Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.
These are set up so there is 6.6 mA from a pair of 6922 cathodes flowing
down to -160V
via 16k and the drain to source. A pot varies the source resistor to the
-160V. The gate is taken directly to
the -160V.

Both of these have failed to a short circuit.

The schematic can be inspected at
http://www.freeinfosociety.com/elect...ew.php?id=2332

ARC do not tell us what sort of mosfet or type number the mosfet are and
there are no numbers on the
mosfet, and only ARC's secret colour code so I am forced to deal with
ARC, which isn't what i want to do
They want usd $80 per hour for their labour rate, which is 20dB more
than I get.

Any idea on a suitable generic replacement for the mosfets?
The op conditions will be

drain voltage -60V, source voltage -158.5V approx,
gate voltage -160V, Ids = 6.6mA. ( this means a voltage of between 1V
and 2V exists between gate and source
with gate negative in respect to source, like a tube.

I searched a few mosfets from Farnell but all required a gate voltage
which goes positive with respect to the
source. I tried BS107A and found it didn't work as the gate voltage
cannot be set above the source voltage.

I do not want to alter the double sided board to accommodate some other
device such as a common bjt
for the CCS.

Anybody got any answers?

TIA,

Patrick Turner.

Patrick Turner wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.
These are set up so there is 6.6 mA from a pair of 6922 cathodes flowing
down to -160V
via 16k and the drain to source. A pot varies the source resistor to the
-160V. The gate is taken directly to
the -160V.

Both of these have failed to a short circuit.

The schematic can be inspected at
http://www.freeinfosociety.com/elect...ew.php?id=2332

ARC do not tell us what sort of mosfet or type number the mosfet are and
there are no numbers on the
mosfet, and only ARC's secret colour code so I am forced to deal with
ARC, which isn't what i want to do
They want usd $80 per hour for their labour rate, which is 20dB more
than I get.

Any idea on a suitable generic replacement for the mosfets?
The op conditions will be

drain voltage -60V, source voltage -158.5V approx,
gate voltage -160V, Ids = 6.6mA. ( this means a voltage of between 1V
and 2V exists between gate and source
with gate negative in respect to source, like a tube.

I searched a few mosfets from Farnell but all required a gate voltage
which goes positive with respect to the
source. I tried BS107A and found it didn't work as the gate voltage
cannot be set above the source voltage.

I do not want to alter the double sided board to accommodate some other
device such as a common bjt
for the CCS.

Anybody got any answers?

I think I have an answer why they failed. There's ~ 660mW dissipated in these
and that's typically over the limit in TO-92 even at Ta = 25C.

In order to be reliable I reckon you should be looking at a part in TO-220 most
likely

Graham

Patrick Turner wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.

You could try BSP245A http://www.farnell.com/datasheets/59098.pdf

It is TO-92 with a rated Pd of 1W @ Ta = 25C so it's right on the limit. Maybe a
heatsink clip would be in order.

Otherwise take your pick.....
http://uk.farnell.com/jsp/endecaSear...io n&x=37&y=6

I expect that'll wrap !

Graham

On Thu, 01 Jun 2006 14:27:32 GMT, Patrick Turner
wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.
These are set up so there is 6.6 mA from a pair of 6922 cathodes flowing
down to -160V
via 16k and the drain to source. A pot varies the source resistor to the
-160V. The gate is taken directly to
the -160V.

Both of these have failed to a short circuit.

The schematic can be inspected at
http://www.freeinfosociety.com/elect...ew.php?id=2332

ARC do not tell us what sort of mosfet or type number the mosfet are and
there are no numbers on the
mosfet, and only ARC's secret colour code so I am forced to deal with
ARC, which isn't what i want to do
They want usd $80 per hour for their labour rate, which is 20dB more
than I get.

Any idea on a suitable generic replacement for the mosfets?
The op conditions will be

drain voltage -60V, source voltage -158.5V approx,
gate voltage -160V, Ids = 6.6mA. ( this means a voltage of between 1V
and 2V exists between gate and source
with gate negative in respect to source, like a tube.

I searched a few mosfets from Farnell but all required a gate voltage
which goes positive with respect to the
source. I tried BS107A and found it didn't work as the gate voltage
cannot be set above the source voltage.

I do not want to alter the double sided board to accommodate some other
device such as a common bjt
for the CCS.

Anybody got any answers?

TIA,

Patrick Turner.

Patrick, the BS107A is a standard MOSFET, unless I'm missing something
with what you wrote you want a *depletion mode* (normally on) MOSFET.
You may end up tweaking your source resistor up or down to trim the
exact constant current, but something like this comes close

http://www.supertex.com/pdf/datasheets/DN3545.pdf

Or google "high voltage depletion mode"

On Thu, 01 Jun 2006 10:22:45 -0600, Gilbert Bates
wrote:

On Thu, 01 Jun 2006 14:27:32 GMT, Patrick Turner
wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.
These are set up so there is 6.6 mA from a pair of 6922 cathodes flowing
down to -160V
via 16k and the drain to source. A pot varies the source resistor to the
-160V. The gate is taken directly to
the -160V.

Both of these have failed to a short circuit.

The schematic can be inspected at
http://www.freeinfosociety.com/elect...ew.php?id=2332

ARC do not tell us what sort of mosfet or type number the mosfet are and
there are no numbers on the
mosfet, and only ARC's secret colour code so I am forced to deal with
ARC, which isn't what i want to do
They want usd $80 per hour for their labour rate, which is 20dB more
than I get.

Any idea on a suitable generic replacement for the mosfets?
The op conditions will be

drain voltage -60V, source voltage -158.5V approx,
gate voltage -160V, Ids = 6.6mA. ( this means a voltage of between 1V
and 2V exists between gate and source
with gate negative in respect to source, like a tube.

I searched a few mosfets from Farnell but all required a gate voltage
which goes positive with respect to the
source. I tried BS107A and found it didn't work as the gate voltage
cannot be set above the source voltage.

I do not want to alter the double sided board to accommodate some other
device such as a common bjt
for the CCS.

Anybody got any answers?

TIA,

Patrick Turner.

Patrick, the BS107A is a standard MOSFET, unless I'm missing something
with what you wrote you want a *depletion mode* (normally on) MOSFET.
You may end up tweaking your source resistor up or down to trim the
exact constant current, but something like this comes close

http://www.supertex.com/pdf/datasheets/DN3545.pdf

Or google "high voltage depletion mode"


This one has a little *closer* transfer characteristics based on the
size of your existing source resistor.

http://www.supertex.com/pdf/datasheets/DN2540.pdf

Gilbert Bates wrote:

On Thu, 01 Jun 2006 16:44:38 +0100, Pooh Bear
wrote:



Patrick Turner wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.

