Anyone recognize this sort of distortion going through NPN?

[cincin]

Working on a Sony TC-105 reel to reel tape deck. (see attached schematics)

I injected a 1KHz 1V sine wave through the AUX IN, and the signal at the collector of X3 (a 2SC402 NPN) gets distorted in a strange way. With the original 2SC402 in, the botton half of the sine wave would get squished out of shape. So I replaced the NPN with a 2N3904 with twisted legs (after replacing C14 to be sure), and with that one in the sine wave gets squished in both directions (see attached oscope screen grab).

The signal at this point has a 3.9V DC bias (oscope is AC coupled), schematics says it should be 4.5V but I don't think that's the problem (?) and the signal at both the base and the emitter is nice and sinusoidal.

Does anyone recognize this shape distortion? What could be causing this?

Thanks!

[Xena E]

You're within the output stage negative feedback loop there.

What does the output look like?

E

[wasedadoc]

Why not just use the output from the J4 MONITOR socket? No buffer required.

[cincin]

Why not just use the output from the J4 MONITOR socket? No buffer required.

Because it's at speaker level and expecting 8 Ohm? You would know better than me, but that was my reasoning.
If I can just plug the MONITOR OUT from the tape deck into my mixer, with good results, then I don't need buffers at all!

[cincin]

You're within the output stage negative feedback loop there.

What does the output look like?

E

I will be checking on that very soon.
I have too may concurrent projects...

[srb1954]

Does anyone recognize this shape distortion? What could be causing this?

Try reducing the input level a little bit and see if the observed waveform changes dramatically.

The O/P stage might be just going into clipping and, through the effects of the negative feedback, the driver has to produce a pre-distorted waveform to try and get a pure sine wave on the amplifier O/P.

[magic]

You're within the output stage negative feedback loop there.

Exactly, and this means that X3 collector voltage will be whatever happens to be required to drive X4 in such way to produce whatever current that's required in T1 to get the desired output from the final power NPNs. The waveform is distorted to counter an opposite distortion in the rest of the amplifier. I'm not sure why the distortion changed with X3 swap because it really shouldn't change much, maybe you changed or repaired something else too?

As for the root cause of the distortion, it's probably something in the output stage because X4 is supposed to be running at ~100mA bias, likely quite deep in class A. The amount of distortion should change with output signal level. Maybe this explains why it was different in your earlier tests?

As for X3, its quiescent current depends on β because resistance of R20/R21 is simply much higher than input resistance of the base, so they act more as a current source than a voltage divider. So no surprise to see that X3 collector current and voltage across R23 changed when X3 was replaced.


You could simply use a voltage divider to reduce the final output back to line level. Picking up audio signal from X3 collector is completely nuts. If anything, you could have more luck with the emitter, but note that any nonlinear loading added there may cause distortion in the speaker output and excessive capacitive loading may imperil stability.

[wasedadoc]

Why not just use the output from the J4 MONITOR socket? No buffer required.

Because it's at speaker level and expecting 8 Ohm? You would know better than me, but that was my reasoning.
If I can just plug the MONITOR OUT from the tape deck into my mixer, with good results, then I don't need buffers at all!

It is not expecting 8 Ohm. There is a 82 Ohm series resistor. You csn reduce the level with a couple of resistors.

[jzx]

It is an AB class amplifier, with a transformer, the feedback may be trying to compensate for this, if the output is good, thats is normal.

If you want a line output, why you do not take the signal from the volume potentiometer (upper side)? This way, the line out is independent from the volume of the speaker.
If you dont want to load this point, or if the level is not correct for your use, use a operational amplifier.

[cincin]

What does the output look like?

The output is pretty distorted too but evenly: both the top and the bottom of my sine wave get squared off. See the last oscope screen cap from the T1 transformer.

Try reducing the input level a little bit and see if the observed waveform changes dramatically.

I did, and under 600mV or so (through the AUX IN) the signal pretty much disappears. I also tried raising the signal's amplitude to 2V and the distortion is more of a square wave clipping on the lower half of the signal. Top half is still good at 2V.

I'm not sure why the distortion changed with X3 swap because it really shouldn't change much, maybe you changed or repaired something else too?

