especially power supply connections
and the feedback to the amp from the speaker protection fuse. However
because your problem is only with one channel I doubt anything is wrong with
the power supply unless it is a power supply connection to the card. You
could monitor the voltages at the + and - supply rails on the card as you
tap its components and its connections back to the PS. They should be ~65
+/- VDC.

5. You should lay out both heat sinks to enable access to both circuit
cards. Then compare voltages at suspicious parts by noting voltages on the
good circuit and the same spot on the bad card. But, of course, this
prolonged comparison of voltages works best when a variac is used so you can
keep the bad card at energized at a low voltage that precludes blowing up or
frying the rest of its channel.

Well, that's enough for now and I am sure enough to draw some flames from
others. Tell me the serial number and I will make other suggestions.

Dick

"Colin B." wrote:

John A. Weeks III wrote:

Did the original poster ever remove the MOSFET's from the unit. If
so, were they put back in exactly the same place? If not, it is
possible that the matched sets were mismatched in the process.
That would be enough to cause the problem that Dick is talking about.

Here I am, the original poster! :-)

I removed the MOSFETS at one point, and they all went back into the same
slots. Exactly.

And as for Dick's suggestions, let me recap and intersperse a few
questions.

1) Q9, the transistor in the bias circuit, was definitely fried. It was
measuring a hfE between 200 and 1700, depending on the time of day,
when we tested it out of circuit. (specs for it should be about 60).
And yes, the reading changed drastically from hour to hour.
2) Replacing it and fixing a broken solder joint seemed to fix things,
but after something less than three hours of playing, it died again.
This time the speaker fuse was shot as well, and when I replaced it,
I was back to the same scenario I had before replacing Q9. The
symptoms a
-20VDC at the speaker outputs.
No sound to the speaker. (But I'm not about to hook them up again!)
3) And for the record, for that brief while that the amp was working,
the DC output at the speaker terminals was tiny--definitely less
than 80mV. (with no input signal, of course)

So my plan is to check Q9 again, and I'm expecting to find it dead.
After that, I'll go after the Q1/Q2 pair, and see if they've drifted
apart. Next will be the opposite pair, Q5 and Q6 I believe (I don't
have the schematic in front of me). After that, it's going to be a
process of testing every resistor, diode, and cap on the board.

Then--MAYBE--I'll plug it in again.

Colin

Colin, this is Dick, the "HaflerLover." Thanks for summarizing your
progress.

Here are some suggestions and opinions:

1. I doubt Q9 is the main offender. Its hfe should be around 150 but a
modestly lower or higher hfe will result mostly in the inability of P1 to
adjust the amount of bias current. You seem to be getting dramatic changes
in voltages as a function of time (and probably temperature) too large to be
related just to bias adjustment.

2. Dramatic changes in parameters that seem related to temperature could be
caused by a break (a lesion) in a trace, a solder joint, or even an internal
connection in an active device (transistor or diode) or a resistor. Things
expand and contract due to temperature and only a microscopic circuit break
is required to produce significant changes in output. But, without better
control in input AC voltage it is difficult to track down this type of
problem. You need a Variac. However a change in DC offset from 80 mV to 20V
is dramatic and makes me think you have an either/or thing going on, such as
a temperature related break in the circuit or a component.

Here is a repair strategy to consider:

1. Get a Variac or wire up 2-3 60 watt light bulbs in series with the input
AC. By switching a bypass to each bulb in turn you can slowly ramp up the
input AC surge and voltage -- giving you some control over events in the
circuit.

2. Use two meters, a Variac, a can of "cool-it" and a heat gun (hair dryer).
Monitor bias current and DC offset voltage with the two meters as you slowly
ramp up the incoming AC voltage. Put a 10 ohm 10 watt resistor across the
speaker outputs in the channel under test. This will help you more reliably
test for DC offset when you connect meter probes. And, make all voltage
measurements related to circuit ground -- the connection between the two
power supply filter caps.

Selectively heat and cool various transistors and monitor the effect of
these events by watching the changes in bias and, especially, the DC offset.
This will help you make some assumptions about which part of the circuit is
bad.

Also, use a plastic tool (toothbrush handle) to tap or gently strike various
components and circuit traces to see if doing so changes the readings on the
meters. This process may help you find any small circuit breaks that might
exist. On one occasion I took the time to reflow all solder joints on the
back of one circuit card that was causing a problem, and doing this fixed
the circuit.

3. What color is your circuit card? Is it phenolic board or the gray-green
fiberglass? What is the serial number of your amp? It is on a 2" square
white sticker on the bottom or back side of your amp. There were 3 revisions
to the DH-200 and I can tell you which you have based on the serial number.
Earlier amps with phenolic cards often had microscopic dry solder joints
around the driver and pre-driver transistors (the 4 with the cooling fins).
This was because the solder bath temp and time this card could tolerate was
too low to produce good joints all of the time. Later versions of the amp
used a small plastic washer at the bottom of each of these 4 transistors and
the attachment of the traces on the rear of the card was improved (better
glue), which resulted in higher tolerances in solder bath time and
temperatures which resulted in improved solder connections without heat
damage to components.

As a minimum I would re-flow the solder joints around the BCE of these four
transistors, the ones with the small metal can tops.

4. And, of course, you should check for the obvious. Broken wires or bad
solder joints in other parts of the amp, especially power supply connections
and the feedback to the amp from the speaker protection fuse. However
because your problem is only with one channel I doubt anything is wrong with
the power supply unless it is a power supply connection to the card. You
could monitor the voltages at the + and - supply rails on the card as you
tap its components and its connections back to the PS. They should be ~65
+/- VDC.

5. You should lay out both heat sinks to enable access to both circuit
cards. Then compare voltages at suspicious parts by noting voltages on the
good circuit and the same spot on the bad card. But, of course, this
prolonged comparison of voltages works best when a variac is used so you can
keep the bad card at energized at a low voltage that precludes blowing up or
frying the rest of its channel.

Well, that's enough for now and I am sure enough to draw some flames from
others. Tell me the serial number and I will make other suggestions.

Dick

"Colin B." wrote:

John A. Weeks III wrote:

Did the original poster ever remove the MOSFET's from the unit. If
so, were they put back in exactly the same place? If not, it is
possible that the matched sets were mismatched in the process.
That would be enough to cause the problem that Dick is talking about.

