Well, at 72vac power with no load or input there is 88mvdc at the speaker out on the bad channel and 3mvdc on the good channel. I don't want to turn it up to full power while it's still oscillating. But that does *seem* like it might be better...?
No use to test it with full voltage, to the contrary, stay in safe area.
Check the voltage through the diodes D5 and D6...
It must be about 650mV across each diodes.
It s extremely rare that these diodes fail, but according to
your measurements, the voltage they are feeding to Q4 across R15
is highly suspect.
If one diode is in short circuit, that would explain both the high offset
and the oscillations that occur..
Hope it will be the good point...
Check the voltage through the diodes D5 and D6...
It must be about 650mV across each diodes.
It s extremely rare that these diodes fail, but according to
your measurements, the voltage they are feeding to Q4 across R15
is highly suspect.
If one diode is in short circuit, that would explain both the high offset
and the oscillations that occur..
Hope it will be the good point...
Last edited:
Where is the 1K pot on Q12 set?
The VAS transistor is what pulls the output positive as driven by the first input transistor.
It will clip, if the first transistor is bad - very possible if this transistor caused the VAS to fail,
or if the protection transistor turns on too early - might have been a selected part, or
if the output stage is not providing enough current gain or is damaged in some way.
The Q12 pot set too low reduces the current gain.
It is pulled the other way by the current source, this too could cause the clipping problem
if the transistor does not meet the voltage breakdown spec.
The VAS transistor is what pulls the output positive as driven by the first input transistor.
It will clip, if the first transistor is bad - very possible if this transistor caused the VAS to fail,
or if the protection transistor turns on too early - might have been a selected part, or
if the output stage is not providing enough current gain or is damaged in some way.
The Q12 pot set too low reduces the current gain.
It is pulled the other way by the current source, this too could cause the clipping problem
if the transistor does not meet the voltage breakdown spec.
Last edited:
Many thanks for the response.
Should I check Q1 and Q2 out of circuit?
The pot (P1) is set to where it was when the amp was functional, roughly mid-range. I have cleaned it and rotated it several times. I can see it increase and decrease voltage on the scope, so I think it's good. I haven't pulled it to test it, though.
How would I safely check the output stage to see if it's providing enough current? Would a bad Q11 or Q12 cause the oscillation?
I'm suggesting this assuming that nothing jumps out on measuring all the EBC voltages
on the transistors. You might want to report those here first and let us check them. You
can do the outputs as a group as long as you have tested the .82 ohm resistors.
I would pull Q1 and Q2 and test them.
With 1A output stage fuses:
You can do some testing while you have them (Q1 and Q2) out there will be no base
drive to the VAS so it will turn off, the current source Q4 will supply about 4 mA and should
pull the output stage with no load to the negative rail. Turn it on to about 50V on the rails
and check if the output is within a few volts of the negative rail. If so, I would raise it to
the full voltage, check again, all no load of course. You can leave it there, make sure nothing
get hot, you could check the drop across each of the .82 ohm emitter resistors on the outputs.
There should be very little voltage on them. Report those voltages here.
We are doing this to be sure that all the parts
still meet their breakdown voltage and have reasonable gain without leakage.
If that passes shut it down, bleed down the supply caps. And tack in a 1K ohm resistor
from the VAS collector to the positive supply. There will be about 4 mA through this so
it will pull the output to Vpos - 4V. Bring the supply up to about half check that the amp
output is about Vpos - 5V see if anything gets warm, go to full voltage and check again.
Again no load, watch for smoke and blown 1A fuses. Measure all the .82 ohm resistor
voltages and report here.
Note that since we left the VAS and current source transistors in place they are being tested
for breakdown voltage, they will leak if they do not meet spec. We are simply manually
swinging the output stage from full neg to pos output. If everything checks out, and you
are feeling daring, put the normal fuses back, attach an 8 ohm dummy load and do it again
to see if the output stage has reasonable current gain. I suggest that you do the 50% voltage
test then check in here with the voltages across the output stage emitter resistors. This
will tell us if all the outputs have good beta and are current sharing on the load pulling side
and if any are leaking on the turned off side.