Yes, I did replace C14 to be sure that wasn't the problem AND I managed to install it backwards (!!!) thanks to those particular 60's capacitors having the black band on the positive side somehow (I learned something today!)... That didn't seem to fry anything. Things are back the way they should be with the original 2SC402 back in place, but the distortion is still there, normal or not.

You could simply use a voltage divider to reduce the final output back to line level.

This is starting to sound tempting... the logic of trying to tap the signal before the power amp was to save it from any distortion that the power section might introduce, which still sounds like a good idea given how the output looks?

Picking up audio signal from X3 collector is completely nuts. If anything, you could have more luck with the emitter, but note that any nonlinear loading added there may cause distortion in the speaker output and excessive capacitive loading may imperil stability.

I was going on @wasedadoc's suggestion here <throws wase under the bus> I built a little emitter-follower buffer so as not to load the circuit.

If you want a line output, why you do not take the signal from the volume potentiometer (upper side)? This way, the line out is independent from the volume of the speaker.
If you dont want to load this point, or if the level is not correct for your use, use a operational amplifier.

That was sort of the original plan. but there was some debate about wether to go op-amp or emitter-follower, and where to tap from.

So then, where in this circuit IS the best place to tap the highest quality signal, to be input at line level into a mixer?

Attached are some oscope screen caps, in this order:
1 Emitter of X3
2 Base of X3
3 Collector of X3
4 Base of X4
5 T1 tap

[magic]

So then, where in this circuit IS the best place to tap the highest quality signal, to be input at line level into a mixer?

Before the whole power amplifier.

If you want volume control to work, you need to take signal from the volume pot wiper. Either AC couple to 50% positive rail voltage (like C12 and R20/R21, but equal resistors), or add a negative supply and work with ground-centered signal. Add a buffer and you're done.

If you don't care about volume control, you can save yourself one AC coupling by tapping X2 collector, before C8. This is the signal that goes to the volume pot, but centered around 4.6V.

If you don't care about tone control either, take it straight from the AUX jack

[cincin]

If you don't care about volume control, you can save yourself one AC coupling by tapping X2 collector, before C8. This is the signal that goes to the volume pot, but centered around 4.6V.

If you don't care about tone control either, take it straight from the AUX jack

Ha! Well I do want the tone control, but not the volume. And, of course, the signal I want to output is to be read from a tape in the end. I was just testing with a directly injected signal while waiting for the motor run capacitor to arrive.

[magic]

Then I would try this:

Code: [Select]

X2 collector - NE5532 at unity gain - 100Ω - coupling capacitor - output jack
10μF gives 1.5Hz highpass filter when driving typical 10kΩ input impedance of external gear, so you may make the cap a few μF. Or make it larger if you want to drive lower impedance (headphones?) without significant bass rolloff.

[wasedadoc]

The negative feedback ratio is largely determined by the 27  and 6k8 resistors. Does not mean that the actual closed loop gain is 6827/27 (about 250) but probably not much less than 100. Given that the output stage runs from 24 volts, the maximum possible output peak to peak is 24 volts. A gain of 100 would put the max level from X2 at around 240mV p-p which is 85mV rms.

Need to confirm the actual  level there when replaying a tape.

[cincin]

Ok, I've recorded a 1KHz signal through the Aux In input, and used the unit's VU meter to set the amplitude. According to it, 100mV is what the tape deck wants with the recording level knob at max (not Auto).

When played back from the tape I get a strong, good quality sound from the speaker, and 240mV p-p (on the nose!) at the collector of X2 with a 4.6V offset. And the signal is clean-ish, no distortion. The sine waves are intact, just a bit fuzzy with hiss, as expected.

240mV is almost a usable Line-In level (?) but my mixer's max input level is +22dBu (12V?!). It's probably not necessary to load it up, but I am trying to use the best signal path and levels.

Would it be worth adding some gain to my buffer? Or would one of my mixer's pre-amp likely do a better job?

Code: [Select]

X2 collector - NE5532 at unity gain - 100Ω - coupling capacitor - output jack

Cheers!

[magic]

I'd add 20dB or so to bring it nearer 1V RMS, it's just two extra resistors and any audio opamp should be OK with it. The benefit is reduced sensitivity to external noise sources, such as input noise of downstream gear (maybe your mixer is not too noisy, but maybe you will want to connect other things in the future) or mains hum from ground loops.

[cincin]

This is all in place and working great, thanks for all the help!