Here I am, the original poster! :-)

I removed the MOSFETS at one point, and they all went back into the same
slots. Exactly.

And as for Dick's suggestions, let me recap and intersperse a few
questions.

1) Q9, the transistor in the bias circuit, was definitely fried. It was
measuring a hfE between 200 and 1700, depending on the time of day,
when we tested it out of circuit. (specs for it should be about 60).
And yes, the reading changed drastically from hour to hour.
2) Replacing it and fixing a broken solder joint seemed to fix things,
but after something less than three hours of playing, it died again.
This time the speaker fuse was shot as well, and when I replaced it,
I was back to the same scenario I had before replacing Q9. The
symptoms a
-20VDC at the speaker outputs.
No sound to the speaker. (But I'm not about to hook them up again!)
3) And for the record, for that brief while that the amp was working,
the DC output at the speaker terminals was tiny--definitely less
than 80mV. (with no input signal, of course)

So my plan is to check Q9 again, and I'm expecting to find it dead.
After that, I'll go after the Q1/Q2 pair, and see if they've drifted
apart. Next will be the opposite pair, Q5 and Q6 I believe (I don't
have the schematic in front of me). After that, it's going to be a
process of testing every resistor, diode, and cap on the board.

Then--MAYBE--I'll plug it in again.

Colin

Colin, this is Dick, the "HaflerLover." Thanks for summarizing your
progress.

Here are some suggestions and opinions:

1. I doubt Q9 is the main offender. Its hfe should be around 150 but a
modestly lower or higher hfe will result mostly in the inability of P1 to
adjust the amount of bias current. You seem to be getting dramatic changes
in voltages as a function of time (and probably temperature) too large to be
related just to bias adjustment.

2. Dramatic changes in parameters that seem related to temperature could be
caused by a break (a lesion) in a trace, a solder joint, or even an internal
connection in an active device (transistor or diode) or a resistor. Things
expand and contract due to temperature and only a microscopic circuit break
is required to produce significant changes in output. But, without better
control in input AC voltage it is difficult to track down this type of
problem. You need a Variac. However a change in DC offset from 80 mV to 20V
is dramatic and makes me think you have an either/or thing going on, such as
a temperature related break in the circuit or a component.

Here is a repair strategy to consider:

1. Get a Variac or wire up 2-3 60 watt light bulbs in series with the input
AC. By switching a bypass to each bulb in turn you can slowly ramp up the
input AC surge and voltage -- giving you some control over events in the
circuit.

2. Use two meters, a Variac, a can of "cool-it" and a heat gun (hair dryer).
Monitor bias current and DC offset voltage with the two meters as you slowly
ramp up the incoming AC voltage. Put a 10 ohm 10 watt resistor across the
speaker outputs in the channel under test. This will help you more reliably
test for DC offset when you connect meter probes. And, make all voltage
measurements related to circuit ground -- the connection between the two
power supply filter caps.

Selectively heat and cool various transistors and monitor the effect of
these events by watching the changes in bias and, especially, the DC offset.
This will help you make some assumptions about which part of the circuit is
bad.

Also, use a plastic tool (toothbrush handle) to tap or gently strike various
components and circuit traces to see if doing so changes the readings on the
meters. This process may help you find any small circuit breaks that might
exist. On one occasion I took the time to reflow all solder joints on the
back of one circuit card that was causing a problem, and doing this fixed
the circuit.

3. What color is your circuit card? Is it phenolic board or the gray-green
fiberglass? What is the serial number of your amp? It is on a 2" square
white sticker on the bottom or back side of your amp. There were 3 revisions
to the DH-200 and I can tell you which you have based on the serial number.
Earlier amps with phenolic cards often had microscopic dry solder joints
around the driver and pre-driver transistors (the 4 with the cooling fins).
This was because the solder bath temp and time this card could tolerate was
too low to produce good joints all of the time. Later versions of the amp
used a small plastic washer at the bottom of each of these 4 transistors and
the attachment of the traces on the rear of the card was improved (better
glue), which resulted in higher tolerances in solder bath time and
temperatures which resulted in improved solder connections without heat
damage to components.

As a minimum I would re-flow the solder joints around the BCE of these four
transistors, the ones with the small metal can tops.

4. And, of course, you should check for the obvious. Broken wires or bad
solder joints in other parts of the amp, especially power supply connections
and the feedback to the amp from the speaker protection fuse. However
because your problem is only with one channel I doubt anything is wrong with
the power supply unless it is a power supply connection to the card. You
could monitor the voltages at the + and - supply rails on the card as you
tap its components and its connections back to the PS. They should be ~65
+/- VDC.

5. You should lay out both heat sinks to enable access to both circuit
cards. Then compare voltages at suspicious parts by noting voltages on the
good circuit and the same spot on the bad card. But, of course, this
prolonged comparison of voltages works best when a variac is used so you can
keep the bad card at energized at a low voltage that precludes blowing up or
frying the rest of its channel.

Well, that's enough for now and I am sure enough to draw some flames from
others. Tell me the serial number and I will make other suggestions.

Dick

lid wrote:

Colin, this is Dick, the "HaflerLover." Thanks for summarizing your
progress.

Here are some suggestions and opinions:

1. I doubt Q9 is the main offender. Its hfe should be around 150 but a
modestly lower or higher hfe will result mostly in the inability of P1 to
adjust the amount of bias current. You seem to be getting dramatic changes
in voltages as a function of time (and probably temperature) too large to be
related just to bias adjustment.

Agreed that Q9 isn't the offender--it does, however, seem to be a repeated
victim. Just to expand on my earlier comments...
When we pulled Q9, it was dead. It actually measured the same forwards
and backwards (i.e. BC and CB were identical in a diode tester, as were
BE/EB)

With the new Q9 in place and working, the DC offset at the speakers was
~80mV, and everything was fine. Then the channel died, the fuse blew,
and (after replacing the fuse) the DC offset was -20V. After this, letting
it cool or heat didn't have any effect. That is to say, the DC offset
seems to be caused by Q9 being blown open, not causing it.

So if it's a thermal problem, it's a thermal problem blowing up Q9. The
amp doesn't behave or misbehave based on temperature. (other than that
it took a few hours for the thing to blow up again--which could have been
either thermal, or pushing Q9 a bit harder than allowable, for those
hours, until it failed.)

