[Gareth Magennis]

Hi,

In a typical class B transistor power amp with multiple output devices, what
should be the acceptable variation of bias currents through each output
device?

In this example, 5 parallel NPN and 5 parallel PNP devices connected to
output rail by 0.47 Ohm emitter resistors. The voltage across each 0.47 ohm
gives each device's bias current (not sure what tolerance the resistors are)
which varies from device to device (and globally) quite a bit as temperature
goes up and down. At what point would you consider a single device to be
out of spec or suspect?

This amp very occasionally goes briefly into protect, I'm wondering if there
is a suspect output device and/or if I can find it.


Cheers,



Gareth.

[Randy Yates]

"Gareth Magennis" writes:

Hi,

In a typical class B transistor power amp with multiple output devices, what
should be the acceptable variation of bias currents through each output
device?

In this example, 5 parallel NPN and 5 parallel PNP devices connected to
output rail by 0.47 Ohm emitter resistors. The voltage across each 0.47 ohm
gives each device's bias current (not sure what tolerance the resistors are)
which varies from device to device (and globally) quite a bit as temperature
goes up and down. At what point would you consider a single device to be
out of spec or suspect?

A general rule of thumb in electrical engineering is that accuracy
should be maintained to 10 or 20 percent. If the currents are varying
more than that, I'd be suspect.

This amp very occasionally goes briefly into protect, I'm wondering if there
is a suspect output device and/or if I can find it.

I'm just guessing, but I would suspect the one that has a steeper slope
of current vs. temperature.
--
% Randy Yates % "With time with what you've learned,
%% Fuquay-Varina, NC % they'll kiss the ground you walk
%%% 919-577-9882 % upon."
%%%% % '21st Century Man', *Time*, ELO
http://www.digitalsignallabs.com

[Gareth Magennis]

"Randy Yates" wrote in message
...
"Gareth Magennis" writes:

Hi,

In a typical class B transistor power amp with multiple output devices,
what
should be the acceptable variation of bias currents through each output
device?

In this example, 5 parallel NPN and 5 parallel PNP devices connected to
output rail by 0.47 Ohm emitter resistors. The voltage across each 0.47
ohm
gives each device's bias current (not sure what tolerance the resistors
are)
which varies from device to device (and globally) quite a bit as
temperature
goes up and down. At what point would you consider a single device to be
out of spec or suspect?

A general rule of thumb in electrical engineering is that accuracy
should be maintained to 10 or 20 percent. If the currents are varying
more than that, I'd be suspect.

This amp very occasionally goes briefly into protect, I'm wondering if
there
is a suspect output device and/or if I can find it.

I'm just guessing, but I would suspect the one that has a steeper slope
of current vs. temperature.
--
% Randy Yates



Thanks. this one's kind of tricky because all NPNs from both channels are
on one heatsink the width of the rhe amp, and all PNP's on another the same.
The ones in the middle of the heatsink run at a lower bias than the ones on
the outside, presumably due to the difference in temperature which I haven't
measured. As heatsink temp goes up and down after loading, the bias can
vary from inner to outer up to 100%!
(The top is off the amp for measuring, so heatflow will not be as designed,
so its all a bit wibbly anyway). I've been stepping through the biases but I
havent seen anything like a 20% anomaly in adjacent devices.

The protect happened once yesterday for about a second, after 4 hours on the
bench, reassembled with no load, and the fans weren't on so it wasn't hot.



Nightmare.



Gareth.

[Arny Krueger]

"Gareth Magennis" wrote in
message news
Hi,

In a typical class B transistor power amp with multiple
output devices, what should be the acceptable variation
of bias currents through each output device?

In this example, 5 parallel NPN and 5 parallel PNP
devices connected to output rail by 0.47 Ohm emitter
resistors. The voltage across each 0.47 ohm gives each
device's bias current (not sure what tolerance the
resistors are)

At worst, 20%. Could be as close as 1%. Why not measure them in-circuit
(power removed)?

which varies from device to device (and
globally) quite a bit as temperature goes up and down. At what point would
you consider a single device to be
out of spec or suspect?

I'd get worried if the variations are 20%.