You could try BSP245A http://www.farnell.com/datasheets/59098.pdf

It is TO-92 with a rated Pd of 1W @ Ta = 25C so it's right on the limit. Maybe a
heatsink clip would be in order.

Otherwise take your pick.....
http://uk.farnell.com/jsp/endecaSearch/searchPage2.jsp?Ntk=&comSearch=true&select1=179280 &select1=179816&select1=179818&select1=179828&sele ct1=179841&select1=179883&select1=179906&paramSele ctCount=7&showImages=true&st=parametricSelection&N =401%2B1004689+194090+194091+185620+185636+185637+ 185639+185641+185643+185654+1856I expect that'll wrap !

Graham

He needs a N channel depletion device not a P channel enhancement
device. Think JFET operation.

Crikey. A depletion mode mosfet. Rarer than hens' teeth I should think.

Graham

Gilbert Bates wrote:

On Thu, 01 Jun 2006 10:22:45 -0600, Gilbert Bates
wrote:

On Thu, 01 Jun 2006 14:27:32 GMT, Patrick Turner
wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.
These are set up so there is 6.6 mA from a pair of 6922 cathodes flowing
down to -160V
via 16k and the drain to source. A pot varies the source resistor to the
-160V. The gate is taken directly to
the -160V.

Both of these have failed to a short circuit.

The schematic can be inspected at
http://www.freeinfosociety.com/elect...ew.php?id=2332

ARC do not tell us what sort of mosfet or type number the mosfet are and
there are no numbers on the
mosfet, and only ARC's secret colour code so I am forced to deal with
ARC, which isn't what i want to do
They want usd $80 per hour for their labour rate, which is 20dB more
than I get.

Any idea on a suitable generic replacement for the mosfets?
The op conditions will be

drain voltage -60V, source voltage -158.5V approx,
gate voltage -160V, Ids = 6.6mA. ( this means a voltage of between 1V
and 2V exists between gate and source
with gate negative in respect to source, like a tube.

I searched a few mosfets from Farnell but all required a gate voltage
which goes positive with respect to the
source. I tried BS107A and found it didn't work as the gate voltage
cannot be set above the source voltage.

I do not want to alter the double sided board to accommodate some other
device such as a common bjt
for the CCS.

Anybody got any answers?

TIA,

Patrick Turner.

Patrick, the BS107A is a standard MOSFET, unless I'm missing something
with what you wrote you want a *depletion mode* (normally on) MOSFET.

BS107A also fails on dissipation.

You may end up tweaking your source resistor up or down to trim the
exact constant current, but something like this comes close

http://www.supertex.com/pdf/datasheets/DN3545.pdf

Way too close on dissipation.


Or google "high voltage depletion mode"

This one has a little *closer* transfer characteristics based on the
size of your existing source resistor.

http://www.supertex.com/pdf/datasheets/DN2540.pdf

That looks decent. Advisable still to fit a heatsink clip though.

Graham

Patrick,

If you want I'll call them and ask them . . .

Jon

Pooh Bear wrote:

Patrick Turner wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.
These are set up so there is 6.6 mA from a pair of 6922 cathodes flowing
down to -160V
via 16k and the drain to source. A pot varies the source resistor to the
-160V. The gate is taken directly to
the -160V.

Both of these have failed to a short circuit.

The schematic can be inspected at
http://www.freeinfosociety.com/elect...ew.php?id=2332

ARC do not tell us what sort of mosfet or type number the mosfet are and
there are no numbers on the
mosfet, and only ARC's secret colour code so I am forced to deal with
ARC, which isn't what i want to do
They want usd $80 per hour for their labour rate, which is 20dB more
than I get.

Any idea on a suitable generic replacement for the mosfets?
The op conditions will be

drain voltage -60V, source voltage -158.5V approx,
gate voltage -160V, Ids = 6.6mA. ( this means a voltage of between 1V
and 2V exists between gate and source
with gate negative in respect to source, like a tube.

I searched a few mosfets from Farnell but all required a gate voltage
which goes positive with respect to the
source. I tried BS107A and found it didn't work as the gate voltage
cannot be set above the source voltage.

I do not want to alter the double sided board to accommodate some other
device such as a common bjt
for the CCS.

Anybody got any answers?

I think I have an answer why they failed. There's ~ 660mW dissipated in these
and that's typically over the limit in TO-92 even at Ta = 25C.

In order to be reliable I reckon you should be looking at a part in TO-220 most
likely

Graham

There is a current limiting drain resistance of 16.8k and the voltage I quoted
above was wrong
since there drain operating voltage is actually supposed to be -49V, so Vgs = about
47V
so dissipation = 0.31 watts which is within spec but the heatsinks on the failed
parts were within
12mm from the input tubes which is stupid design.

On a summer's day how could they run cool?

But in this particular amp one channel has the mosfets on one side of the board
away from the tubes, and
the other where the mosfets failed the mosfets are on the same side as the tubes.

But after some thought the T0220 package would have been better of course and it is
penny
pinching by ARC to do things the way they have.

And then when such unusual mosfets fail nobody has the special parts.


Patrick Turner.

Gilbert Bates wrote:

On Thu, 01 Jun 2006 14:27:32 GMT, Patrick Turner
wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.
These are set up so there is 6.6 mA from a pair of 6922 cathodes flowing
down to -160V
via 16k and the drain to source. A pot varies the source resistor to the
-160V. The gate is taken directly to
the -160V.

Both of these have failed to a short circuit.

The schematic can be inspected at
http://www.freeinfosociety.com/elect...ew.php?id=2332

ARC do not tell us what sort of mosfet or type number the mosfet are and
there are no numbers on the
mosfet, and only ARC's secret colour code so I am forced to deal with
ARC, which isn't what i want to do
They want usd $80 per hour for their labour rate, which is 20dB more
than I get.

Any idea on a suitable generic replacement for the mosfets?
The op conditions will be

drain voltage -60V, source voltage -158.5V approx,
gate voltage -160V, Ids = 6.6mA. ( this means a voltage of between 1V
and 2V exists between gate and source
with gate negative in respect to source, like a tube.

I searched a few mosfets from Farnell but all required a gate voltage
which goes positive with respect to the
source. I tried BS107A and found it didn't work as the gate voltage
cannot be set above the source voltage.

I do not want to alter the double sided board to accommodate some other
device such as a common bjt
for the CCS.

Anybody got any answers?

TIA,

Patrick Turner.