In other words, there's no way of going back to a working amp, short
of replacing that transistor. The only way to actually FIX the amp, is
to find what's killing it.

The next time I see my parents, I'll grab my dad's variac and my old
'scope. I wish I hadn't been so optimistic on our repairs, and grabbed
the stuff when we were there last weekend.

Also, use a plastic tool (toothbrush handle) to tap or gently strike various
components and circuit traces to see if doing so changes the readings on the
meters. This process may help you find any small circuit breaks that might
exist. On one occasion I took the time to reflow all solder joints on the
back of one circuit card that was causing a problem, and doing this fixed
the circuit.

Good point. I'm tempted to reflow the board as it is, just because.
When we saw the one bad power lead on the board, we did touch up all
of the connecting wires (none of them looked very good, actually) but
not on the board itself. Speaking of which, it's a phenolic board on
this amp. I don't have the serial number right now, but I'll get it
when I go home. I'll look closely, but I don't think there are any
washers between the transistors and the board, anywhere.

Well, that's enough for now and I am sure enough to draw some flames from
others. Tell me the serial number and I will make other suggestions.

I'll get that tonight.

By the way, thanks to you and all the others who have helped with this
amp. I really appreciate it!

Colin

lid wrote:

Colin, this is Dick, the "HaflerLover." Thanks for summarizing your
progress.

Here are some suggestions and opinions:

1. I doubt Q9 is the main offender. Its hfe should be around 150 but a
modestly lower or higher hfe will result mostly in the inability of P1 to
adjust the amount of bias current. You seem to be getting dramatic changes
in voltages as a function of time (and probably temperature) too large to be
related just to bias adjustment.

Agreed that Q9 isn't the offender--it does, however, seem to be a repeated
victim. Just to expand on my earlier comments...
When we pulled Q9, it was dead. It actually measured the same forwards
and backwards (i.e. BC and CB were identical in a diode tester, as were
BE/EB)

With the new Q9 in place and working, the DC offset at the speakers was
~80mV, and everything was fine. Then the channel died, the fuse blew,
and (after replacing the fuse) the DC offset was -20V. After this, letting
it cool or heat didn't have any effect. That is to say, the DC offset
seems to be caused by Q9 being blown open, not causing it.

So if it's a thermal problem, it's a thermal problem blowing up Q9. The
amp doesn't behave or misbehave based on temperature. (other than that
it took a few hours for the thing to blow up again--which could have been
either thermal, or pushing Q9 a bit harder than allowable, for those
hours, until it failed.)

In other words, there's no way of going back to a working amp, short
of replacing that transistor. The only way to actually FIX the amp, is
to find what's killing it.

The next time I see my parents, I'll grab my dad's variac and my old
'scope. I wish I hadn't been so optimistic on our repairs, and grabbed
the stuff when we were there last weekend.

Also, use a plastic tool (toothbrush handle) to tap or gently strike various
components and circuit traces to see if doing so changes the readings on the
meters. This process may help you find any small circuit breaks that might
exist. On one occasion I took the time to reflow all solder joints on the
back of one circuit card that was causing a problem, and doing this fixed
the circuit.

Good point. I'm tempted to reflow the board as it is, just because.
When we saw the one bad power lead on the board, we did touch up all
of the connecting wires (none of them looked very good, actually) but
not on the board itself. Speaking of which, it's a phenolic board on
this amp. I don't have the serial number right now, but I'll get it
when I go home. I'll look closely, but I don't think there are any
washers between the transistors and the board, anywhere.

Well, that's enough for now and I am sure enough to draw some flames from
others. Tell me the serial number and I will make other suggestions.

I'll get that tonight.

By the way, thanks to you and all the others who have helped with this
amp. I really appreciate it!

Colin

lid wrote:

Colin, this is Dick, the "HaflerLover." Thanks for summarizing your
progress.

Here are some suggestions and opinions:

1. I doubt Q9 is the main offender. Its hfe should be around 150 but a
modestly lower or higher hfe will result mostly in the inability of P1 to
adjust the amount of bias current. You seem to be getting dramatic changes
in voltages as a function of time (and probably temperature) too large to be
related just to bias adjustment.

Agreed that Q9 isn't the offender--it does, however, seem to be a repeated
victim. Just to expand on my earlier comments...
When we pulled Q9, it was dead. It actually measured the same forwards
and backwards (i.e. BC and CB were identical in a diode tester, as were
BE/EB)

With the new Q9 in place and working, the DC offset at the speakers was
~80mV, and everything was fine. Then the channel died, the fuse blew,
and (after replacing the fuse) the DC offset was -20V. After this, letting
it cool or heat didn't have any effect. That is to say, the DC offset
seems to be caused by Q9 being blown open, not causing it.

So if it's a thermal problem, it's a thermal problem blowing up Q9. The
amp doesn't behave or misbehave based on temperature. (other than that
it took a few hours for the thing to blow up again--which could have been
either thermal, or pushing Q9 a bit harder than allowable, for those
hours, until it failed.)

In other words, there's no way of going back to a working amp, short
of replacing that transistor. The only way to actually FIX the amp, is
to find what's killing it.

The next time I see my parents, I'll grab my dad's variac and my old
'scope. I wish I hadn't been so optimistic on our repairs, and grabbed
the stuff when we were there last weekend.

Also, use a plastic tool (toothbrush handle) to tap or gently strike various
components and circuit traces to see if doing so changes the readings on the
meters. This process may help you find any small circuit breaks that might
exist. On one occasion I took the time to reflow all solder joints on the
back of one circuit card that was causing a problem, and doing this fixed
the circuit.

Good point. I'm tempted to reflow the board as it is, just because.
When we saw the one bad power lead on the board, we did touch up all
of the connecting wires (none of them looked very good, actually) but
not on the board itself. Speaking of which, it's a phenolic board on
this amp. I don't have the serial number right now, but I'll get it
when I go home. I'll look closely, but I don't think there are any
washers between the transistors and the board, anywhere.

Well, that's enough for now and I am sure enough to draw some flames from
others. Tell me the serial number and I will make other suggestions.

I'll get that tonight.

By the way, thanks to you and all the others who have helped with this
amp. I really appreciate it!

Colin

lid wrote:

Colin, this is Dick, the "HaflerLover." Thanks for summarizing your
progress.

