Amplifier calibration after transistor replacement

[tohtorizorro]

My active subwoofer has 2 BJTs with broken pins due to a physical impact. I have purchased new ones, but I was wondering if the circuit needs to be calibrated after installing them. The PCB has a multi-turn trimpot that is locked in place with paint which I'm guessing might be used to compensate the characteristic deviations of transistors. However I have no idea what the adjustment should be based on. Fortunately the subwoofer has two additional identical fully functioning amplifiers which could be used as a reference.

The transistors are a complementary pair of TO-220 packaged power transistors. I've tried to include all relevant info in the attached schematic PDF of the amp. I had to draw it myself so it might not be 100% accurate but hopefully close enough. There are also two photos of the PCB, note that it houses two amplifiers (there are three in total). The replaced transistors are the ones on the bottom side, they are normally attached to a massive heatsink and without it seemed to get quite hot almost instantly when the speaker was turned on even with no signal present.



After looking at the schematic and actually thinking about stuff for a moment, I realized that the node most immediately affected by the trimpot would be the one connecting R97, R101, R102, C57 and the dc blocking input cap. So, if the voltage at that point would match the other circuits, everything should be pretty much in order, right? After this revelation the replaced transistors (T6 & T9) seem quite distant to the potentiometer, which makes me think there might be no reason to touch the trimpot after all. But, I really don't understand how the circuit works and would appreciate any comments.

Thanks!

I added a PNG for the schematic since it at least shows up with a thumbnail in the attachments list. The PDF is slightly better quality though. If someone can help me with how to attach images into the message body, please PM me. This:


 SoundProjects_SP4-15_amp_schema.pdf (78.76 kB - downloaded 94 times.)

...does not seem to be working.

[Kim Christensen]

Personally, I'd adjust it if it's "out". I assume the single speaker is driven in bridge mode by the two amplifiers.
Disconnect speaker and let the amplifier warm up. Nothing should be getting hot at this point. If it is, then you have another problem and you'll need to fix that first.

Measure the output DC voltage with no audio input and, if it's more than +-50mV off, adjust the pot for zero VDC to ground. Do this for both amps.
Reconnect the speaker.

[tohtorizorro]

Thanks for the feedback, your suggestion sounds good and something I would never have come up by myself. Just to clarify, the subwoofer has three speaker elements which each seem to have a dedicated amp circuit. The device has four relays that turn on when it is powered up, one for each output channel and maybe one for input. First the one can be heard turning on and half a second later the remaining three all at the same time. My guess would be that the amps work independent of each other ie. the bridge mode thing sounds a bit strange to me but as said, I don't really understand the system too well. Sadly I only have the schematic for the amplifier circuit to offer. All in all, the system seems pretty straightforward (at least for someone who knows his way around these things unlike me) although it does have some sort of limiter protection housed on a separate PCB. I wont be able to get my hands on the device until tomorrow but will report back after I've done the measurements (I'm curious about the results in the channels that should be functional), also I clearly need to observe more closely how hot things get without input.

Based on the circuit diagram and the fact that the hFE values (measured with a cheap chinese component tester but still) of the old vs new transistors are NPN:67 -> 125 and PNP:94 -> 155, how scared should I be turning on the device on after I've replaced them? Am I worrying about a problem that doesn't exist? Thanks a lot for your output!

[andy3055]

Leave the trim pots alone! If you had physical damage, there is no reason to muck around with them. Replacing those transistors will not make major changes in the system. Of course, it is a good idea to check if there is DC at the speaker connections by keeping them disconnected. Unless you have a service manual that explains how to adjust those trim pots,  keep your hands off of them.

[Kleinstein]

The pots in the plan are for the DC level and thus not super critical.  The relaced transistors shoudl not effect the DC level much. There should be no special instractions needed for an adjustment. Zero DC output is the obvious target.

A chance of the transistors could change the standing current in the output stage a little. So the voltage measured across R149-R152 with no load and zero input.  It looks like it is more like set to a low value more in a class B than class AB range. A class AB amplifier should have a trimmer for the standing current.  So the voltage there should be small (e.g. < 5 mV). If higher one should change the Bias1 resistor aross D20 to a smaller value.