Patrick, the BS107A is a standard MOSFET, unless I'm missing something
with what you wrote you want a *depletion mode* (normally on) MOSFET.
You may end up tweaking your source resistor up or down to trim the
exact constant current, but something like this comes close

http://www.supertex.com/pdf/datasheets/DN3545.pdf

Or google "high voltage depletion mode"

The DN345 may indeed work but a mosfet which works with only 7mA Id at about
Vgs = -1.5V is
about what i want.

Patrick Turner.

Is that an original VT100 or a MKII/III? Reason I ask is it looks like
they added a fan on the later models.

http://www.arcdb.ws/VT100/VT100.html

Also, is there room to flip mount the problem MOSFET on the underside
of the board?

No problem. The good channel was originally done like this. Why each channel had the
mosfets
on different sides of the board is a mystery.

This one has no fan, and uses 6922 and mosfets for the input pairs but the later models
are said to have 6H30 plus j-fets..

Patrick Turner.

Pooh Bear wrote:

Patrick Turner wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.

You could try BSP245A http://www.farnell.com/datasheets/59098.pdf

This is a P channel enhancement fet and cannot work in the circuit.
We want N channel depletion type.



It is TO-92 with a rated Pd of 1W @ Ta = 25C so it's right on the limit. Maybe a
heatsink clip would be in order.

That's a P channel mosfet and won't work.

Patrick Turner



Otherwise take your pick.....
http://uk.farnell.com/jsp/endecaSear...43+185654+1856
I expect that'll wrap !

Graham

Farnell do not seem to sell depletion mosfets. The word depletion drew no hits in a search of their site.

I searched elsewhere and found DN2530 looks maybe suitable.

Patrick Turner.

Jon Yaeger wrote:

Patrick,

If you want I'll call them and ask them . . .

Jon

If they condescend to tell you, perhaps you could email me the type
number privately.
We don't know what spies ARC have monitoring our discussions and I don't

want ARC chief phoning Dubbya to send an aircraft carrier to bomb
my house in Canberra.

Patrick Turner.

Patrick Turner wrote:

Pooh Bear wrote:

Patrick Turner wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.

You could try BSP245A http://www.farnell.com/datasheets/59098.pdf

This is a P channel enhancement fet and cannot work in the circuit.
We want N channel depletion type.

Sorry, I missed the apparently obvious. Probably on acount of not expecting to see a depletion mode device.

Could you sub a jfet ?

Graham

Patrick Turner said:


If they condescend to tell you, perhaps you could email me the type
number privately.
We don't know what spies ARC have monitoring our discussions and I don't
want ARC chief phoning Dubbya to send an aircraft carrier to bomb
my house in Canberra.


They'll just pay Phil Allison for another round of insults.

--

- Never argue with idiots, they drag you down their level and beat you with experience. -

Patrick Turner said:


And then when such unusual mosfets fail nobody has the special parts.


Yeah, what's the use of a tube amp anyway when the only solid state
devices inthere crap out on you? ;-)

--

- Never argue with idiots, they drag you down their level and beat you with experience. -

Pooh Bear wrote:

Patrick Turner wrote:

Pooh Bear wrote:

Patrick Turner wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.

You could try BSP245A http://www.farnell.com/datasheets/59098.pdf

This is a P channel enhancement fet and cannot work in the circuit.
We want N channel depletion type.

Sorry, I missed the apparently obvious. Probably on acount of not expecting to see a depletion mode device.

Could you sub a jfet ?

I have never seen a j-fet with a high enough Vds rating.

The Audio Encyclopedia by Tremain from the 1960's shows a tube amp schematic where the input tubes have
all been substituted for j-fets and working with near original B+ rails.
Gain was higher than triodes and since THD is proportional to gain +1, the thd in the j-fet
altered amp was lower. I tried looking up the old j-fet numbers long ago but I ain't never seen any around
Something like mpf102 would be quite unsuitable.

Patrick Turner.


Graham

flipper wrote:

On Fri, 02 Jun 2006 06:45:30 GMT, Patrick Turner
wrote:



Jon Yaeger wrote:

Patrick,

If you want I'll call them and ask them . . .

Jon

If they condescend to tell you, perhaps you could email me the type
number privately.
We don't know what spies ARC have monitoring our discussions and I don't

want ARC chief phoning Dubbya to send an aircraft carrier to bomb
my house in Canberra.

Nah, they'd just use one of the new shoulder fired smart missiles. Saw
them on a show last week and they made a point to mention that,
besides tanks, they're great for blowing up small buildings.

Yeh, but our boomerangs which are carved from a large hard gum tree and aimed
at
anyone trying to
fire a missle would cut the man in half and explode his rocket to spread the
remains........

Beware of trying to bomb Canberra, and the evil spirits leashed by the local
abos
will make a plague of fleas infest your groin, and your arms will wither too
short to be able to
scratch it.

Patrick Turner.




Patrick Turner.

Sander deWaal wrote:

Patrick Turner said:

And then when such unusual mosfets fail nobody has the special parts.

Yeah, what's the use of a tube amp anyway when the only solid state
devices inthere crap out on you? ;-)


Well in this case methinks a TO220 bjt would have been a lot better device to use.
But for some reason bjts are seen by the High Priests of Audio Waffle seem to think
bjts stink, or are audiologically unacceptable.
Such 'know-nothing-technical' types describe the front end of the ARC amps as being hybrid
with j-fets or mosfets and tubes contributing to the sound etc.
But the mosfets and j-fets are used as constant current sources, and have virtually zero
contribution
to the sound and that's why they are used, and acceptable; the tubes do 99.99999% of the
signal handling.
A bjt would work fine as a CCS. But maybe they'd lose sales if they put bjts in there.
People would not accept that there were transistors in the circuit.

The VT100 has 5 mosfet CCS in cathode current sinks, and has a total of 8 triodes in the
input/driver amp.
They could have used a simpler line up of tubes but they
wanted the benefits of fully balanced circuitry including balanced inputs and balanced global
FB paths.
The alternative to using SS as CCS would be to use tube CCS which is entirely unecessary
because whatever active CCS is used becomes such a high dynamic impedance its "sonic
signature"
is reduced because it feeds low impedance cathode circuits. Cathode followers drive the
output stage with fixed bias
so allowing slightly class AB2 operation and presumably a soft clip because the CF drivers
have series R to
limit output stage grid current.
But why would anyone want to run such amps up to their 100W/channel capacity?
I guess someone might, and for that fussy man he gets what he wants.

Patrick Turner.


--

- Never argue with idiots, they drag you down their level and beat you with experience. -

Patrick Turner wrote:

Pooh Bear wrote:

Patrick Turner wrote:

Pooh Bear wrote:

Patrick Turner wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.