Here are some suggestions and opinions:

1. I doubt Q9 is the main offender. Its hfe should be around 150 but a
modestly lower or higher hfe will result mostly in the inability of P1 to
adjust the amount of bias current. You seem to be getting dramatic changes
in voltages as a function of time (and probably temperature) too large to be
related just to bias adjustment.

Agreed that Q9 isn't the offender--it does, however, seem to be a repeated
victim. Just to expand on my earlier comments...
When we pulled Q9, it was dead. It actually measured the same forwards
and backwards (i.e. BC and CB were identical in a diode tester, as were
BE/EB)

With the new Q9 in place and working, the DC offset at the speakers was
~80mV, and everything was fine. Then the channel died, the fuse blew,
and (after replacing the fuse) the DC offset was -20V. After this, letting
it cool or heat didn't have any effect. That is to say, the DC offset
seems to be caused by Q9 being blown open, not causing it.

So if it's a thermal problem, it's a thermal problem blowing up Q9. The
amp doesn't behave or misbehave based on temperature. (other than that
it took a few hours for the thing to blow up again--which could have been
either thermal, or pushing Q9 a bit harder than allowable, for those
hours, until it failed.)

In other words, there's no way of going back to a working amp, short
of replacing that transistor. The only way to actually FIX the amp, is
to find what's killing it.

The next time I see my parents, I'll grab my dad's variac and my old
'scope. I wish I hadn't been so optimistic on our repairs, and grabbed
the stuff when we were there last weekend.

Also, use a plastic tool (toothbrush handle) to tap or gently strike various
components and circuit traces to see if doing so changes the readings on the
meters. This process may help you find any small circuit breaks that might
exist. On one occasion I took the time to reflow all solder joints on the
back of one circuit card that was causing a problem, and doing this fixed
the circuit.

Good point. I'm tempted to reflow the board as it is, just because.
When we saw the one bad power lead on the board, we did touch up all
of the connecting wires (none of them looked very good, actually) but
not on the board itself. Speaking of which, it's a phenolic board on
this amp. I don't have the serial number right now, but I'll get it
when I go home. I'll look closely, but I don't think there are any
washers between the transistors and the board, anywhere.

Well, that's enough for now and I am sure enough to draw some flames from
others. Tell me the serial number and I will make other suggestions.

I'll get that tonight.

By the way, thanks to you and all the others who have helped with this
amp. I really appreciate it!

Colin

In article ,
Colin B. wrote:

Good point. I'm tempted to reflow the board as it is, just because.
When we saw the one bad power lead on the board, we did touch up all
of the connecting wires (none of them looked very good, actually) but
not on the board itself. Speaking of which, it's a phenolic board on
this amp. I don't have the serial number right now, but I'll get it
when I go home. I'll look closely, but I don't think there are any
washers between the transistors and the board, anywhere.

I had (have) a Van Alstine Omega 150 amp which developed some odd
symptoms - a sharp POP in one channel, anywhere from 15 seconds to a
couple of minutes after it was powered on. This occurred
intermittently for a couple of years, and then the channel suddenly
became extremely noisy (hiss).

I phoned AVA, and Frank Van Alstine came on the phone and explained
that the problem was of basically the same nature as has been
suggested for your Hafler amp - bad solder joints on some of the
connections between the driver transistors and the board. Repeated
thermal stress could eventually cause the joint to become intermittent
or open. It was apparently due to the soldering process used... AVA
changed the process once they learned of the problem with this batch
of amps.

He recommended that I reflow the individual solder joints and add a
bit of good-quality new solder to each. I did, and it fixed the
problem completely.

--
Dave Platt AE6EO
Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!

In article ,
Colin B. wrote:

Good point. I'm tempted to reflow the board as it is, just because.
When we saw the one bad power lead on the board, we did touch up all
of the connecting wires (none of them looked very good, actually) but
not on the board itself. Speaking of which, it's a phenolic board on
this amp. I don't have the serial number right now, but I'll get it
when I go home. I'll look closely, but I don't think there are any
washers between the transistors and the board, anywhere.

I had (have) a Van Alstine Omega 150 amp which developed some odd
symptoms - a sharp POP in one channel, anywhere from 15 seconds to a
couple of minutes after it was powered on. This occurred
intermittently for a couple of years, and then the channel suddenly
became extremely noisy (hiss).

I phoned AVA, and Frank Van Alstine came on the phone and explained
that the problem was of basically the same nature as has been
suggested for your Hafler amp - bad solder joints on some of the
connections between the driver transistors and the board. Repeated
thermal stress could eventually cause the joint to become intermittent
or open. It was apparently due to the soldering process used... AVA
changed the process once they learned of the problem with this batch
of amps.

He recommended that I reflow the individual solder joints and add a
bit of good-quality new solder to each. I did, and it fixed the
problem completely.

--
Dave Platt AE6EO
Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!

In article ,
Colin B. wrote:

Good point. I'm tempted to reflow the board as it is, just because.
When we saw the one bad power lead on the board, we did touch up all
of the connecting wires (none of them looked very good, actually) but
not on the board itself. Speaking of which, it's a phenolic board on
this amp. I don't have the serial number right now, but I'll get it
when I go home. I'll look closely, but I don't think there are any
washers between the transistors and the board, anywhere.

I had (have) a Van Alstine Omega 150 amp which developed some odd
symptoms - a sharp POP in one channel, anywhere from 15 seconds to a
couple of minutes after it was powered on. This occurred
intermittently for a couple of years, and then the channel suddenly
became extremely noisy (hiss).

I phoned AVA, and Frank Van Alstine came on the phone and explained
that the problem was of basically the same nature as has been
suggested for your Hafler amp - bad solder joints on some of the
connections between the driver transistors and the board. Repeated
thermal stress could eventually cause the joint to become intermittent
or open. It was apparently due to the soldering process used... AVA
changed the process once they learned of the problem with this batch
of amps.

He recommended that I reflow the individual solder joints and add a
bit of good-quality new solder to each. I did, and it fixed the
problem completely.

--
Dave Platt AE6EO
Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!

In article ,
Colin B. wrote:

Good point. I'm tempted to reflow the board as it is, just because.
When we saw the one bad power lead on the board, we did touch up all
of the connecting wires (none of them looked very good, actually) but
not on the board itself. Speaking of which, it's a phenolic board on
this amp. I don't have the serial number right now, but I'll get it
when I go home. I'll look closely, but I don't think there are any
washers between the transistors and the board, anywhere.

