Getting lost with analysing Class AB amplifier circuit

[bborisov567]

I have some questions regarding how the given circuit works. Earlier today i was poking with the scope following the signal path an i noticed the signal is coming from the base of T602, not the collector like usually is done with differential pairs. How exactly is working that differential pair? Later T607 and T609 form a Sziklai pair. Same is valid for T606 and T608. Both of these pairs work as a push-pull stage (which is amplifying only current?). Then T2 is connected in a common collector configuration (forming a Darlignton pair with the push-pull Sizklai pair T607 and T609?). T1 is connected in a common emitter configuration (forming another Sizkai pair with the previous Sizklai pair T606 and T608?) So in the end there is only half stage in common emitter configuration amplifying voltage an current? I know that i am missing something. The more i am looking into it the more confused i get. 

[fourfathom]

T601 and T602 are indeed a diff pair.  The base of T602 is getting negative feedback from the output, the feedback value set by R608 and R607.  The overall amplifier gain is positive (positive going at the C601 input fives a positive going output at testpoint #8.  When the output goes positive, the emitter of T602 is driven more positive (T602 draws less current), providing negative feedback.

Work from there and see if it makes sense.

[magic]

T602 base is feedback from the (divided by resistors) output. T601 and T602 are an ordinary diff pair. T601 drives the T603 VAS, loaded with R611 bootstrapped to the output by C604, a common trick back in the day.

The output stage is fundamentally a complementary Sziklai pair stage (aka CFP output stage), but with a kind of quasi-complementary twist.

T606,T607 are emitter followers and their collector currents are multiplied by the combined beta of the Sziklai pair T609,T2 on the high side and the Darlington pair T608,T1 on the low side. Each of T606,T607 forms a Sziklai pair with its corresponding pair of "current booster" transistors.

The whole output stage is of course unity gain. Chances are it may have less distortion than an ordinary quasi-comp. Current gain of the halves is matched if the drivers and pre-drivers are matched, because the final outputs are of the same type on each side.

Not sure why BD140 is in Latin but BD139 in Cyrillic

[gcewing]

Not sure why BD140 is in Latin but BD139 in Cyrillic

Russia was out of BD140s, so they had to use a Western one?

[gbaddeley]

That brings back memories of late 1970s designs. Good old 2N3055. No degeneration or CCS on input stage emitters. Boot strapped driver stage.

[magic]

Don't forget the single resistor in the collectors. Yeah, the input stage is pretty lame

[dietert1]

There is nothing wrong with the circuit. It has adjustment pots for bias current (R602) and output offset voltage (R601).
You need to check C602. That cap won't last as it typically rests with reverse voltage after turning off the amplifier... If you can afford, you can replace it by a bipolar cap or a film cap.
One slight risk in those amplifier circuits is RF oscillation - i mean it it destroys speakers or overheats for no obvious reason you know what to look for.

Regards, Dieter

[magic]

BTW, what's the supply voltage and the value of that D601 zener?

[bborisov567]

Thanks for the replies. The voltage of D601 is 12v. So if the output stage is unity gain then the voltage amplification is done by the differential pair, right? I will replace C604. Somebody actually has replaced it on one of the channels. Also what is the purpose of C602? Regarding the bias set of the differential pair i should be adjusting R601 until i get around 0.6v between base and emitter of T601 so that the transistor is in slight conduction? Regarding the BD marking i just noticed it but not surprised at all. The base and emitter markings on the solder mask are reversed on the actual board. The whole amplifier is quite a sloppy device.

EDIT: I just measured base-emitter voltage of T601 while adjusting R601 and is is not changing.

[magic]

Voltage amplification is done primarily by the voltage amplification stage, obviously, which is T603 in this particular amplifier

I'm not sure what to recommend for learning basics of power amp design, maybe this would help but I haven't read it in years so not sure if it's sufficiently beginner friendly
https://sound-au.com/amp_design.htm

What's wrong with C604?

The purpose of C602 is to reduce gain at subsonic frequencies to unity, so that the amplifier doesn't amplify its offset voltage.

R601 is for offset voltage adjustment. Tweak it to get zero volts DC on the output with no signal. Do it on a warmed up amplifier because it may drift with temperature.

What's wrong with this amplifier and why are you messing with it?

[fourfathom]

Also what is the purpose of C602?

It reduces the amplifier gain to unity at DC and very low frequencies (4 Hz is the corner frequency).

EDIT: I just measured base-emitter voltage of T601 while adjusting R601 and is is not changing.

R601 is used to compensate for DC gain mismatch between T601 and T602.  R601 and R602 provide base current to T601, R608 provides base current to T602.  Look at the amplifier output DC voltage as you adjust R601, you are trying for 0V at the output.  There may not be much change, it depends on the gains of the transistors.

