Assymetrical correction (?) waveform

[Helmholtz]

They are at U101 output, first a lower power then at higher output levels.

Did you have a load at the speaker output?
Without a load the NFB signal might look different.

[J M Fahey]

1) the waveforms you show are normal and to be expected IF NFB ones, because to get the nice output sinewave the outside World expects, internally NFB will do anything, I mean ANYTHING to correct it, obsessively .
Thatīs its job.

Correction waveforms normally look *ugly*

2) that asymmetrical waveform was common in old "quasi"-complementary power amps (all NPN) because top and bottom halves are different, but modern amps are all complementary, so didnīt expect that.

But the moment you mentioned "limiter"and knowing this is a Japanese (or Korean, same thing) amp I started guessing a culprit, and I found it.

That distortion comes from the *cheesy* limiter.

Normal would have been a FET limiter (which is not perfect, requires internal NFB for linearity, but at least can handle 1 or 2 V RMS signal

Instead they used the CHEESY bipolar transistor type used from the Stone Age (ok, since the 60s) in cheap ubiquitous Japanese cassette recorders (or cheap tape recorders).

Remember typical cassette recorders did NOT have a Record level control, only a big red "Record" button or key.
Horrible "pumping" since time constants were way too long (by sheer necessity) and since they expected voice input from a few inches to a couple feet away levels, they murdered any live show or concert you wanted to record.

It used to be called ALC (automatic level control) by Sony, Aiwa, National Panasonic, etc.

We all have a couple of poorly sounding old band rehearsal or show cassettes stashed away somewhere

A bipolar transistor *can* be used as a variable resistor BUT:

* it is VERY nonlinear (hint hint)

* it can only handle very low signal levels, say 50mV or less.

* control signal feeds through into output, so time constants need to be sllloooowwww



Input signal comes through P301 IN pin.

Is heavily attenuated by 22k R310 and 750 ohm R311 1V becomes 35 mV

Q302 is a (cheesy and non linear) variable resistor, current controlled through its Base.

Said signal is then re-amplified to former around 1V level by U301, which not surprisingly has exact same Gain resistors as the attenuator : 22k R312 and 750 ohm R313

Output signal leaves through P302 OUT pin.

Triggering signal:

Distortion is detected at Power amp (drive signal goes wild trying to further drive amp beyond a certain point) , similar to Crown and Peavey DDT (distortion detection) , it comes to limiter through P302 "Limit" pin, and goes to window comparator U301 B/C

Any of them will charge 22uF C301 through 33k R305 which sets the attack time. to 0.7 seconds
Slow but acceptable for a subwoofer.
Release (discharge) time through 470k R307 is some 10 seconds, unbearably slow but acceptable here.

For full audio limiting attack "should" be quite faster to catch cymbal crashes or drum kicks and such.

Q301 is prebiased to about "2 diodes" so limiter avoids hiccups.

A lot of complication to save on a FET (I just bought 1000 J112 at Mouser for 10 cents each) but hey ..... a bean is a bean, said the bean counter

[Helmholtz]

There's no signal across the limiter transistor Q101. It only acts on the DC bias of Q103 and Q104.
So its nonlinearity shouldn't cause signal distortion.
(Same with Q102 which sets the bias voltage between the bases of Q103/104.)
But by lowering the DCV between the bases of Q103/104 it can make those transistors distort.
This DC bias should not drop before the amp reaches its max. output power.

Is C105 good?

[g1]

Just want to point out here to casual readers that the limiter circuit JM Fahey spoke of is on a separate board and different from the Q101 limiter Helmholtz mentioned.

Helmholtz , yes those last images were with load connected. I am planning on checking the rest of the power amp electrolytic caps before letting this one go. Very poor design with caps placed right up against very hot transistors and resistors was what I think caused C129 and C130 failure. I also upped the voltage rating as 10V for those seemed marginal.

[x-pro]

Q101 is an ordinary dynamic spacer (Russian slang for technicians). Analog of a zener diode. It is installed near the output transistors to react to temperature. It can be called a limiter with a big stretch.
Q102 - standby key.
C129, C130 at 10 V is sufficient.
g1, we feed the signal directly to R103.
And in general, what are we looking for? I didn't see it in the topic.
Did g1 solve the problem?

[Helmholtz]

Just want to point out here to casual readers that the limiter circuit JM Fahey spoke of is on a separate board and different from the Q101 limiter Helmholtz mentioned.
l.

So Q101 is only used for standby?
Do you have the limiter schematic?

[x-pro]

So Q101 is only used for standby?
Do you have the limiter schematic?

Helmholtz, schematic Limiter in the post 17.
Q102 - for standby, to the base of which D101 is connected

[g1]

r1, we feed the signal directly to R103.
And in general, what are we looking for? I didn't see it in the topic.
Did r1 solve the problem?

Not sure what r1 refers to, but I think you are offering it as a test solution to bypass the limiter board?
The problem was that the amp would not get to full power. The fault turned out to be C129. It was reducing the gain of the amp and the limiter was engaging because the input signal was large enough that it should have been producing full power.
Along the way we were looking at various waveforms at the output pin of U101.

[Helmholtz]

Helmholtz, schematic Limiter in the post 17.
Q102 - for standby, to the base of which D101 is connected

I think you're mixing up Q101 and Q102 (hard to decipher from schematic).
Otherwise agree.

[x-pro]

g1 sorry, I didn't switch the translator. That g1 showed up on the keyboard as r1. The appeal was to you.

[Helmholtz]

The rules of designation of position numbers of elements on schemes from left to right, from top to bottom.

Q102 is more left, so not sure.

[g1]

Sorry, I had mentioned before (post #9), but it got buried in other stuff. Q101 is on the left and Q102 on the right in the schematic. Q101 is mounted on the heatsink. Q102 gets the standby command.

[Helmholtz]

Sorry, I didn't realize that JMF showed the actual limiter circuit of the amp.
I wrongly assumed he just used an other limiter example.

[J M Fahey]

There's no signal across the limiter transistor Q101.

Q101 is not a signal or current limiter transistor of any kind but just the Bias one.
It "should" be in thermal contact with heat sink.

It only acts on the DC bias of Q103 and Q104.
So its non-linearity shouldn't cause signal distortion.

It does not handle any signal so linearity is of no importance here, only thermal tracking.

Given the biasing resistors used, it drops slightly more than "2 diodes"
Exactly whatīs needed to slightly bias Q103/104 ON which in due time bias Q105/106 which drive and bias Q107/108

(Same with Q102 which sets the bias voltage between the bases of Q103/104.)

No, biasing was already done by Q101.

Q102 is a Mute/Standby switch, used by whatīs called "Auto turn on".

In fact PSU is always ON (thatīs why idle power is very high:13W)

U203a/b detect input signal (together with circuit around them) and triggers Q102. OFF so amp can be driven.

But by lowering the DCV between the bases of Q103/104 it can make those transistors distort.

It only works when amp is muted; otherwise it is biased OFF and becomes inoperative.

This DC bias should not drop before the amp reaches its max. output power.

Canīt imagine any mechanism by which it would drop.
Except by heatsink temperature, which is its job.

This power amp does not even have short circuit current limiters, but has very Japanese relay protection.

In theory it acts within milliseconds.

[x-pro]

As J M Fahey writes "U203a/b detect input signal (together with circuit around them) and triggers Q102. OFF so amp can be driven"
Of course, to test the amplifier in test mode by applying a signal to R103, Q102 should be temporarily desoldered.