I had (have) a Van Alstine Omega 150 amp which developed some odd
symptoms - a sharp POP in one channel, anywhere from 15 seconds to a
couple of minutes after it was powered on. This occurred
intermittently for a couple of years, and then the channel suddenly
became extremely noisy (hiss).

I phoned AVA, and Frank Van Alstine came on the phone and explained
that the problem was of basically the same nature as has been
suggested for your Hafler amp - bad solder joints on some of the
connections between the driver transistors and the board. Repeated
thermal stress could eventually cause the joint to become intermittent
or open. It was apparently due to the soldering process used... AVA
changed the process once they learned of the problem with this batch
of amps.

He recommended that I reflow the individual solder joints and add a
bit of good-quality new solder to each. I did, and it fixed the
problem completely.

--
Dave Platt AE6EO
Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!

Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

If the bias measured the same for both channels, then I would look again for
a DC offset. This assumes a load was connected. You might start by adjusting
the bias with NO load. There could be a problem on the load side at this
point.

Mark Z.

--
Please reply only to Group. I regret this is necessary. Viruses and spam
have rendered my regular e-mail address useless.


"Colin B." wrote in message
...
Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

If the bias measured the same for both channels, then I would look again for
a DC offset. This assumes a load was connected. You might start by adjusting
the bias with NO load. There could be a problem on the load side at this
point.

Mark Z.

--
Please reply only to Group. I regret this is necessary. Viruses and spam
have rendered my regular e-mail address useless.


"Colin B." wrote in message
...
Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

If the bias measured the same for both channels, then I would look again for
a DC offset. This assumes a load was connected. You might start by adjusting
the bias with NO load. There could be a problem on the load side at this
point.

Mark Z.

--
Please reply only to Group. I regret this is necessary. Viruses and spam
have rendered my regular e-mail address useless.


"Colin B." wrote in message
...
Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

If the bias measured the same for both channels, then I would look again for
a DC offset. This assumes a load was connected. You might start by adjusting
the bias with NO load. There could be a problem on the load side at this
point.

Mark Z.

--
Please reply only to Group. I regret this is necessary. Viruses and spam
have rendered my regular e-mail address useless.


"Colin B." wrote in message
...
Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET
-------------------------------------------------------

This DH-200 DC nulling circuit was described by
Walt Jung in The Audio Amateur 1/83, page 56.
It is simple and uses a small variable resistor that
mounts nicely on the circuit card by drilling 3 small holes
with a wire gauge drill (like for drilling holes in a PCB).
I drilled my 3 small holes in a vertical line beginning just
above where R8 is attached to the + PS rail.
The bottom leg of the pot was solder tacked to the + PS rail,
the other leg (top pin) is attached to the - PS rail by a short
length of insulated wire. The middle pin (wiper)
is attached to the junction of R4 and R5 via a
2 megohm 1/4 watt metal film resistor.

Just attach a 100Kohm pot between the two power
supply rails and feed the output of its wiper through
a 2 megohm resistor to where R4 and R5 touch the
input trace. I used a miniature 1/2 Watt multi-turn cermet
variable resistor. A small value film cap across the
legs of the pot would be a good idea to help cancel
any noise riding on the PS rails.

This little circuit solves a bunch of problems and
I don't know why more people don't use it. It is not a
destructive mod, just a simple add-on.

The pot allows inserting through the 2 meg ohm
resistor a very small "sample" of either + or - voltage
to the amp's input to null DC offset. I've seen more than
one schematic for other amps that routinely do the same thing.

Hope this helps. Spread the word!

Dick

==============================
"Colin B." wrote:

Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET
-------------------------------------------------------

This DH-200 DC nulling circuit was described by
Walt Jung in The Audio Amateur 1/83, page 56.
It is simple and uses a small variable resistor that
mounts nicely on the circuit card by drilling 3 small holes
with a wire gauge drill (like for drilling holes in a PCB).
I drilled my 3 small holes in a vertical line beginning just
above where R8 is attached to the + PS rail.
The bottom leg of the pot was solder tacked to the + PS rail,
the other leg (top pin) is attached to the - PS rail by a short
length of insulated wire. The middle pin (wiper)
is attached to the junction of R4 and R5 via a
2 megohm 1/4 watt metal film resistor.

Just attach a 100Kohm pot between the two power
supply rails and feed the output of its wiper through
a 2 megohm resistor to where R4 and R5 touch the
input trace. I used a miniature 1/2 Watt multi-turn cermet
variable resistor. A small value film cap across the
legs of the pot would be a good idea to help cancel
any noise riding on the PS rails.

This little circuit solves a bunch of problems and
I don't know why more people don't use it. It is not a
destructive mod, just a simple add-on.

The pot allows inserting through the 2 meg ohm
resistor a very small "sample" of either + or - voltage
to the amp's input to null DC offset. I've seen more than
one schematic for other amps that routinely do the same thing.

Hope this helps. Spread the word!

Dick

==============================
"Colin B." wrote:

Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET
-------------------------------------------------------

This DH-200 DC nulling circuit was described by
Walt Jung in The Audio Amateur 1/83, page 56.
It is simple and uses a small variable resistor that
mounts nicely on the circuit card by drilling 3 small holes
with a wire gauge drill (like for drilling holes in a PCB).
I drilled my 3 small holes in a vertical line beginning just
above where R8 is attached to the + PS rail.
The bottom leg of the pot was solder tacked to the + PS rail,
the other leg (top pin) is attached to the - PS rail by a short
length of insulated wire. The middle pin (wiper)
is attached to the junction of R4 and R5 via a
2 megohm 1/4 watt metal film resistor.

Just attach a 100Kohm pot between the two power
supply rails and feed the output of its wiper through
a 2 megohm resistor to where R4 and R5 touch the
input trace. I used a miniature 1/2 Watt multi-turn cermet
variable resistor. A small value film cap across the
legs of the pot would be a good idea to help cancel
any noise riding on the PS rails.

This little circuit solves a bunch of problems and
I don't know why more people don't use it. It is not a
destructive mod, just a simple add-on.

The pot allows inserting through the 2 meg ohm
resistor a very small "sample" of either + or - voltage
to the amp's input to null DC offset. I've seen more than
one schematic for other amps that routinely do the same thing.