If still there an old original channel could give a clou on the suitable biasing current range.

[Kim Christensen]

Thanks for the feedback, your suggestion sounds good and something I would never have come up by myself. Just to clarify, the subwoofer has three speaker elements which each seem to have a dedicated amp circuit. The device has four relays that turn on when it is powered up, one for each output channel and maybe one for input. First the one can be heard turning on and half a second later the remaining three all at the same time. My guess would be that the amps work independent of each other ie. the bridge mode thing sounds a bit strange to me but as said, I don't really understand the system too well.

It won't be using bridge mode if there are 3 speakers and 3 amplifiers. I assumed wrong there.

Based on the circuit diagram and the fact that the hFE values (measured with a cheap chinese component tester but still) of the old vs new transistors are NPN:67 -> 125 and PNP:94 -> 155, how scared should I be turning on the device on after I've replaced them? Am I worrying about a problem that doesn't exist? Thanks a lot for your output!

If you're worried, disconnect the speaker associated with the amplifier that you repaired. If you're super paranoid, disconnect all 3 speakers before powering back on. But it sounds like that speaker/amp might have built in protection against a high DC offset at the output of the amp, as well as a simple delay timer to eliminate turn on pop. Or those relays are only on a delay timer. Hard to tell without seeing the circuit.

The replaced transistors are the ones on the bottom side, they are normally attached to a massive heatsink and without it seemed to get quite hot almost instantly when the speaker was turned on even with no signal present.

You have to remember, with 90V across them, it wouldn't take much current for those transistors to get hot without a heatsink. Plus, the hotter they'd get, the more current they'd draw since they are now not as thermally linked to the NTC. (I would have thought they would have attached the NTC to the heatsink better) Maybe they are relying on the thermal mass of the heatsink to give time for the system to reach thermal equilibrium.

[tohtorizorro]

First of all, you guys are awesome! I really appreciate the polite and informative replies.

I replaced the BJT's. Before I had the broken legged ones stitched up with solder and installed to take the original readings:

ch1 output: -8mV
ch2 output: 0V
ch2 output: 0V


I ended up replacing them for both amp circuits on the PCB (2 and 3) since both had fractured pins at this point. No big changes:

ch1 output: -8mV
ch2 output: +1mV   *this had BJTs replaced
ch3 output: -1mV    *this had BJTs replaced

So, I guess that seems still pretty ok?

The voltage over R149-R152 was 1-2mV for the ones that I could reach (they are in a difficult place once everything is hooked up).

Another source, supposedly experienced with the particular subwoofer (Let's call him the expert I probably shouldn't have made this forum post before waiting a few days for him to respond but I'm a restless soul and here we are), advised me to measure voltages on T14 & T15 collectors saying they should be around 0.8-1V and equal. Now, before the replacement, the situation was same on all the amp circuits: almost spot on -1V for T14 and +1V for T15. I thought he meant they should be roughly at the same voltage but maybe he meant that they are mirrored across 0V? Anyways, after the replacement ch2 now has T14 at -1.57 and T15 at 0.47V, so still the same ~2V apart but shifted down half a Volt. Soo, maybe it back to -1V/+1V?


Oh, and the thermal situation seemed to be under control, without heatsink the to-be-replaced transistors got quite warm but still easy to touch after 15min of power on. The new ones are attached to the heatsink now.

I'll try driving a speaker with a signal through each channel after determining how to deal with the changed collector voltages, i'd say thing's are looking quite good though.

   

[Kim Christensen]

I ended up replacing them for both amp circuits on the PCB (2 and 3) since both had fractured pins at this point. No big changes:
ch1 output: -8mV
ch2 output: +1mV   *this had BJTs replaced
ch3 output: -1mV    *this had BJTs replaced
So, I guess that seems still pretty ok?

That's perfect.