This amp very occasionally goes briefly into protect, I'm
wondering if there is a suspect output device and/or if I
can find it.

Under what conditions?

Usually, transistor failures of this magnitude are a one-way trip from good
to permanently bad.

[Arny Krueger]

"Gareth Magennis" wrote in
message

The protect happened once yesterday for about a second,
after 4 hours on the bench, reassembled with no load,
and the fans weren't on so it wasn't hot.

That clarifies a lot. Obviously, there should be no protected condition.

OTOH, what about the power? I recall that most protect circuits come on
during the power-up cycle. A momentary loss of power could cause the symptom
you observe.

The next question would be about the logic that goes into detecting a
protected state.

I've seen some moderately complex circuits for turning on the "protect"
light. The fault could be in a part in the protection circuit, not a fault
of the circuit(s) being protected.

[Gareth Magennis]

"Arny Krueger" wrote in message
. ..
"Gareth Magennis" wrote in
message

The protect happened once yesterday for about a second,
after 4 hours on the bench, reassembled with no load,
and the fans weren't on so it wasn't hot.

That clarifies a lot. Obviously, there should be no protected condition.

OTOH, what about the power? I recall that most protect circuits come on
during the power-up cycle. A momentary loss of power could cause the
symptom you observe.
The next question would be about the logic that goes into detecting a
protected state.

I've seen some moderately complex circuits for turning on the "protect"
light. The fault could be in a part in the protection circuit, not a fault
of the circuit(s) being protected.


I spoke to a tech who fixes a lot of these, and he suggested an output
device as the most likely culprit, so I am following up this line of enquiry
first, as well as learning a few things here along the way. The next step
will be to remove one power amp from circuit and wait (another whole day for
God's sake) for it to go into protect or not, then try the other. If I
could force this fault to happen it would make things a lot easier.


As far as your power sugestion goes, I might look a lot more closely at the
low voltage rails. A blip in one of those could well be the culprit.



Cheers,



Gareth.

[Eeyore]

Gareth Magennis wrote:

Hi,

In a typical class B transistor power amp with multiple output devices, what
should be the acceptable variation of bias currents through each output
device?

It's not highly critical.


In this example, 5 parallel NPN and 5 parallel PNP devices connected to
output rail by 0.47 Ohm emitter resistors. The voltage across each 0.47 ohm
gives each device's bias current (not sure what tolerance the resistors are)

Typically 5 or 10%


which varies from device to device (and globally) quite a bit as temperature
goes up and down. At what point would you consider a single device to be
out of spec or suspect?

This amp very occasionally goes briefly into protect, I'm wondering if there
is a suspect output device and/or if I can find it.

It certainly won't be that.

Graham

[Eeyore]

Arny Krueger wrote:

"Gareth Magennis" wrote:

The protect happened once yesterday for about a second,
after 4 hours on the bench, reassembled with no load,
and the fans weren't on so it wasn't hot.

That clarifies a lot. Obviously, there should be no protected condition.

OTOH, what about the power? I recall that most protect circuits come on
during the power-up cycle. A momentary loss of power could cause the symptom
you observe.

The next question would be about the logic that goes into detecting a
protected state.

I've seen some moderately complex circuits for turning on the "protect"
light. The fault could be in a part in the protection circuit, not a fault
of the circuit(s) being protected.

I agree.

Graham

[Eeyore]

Gareth Magennis wrote:

I spoke to a tech who fixes a lot of these, and he suggested an output
device as the most likely culprit

Sounds like he's an idiot.

Graham

[gaby de wilde]

On May 21, 4:16 pm, Eeyore
wrote:
Gareth Magennis wrote:
I spoke to a tech who fixes a lot of these, and he suggested an output
device as the most likely culprit

Sounds like he's an idiot.

Graham

And you are a Facist

[jakdedert]

gaby de wilde wrote:
On May 21, 4:16 pm, Eeyore
wrote:
Gareth Magennis wrote:
I spoke to a tech who fixes a lot of these, and he suggested an output
device as the most likely culprit
Sounds like he's an idiot.

Graham

And you are a Facist

I've heard that about him. He's absolutely obsessed with faces...looks
at his own in the mirror at least once a day, as he shaves it. He's
been known to study others' faces--both on the street and inside public
facilities--and not just for purposes of recognition.