Hope this helps. Spread the word!

Dick

==============================
"Colin B." wrote:

Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET
-------------------------------------------------------

This DH-200 DC nulling circuit was described by
Walt Jung in The Audio Amateur 1/83, page 56.
It is simple and uses a small variable resistor that
mounts nicely on the circuit card by drilling 3 small holes
with a wire gauge drill (like for drilling holes in a PCB).
I drilled my 3 small holes in a vertical line beginning just
above where R8 is attached to the + PS rail.
The bottom leg of the pot was solder tacked to the + PS rail,
the other leg (top pin) is attached to the - PS rail by a short
length of insulated wire. The middle pin (wiper)
is attached to the junction of R4 and R5 via a
2 megohm 1/4 watt metal film resistor.

Just attach a 100Kohm pot between the two power
supply rails and feed the output of its wiper through
a 2 megohm resistor to where R4 and R5 touch the
input trace. I used a miniature 1/2 Watt multi-turn cermet
variable resistor. A small value film cap across the
legs of the pot would be a good idea to help cancel
any noise riding on the PS rails.

This little circuit solves a bunch of problems and
I don't know why more people don't use it. It is not a
destructive mod, just a simple add-on.

The pot allows inserting through the 2 meg ohm
resistor a very small "sample" of either + or - voltage
to the amp's input to null DC offset. I've seen more than
one schematic for other amps that routinely do the same thing.

Hope this helps. Spread the word!

Dick

==============================
"Colin B." wrote:

Bugger! I had a nice long post, and most of it got eaten.

Oh well.

Here's the weekend's update on the amp.

First of all, in answer to Dick's question, it's an early amp, with the
phenolic board and S/N 3113059.

Now the long and short of my work is that I pulled Q9 and measured it
on the bench. It kept coming and going, so I splurged on the $0.30,
and put a new one in. Then I went over the thing with a voltmeter, and
measured everything on the board. If there was an odd-looking component,
I pulled it out and measured it on the bench. If there was an ugly
solder joint, I touched it up. Then I went over pretty much all of the
traces on the board, paying special attention to the driver transistor
leads. Measured the power supply incidentally, and it's sitting at
+/-58VDC, on both channels.

So far so good. There were some bad solder joints, but everything seems
to be working properly. Not much to do but turn it on and see if it works.
(as an aside, I am currently lacking both variac and oscilloscope, and
won't get them back for a few weeks at least). The results were fairly
promising, but first--a diversion!

I hadn't noticed before, but Q5 and Q6 have been replaced before. I'm
wondering if they're not particularly well balanced, as Dick suggested.

Furthermore, I had pulled Q9 and replaced it with an identical...2N5550.
Today I noticed on the schematic, that it's supposed to be an NP2222!
Checked the other channel, and it too had a 5550 in the bias loop. Both
channels' transistors were original, so there was a change at some point
between the circuit being made, and the schematic being written up.

Now as I said, I turned on the amp to measure any DC offset at the
outputs. Everything seemed fine, so I then set the bias back to 220mA,
same as I've got on the other channel, and let it run for a while,
about 10 minutes. Everything still measured fine after that, but the
heat sink and output transistors were decidedly warmer on this channel
than the other one. That was enough for me--the amp got turned OFF
again.

So we have about 30mVDC more on this channel than the other one, and
hotter heat sinks for the same bias voltage. The first could be
explained by a (fairly small) imbalance between Q5/Q6, but how much
heat would that generate?

Any guesses folks? I'll probably pull the pair out and measure the hFe,
and if they're out of whack, get a matched pair. (The local store
has the 2N5550 in bags of 5/$1.49, so I'm hoping to get the 2N5401
similarly packaged, and measure them closely to find a pair).

So what next?

Thanks,
Colin

lid wrote:
Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

No problem for me.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

First of all, I've been doing much of my 'rough 'n' ready' measurements
with no load. When I get closer, I toss in an ancient speaker of
suspicious origins, but a nominal resistance of 8ohm.

As for Q9 not being a NP2222, I didn't think that it would be an issue,
but it was rather interesting (not to mention startling, when I first
noticed the discrepancy! :-).

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

Thanks much! Very cool to know.

Now we're all on the same page about there being an undesirable DC offset
happening, and the real trick is finding it. You had mentioned the Q1/Q2
and Q5/Q6 differential amp pairs, and so I went after them.
Q1 and Q2 were matched almost perfectly, (gain of 78 I think), but
Q5 and Q6 were quite different: hFe was 70 and 115 respectively, which is
a LONG way from 10%! Well, I swapped them with a matched pair and the
no-load offset at the speaker terminal went from 55mV down to 7mV. Seven!

Everything looks good on the bench, so next step is to plug it into
the stereo and let it sit for a while before buttoning it up.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

I'm hoping that this isn't necessary. It looks like it won't be, so I've
got my fingers crossed.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET

This is sooo simple that I'll probably do it as soon as I've got all of my
other projects bottled up and out of the way. Brilliant solution to the
problem here!

So we may (fingers crossed) have a working and stable amp--and in better
shape than when I got it. I'll keep you posted.

Colin

lid wrote:
Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

No problem for me.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

First of all, I've been doing much of my 'rough 'n' ready' measurements
with no load. When I get closer, I toss in an ancient speaker of
suspicious origins, but a nominal resistance of 8ohm.

As for Q9 not being a NP2222, I didn't think that it would be an issue,
but it was rather interesting (not to mention startling, when I first
noticed the discrepancy! :-).

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

Thanks much! Very cool to know.

Now we're all on the same page about there being an undesirable DC offset
happening, and the real trick is finding it. You had mentioned the Q1/Q2
and Q5/Q6 differential amp pairs, and so I went after them.
Q1 and Q2 were matched almost perfectly, (gain of 78 I think), but
Q5 and Q6 were quite different: hFe was 70 and 115 respectively, which is
a LONG way from 10%! Well, I swapped them with a matched pair and the
no-load offset at the speaker terminal went from 55mV down to 7mV. Seven!

Everything looks good on the bench, so next step is to plug it into
the stereo and let it sit for a while before buttoning it up.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

I'm hoping that this isn't necessary. It looks like it won't be, so I've
got my fingers crossed.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET

This is sooo simple that I'll probably do it as soon as I've got all of my
other projects bottled up and out of the way. Brilliant solution to the
problem here!