A good tech could do this dynamically with an AC signal and the scope set up for differential
voltage measurement on the emitter resistors, but this is much easier. Note that your dummy
load will get HOT so turn it on for 30 sec or so, and take a few measurements then let it cool.
The other possibility is to just shot gun the input section as we say. Change all the semis,
it is the output stage that is expensive. If you do good clean work and check all semis
before installing them it should work - assuming the output stage is fine. I would do the
drivers to the output stage also if they are available. I would check as many resistors as you
can while you have various semis out if you choose to do this. If the output stage checks out
this makes a lot of sense actually.
on the transistors. You might want to report those here first and let us check them. You
can do the outputs as a group as long as you have tested the .82 ohm resistors.
I would pull Q1 and Q2 and test them.
With 1A output stage fuses:
You can do some testing while you have them (Q1 and Q2) out there will be no base
drive to the VAS so it will turn off, the current source Q4 will supply about 4 mA and should
pull the output stage with no load to the negative rail. Turn it on to about 50V on the rails
and check if the output is within a few volts of the negative rail. If so, I would raise it to
the full voltage, check again, all no load of course. You can leave it there, make sure nothing
get hot, you could check the drop across each of the .82 ohm emitter resistors on the outputs.
There should be very little voltage on them. Report those voltages here.
We are doing this to be sure that all the parts
still meet their breakdown voltage and have reasonable gain without leakage.
If that passes shut it down, bleed down the supply caps. And tack in a 1K ohm resistor
from the VAS collector to the positive supply. There will be about 4 mA through this so
it will pull the output to Vpos - 4V. Bring the supply up to about half check that the amp
output is about Vpos - 5V see if anything gets warm, go to full voltage and check again.
Again no load, watch for smoke and blown 1A fuses. Measure all the .82 ohm resistor
voltages and report here.
Note that since we left the VAS and current source transistors in place they are being tested
for breakdown voltage, they will leak if they do not meet spec. We are simply manually
swinging the output stage from full neg to pos output. If everything checks out, and you
are feeling daring, put the normal fuses back, attach an 8 ohm dummy load and do it again
to see if the output stage has reasonable current gain. I suggest that you do the 50% voltage
test then check in here with the voltages across the output stage emitter resistors. This
will tell us if all the outputs have good beta and are current sharing on the load pulling side
and if any are leaking on the turned off side.
A good tech could do this dynamically with an AC signal and the scope set up for differential
voltage measurement on the emitter resistors, but this is much easier. Note that your dummy
load will get HOT so turn it on for 30 sec or so, and take a few measurements then let it cool.
The other possibility is to just shot gun the input section as we say. Change all the semis,
it is the output stage that is expensive. If you do good clean work and check all semis
before installing them it should work - assuming the output stage is fine. I would do the
drivers to the output stage also if they are available. I would check as many resistors as you
can while you have various semis out if you choose to do this. If the output stage checks out
this makes a lot of sense actually.
Last edited:
Tip > you can get the amp working with 1 set of TO-03 outputs, once working put the other 3 sets in, saves $$
The TO-220 drivers need to be upgraded from stock part no's.
The TO-220 drivers need to be upgraded from stock part no's.
1 set of outputs is fine but don't test with an 8 ohm load use something more like 50 to 100 ohms.
And it will alter the stability.
Also note that you can probe for oscillations with the scope if you do the manual test
that I mentioned. The global feedback loop is open now so if there are any they are
local to the output stage - very unlikely.
And it will alter the stability.
Also note that you can probe for oscillations with the scope if you do the manual test
that I mentioned. The global feedback loop is open now so if there are any they are
local to the output stage - very unlikely.
Last edited:
I'll just pull these and replace them since I have replacements. I'll test the old ones to see if they're good.
They're only 150V breakdown, you need 2X the supply voltage and they
are 8A the originals were 1.5A 3A peak. I expect them to be much slower.
I prefer the 15034/35, they are faster and all around better.
I have to wonder if there are any capacitor changes for stability with the factory upgrade.
Edit: Just noticed that the schematic shows the 2SB595 as the alternate and this
is a 5A/40W part. The data sheet that I found for this lists it as a 100V part which does
not seem right at all.