[bborisov567]

Thank you for the link. I just found started reading it just before you posted it. Currently reading The Art of Electronics which has a more general approach. But is there a book specific to amplifiers or something, i couldn't seem to find one. I am messing because i noticed it was heating when idle and also i thought it is going to be a good exercise of bjt circuits.

 I will adjust R601 tomorrow. Now to talk about quiescent current. Not exactly sure but i tweaked R603 so that i have 0.55v base-emitter voltage on T609. Same voltage comes on T608 . I did some more test and found that the transistors start to heat up around 0.6v. Also, just for test purposes i set the voltage to 0.1v, hooked the scope while having sine wave at the input and there no sign of crossover distortion. What is going on?

And last D604 and D605 are for temperature compensations, right? In theory they should be on the heat sink, in realty they are just on the board. I did some temperature measurement and under load the output transistors are around 65 degrees while the diodes on the board are around 35. 30 degrees difference would lead to about 60 mv decrease in Vbe and roughly 3 times increase in collector current. Feel free to correct my calculations. Isn't that that a really bad design that can lead to a thermal runaway?

[magic]

This may work if you want half thousand pages of dead tree about power amplifiers and only power amplifiers:
http://douglas-self.com/ampins/books/apad.htm

This output is really a Sziklai pair output, ignore those stupid 2N3055 for a moment. T606,T607 are the "drivers" and it is their Vbe which needs to be compensated by the diodes, as in every other Sziklai output stage. So yes, the diodes belong on the PCB, near those transistors. Nothing wrong with it at all.

Quiescent current is determined by voltage across R626,R627. This voltage is the voltage across the diodes minus Vbe of T606,T607.

Crossover distortion may only show under load. Note that even if T608,T609,T1,T2 are completely off, T606,T607 still drive the output. And even under load, as soon as T606 current increases enough, the rest of the upper side will turn on and supply full current.

[bborisov567]

Quiescent current is determined by voltage across R626,R627. This voltage is the voltage across the diodes minus Vbe of T606,T607.

So in theory voltage across R626 and R627 should be 0 whereas in practice is the difference between the voltage drops of T606, T607 and the diodes? I have to measure the drop of the diodes then the drop of the transistors and set the voltage across both resistors to be the difference?

And last question - i have noticed some amplifiers have complimentary output stage while others like this one have just two NPNs. Is it just a design decision to cut cost back in the day when PNPs were difficult to produce or has something to do with the actual performance of the amplifier?

[magic]

Absolutely motivated by cost/availability of power PNPs and its influence on performance tends to be negative, that's why the designers bothered with that relatively complex output stage with three transistors per side and local feedback.

Voltage across 0.33Ω resistors definitely shouldn't be zero, that means no bias at all - you know the Ohm's law, I hope. Adjust R609 for the same voltage across resistors as on the other channel and be careful because it may increase or decrease after warmup if thermal compensation isn't perfect. Alternatively, if you can find complete service manual, it should tell how much bias to set. And you could test those resistors if their value is right.

edit
Now that I think about it, if both channels are set for the same bias and run at the same temperature when idle, but one gets hotter under load, then C604 may potentially be the culprit. It could result in temporary increase of D604,D605 current during negative output cycles and some corresponding increase in bias, maybe(?). I'm not entirely sure, but it's a cheap and easy component to replace...

[bborisov567]

After more messing i found that there is something wrong with one of the channels. First the offset voltage - I can't get it below 90mv even with R601 turned all the way. Also i noticed several times , T609 and T2 are getting hotter than T608 and T1. C604 an C602 are already replaced. What might causing more current to flow on the one side? Maybe D606-607?

[magic]

What might causing more current to flow on the one side?

Offset voltage.

And if the amp works otherwise, I would look at the input stage for the culprit.
What are the voltages at T601,T602 bases and across R604,R605,R610? Are those resistors of right value?

[Kleinstein]

D606 and D607 should normally only conduct when he overcurrent limit engages. So they should not cause an offset.

90 mV of offset ist quite a lot, more than the trim range of R601 (estimated  to be some +-20 mV). The offset could explain why some transistors get warmer than others.

C601 may not see a correct bias and may thus be faulty / leaking. A wrong voltage at D601 could be a problem too. My rough estimate is at some 6 V
A drift or large difference  in the transistor gain / BE voltage for the input stage could cause an offset. This could happen with parts from different sources (replaced just 1).

The Sziklai stages are a bit prone to oscialltion. So worth a test of there is high frequency oscillation.

[gbaddeley]

Yeah, no current mirror. The simplicity of the IPS is at odds with the complexity of the OPS.

THD and bandwidth of this amp is not going to be very good, but it was acceptable 50 years ago.