So we may (fingers crossed) have a working and stable amp--and in better
shape than when I got it. I'll keep you posted.

Colin

lid wrote:
Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

No problem for me.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

First of all, I've been doing much of my 'rough 'n' ready' measurements
with no load. When I get closer, I toss in an ancient speaker of
suspicious origins, but a nominal resistance of 8ohm.

As for Q9 not being a NP2222, I didn't think that it would be an issue,
but it was rather interesting (not to mention startling, when I first
noticed the discrepancy! :-).

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

Thanks much! Very cool to know.

Now we're all on the same page about there being an undesirable DC offset
happening, and the real trick is finding it. You had mentioned the Q1/Q2
and Q5/Q6 differential amp pairs, and so I went after them.
Q1 and Q2 were matched almost perfectly, (gain of 78 I think), but
Q5 and Q6 were quite different: hFe was 70 and 115 respectively, which is
a LONG way from 10%! Well, I swapped them with a matched pair and the
no-load offset at the speaker terminal went from 55mV down to 7mV. Seven!

Everything looks good on the bench, so next step is to plug it into
the stereo and let it sit for a while before buttoning it up.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

I'm hoping that this isn't necessary. It looks like it won't be, so I've
got my fingers crossed.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET

This is sooo simple that I'll probably do it as soon as I've got all of my
other projects bottled up and out of the way. Brilliant solution to the
problem here!

So we may (fingers crossed) have a working and stable amp--and in better
shape than when I got it. I'll keep you posted.

Colin

lid wrote:
Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

No problem for me.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

First of all, I've been doing much of my 'rough 'n' ready' measurements
with no load. When I get closer, I toss in an ancient speaker of
suspicious origins, but a nominal resistance of 8ohm.

As for Q9 not being a NP2222, I didn't think that it would be an issue,
but it was rather interesting (not to mention startling, when I first
noticed the discrepancy! :-).

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

Thanks much! Very cool to know.

Now we're all on the same page about there being an undesirable DC offset
happening, and the real trick is finding it. You had mentioned the Q1/Q2
and Q5/Q6 differential amp pairs, and so I went after them.
Q1 and Q2 were matched almost perfectly, (gain of 78 I think), but
Q5 and Q6 were quite different: hFe was 70 and 115 respectively, which is
a LONG way from 10%! Well, I swapped them with a matched pair and the
no-load offset at the speaker terminal went from 55mV down to 7mV. Seven!

Everything looks good on the bench, so next step is to plug it into
the stereo and let it sit for a while before buttoning it up.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

I'm hoping that this isn't necessary. It looks like it won't be, so I've
got my fingers crossed.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET

This is sooo simple that I'll probably do it as soon as I've got all of my
other projects bottled up and out of the way. Brilliant solution to the
problem here!

So we may (fingers crossed) have a working and stable amp--and in better
shape than when I got it. I'll keep you posted.

Colin

Colin,

SUCCESS! Experience has taught me a good troubleshooting strategy for these amps
and the first 4 transistors are almost always the main culprits when too much DC
offset is present. And, as you found, the solution is relatively simple for those
who have good soldering skills.

Dick

=========================
"Colin B." wrote:

lid wrote:
Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

No problem for me.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

First of all, I've been doing much of my 'rough 'n' ready' measurements
with no load. When I get closer, I toss in an ancient speaker of
suspicious origins, but a nominal resistance of 8ohm.

As for Q9 not being a NP2222, I didn't think that it would be an issue,
but it was rather interesting (not to mention startling, when I first
noticed the discrepancy! :-).

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

Thanks much! Very cool to know.

Now we're all on the same page about there being an undesirable DC offset
happening, and the real trick is finding it. You had mentioned the Q1/Q2
and Q5/Q6 differential amp pairs, and so I went after them.
Q1 and Q2 were matched almost perfectly, (gain of 78 I think), but
Q5 and Q6 were quite different: hFe was 70 and 115 respectively, which is
a LONG way from 10%! Well, I swapped them with a matched pair and the
no-load offset at the speaker terminal went from 55mV down to 7mV. Seven!

Everything looks good on the bench, so next step is to plug it into
the stereo and let it sit for a while before buttoning it up.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

I'm hoping that this isn't necessary. It looks like it won't be, so I've
got my fingers crossed.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET

This is sooo simple that I'll probably do it as soon as I've got all of my
other projects bottled up and out of the way. Brilliant solution to the
problem here!

So we may (fingers crossed) have a working and stable amp--and in better
shape than when I got it. I'll keep you posted.

Colin

Colin,

SUCCESS! Experience has taught me a good troubleshooting strategy for these amps
and the first 4 transistors are almost always the main culprits when too much DC
offset is present. And, as you found, the solution is relatively simple for those
who have good soldering skills.

Dick

=========================
"Colin B." wrote:

lid wrote:
Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

No problem for me.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

First of all, I've been doing much of my 'rough 'n' ready' measurements
with no load. When I get closer, I toss in an ancient speaker of
suspicious origins, but a nominal resistance of 8ohm.

As for Q9 not being a NP2222, I didn't think that it would be an issue,
but it was rather interesting (not to mention startling, when I first
noticed the discrepancy! :-).

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

Thanks much! Very cool to know.

Now we're all on the same page about there being an undesirable DC offset
happening, and the real trick is finding it. You had mentioned the Q1/Q2
and Q5/Q6 differential amp pairs, and so I went after them.
Q1 and Q2 were matched almost perfectly, (gain of 78 I think), but
Q5 and Q6 were quite different: hFe was 70 and 115 respectively, which is
a LONG way from 10%! Well, I swapped them with a matched pair and the
no-load offset at the speaker terminal went from 55mV down to 7mV. Seven!

Everything looks good on the bench, so next step is to plug it into
the stereo and let it sit for a while before buttoning it up.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

I'm hoping that this isn't necessary. It looks like it won't be, so I've
got my fingers crossed.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET

This is sooo simple that I'll probably do it as soon as I've got all of my
other projects bottled up and out of the way. Brilliant solution to the
problem here!

So we may (fingers crossed) have a working and stable amp--and in better
shape than when I got it. I'll keep you posted.

Colin

Colin,

SUCCESS! Experience has taught me a good troubleshooting strategy for these amps
and the first 4 transistors are almost always the main culprits when too much DC
offset is present. And, as you found, the solution is relatively simple for those
who have good soldering skills.