You could try BSP245A http://www.farnell.com/datasheets/59098.pdf

This is a P channel enhancement fet and cannot work in the circuit.
We want N channel depletion type.

Sorry, I missed the apparently obvious. Probably on acount of not expecting to see a depletion mode device.

Could you sub a jfet ?

I have never seen a j-fet with a high enough Vds rating.

Good point.

The Audio Encyclopedia by Tremain from the 1960's shows a tube amp schematic where the input tubes have
all been substituted for j-fets and working with near original B+ rails.
Gain was higher than triodes and since THD is proportional to gain +1, the thd in the j-fet
altered amp was lower. I tried looking up the old j-fet numbers long ago but I ain't never seen any around
Something like mpf102 would be quite unsuitable.

ISTR that at one time you could buy fet replacements for tubes in B9A etc. There was some trade name IIRC but
forget what it is now.

DN2535 or 2540 comes back again as a best bet.

Infineon have some stuff but only in SMT !

Graham

Patrick Turner said:


The VT100 has 5 mosfet CCS in cathode current sinks, and has a total of 8 triodes in the
input/driver amp.
They could have used a simpler line up of tubes but they
wanted the benefits of fully balanced circuitry including balanced inputs and balanced global
FB paths.
The alternative to using SS as CCS would be to use tube CCS which is entirely unecessary
because whatever active CCS is used becomes such a high dynamic impedance its "sonic
signature"
is reduced because it feeds low impedance cathode circuits. Cathode followers drive the
output stage with fixed bias
so allowing slightly class AB2 operation and presumably a soft clip because the CF drivers
have series R to
limit output stage grid current.
But why would anyone want to run such amps up to their 100W/channel capacity?
I guess someone might, and for that fussy man he gets what he wants.


Couldn't you just convert to BJT and be done with it?
MJE340 or 350 could do it, I guess.

A depletion P channel MOSFET is as rare as an Allison post without
swearing.

--

- Never argue with idiots, they drag you down their level and beat you with experience. -

Pooh Bear wrote:

Patrick Turner wrote:

Pooh Bear wrote:

Patrick Turner wrote:

Pooh Bear wrote:

Patrick Turner wrote:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.

You could try BSP245A http://www.farnell.com/datasheets/59098.pdf

This is a P channel enhancement fet and cannot work in the circuit.
We want N channel depletion type.

Sorry, I missed the apparently obvious. Probably on acount of not expecting to see a depletion mode device.

Could you sub a jfet ?

I have never seen a j-fet with a high enough Vds rating.

Good point.

The Audio Encyclopedia by Tremain from the 1960's shows a tube amp schematic where the input tubes have
all been substituted for j-fets and working with near original B+ rails.
Gain was higher than triodes and since THD is proportional to gain +1, the thd in the j-fet
altered amp was lower. I tried looking up the old j-fet numbers long ago but I ain't never seen any around
Something like mpf102 would be quite unsuitable.

ISTR that at one time you could buy fet replacements for tubes in B9A etc. There was some trade name IIRC but
forget what it is now.

DN2535 or 2540 comes back again as a best bet.

I have some DN3545 on the way here.

Maybe thye work OK with low current.

The data drain resistance curves for all these devices is for up to an amp and it is difficult to see
what performance is possible at Id = 6mA.

Anyway, if they don't work maybe something else will.



Infineon have some stuff but only in SMT !

Yeah, quite a few depletion mosfets in surface mount but not much in TO92 or TO220 or TO126 packages.

Patrick Turner.



Graham

Sander deWaal wrote:

Patrick Turner said:

The VT100 has 5 mosfet CCS in cathode current sinks, and has a total of 8 triodes in the
input/driver amp.
They could have used a simpler line up of tubes but they
wanted the benefits of fully balanced circuitry including balanced inputs and balanced global
FB paths.
The alternative to using SS as CCS would be to use tube CCS which is entirely unecessary
because whatever active CCS is used becomes such a high dynamic impedance its "sonic
signature"
is reduced because it feeds low impedance cathode circuits. Cathode followers drive the
output stage with fixed bias
so allowing slightly class AB2 operation and presumably a soft clip because the CF drivers
have series R to
limit output stage grid current.
But why would anyone want to run such amps up to their 100W/channel capacity?
I guess someone might, and for that fussy man he gets what he wants.

Couldn't you just convert to BJT and be done with it?
MJE340 or 350 could do it, I guess.

Then it becomes a ARCner amp, or a TurnARC, and folks may not want to pay big prices for
incorribly modded second hand hi-end amps.
I am sure the owner would like to see it kept as original as possible.

Otherwise I agree, but I would have to replace all the mosfets, and establish a stable base
voltage and solder in suitable value emitter resistors and there isn't must space anywhere on the
boards and its actually a lot of farnarkling around, and since the majority of the mosfets have
survived because they are located in cooler locations then perhaps they will survive indefinately.
I just wanna repair it, and get it off the bench.
The location of the failed mosfets beside tubes that get rather warm in the middle of summer was
unfortunate.
When relocated to the other side of the board they will probably reduce their temperature by 50
degrees which
should allow them to survive OK.

ARC temp fate with the location of the heatsink for the regulator for the driver amp B+
being very close to a 6550 tube. Its madness. Lucky the reg and device doesn't have to do any more
than
pass a fairly small current, but it must run way hotter than it has to.

The reliablity of solid state declines by 10% for each 10 degrees C above room temperature.
So with a rise of 100C, an amplifier is 100% unreliable. It will fail.


Patrick Turner.

Patrick,

I called ARC. If the Mosfets in question are green/white/orange they are
charging $15.50 each.

Let me know if you want me to help you get a pair.

Jon






in article , Patrick Turner at
wrote on 6/1/06 10:27 AM:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.
These are set up so there is 6.6 mA from a pair of 6922 cathodes flowing
down to -160V
via 16k and the drain to source. A pot varies the source resistor to the
-160V. The gate is taken directly to
the -160V.

Both of these have failed to a short circuit.

The schematic can be inspected at
http://www.freeinfosociety.com/elect...ew.php?id=2332

ARC do not tell us what sort of mosfet or type number the mosfet are and
there are no numbers on the
mosfet, and only ARC's secret colour code so I am forced to deal with
ARC, which isn't what i want to do
They want usd $80 per hour for their labour rate, which is 20dB more
than I get.

Any idea on a suitable generic replacement for the mosfets?
The op conditions will be

drain voltage -60V, source voltage -158.5V approx,
gate voltage -160V, Ids = 6.6mA. ( this means a voltage of between 1V
and 2V exists between gate and source
with gate negative in respect to source, like a tube.