In fact, I hear that he owns and employs a device known as a
'television' which displays peoples' faces on a screen inside of his home.

Oh, the humanity!

jak

[Arny Krueger]

"gaby de wilde" wrote in message

On May 21, 4:16 pm, Eeyore
wrote:
Gareth Magennis wrote:
I spoke to a tech who fixes a lot of these, and he
suggested an output device as the most likely culprit

Sounds like he's an idiot.

Graham

And you are a Facist

I suspect you mean Fascist. It's always funny when someone engages in
childish name-calling and then pees on themselves in the process.

[Eeyore]

gaby de wilde wrote:

Eeyore wrote:
Gareth Magennis wrote:
I spoke to a tech who fixes a lot of these, and he suggested an output
device as the most likely culprit

Sounds like he's an idiot.

Graham

And you are a Facist

And you are a MORON.

[Mark D. Zacharias[_2_]]

"Gareth Magennis" wrote in message
...

"Randy Yates" wrote in message
...
"Gareth Magennis" writes:

Hi,

In a typical class B transistor power amp with multiple output devices,
what
should be the acceptable variation of bias currents through each output
device?

In this example, 5 parallel NPN and 5 parallel PNP devices connected to
output rail by 0.47 Ohm emitter resistors. The voltage across each 0.47
ohm
gives each device's bias current (not sure what tolerance the resistors
are)
which varies from device to device (and globally) quite a bit as
temperature
goes up and down. At what point would you consider a single device to
be
out of spec or suspect?

A general rule of thumb in electrical engineering is that accuracy
should be maintained to 10 or 20 percent. If the currents are varying
more than that, I'd be suspect.

This amp very occasionally goes briefly into protect, I'm wondering if
there
is a suspect output device and/or if I can find it.

I'm just guessing, but I would suspect the one that has a steeper slope
of current vs. temperature.
--
% Randy Yates



Thanks. this one's kind of tricky because all NPNs from both channels are
on one heatsink the width of the rhe amp, and all PNP's on another the
same. The ones in the middle of the heatsink run at a lower bias than the
ones on the outside, presumably due to the difference in temperature which
I haven't measured. As heatsink temp goes up and down after loading, the
bias can vary from inner to outer up to 100%!
(The top is off the amp for measuring, so heatflow will not be as
designed, so its all a bit wibbly anyway). I've been stepping through the
biases but I havent seen anything like a 20% anomaly in adjacent devices.

The protect happened once yesterday for about a second, after 4 hours on
the bench, reassembled with no load, and the fans weren't on so it wasn't
hot.



Nightmare.



Gareth.







Well, we haven't been given the amp model, and have no schematic here, so
generalizations are all that is possible but I have some experience
troubleshooting problems like this.

It is not a sin to defeat the protection circuit temporarily for the purpose
of troubleshooting. For example, there could be a pull-up line to the
protection circuit (or microprocessor as the case may be) which is pulling
low momentarily. Opening this line up might help determine whether it's the
amp or the protect circuit - ditto with the DC sensing resistor, it can be
disconnected as well.

Possibly a bad driver transistor clips the waveform briefly, triggering a DC
protect. This would most often be a bad emitter-base junction on a TO-220
package driver. The DC voltage across this junction can be observed and of
course if the voltage jumps from the normal .6 volts or so to more than a
volt, or several volts obviously that transistor is bad. If the problem only
occurs with a load connected this is a pretty likely scenario.

Many protection circuits also monitor the power supplies as well; often the
sum of the voltages, as seen through a voltage divider network, should be
near zero volts. If for example a voltage regulator drops in output, the sum
is no longer zero, and the circuit triggers. Rarely a resistor in the
network could be bad or out of spec.

Many protection circuits also feature a "fast shut-off" so that when the
power is cut the speaker relay cuts off immediately, preventing the amp
clipping or popping audibly at turnoff. This circuit usually consists of a
small electrolytic lightly filtering a half-wave (single diode) line off the
power transformer, going from there to the base of one of the transistors in
the relay drive circuit. If the cap is totally bad, the amp would never
un-mute, but it could be marginal and not totally open.


Mark Z.