Dick

=========================
"Colin B." wrote:

lid wrote:
Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

No problem for me.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

First of all, I've been doing much of my 'rough 'n' ready' measurements
with no load. When I get closer, I toss in an ancient speaker of
suspicious origins, but a nominal resistance of 8ohm.

As for Q9 not being a NP2222, I didn't think that it would be an issue,
but it was rather interesting (not to mention startling, when I first
noticed the discrepancy! :-).

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

Thanks much! Very cool to know.

Now we're all on the same page about there being an undesirable DC offset
happening, and the real trick is finding it. You had mentioned the Q1/Q2
and Q5/Q6 differential amp pairs, and so I went after them.
Q1 and Q2 were matched almost perfectly, (gain of 78 I think), but
Q5 and Q6 were quite different: hFe was 70 and 115 respectively, which is
a LONG way from 10%! Well, I swapped them with a matched pair and the
no-load offset at the speaker terminal went from 55mV down to 7mV. Seven!

Everything looks good on the bench, so next step is to plug it into
the stereo and let it sit for a while before buttoning it up.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

I'm hoping that this isn't necessary. It looks like it won't be, so I've
got my fingers crossed.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET

This is sooo simple that I'll probably do it as soon as I've got all of my
other projects bottled up and out of the way. Brilliant solution to the
problem here!

So we may (fingers crossed) have a working and stable amp--and in better
shape than when I got it. I'll keep you posted.

Colin

Colin,

SUCCESS! Experience has taught me a good troubleshooting strategy for these amps
and the first 4 transistors are almost always the main culprits when too much DC
offset is present. And, as you found, the solution is relatively simple for those
who have good soldering skills.

Dick

=========================
"Colin B." wrote:

lid wrote:
Pardon me for top posting but I want this material to be "up front" so others
might see it, especially the part about a mod to fix DC offset in these amps.

No problem for me.

Your soldering skills are excellent and it seems you are zeroing in on
success. Did you state you still have a DC offset of 30 mV? Did you "load"
that channel with a ~10 ohm 10 watt resistor and then measure for DC offset at
the positive speaker terminal referenced to circuit ground? Don't be confused
by the NP2222 or PN2222 labels. They are a poor choice in labeling and create
confusion. Most any NPN transistor with an hfe of 100 or more should be OK at
that part of the circuit.

First of all, I've been doing much of my 'rough 'n' ready' measurements
with no load. When I get closer, I toss in an ancient speaker of
suspicious origins, but a nominal resistance of 8ohm.

As for Q9 not being a NP2222, I didn't think that it would be an issue,
but it was rather interesting (not to mention startling, when I first
noticed the discrepancy! :-).

Your sr.# shows your Hafler kit was readied for shipment during the 13th week
of 1981 and was a kit, not factory assembled.

Thanks much! Very cool to know.

Now we're all on the same page about there being an undesirable DC offset
happening, and the real trick is finding it. You had mentioned the Q1/Q2
and Q5/Q6 differential amp pairs, and so I went after them.
Q1 and Q2 were matched almost perfectly, (gain of 78 I think), but
Q5 and Q6 were quite different: hFe was 70 and 115 respectively, which is
a LONG way from 10%! Well, I swapped them with a matched pair and the
no-load offset at the speaker terminal went from 55mV down to 7mV. Seven!

Everything looks good on the bench, so next step is to plug it into
the stereo and let it sit for a while before buttoning it up.

At this point I would suggest swapping the MOSFETs from one channel to
another. It has been my experience that some of the older MOSFETs become
"leaky" which could influence bias settings and cause extra heat. I've had
some of the P-channel devices that no longer matched the original Hafler grade
number and had gone down to a grade level of 1. Also, when checked for leakage
they turned out to be too far out of spec to be usable. I have an original
Hafler MOSFET checker/grader meter here and it has been helpful. Let us know
if the MOSFET swaps from one channel to the other make a difference in bias
current and DC offset and heat sink temps.

I'm hoping that this isn't necessary. It looks like it won't be, so I've
got my fingers crossed.

Now, here is the "magic answer" to fixing DC offsets of around 200 mV or less:

A CIRCUIT MOD TO FIX DC OFFSET

This is sooo simple that I'll probably do it as soon as I've got all of my
other projects bottled up and out of the way. Brilliant solution to the
problem here!

So we may (fingers crossed) have a working and stable amp--and in better
shape than when I got it. I'll keep you posted.

Colin

A bit of a belated response here...

lid wrote:
Colin,

SUCCESS! Experience has taught me a good troubleshooting strategy for these amps
and the first 4 transistors are almost always the main culprits when too much DC
offset is present. And, as you found, the solution is relatively simple for those
who have good soldering skills.

Indeed, success at last! My soldering skills were a bit rusty, but have
come back with a vengeance. My next plan is to replace my crappy iron
with a nice new station (Weller WES51, or Hakko 936--any opinions folks?)
and then start putting all of my other projects back together.

As an aside, this amp may be up for sale before too long. Just a heads
up.

Thanks again, everyone who helped.

Colin

A bit of a belated response here...

lid wrote:
Colin,

SUCCESS! Experience has taught me a good troubleshooting strategy for these amps
and the first 4 transistors are almost always the main culprits when too much DC
offset is present. And, as you found, the solution is relatively simple for those
who have good soldering skills.

Indeed, success at last! My soldering skills were a bit rusty, but have
come back with a vengeance. My next plan is to replace my crappy iron
with a nice new station (Weller WES51, or Hakko 936--any opinions folks?)
and then start putting all of my other projects back together.

As an aside, this amp may be up for sale before too long. Just a heads
up.

Thanks again, everyone who helped.

Colin

A bit of a belated response here...

lid wrote:
Colin,

SUCCESS! Experience has taught me a good troubleshooting strategy for these amps
and the first 4 transistors are almost always the main culprits when too much DC
offset is present. And, as you found, the solution is relatively simple for those
who have good soldering skills.

Indeed, success at last! My soldering skills were a bit rusty, but have
come back with a vengeance. My next plan is to replace my crappy iron
with a nice new station (Weller WES51, or Hakko 936--any opinions folks?)
and then start putting all of my other projects back together.

As an aside, this amp may be up for sale before too long. Just a heads
up.

Thanks again, everyone who helped.

Colin