I searched a few mosfets from Farnell but all required a gate voltage
which goes positive with respect to the
source. I tried BS107A and found it didn't work as the gate voltage
cannot be set above the source voltage.

I do not want to alter the double sided board to accommodate some other
device such as a common bjt
for the CCS.

Anybody got any answers?

TIA,

Patrick Turner.

Patrick Turner wrote:


Pooh Bear wrote:


Patrick Turner wrote:


Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.

You could try BSP245A http://www.farnell.com/datasheets/59098.pdf


This is a P channel enhancement fet and cannot work in the circuit.
We want N channel depletion type.


If you end up with no other options, you could build a small daughter
board to use the P channel fet and associated circuitry "backwards" to
create the constant current source. The daughter board is suggested so
you can avoid cutting traces on the existing board.

On Fri, 02 Jun 2006 19:24:16 GMT, robert casey
wrote:

Patrick Turner wrote:


Pooh Bear wrote:


Patrick Turner wrote:


Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.

You could try BSP245A http://www.farnell.com/datasheets/59098.pdf


This is a P channel enhancement fet and cannot work in the circuit.
We want N channel depletion type.


If you end up with no other options, you could build a small daughter
board to use the P channel fet and associated circuitry "backwards" to
create the constant current source. The daughter board is suggested so
you can avoid cutting traces on the existing board.

I believe you run into the same problem but in reverse. The CCS
produces a neg gate voltage across the source resistor with respect to
the source. The N channel enhancement device requires a positive gate
voltage and won't work in the standard configuration.

A P channel enhancement device requires a negative gate-source
voltage. But to run it 'upside-down' you end up with a positive Vgs
with respect to the source. Check it out, same problem, different
polarity. Only a depletion mode device will work with the classic CCS
setup were Vgs is derived from a source resistor.

On Fri, 02 Jun 2006 17:36:10 GMT, Patrick Turner
wrote:

ARC temp fate with the location of the heatsink

Nominated for best typo of 2006. But you planned it,
didn't ya?

All good fortune,

Chris Hornbeck

Jon Yaeger wrote:

Patrick,

I called ARC. If the Mosfets in question are green/white/orange they are
charging $15.50 each.

Let me know if you want me to help you get a pair.

Thanks Jon. I knew ARC would tell you that and try to have us pay 20dB more
for something
available from a supplier elsewhere.

The green orange black is a type no I have not heard of but yes, that is the
colour used to id the mosfet.

usd $15.5 is about aud $25 and then i have to arrange an international
payment.

You'd think ARC would just give you a pair for goodwill.

I have some alternatives on the way. If they work OK then they would be close
to whatever ARC have used.

Patrick Turner.



Jon

in article , Patrick Turner at
wrote on 6/1/06 10:27 AM:

Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.
These are set up so there is 6.6 mA from a pair of 6922 cathodes flowing
down to -160V
via 16k and the drain to source. A pot varies the source resistor to the
-160V. The gate is taken directly to
the -160V.

Both of these have failed to a short circuit.

The schematic can be inspected at
http://www.freeinfosociety.com/elect...ew.php?id=2332

ARC do not tell us what sort of mosfet or type number the mosfet are and
there are no numbers on the
mosfet, and only ARC's secret colour code so I am forced to deal with
ARC, which isn't what i want to do
They want usd $80 per hour for their labour rate, which is 20dB more
than I get.

Any idea on a suitable generic replacement for the mosfets?
The op conditions will be

drain voltage -60V, source voltage -158.5V approx,
gate voltage -160V, Ids = 6.6mA. ( this means a voltage of between 1V
and 2V exists between gate and source
with gate negative in respect to source, like a tube.

I searched a few mosfets from Farnell but all required a gate voltage
which goes positive with respect to the
source. I tried BS107A and found it didn't work as the gate voltage
cannot be set above the source voltage.

I do not want to alter the double sided board to accommodate some other
device such as a common bjt
for the CCS.

Anybody got any answers?

TIA,

Patrick Turner.

robert casey wrote:

Patrick Turner wrote:


Pooh Bear wrote:


Patrick Turner wrote:


Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.

You could try BSP245A http://www.farnell.com/datasheets/59098.pdf


This is a P channel enhancement fet and cannot work in the circuit.
We want N channel depletion type.


If you end up with no other options, you could build a small daughter
board to use the P channel fet and associated circuitry "backwards" to
create the constant current source. The daughter board is suggested so
you can avoid cutting traces on the existing board.

Yes but the pchannel wouldn't work as well for CCS.
The N channel has its resistor in the source circuit, thereby self regulating
the CCS.

N type bjts would be far easier if I have to do something even though a board
is involved.

Patrick Turner.

Gilbert Bates wrote:

On Fri, 02 Jun 2006 19:24:16 GMT, robert casey
wrote:

Patrick Turner wrote:


Pooh Bear wrote:


Patrick Turner wrote:


Hi All,

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.

You could try BSP245A http://www.farnell.com/datasheets/59098.pdf


This is a P channel enhancement fet and cannot work in the circuit.
We want N channel depletion type.


If you end up with no other options, you could build a small daughter
board to use the P channel fet and associated circuitry "backwards" to
create the constant current source. The daughter board is suggested so
you can avoid cutting traces on the existing board.

I believe you run into the same problem but in reverse. The CCS
produces a neg gate voltage across the source resistor with respect to
the source. The N channel enhancement device requires a positive gate
voltage and won't work in the standard configuration.

A P channel enhancement device requires a negative gate-source
voltage. But to run it 'upside-down' you end up with a positive Vgs
with respect to the source. Check it out, same problem, different
polarity. Only a depletion mode device will work with the classic CCS
setup were Vgs is derived from a source resistor.

Well, an enhancenment mosfet would work if I had a positive voltage supply of
about +3V above the -160V supply.
The source would then be at about +1.5V above the -160V supply and the existing
adjustable source resistance could be retained but there is still some circuit
mod to be done. The depletion mosfet allows the simplest arrangement
since the gate is taken direct to -160V and the source is at about =2V above
160V, or at -158V, with a multiturn adjust pot to set the source current.

Patrick Turner.

Chris Hornbeck wrote:

On Fri, 02 Jun 2006 17:36:10 GMT, Patrick Turner
wrote:

ARC temp fate with the location of the heatsink

Nominated for best typo of 2006. But you planned it,
didn't ya?

Degrees are on me brain, and I get all hot up,
I have to deal with makers faults
and put up with all their crup.