[bborisov567]

D601 is 12v, already checked and it is fine. Upon closer inspection i noticed one of the differential transistors was replaced. Measuring base-emitter voltage gave about 20 mv difference between the two transistors. I desoldered both of them and tested them - one has 1.06M hFE, the other 1.08M. I know differential pairs should be closely matched but could such a small difference be causing problem? Also about the quiescent current - at about 10 mA the BDs start to heat up a lot. I am out of luck finding any kind of truthful information about the specified current from the manufacturer. Upon searching in local forums somebody had suggested about 25-30 mA but that seems awful lot of current.

[Kleinstein]

The gain is quite high and the adjustment range of R601 is thus quite small, as the adjustment is via the input bias current and input resistors.
A 0.2 mV VBE matching would be very good - like a lucky concidence or a really matched pair.

The base of T603 should be at some 700-900 mV from the negative supply (depends a little on the current through R611/R612 / supply voltage).
To get equal current for the 2 transistors in the input stage, it would take a emittter resistor (R605) to be equal to 1 half of R604 (3.3. K in the plan) times  U(D601) divides by the drop at R605.  So one could get a better offset adjustment with a slightly different value for D601 or R604.

If C602 is OK, the DC gain should be at x1, so the ouput offset should be close to the error from the input stage. Depending on the polarity even new caps can show quite some leakage at the wrong polarty. Ideally C602 should be non polarized (or maybe 2x100 µF back to back as a substitute). Leakage at C602 can amplifiy the offset.

For the ouput stage, I would expect it to get away with a relatively low standing current. Some 10 mA would be needed to get 400 mV at 39 Ohms and thus the start of the active region for the 2N3055. So 10 mA are probably a lower limit for low distortion. With the 3 transistor combinations the cross over is quite sudden anyway, so not that much option to fine tune the THD by choosing just the right current.

[bborisov567]

I just notice a huge mistake - instead 20mv i had written 0.2 for some reason. Sorry for that. C602 is replaced so shouldn't cause any trouble.

[Kleinstein]

The way the circuit is made C602 sees a low voltage of unknown polarity. Even a new cap (especially low ESR types) can show high leakage if the voltage is in the reverse direction - with some caps even some 10 mV reverse can result in significant leakage ( > 1 µA). The old caps may have been more tolerant in this respect. Depending on the offset of the transistors it may be better to change the polarity of C602.

20 mV offset for VBE is more like the expected order of magnitude and more than can be adjusted by the trimmer (R601). One could do a little extra trim by changing R604 or a larger value for R602 if suitable. There is no real need to have Hfe mathicng for T601/T602 it would be more about VBE matching and a reasonable high Hfe.

[dietert1]

Bias current is adjusted by looking at the voltage over the emitter resistors R626 and R627. The combined voltage should be 20 or 30 mV. It really isn't that precise and it will change with room temperature.
Some audio freaks search for solid amplifiers with large heat sinks and tune them to run at higher bias currents - like 100 mA. Then at the usual listening levels inside a room the amplifier operates as class-A most of the time. This method usually gives good results, but does not replace a THD measurement. A normal listener will not be able to tell the difference between 10 mA and 50 mA bias current. Except with 10 mA there is a small risk it will go to zero with10 °C more or 10 °C less ambient temperature or when the amplifier runs hot.
Once i fixed a chinese amplifier "AXPS-400" where they had wired the negative feedback from the wrong end of one of the emitter resistors - probably a copy/re-engineering mistake. Terrible THD and they sold thousands of those.

Regards, Dieter

[magic]

D601 is 12v, already checked and it is fine. Upon closer inspection i noticed one of the differential transistors was replaced. Measuring base-emitter voltage gave about 20 mv difference between the two transistors. I desoldered both of them and tested them - one has 1.06M hFE, the other 1.08M.

Was the 20mV difference measured in circuit or out of circuit?
It could either be because of transistor mismatch or about 2:1 mismatch of collector currents due to a problem elsewhere (if in circuit).

Not sure what 1.06M or 1.08M hFE is supposed to mean?

I know differential pairs should be closely matched but could such a small difference be causing problem?

Offset voltage is always due to input stage and resistor mismatch (unless the amp is completely nonoperational). 20mV Vbe difference translates directly to 20mV offset voltage. But there is 90mV, so the rest likely comes from voltage drop across resistors due to base current: T601 base current flows through R601~R603 to ground, T602 base current flows through R608 to the output.

20mV is not a great match, with transistors from one production batch a few mV is common. Even in Chinese junk transistors from auction sites there is maybe less than 25% outliers which are worse than that. Test the other channel which hasn't been monkeyed with and see yourself.

Once i fixed a chinese amplifier "AXPS-400" where they had wired the negative feedback from the wrong end of one of the emitter resistors - probably a copy/re-engineering mistake. Terrible THD and they sold thousands of those.

LOL.
And for the record, neither end of neither emitter resistor is a good place for feedback takeoff, although some ends are surely worse than others