The 15034/35 is closer to this part and is what I would use.
If you look at the SOA of the 2SA1011 at 70V it is 70 mA.
The 15034/35 will do 400 mA at 70V so they are a significant upgrade.
are 8A the originals were 1.5A 3A peak. I expect them to be much slower.
I prefer the 15034/35, they are faster and all around better.
I have to wonder if there are any capacitor changes for stability with the factory upgrade.
Edit: Just noticed that the schematic shows the 2SB595 as the alternate and this
is a 5A/40W part. The data sheet that I found for this lists it as a 100V part which does
not seem right at all.
The 15034/35 is closer to this part and is what I would use.
If you look at the SOA of the 2SA1011 at 70V it is 70 mA.
The 15034/35 will do 400 mA at 70V so they are a significant upgrade.
Last edited:
Yes, looks like Ft is the same, better beta on the others and voltage specs with lots of margin. The die size looks to be the same based on the thermal resistance.
Yes, lower voltage devices display a mini beta of 40 at 2A, compared to
minimum beta of 10 at same current for the high voltage siblings..
I suppose that it s not easy to have both of best worlds...
minimum beta of 10 at same current for the high voltage siblings..
I suppose that it s not easy to have both of best worlds...
Hi Fred,
Too much advice from all quarters again. Infinia and Pete (PB2) are giving you the most accurate advice so far. Time to slow down and think about things a little.
What Infinia said in post #213 is correct on all counts. And yes, those 2SD525 and 2SB595 drivers were hand selected - believe it or not! These days there are a better selection of driver transistors that do not require much in the way of hand selection with a curve tracer. MJE15032 and MJE15033 are what I use in these amps. I would worry about using the MJE15034 and MJE15035 as drivers. Into low resistance loads and high power, they may fail due to excessive current. This has been a problem Adcom has had in the past with different driver transistors.
Remember to test the resistors in areas where you suspect damage. Resistors often change value or open up without leaving much evidence. Often, the resistor may exhibit a burn mark on the surface closest to the PCB - under the part. So never assume a resistor (or any other part) is fine as the result of a visual check. I've had open fuses with an intact element. That cost me time.
What do we know right now about what is going on with this amplifier?
We know that the positive drive is limited, and that it breaks the feedback loop at this point. Your DC offset is the average voltage caused by the missing positive portions of the sine wave. This is expected with a clipped wave form.
Have you tested Q9 and Q10, plus associated parts? You can pull these transistors and run the amp during testing safely. Don't forget to put them back in when you have the amp running normally. R17 may have gone up in value, and all the emitter resistors may have opened on the positive bank. A base-emitter short or leakage in any one of Q17 ~ Q20 can cause similar behavior.
Relax, write out what you are going to test, then mark the results of each test as you do them. Those will be your working facts, or things we know for sure to be correct.
Hi wahab,
We already know these diodes are working correctly. The diff amp is operating. Failure of this part of the amplifier will shut it down, look at where the thermal cutout is located. Also, the existing compensation is not a concern here at all. Never mess with compensation components in a successful commercial amplifier unless you have a very good reason to do so. This situation hardly qualifies.
There are times when output transistors are changed to a different type / part number and oscillation then occurs. Since the high frequency may have changed, the compensation may also need to be adjusted. This is also true when replacing an STK power IC with a newer part numbered the same.
Worrying about absolute beta as stated on the data sheet is counter productive unless you are currently looking for a replacement. Keep in mind that figures quoted in a data sheet are very rough and taken under conditions that may be different in this amplifier. Yes, details can matter - but don't sweat them to this degree at this point in a repair.
Hi infinia,
Generally speaking, testing an amplifier during service should use either all the output transistors, or none of them. Pete is right when he says the compensation will not be correct any more. This may possibly cause oscillation to occur. This depends on the amplifier of course. Running fewer outputs and causing oscillation will only cause lost time a you try to figure that out, and it doesn't happen every time as you know.
Happy hunting! -Chris
Too much advice from all quarters again. Infinia and Pete (PB2) are giving you the most accurate advice so far. Time to slow down and think about things a little.