Patrick Turner.



All good fortune,

Chris Hornbeck

On Sat, 03 Jun 2006 03:42:56 GMT, Patrick Turner
wrote:

Well, an enhancenment mosfet would work if I had a positive voltage supply of
about +3V above the -160V supply.

Battery?

All good fortune,

Chris Hornbeck

Chris Hornbeck wrote:

On Sat, 03 Jun 2006 03:42:56 GMT, Patrick Turner
wrote:

Well, an enhancenment mosfet would work if I had a positive voltage supply of
about +3V above the -160V supply.

Battery?

It would eventually go flat.
Nope, there needs to be a divider and a large electro cap.

Patrick Turner.



All good fortune,

Chris Hornbeck

Patrick Turner wrote

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.
These are set up so there is 6.6 mA from a pair of 6922 cathodes flowing
down to -160V
via 16k and the drain to source. A pot varies the source resistor to the
-160V. The gate is taken directly to
the -160V.

Both of these have failed to a short circuit.

The schematic can be inspected at
http://www.freeinfosociety.com/elect...ew.php?id=2332

ARC do not tell us what sort of mosfet or type number the mosfet are and
there are no numbers on the
mosfet, and only ARC's secret colour code so I am forced to deal with
ARC, which isn't what i want to do
They want usd $80 per hour for their labour rate, which is 20dB more
than I get.

Any idea on a suitable generic replacement for the mosfets?
The op conditions will be

drain voltage -60V, source voltage -158.5V approx,
gate voltage -160V, Ids = 6.6mA. ( this means a voltage of between 1V
and 2V exists between gate and source
with gate negative in respect to source, like a tube.

I searched a few mosfets from Farnell but all required a gate voltage
which goes positive with respect to the
source. I tried BS107A and found it didn't work as the gate voltage
cannot be set above the source voltage.

I do not want to alter the double sided board to accommodate some other
device such as a common bjt
for the CCS.

Anybody got any answers?

Not me, but thanks for the circuit.

It follows your practice of putting a low value resistor between signal ground
and earth. Why is it in one channel only?

Are there other examples of using 2 LTP in parallel like that? Why use two
independent CCS rather than one and a mirror?

Presumably you could use the opposite type of mosfet if you add another
transistor to make a "ring of two"? A bit late, I know...and the original
circuit is so elegant.

Why did the originals fail? I'm still struggling with transistors, and in
particular with judgements about reliability in valve circuits. How can I ensure
that an SS CCS fails to a short rather than an open circuit, or vice-versa? Is
there any inherent difference in the reliability of bipolars and mosfets?

Some web reference, along the lines of "how to make reliable SS circuits" would
be handy.

cheers, Ian

Ian Iveson wrote:

Why did the originals fail?

Very likely due to over dissipation / over temperature due to a poor selection of
device ( TO-92 package ) and placement ( in a hot place ).

I'm still struggling with transistors, and in
particular with judgements about reliability in valve circuits. How can I ensure
that an SS CCS fails to a short rather than an open circuit, or vice-versa?

You can't.

Is there any inherent difference in the reliability of bipolars and mosfets?

No.

Some web reference, along the lines of "how to make reliable SS circuits" would
be handy.

Don't overvolt, overcurrent, overtemp.

Graham

Ian Iveson wrote:

Patrick Turner wrote

I have an ARC VT100 amplifier needing repairs.
There are two TO92 pacakge mosfets used as constant current sources for
a pair ot long tail pair input differential amplifiers.
These are set up so there is 6.6 mA from a pair of 6922 cathodes flowing
down to -160V
via 16k and the drain to source. A pot varies the source resistor to the
-160V. The gate is taken directly to
the -160V.

Both of these have failed to a short circuit.

The schematic can be inspected at
http://www.freeinfosociety.com/elect...ew.php?id=2332

ARC do not tell us what sort of mosfet or type number the mosfet are and
there are no numbers on the
mosfet, and only ARC's secret colour code so I am forced to deal with
ARC, which isn't what i want to do
They want usd $80 per hour for their labour rate, which is 20dB more
than I get.

Any idea on a suitable generic replacement for the mosfets?
The op conditions will be

drain voltage -60V, source voltage -158.5V approx,
gate voltage -160V, Ids = 6.6mA. ( this means a voltage of between 1V
and 2V exists between gate and source
with gate negative in respect to source, like a tube.

I searched a few mosfets from Farnell but all required a gate voltage
which goes positive with respect to the
source. I tried BS107A and found it didn't work as the gate voltage
cannot be set above the source voltage.

I do not want to alter the double sided board to accommodate some other
device such as a common bjt
for the CCS.

Anybody got any answers?

Not me, but thanks for the circuit.

It follows your practice of putting a low value resistor between signal ground
and earth. Why is it in one channel only?

Both channels are in the one case so there is only one connection of the single 0V
rail to the chassis.



Are there other examples of using 2 LTP in parallel like that? Why use two
independent CCS rather than one and a mirror?

I am not aware of any other maker using ARC's method.
The set up allows balanced global NFB and balanced or non balanced inputs.
Only two anode outputs which will always be balanced and there are 4 inputs.



Presumably you could use the opposite type of mosfet if you add another
transistor to make a "ring of two"? A bit late, I know...and the original
circuit is so elegant.

No need to do it any other way for ARC.
Maybe a floating NFB coil in the OPT would allow direct injection of NFB voltage
cathode to cathode
of an LTP with two triodes and one CCS would allow balanced or unbalanced inputs and
reduce the complexity.



Why did the originals fail? I'm still struggling with transistors, and in
particular with judgements about reliability in valve circuits. How can I ensure
that an SS CCS fails to a short rather than an open circuit, or vice-versa? Is
there any inherent difference in the reliability of bipolars and mosfets?

Very often if not usually, an SS device fails to a short unless the current x
voltage power diss
heats the device and explodes it.

The CCS in this case had 14mm x 14mm blade type heatsinks and were 12mm away from
the input tubes so after
so many heat cycles they fatigue and just expire.




Some web reference, along the lines of "how to make reliable SS circuits" would
be handy.

Read Ben Duncan's 1996 book about SS amplifiers.
Its a must for a broard view about SS audio challenges.
He is another great British audio writer.

Patrick Turner.



cheers, Ian

Pooh Bear wrote:

Ian Iveson wrote:

Why did the originals fail?

Very likely due to over dissipation / over temperature due to a poor selection of
device ( TO-92 package ) and placement ( in a hot place ).

I'm still struggling with transistors, and in
particular with judgements about reliability in valve circuits. How can I ensure
that an SS CCS fails to a short rather than an open circuit, or vice-versa?