Another source, supposedly experienced with the particular subwoofer (Let's call him the expert I probably shouldn't have made this forum post before waiting a few days for him to respond but I'm a restless soul and here we are), advised me to measure voltages on T14 & T15 collectors saying they should be around 0.8-1V and equal. Now, before the replacement, the situation was same on all the amp circuits: almost spot on -1V for T14 and +1V for T15. I thought he meant they should be roughly at the same voltage but maybe he meant that they are mirrored across 0V? Anyways, after the replacement ch2 now has T14 at -1.57 and T15 at 0.47V, so still the same ~2V apart but shifted down half a Volt. Soo, maybe it back to -1V/+1V?

Yea, he definitely meant "mirrored" which would make sense. (-1V for T14 and +1V for T15)
The fact that ch2's output is sitting at +1mV is good. But the voltages on T14 & T15 shouldn't be that far off. There is probably another faulty component in ch2 still. Check the voltage drops across R75 and also R78. The drops should be more or less equal. I'm guessing around 0.4V or so.  Think you might have a bad T8 or T8A (Possible shorted base-emitter junction)

EDIT: With the power off, you could also measure the resistance of R75 and R78. A reading near zero ohms for R78 would point towards a bad T8 or T8A. I would expect them to measure around 100ohms in circuit. Maybe a bit lower but not much. You can compare with the other channels.

Also, I'm assuming you've replaced with the exact same transistors as the originals and gotten them from a reputable source?

 

[Kleinstein]

A low voltage at R149 - R152 is good and shows the expected low biasing current.

The voltages at the collectors of T14 and T15 with a zero input can vary a bit, depending one the sign of the offset.
The +-1 V are about what is expects, so they should be about mirrored and some 2-2.5 V appart.

A little asymmetry (e.g. some 200 mV, dependind on the diode types and transistors and temperature) is normal for a class B amplifier, as there is some kind of dead zone.
The sign of that asymmery can change with only a minor change (e.g. -0.1 to +0.1 mV) in the offset or loading. Possibly even just the DMM and hum from an open input  may be enough to change the voltage.
If the output offset and votlage over R149-152 is OK there is not need to worry about the voltage at the T14 T15 collectors.

If the offset is OK (< 50 mV should be OK) there is no need to adjust the trimmer. So if at all it would be channel 1  where one could consider adjusting the offst.

[magic]

No, this is a pathological result.

Somebody suggested that T8 or T9 is shorted - this could be it, if not for the very low output stage bias current. I suspect the opposite is the case - T7 or T6 is failed open circuit, or disconnected, or maybe some resistor is disconnected.


Please measure and report voltages at bases and emitters of T6, T7, T8, T9. They all should be symmetric around ground, with maybe ±0.4V at the emitters of T6/T9 and bases of T7/T8 and the familiar ±1V at the bases of T6/T9. Check R75 and R78 as suggested by others.

[tohtorizorro]

I guess this would be a good moment to cover the events leading to the situation in the first post, sorry if it’s a bit lengthy:

The subwoofer was brought to me because one of the channels had started behaving badly. When I tested it, the faulty channel seemed to work OK without a load, ie. without the speaker connected a signal fed to the input could be measured clean from the output. Without a signal present everything was ok even with the speaker connected, but as soon as I would feed the weakest sine wave to the input, the speaker cone would first slam to the inward extreme position, then outward as if outputting a square wave of around 2Hz or so. I‘m not sure if the speaker actually moved between the extreme positions but each transition was accompanied with such a loud snap that I worried it would damage the speaker. Oscilloscope showed there was more going on, see the attached screen capture.  I think the oscillation before the 'rising edge' was at least ~100kHz, well above the audio range in any case.

I removed the PCB in order to access the components which was a bit of a pain thanks to the BJTs being fixed in the heat sink. I needed to bend the legs of T6 and T9 and some of them came loose at this point. I inspected the circuit quite thoroughly but couldn't spot anything broken, so I painstakingly soldered the broken pins back to the transistors and installed them back on. The point was to compare measurements with the functioning circuits, but to my surprise the faulty channel had started working (or so I thought since the aforementioned loud snaps were missing).