What Infinia said in post #213 is correct on all counts. And yes, those 2SD525 and 2SB595 drivers were hand selected - believe it or not! These days there are a better selection of driver transistors that do not require much in the way of hand selection with a curve tracer. MJE15032 and MJE15033 are what I use in these amps. I would worry about using the MJE15034 and MJE15035 as drivers. Into low resistance loads and high power, they may fail due to excessive current. This has been a problem Adcom has had in the past with different driver transistors.
Remember to test the resistors in areas where you suspect damage. Resistors often change value or open up without leaving much evidence. Often, the resistor may exhibit a burn mark on the surface closest to the PCB - under the part. So never assume a resistor (or any other part) is fine as the result of a visual check. I've had open fuses with an intact element. That cost me time.
What do we know right now about what is going on with this amplifier?
We know that the positive drive is limited, and that it breaks the feedback loop at this point. Your DC offset is the average voltage caused by the missing positive portions of the sine wave. This is expected with a clipped wave form.
Have you tested Q9 and Q10, plus associated parts? You can pull these transistors and run the amp during testing safely. Don't forget to put them back in when you have the amp running normally. R17 may have gone up in value, and all the emitter resistors may have opened on the positive bank. A base-emitter short or leakage in any one of Q17 ~ Q20 can cause similar behavior.
Relax, write out what you are going to test, then mark the results of each test as you do them. Those will be your working facts, or things we know for sure to be correct.
Hi wahab,
The precise voltage drop across these diodes is dependent on ambient temperature. They roughly track the Vbe changes in Q3 and Q4. The actual voltage drop may be a bit different from what you are citing as the voltage to expect.
We already know these diodes are working correctly. The diff amp is operating. Failure of this part of the amplifier will shut it down, look at where the thermal cutout is located. Also, the existing compensation is not a concern here at all. Never mess with compensation components in a successful commercial amplifier unless you have a very good reason to do so. This situation hardly qualifies.
There are times when output transistors are changed to a different type / part number and oscillation then occurs. Since the high frequency may have changed, the compensation may also need to be adjusted. This is also true when replacing an STK power IC with a newer part numbered the same.
Worrying about absolute beta as stated on the data sheet is counter productive unless you are currently looking for a replacement. Keep in mind that figures quoted in a data sheet are very rough and taken under conditions that may be different in this amplifier. Yes, details can matter - but don't sweat them to this degree at this point in a repair.
Hi infinia,
Generally speaking, testing an amplifier during service should use either all the output transistors, or none of them. Pete is right when he says the compensation will not be correct any more. This may possibly cause oscillation to occur. This depends on the amplifier of course. Running fewer outputs and causing oscillation will only cause lost time a you try to figure that out, and it doesn't happen every time as you know.
Happy hunting! -Chris
My spec sheet shows 20 at 2A, not 10 as you claim, and that is for a worst case temp.
You have chosen an operating point that is beyond the 1.5 A spec of the original parts. Even these better parts will only support 400 mA at 70V so I believe a better point to pick is 1A or less and there the parts that I suggest are better with beta closer to 100. The original drivers showed high IC beta droop at .7A, the 15034/35 at 1A and the 15030 type at 2A.
The OP can choose whatever he wants, just my suggestion.
Back when this amp was popular I remember our local hi end shop claiming that it would drive 1 ohm loads with a certain mod. I heard later that it was just an increase in the rail fuses but does anyone know if this was when the driver upgrade was suggested?
I'd reckon if it oscillates with just one set of outputs then the phase margin is mostably marginal. I was just speaking of getting major faults fixed here. Most operational testing like this will be done not loaded or very lightly. With all the part # replacements to the outputs and drivers Ft's here, the resultant phase margin will be changed from the original design anyhow. So changing the compensation may/should be within the realm of a complete fix after all damaged and left off parts are completely repaired. Lastly full load and square wave testing should be done to check stability.
Chris it wouldn't surprise me that there are a couple of odd ball amps that would challenge this technique, but very few IMO. All the burned output device savings would far outweigh that gamble in my thinking. Now if the topology was much different than this, say like a transconductance output I wouldn't advise it.