You can't.

Is there any inherent difference in the reliability of bipolars and mosfets?

No.

Some web reference, along the lines of "how to make reliable SS circuits" would
be handy.

Don't overvolt, overcurrent, overtemp.

That's right for all electronics. Its just that tube will sweat it out for about long
enough so you can find what may be the problem if there is one.
The SS has a problem and POOF!, its outa there; no chance to find out why while the
**** happens.
Often the last thing to blow is the fuse.
Direct coupling and double sided boards for SS makes it 40dB more difficult to catch
anything in the act
of going wrong. So one has to replace the parts until it works.
At least with simple tube circuits cap coupled each tube is seperate and than be
easily
prodded to know what's going on...

Fixin SS gear is like cleaning toilets to me. Sometimes I gotta do it though.

Patrick Turner.





Graham

"Patrick Turner" a écrit dans le message de news:
...


Pooh Bear wrote:

Ian Iveson wrote:

Why did the originals fail?

Very likely due to over dissipation / over temperature due to a poor
selection of
device ( TO-92 package ) and placement ( in a hot place ).

I'm still struggling with transistors, and in
particular with judgements about reliability in valve circuits. How can
I ensure
that an SS CCS fails to a short rather than an open circuit, or
vice-versa?

You can't.

Is there any inherent difference in the reliability of bipolars and
mosfets?

No.

Some web reference, along the lines of "how to make reliable SS
circuits" would
be handy.

Don't overvolt, overcurrent, overtemp.

That's right for all electronics. Its just that tube will sweat it out for
about long
enough so you can find what may be the problem if there is one.
The SS has a problem and POOF!, its outa there; no chance to find out why
while the
**** happens.
Often the last thing to blow is the fuse.
Direct coupling and double sided boards for SS makes it 40dB more
difficult to catch
anything in the act
of going wrong. So one has to replace the parts until it works.

Bah !
Since you are fixing a "presumably well enginered" unit, just replace ALL
that sand parts for new ones.
More than ever the only unreplaced part will destroy the new ones you just
intalled .:(
I did that in a QUAD 405, after what I discovered that just a 0$ 1N4002 was
open, but I've trashed ALL removed parts.
Customer does not need to know and you sleep better.
At end, you save time and money.

Once it runs, then you may try to improve !

Yves.

At least with simple tube circuits cap coupled each tube is seperate and
than be
easily
prodded to know what's going on...

Fixin SS gear is like cleaning toilets to me. Sometimes I gotta do it
though.

Patrick Turner.





Graham

Patrick Turner wrote

It follows your practice of putting a low value resistor between signal
ground
and earth. Why is it in one channel only?

Both channels are in the one case so there is only one connection of the
single 0V
rail to the chassis.

OK. I missed the link to the other channel's 0V.

Are there other examples of using 2 LTP in parallel like that? Why use two
independent CCS rather than one and a mirror?

I am not aware of any other maker using ARC's method.
The set up allows balanced global NFB and balanced or non balanced inputs.
Only two anode outputs which will always be balanced and there are 4 inputs.

Presumably you could use the opposite type of mosfet if you add another
transistor to make a "ring of two"? A bit late, I know...and the original
circuit is so elegant.

No need to do it any other way for ARC.
Maybe a floating NFB coil in the OPT would allow direct injection of NFB
voltage
cathode to cathode
of an LTP with two triodes and one CCS would allow balanced or unbalanced
inputs and
reduce the complexity.

Yes, Morgan Jones does something similar, but without the complication of
combining the global feedback with the output stage CFB. One of the interesting
aspects of the ARC method is that the global feedback is not directly compared
with the input signal. But the feedback will be distorted in a similar but
opposite way to the input, so they will partly cancel each other. It would be
interesting to see what the remaining distortion looks like.

Speaking of Morgan Jones, his control circuit uses a bipolar as a series
regulator for the 5V supply. A momentary short of the relay coil connections
blows out every device and electro cap in the circuit in a flash.

Why did the originals fail? I'm still struggling with transistors, and in
particular with judgements about reliability in valve circuits. How can I
ensure
that an SS CCS fails to a short rather than an open circuit, or vice-versa?
Is
there any inherent difference in the reliability of bipolars and mosfets?

Very often if not usually, an SS device fails to a short unless the current x
voltage power diss
heats the device and explodes it.

That is useful to know. I have never blown a mosfet, but have wrecked a few
bipolars, always to a c-e short. Never checked whether the b-e and b-c were also
shorted. That becomes significant when they are connected together, I am
discovering.

I assume therefore that if I want a voltage source to fail to the supply rail
rather than to ground, I should use series regulation? I am trying to construct
a relatively fail-safe grid bias supply controlled by a 5V DAC. I wonder also
about the probability of failure cascading through the control circuit and
taking out the DAC.

The CCS in this case had 14mm x 14mm blade type heatsinks and were 12mm away
from
the input tubes so after
so many heat cycles they fatigue and just expire.

Odd they have both failed at the same time, considering they are electrically
independent.

Some web reference, along the lines of "how to make reliable SS circuits"
would
be handy.

Read Ben Duncan's 1996 book about SS amplifiers.
Its a must for a broard view about SS audio challenges.
He is another great British audio writer.

Thanks, I'll see if I can find a copy.

cheers, Ian

Ian Iveson wrote:

Patrick Turner wrote

It follows your practice of putting a low value resistor between signal
ground
and earth. Why is it in one channel only?

Both channels are in the one case so there is only one connection of the
single 0V
rail to the chassis.

OK. I missed the link to the other channel's 0V.

Are there other examples of using 2 LTP in parallel like that? Why use two
independent CCS rather than one and a mirror?

I am not aware of any other maker using ARC's method.
The set up allows balanced global NFB and balanced or non balanced inputs.
Only two anode outputs which will always be balanced and there are 4 inputs.

Presumably you could use the opposite type of mosfet if you add another
transistor to make a "ring of two"? A bit late, I know...and the original
circuit is so elegant.

No need to do it any other way for ARC.
Maybe a floating NFB coil in the OPT would allow direct injection of NFB
voltage
cathode to cathode
of an LTP with two triodes and one CCS would allow balanced or unbalanced
inputs and
reduce the complexity.

Yes, Morgan Jones does something similar, but without the complication of
combining the global feedback with the output stage CFB.

Morgan Jones I believe had the OPT secondary floating but with low enough valued
resistance dividers to
get the global NFB voltage back to the two cathodes of an input LTP.
This works fine. Its simple and effective except for the fact that if there is ever
a leak of tyhe B+
of the primary to secondary, then +500V could appear at the input tube cathodes and
if someone
touched a speaker lead and an earth point they would get a terrible shock.
Thanks Morgan, but your nice idea could ruin any maker foolish enough to try it.