Now, due to the way they are fixed on the PCB and heatsink, the pins of T6 and T9 might be the ones to receive most of the blow in case the subwoofer took a hit, which the owner suspected might've happened. So, my conclusion was that the earlier behavior was caused by fractures in the solder joints which got fixed during the resoldering. Now I only needed to replace the broken legged BJTs with fresh ones to finally get this piece of junk out of my life. I really wanted to believe and thus condensed the above into "broken pins due physical impact" which, I admit, was a bit of a stretch.   

[tohtorizorro]

I just realized I had made a crucial error right at the beginning while tracking the defective channel. All this time I was trying to spot a fault on the 3rd amp circuit and in fact it was the 2nd one having the problems all along (the numbering is my own but consistent through the posts). I am kind of relieved since I was going slightly mad after going through the whole circuit without finding anything. I’m clearly out of my league with this ordeal but might be able to sort it out with your kind help.

I'm assuming you've replaced with the exact same transistors as the originals and gotten them from a reputable source?

Yes, most of the components on the device seem to be well available and BJTs were ordered from a well-known supplier in Finland.


I tried adjusting the pot but it had minimal effect on T14/T15 collector voltages and it caused a dc voltage on the amp output, so I dialed it back so that the output was at 0V again when idle.

I fed a 65Hz sine wave to the input. Again, no loud snaps and the frequency can be recognized on the speaker but distorted. When the amplitude of output signal exceeds 1Vpk-pk, a strange high frequency “ripple” appears at the positive peaks. This produces a waveform with a weird blob that grows as the signal amplitude goes up. When zoomed in the ‘ripple’ seems to be sinusoidal and in the Mhz range. Screenshots attached.

I checked T7/A & T8/A with my component tester and it recognized them all correctly so I suppose they are ok. I’ll report back once I got the measurements y'all have suggested. Anything else I should check before I pull the PCB out for closer inspection (again)?

Thanks!

[Kim Christensen]

I checked T7/A & T8/A with my component tester and it recognized them all correctly so I suppose they are ok.

Is it one of those TC-1 testers or equivalent?  They can be handy for IDing parts, but you can't really trust them to test power transistors. Even testing Hfe of small signal transistors (Like a 2n3904) is not very accurate due to the small currents it puts through the device. And since the voltages it supplies is very low (3V) it won't really tell you if the transistor is failing at higher voltages either.
 
For a quick and dirty go-no-go power-transistor test I've used a small power supply with current limiting set to 2A and low voltage (5V). Then I'd connect the leads across the Emitter - Collector, with correct polarity of course, and feed in a low base current of 10mA (either via a resistor and another supply, or just a resistor (430 ohm) across collector-base) and see how much collector current flows. Some quick math (Ic/Ib) will tell you the beta at that current. Clamping the transistor to a temporary heatsink is a good idea unless you're really quick with your test (under a second or so of current flow).

[tohtorizorro]

Is it one of those TC-1 testers or equivalent?  They can be handy for IDing parts, but you can't really trust them to test power transistors. Even testing Hfe of small signal transistors (Like a 2n3904) is not very accurate due to the small currents it puts through the device. And since the voltages it supplies is very low (3V) it won't really tell you if the transistor is failing at higher voltages either.

Yep, that's exactly what I have. I don't trust the values other than for comparing components/measurements with each other, but up until now I have thought that it couldn't recognize a transistor correctly (type and pinout) unless the transistor was OK. However, it turns out there is something strange going on with T7/T8 transistors of the faulty channel:

The voltages across R149-152 are within 0-2mV on all channels so no surprises on T7/T8 emitters either (all near 0V). But, while the healthy channels have T7/T8 base voltages of ~+/-0.44V, on the faulty circuit all the bases are at at 0V too (or thereabouts).

a quick and dirty go-no-go power-transistor test I've used a small power supply with current limiting set to 2A and low voltage (5V). Then I'd connect the leads across the Emitter - Collector, with correct polarity of course, and feed in a low base current of 10mA (either via a resistor and another supply, or just a resistor (430 ohm) across collector-base) and see how much collector current flows. Some quick math (Ic/Ib) will tell you the beta at that current. Clamping the transistor to a temporary heatsink is a good idea unless you're really quick with your test (under a second or so of current flow).