A small turn fine wire tertiary feedback winding wound bifilar over the top of a
grounded secondary producing only about 1Vrms at full power at very low impedance
can be used as the floating NFB winding to get NFB to the input stage.
The Vgg to the input stage when open loop is abpout 0.1Vrms, so the 1V NFB would
give 20dB of NFB,
without any losses of open loop gain in cathode resistances due to local current FB.

A couple of 470 ohm R across such a winding slung between cathodes of the input LTP
and with a CCS taken
from the R CT to a -ve supply ( depletion mosfet is ideal ) will make a safer and
very nice way to apply NFB
in an input LTP which can have balanced or unbalanced inputs.

The idea is a lot simpler than ARC's method and uses 1 twin triode less in the input
amp.

The ARC and CJ amps seem to have a theme of consciousness which is along the lines
of

""Look at me! Ain't I really smart to use so many tubes to get the job done"
In fact far fewer devices and tubes are required.



One of the interesting
aspects of the ARC method is that the global feedback is not directly compared
with the input signal.

But it is compared. Like in any conventionally unbalanced or balanced amp.
Input id say 1V. FB is say 0.9V. 0.1V is applied differentially and that is what
gets amplied with any
distortion voltage to oppose itself.


But the feedback will be distorted in a similar but
opposite way to the input, so they will partly cancel each other. It would be
interesting to see what the remaining distortion looks like.

The ARC distortion is no better worse or different to any other well made tube amp
using 4 x 6550 in UL PP and with the amount of NFB applied.
The loading on the tubes is rather low, 1.6k a-a for 8 ohms connected to the 8 ohm
outlet
I strongly advise anyone with a VT100 to always ignore the 8 ohm outlets and only
ever connect speakers to the 4 ohm outlet.
This will mean output power will be slightly less that 100 watts but the class A
power will be greater.
ARC like everyone high end tries to satisfy the high end brigade who MUST have 100
watts.
So they give it to them but the amp is working very hard to make that 100 watts.

Most only ever use 2 watts average power but nevertheless using only the 4 ohm
outlet
will give 1/2 the Rout and half the thd of the 8 ohm outlet, so who cares about the
slight
loss of max power? 4 x 6550/KT88/KT90 is usually plenty for anyone
even when allowed to work in class A and not forced to work in AB after a very low
threshold
of class A power.
The use of the 4 ohm connection also allows the biasing of output tubes to be
reduced
And what many people don't know is that were you to adjust the bias so 1/2 the bias
current flowed
so halving the heat dissipation of the output tubes is that they would not hear any
sound change whatsoever.
People say the tubes must run with 37 watts of Pda but expect short tube life and no
extra music.
I don't know how hard the VT100 tubes run because I have not measured them but later
models of the amp
featured a fan in them .

**A Fan!!!??** For crying out loud WHY? A fan in a hi-fi amp is an admission of
failure in my eyes.
Its like the maker is saying " Well folks we didn't get the design right and to
ameliorate high temperatures likely
to cause parts failures we have decided to give it a blow job."

Fans belong only in PA amps.



Speaking of Morgan Jones, his control circuit uses a bipolar as a series
regulator for the 5V supply. A momentary short of the relay coil connections
blows out every device and electro cap in the circuit in a flash.

I have not inspected the schematic of that.



Why did the originals fail? I'm still struggling with transistors, and in
particular with judgements about reliability in valve circuits. How can I
ensure
that an SS CCS fails to a short rather than an open circuit, or vice-versa?
Is
there any inherent difference in the reliability of bipolars and mosfets?

Very often if not usually, an SS device fails to a short unless the current x
voltage power diss
heats the device and explodes it.

That is useful to know. I have never blown a mosfet, but have wrecked a few
bipolars, always to a c-e short. Never checked whether the b-e and b-c were also
shorted. That becomes significant when they are connected together, I am
discovering.

I have serviced a few amp with bjts which heated up and internally vaporized the
plastic
which led to a minor explosion. Bits all around the case. Usually the result of one
channel being
expected to power into shorted speaker leads at low levels
when the NFB forces the amp to vainly produce output voltage across the short, but
not enough current
to blow fuses or trigger protect circuits. Pd = average I x Rail Votlage
so 5A x 50V = 250W; device go BLAT after 30 seconds or less.



I assume therefore that if I want a voltage source to fail to the supply rail
rather than to ground, I should use series regulation? I am trying to construct
a relatively fail-safe grid bias supply controlled by a 5V DAC. I wonder also
about the probability of failure cascading through the control circuit and
taking out the DAC.

So far, I don't know of anyone who has built a protect circuit for SS or for any
other amp
which senses the load value so that whenever a load of say 2ohms or less appears for

longer than 0.5 seconds, the amp is turned off.

Mainstream SS amp makers continue their harrassment of the general public by
continuing to instal
speaker connectors 12mm apart at the rear of amps instead of 50mm so that SS amps
WILL fail
at some time and the owner is forced into a fix or better still for the makers, a
new amp.

It would be so easy to make a board costing 50c to provide load sensing but again no
makers I know of have this feature and its because the makers really
WANT YOUR AMP TO **** ITELF A FEW TIMES SO YOU'LL BUY A NEW ONE.

For the solid state amp makers reading here, I make no apology.



The CCS in this case had 14mm x 14mm blade type heatsinks and were 12mm away
from
the input tubes so after
so many heat cycles they fatigue and just expire.

Odd they have both failed at the same time, considering they are electrically
independent.

Maybe one get slightly more loding after one blows, and they were on the brink
anyway.
But even with both shorted there was 12 watts of relatively undistorted output!
Many owners would have missed there was a problem other than one channel was "down a
bit".
Maybe one blew up and the other eventually blew later; who knows.

I have often used MJE340 and MJE350 as CCS, and these I feel are very rugged,
and where there is about 30mA the device is bolted to a cool part of the chassis.
They can be purchased anywhere.



Some web reference, along the lines of "how to make reliable SS circuits"
would
be handy.

Read Ben Duncan's 1996 book about SS amplifiers.
Its a must for a broard view about SS audio challenges.
He is another great British audio writer.

Thanks, I'll see if I can find a copy.

Most Tech libraries could have a copy; certainly in the uk anyway.
An electronics enginneer who lives next door to me lent me his copy back in '94
when I was learning and when i built a couple of SS amps. It was a terrific read for
me at that time.

Patrick Turner



cheers, Ian