Now that I took the PCB out once more to inspect the real faulty circuit, the T7/T8 transistors are disconnected and also fixed to the heatsink so it would be quite convenient to test them. I don't have a current limiting PSU at the moment but I'll try to come up with some sort of test setup.

Any ideas what might be causing the 0V on the bases?

Thanks to everyone contributing to the thread so far!
 

[magic]

T8 has 0V on its base because there is 0.47V on the base of T9 (presumably, go check it).

T7 has 0V on its base because, as I suspected, there must be lack of continuity somewhere on the T14, RX2, T6, T7 path. Maybe a broken trace, bad solder joint or a blown component.

Compare all voltages there with a good channel to locate the culprit.

[Kim Christensen]

Now that I took the PCB out once more to inspect the real faulty circuit, the T7/T8 transistors are disconnected and also fixed to the heatsink so it would be quite convenient to test them. I don't have a current limiting PSU at the moment but I'll try to come up with some sort of test setup.

Even a simple power supply and some power resistors (even small 12V bulbs), for the collector load so the max current is limited should suffice for a go-no-go test.
You could also check the B-E & B-C junctions on the transistors with a DMM on diode check. A reading around 0.6V -0.7V when forward and open in reverse is an indication that it's not completely broken. (Not an indication that all is good either). But a reading of a short or open is a fail. (I would have thought the component tester would have caught a short)

Any ideas what might be causing the 0V on the bases?

Is this only on the T8 & T8A bases or was it also zero on the T7 & T7A bases?

EDIT: Because if T7 is also zero I concur with magic's conclusion about the lack of continuity on the T14, RX2, T6, T7 path. (Probably RX2 is open)

[tohtorizorro]

T8 has 0V on its base because there is 0.47V on the base of T9 (presumably, go check it).

Yes! As I reported earlier, the voltage at T15 collector was at 0.47V which is probably too low to turn on T15. So yes, this makes sense.

T7 has 0V on its base because, as I suspected, there must be lack of continuity somewhere on the T14, RX2, T6, T7 path. Maybe a broken trace, bad solder joint or a blown component.

Compare all voltages there with a good channel to locate the culprit.

I have the PCB removed from the the system at the moment and the connections between T14 and T7 seem intact. And, the collector voltage of T14 was well below 1V so it should definitely be enough to create a current from T6 emitter to its base, so the missing voltage on T7 base is rather mysterious. I guess I need to hook the PCB up again and check the voltages around those parts. It sounds like some of my earlier measurements is off or something.

I wonder what caused the shift in T14/T15 collector voltages. I'm pretty sure they were +/-1V before I replaced T6 & T9 transistors. I'll try swapping them between ch2 (faulty) and ch3 (both these channels had them replaced).

Is this only on the T8 & T8A bases or was it also zero on the T7 & T7A bases?

EDIT: Because if T7 is also zero I concur with magic's conclusion about the lack of continuity on the T14, RX2, T6, T7 path. (Probably RX2 is open)

All of the four transistors' bases are at 0V.

The RX resistors are kind of interesting. The PCB looks as they were originally meant to be SMD parts. In fact on ch1 which is on a separate PCB they are SMD resistors. Ch2 and ch3 have through hole resistors, but (since there are no holes to go through) soldered in surface mount style on the topside trace. So, clearly less securely bonded than the other components and more vulnerable to the vibrations too. I'll retouch the joints before putting the PCB back in.

Thanks a lot guys!

[Kleinstein]

Having the BE junctions of the power transistors shorted and the rest still connected would destroy T6 and T9 from overload quite fast. So the zero voltage reading would more point to open connections at T6 or T9, the Rx resistors or just the connection to the base.
The string of 4 diodes and the 4 transitors junctions set the standing current / dead zone for the class B amplifier. If only 1 transistor is bad and act like a BE short there would be quite a lot of current flow (some 0.5 V of excess voltage with only some 0.1 ohm of resistance), possibly leading to thermal damage.

The scope traces show an onset of oscillation. So the amplifier seems to be just at the edge of stabilit - OK at 0 signal bit oscillating with a larger positive signal.
There is a slight chance that the higher gain of the replacement transistors could bring it even closer to the edge. In the plan the capacitor values for C52 and C54 look like a typo: 220 nF would be way to large. 220 pF would be more realistic, though maybe a bit on the small side.

There seem to be a 2nd problem with poor linearity and quite some distortion. The way the amplifier is build it is expected to get a more mediocre THD, though I would not expect it to be that bad. With the 68 ohm resistors R81 and R83 at the emitters of the VAS stage the loop gain for the feedback is small. The loading from the output transistors can thus be quite significant and the transistor gain is usually not constant (would help with linearity). A class B design also has a bit more distrtion to start with. I don't think that the higher gain of the relacement transistors would make the linearity worse.

I am still not sure the bad linearity is by design or a defect. There is a chance it could be by design:
The ear percieves the low tones as more loud when there are additional harminics and the ear and speaker can be more efficient with the higher frequency. In the low end audio range such kind of cheating would be relatively smart and not surprising. In part the relatively high THD makes up the tube sound.

[tohtorizorro]

Having the BE junctions of the power transistors shorted and the rest still connected would destroy T6 and T9 from overload quite fast. So the zero voltage reading would more point to open connections at T6 or T9, the Rx resistors or just the connection to the base.

I'm quite certain there are no shorts on the power transistors, the component tester couldn't have recognized them if there were.

I'm starting to suspect the RX resistors more and more. While checking continuity it seems like the solder on them would be extra oxidized as I really need shove the DMM probe in it to get the beeb. Maybe I've been closing a broken connection while doing so, as said I'll reflow the solder on them and see if anything changes. Funny if they would turn out to be the culprit since they were the first thing caught in my eye. In fact, one of them in fact looks like it's gone through some rough times. Here's a photo, its the RX1 from the ch3, the one I originally thought was faulty.

The scope traces show an onset of oscillation. So the amplifier seems to be just at the edge of stabilit - OK at 0 signal bit oscillating with a larger positive signal.
There is a slight chance that the higher gain of the replacement transistors could bring it even closer to the edge. In the plan the capacitor values for C52 and C54 look like a typo: 220 nF would be way to large. 220 pF would be more realistic, though maybe a bit on the small side.

You're right, some of the capacitors should have picoFarads instead of nanos. I noticed this mistake just now as I was inspecting the amp circuit.

There seem to be a 2nd problem with poor linearity and quite some distortion. The way the amplifier is build it is expected to get a more mediocre THD, though I would not expect it to be that bad. With the 68 ohm resistors R81 and R83 at the emitters of the VAS stage the loop gain for the feedback is small. The loading from the output transistors can thus be quite significant and the transistor gain is usually not constant (would help with linearity). A class B design also has a bit more distrtion to start with. I don't think that the higher gain of the relacement transistors would make the linearity worse.

The signal on the healthy channels is audibly much less distorted, I haven't looked at them on the scope but would suppose the sinewave could still be recognized as one on the output side on them.

[magic]

I wonder what caused the shift in T14/T15 collector voltages. I'm pretty sure they were +/-1V before I replaced T6 & T9 transistors.

Feedback caused it.

This circuit drives T14/T15 collector voltage to whatever it takes to obtain output = input. If the output stage works normally, this is ±1V. But something is wrong with the output stage, so internal voltages are automatically corrected to get zero on the final output regardless.

I'll try swapping them between ch2 (faulty) and ch3 (both these channels had them replaced).

I wouldn't bother. Just measure voltages at all pins of suspected devices and it will be clear where the continuity breaks.

[tohtorizorro]

Great news, the subwoofer is working! 🥳

I retouched the joints of RX1 & RX2 and the voltage balance at T14/T15 collectors was restored. So, RX2 was the culprit (RX1 might’ve also been causing problems). They seem to be added to the circuit after the PCB design was finished (I’ve labeled them with the X since they have no designators on the board). Ch1 (on a separate PCB) has surface mount resistors for RX1 & RX2.  If ch2 & ch3 had been done in the same way, this problem would have never occurred. Low freq vibrations probably fractured the joint(s) which were done somewhat poorly in the first place. It would be best to replace them with SMD resistors too but I think the added solder shall do for now. I checked the outputs and ch2 now matches the 1 & 3 so everything is in order.


Thank you for everyone contributing to this thread, I could never have spotted it by myself!


I envy your ability to decipher the workings of such circuit by just looking at a diagram drawn by some noob. Do you think it would be hard to explain the system in, say 15-30min youtube video to a beginner like me? That would be the kind of material I’ve been trying to find recently but found very little. Eg. Big Clive has done a lot of reverse engineering videos but they deal mostly with very simple gadgets like rechargeable flashlights. I’d like to find something just a bit more elaborate, the amp circuit here being a good example. Written articles with photos/pictures would be fine too.

Unless the schematic isn’t drawn in a way that makes it too laborious, maybe someone here could mark the different functional blocks in the schematic for me? After staring at it for so long, it would be nice to make some sense out of it.


Thanks!!!!

Mikko

[Kim Christensen]

I envy your ability to decipher the workings of such circuit by just looking at a diagram drawn by some noob. Do you think it would be hard to explain the system in, say 15-30min youtube video to a beginner like me?

Well, you did a good job of the schematic, measurements, etc which helped us help you. Not everyone does that, but it raises the chances of success by a lot.

There are some variations, but most of these transistor complementary push-pull audio amplifiers follow very a similar design to yours.
I've annotated your schematic with some labels which should help you Google the basic building blocks. Once you learn these, analyzing schematics such as yours will become 2nd nature. Though there'll always be some weirdness out there that'll make you scratch your head.

[tohtorizorro]

I've annotated your schematic with some labels which should help you Google the basic building blocks. Once you learn these, analyzing schematics such as yours will become 2nd nature.

That is perfect, thank you! I’m a little familiar with most of the stuff you’ve labeled there, and recognized some of the configurations but they just seemed to be connected so weirdly together that I had hard time understanding how the signal even gets from input to output and doubted the accuracy of my drawing. But, just now as I noticed the ‘polarity’ of the transistors in the differential pair and things started to make more sense.


The problem got solved and I’m super happy to get the subwoofer out of my hands, but the best part of such projects is the knowledge gained during them. So, I hope you don’t mind me asking a few questions more about the circuit:

1) I assume the connection from output to C56/R89 is providing some sort of (negative) feedback. The part left between the differential pair and PNP common emitter amp looks like a filter, is the purpose of the feedback to filter the signal?

The cascode amplifier is the most obscure part for me. After reading about it I understand that it is mostly used in place of regular common emitter configuration in order to suppress the Miller effect. However, the gain dependent base-collector capacitance should attenuate the higher frequencies but that seems more like an advantage for an amplifier that supposedly operates on a quite narrow band of low frequencies.

2) So, why is the cascode amplifier there in the first place?

3) The cascode amplifier output is inverted, so if it would be removed from the circuit, could the T15 base be driven with the same signal as T14 (ie. from the T16 collector)?

4) Are the common emitter amplifiers there to simply add more gain?

5) Would it be correct to say that the push-pull amplifier is constructed in a Darlingtonish manner?

6) Would the circuit still work, albeit with weaker output, if they were removed and the RX resistors connected straight to T7/T8 bases?

7) I understand diodes D21 & D39 are there for the class-AB bias but what about NTC1, BIAS1 and the extra diodes? Are they just ensuring that T6 & T9 are active?


I hope the questions aren't too silly or fatiguing, I’m thrilled to be making some sense of the circuit all of a sudden. I promise I’ll move on after this but you might be seeing me a lot on this forum in the future 🤓

Thanks!

[Kim Christensen]

1) I assume the connection from output to C56/R89 is providing some sort of (negative) feedback. The part left between the differential pair and PNP common emitter amp looks like a filter, is the purpose of the feedback to filter the signal?

Yes, those components form the negative feedback path that reduces distortion, sets the gain, and set the audio bandwidth. Without going into too much detail, the ratio of R89 to R92 sets the gain of the amplifier. C56 & R91 provide extra feedback at higher frequencies to reduce gain beyond the audio range which helps with stability. C55 rolls off the lower end of the audio spectrum and reduces gain to 1 at DC.

2) So, why is the cascode amplifier there in the first place?

The only reason for it, that I can see, is that it reduces the voltage across T12 & T13 (currently 90V each). If T12 was eliminated, and the collector of T13 was directly connected to the base of T15, then T13 would have 180V across it, which would double it's power dissipation.

3) The cascode amplifier output is inverted, so if it would be removed from the circuit, could the T15 base be driven with the same signal as T14 (ie. from the T16 collector)?

No... T15's base is at +89.3V and T14's base is at -89.3V so connecting them together would be disastrous. Technically, you are correct in that the audio signals are in phase, but DC wise it just doesn't work out for biasing very well. And since this is a DC coupled amplifier, using coupling capacitors to get around that (T14 vs T15 DC base voltages) would break the DC feedback path.

4) Are the common emitter amplifiers there to simply add more gain?

Yes. Gain and phase inversion.

5) Would it be correct to say that the push-pull amplifier is constructed in a Darlingtonish manner?

Yes. T9 and the T8-pair make a NPN Darlington. And the T6 and T7-pair make a PNP one.

6) Would the circuit still work, albeit with weaker output, if they were removed and the RX resistors connected straight to T7/T8 bases?

By "they" I assume you mean T6 & T9... Technically yes, though you'd have to change the bias circuit a bit (Short D39 & D21) and increase R75/R78... T15 and T14 will also get a bit hot trying to supply more current as you cranked up the volume. Magic smoke may leak.

7) I understand diodes D21 & D39 are there for the class-AB bias but what about NTC1, BIAS1 and the extra diodes? Are they just ensuring that T6 & T9 are active?

NTC1 is there to ensure that the bias current in the output stage stays steady as the amplifier heats up. The BIAS1 resistor reduces the voltage a bit to tweak the bias current a bit lower.
There are 4 diodes because there are essentially 4 base-emitter junctions in series that are being forward biased. (T9-T8-T7-T6)... Since a diode junction drops approximately the same voltage as a base-emitter junction, and that voltage tracks with temperature, they use 4 diodes in all.

[magic]

they just seemed to be connected so weirdly together

They are.

I don't think this circuit was designed to realize any clear and sound principle of operation, more like someone threw parts of various designs at it until it sort of worked.
Which is reportedly not an entirely uncommon approach in the audio industry.

2) So, why is the cascode amplifier there in the first place?

The only reason for it, that I can see, is that it reduces the voltage across T12 & T13 (currently 90V each). If T12 was eliminated, and the collector of T13 was directly connected to the base of T15, then T13 would have 180V across it, which would double it's power dissipation.

It looks like T13 is running on less than 1mA bias (it should approximately equal T16 current*), so power dissipation wouldn't be too bad.


A conventional design would make T15 a constant current source like T18. The whole T13 stage is very problematic here, because it makes it an amplifier with three voltage gain stages (T17, T13, T15) rather than two in addition to the usual two (T16, T14). Three stages make for harder compensation. There is some weird inductor there (L2) and I suspect C56 was necessary because the amplifier has too bad phase margin to maintain flat closed loop gain at high frequencies.

Maybe T12 was also actually necessary for stability too, or maybe mine or the designer's estimation of T13 power dissipation was wrong.


*Which is very likely significantly different than T17 current, no wonder they needed offset adjust pots