Taming Ringing...Square-Wave Response

[EchoWars]

Dead amp that I brought back to life with a few new transistors, and new output devices (originals were toast).

As usual, I pumped in a square wave to look for anything funny, and was greeted with pronounced ringing on both the positive and negative edge at only 1KHz and a fairly low amplitude (a few volts output). At 20KHz, the ringing doesn't even have a chance to settle down before the square wave switches polarity.

Was quite suprised to see this, and have never had to battle something like this before. I wish it was as simple as a failed component, but both channels exhibit identical behavior, so I think that is out.

Perhaps someone can enlighten me as to the mechanism behind a ringing square wave...and perhaps suggest a remedy? Is this design so inherently unstable that the replacement transistors are too fast? Or are the replacements too slow?

[SY]

Might I guess that the replacement transistors had a higher ft than the old ones? If so, you'll have to diddle around with the compensation.

[subwo1]

Maybe try some compensation from collector of Q5 to the base of Q2 and maybe half that amount from the output (before the output inductor) to base of Q2 while also removing the miller cap from Q5. The first cap may be 4.7-20pf while the 2nd may be 2.2-10pF, or so, in about a 2 to 1 proportion.

[johnnyx]

Could it be the output inductor ringing with soe load capacitance? If the frequency of the ringing is fairly low then this is probably it. Instability from the change of components would ring at a few megahertz.

[slowhands]

Originally posted by EchoWars
...I pumped in a square wave to look for anything funny, and was greeted with pronounced ringing on both the positive and negative edge at only 1KHz and a fairly low amplitude (a few volts output). At 20KHz, the ringing doesn't even have a chance to settle down before the square wave switches polarity.

Was quite suprised to see this, and have never had to battle something like this before. I wish it was as simple as a failed component, but both channels exhibit identical behavior, so I think that is out.

Perhaps someone can enlighten me as to the mechanism behind a ringing square wave...and perhaps suggest a remedy? Is this design so inherently unstable that the replacement transistors are too fast? Or are the replacements too slow?

My guess is the output transistors are faster, and are revealing mild instability that was masked before. If this is not an audible problem, don't worry about it. If you feel you must quell it, there are so many possible sources.

Here are some ideas. First I would try to figure out which stage is causing it, by looking down the chain from output to input until the ringing disappears. It is likely the previously inspected stage is the source and you can focus there.

My gut feel is that it originates in the output EF triple. I don't see any base stopper resistors between the predrivers and drivers, sometimes that can lead to mild instability. Sometimes just a 10 ohm resistor to the base will shut that down. Alternatively, you can try slowing down the drivers (Q9, Q10) with say 100 pF caps between base and collector, although that might just mask the problem you have revealed. You might consider those options.

If it seems the problem is elsewhere, there are a few other areas that seem a little out of control. The VAS seems undercompensated, that 10 pf (C7) on Q5 looks like eyewash to me (I would expect 100 pF, and it's pair mate has 47 pF). Sub in 100 pF as an experiment.

Some other random thoughts:

It could be the output inductor is underdamped, try a 1 ohm 5 watt there instead of 10 ohm (R41).

Is R42 open, from previous oscillations? If the amp was oscillating before, this could cook, effectively removing the cap and that cap may be needed for stability.

The input filter has only a 47pf cap, that does not do much to limit high frequencies coming in. Consider a 390 or 470 pF low pass input cap (C3).

Those are some thoughts. Good luck!

[EchoWars]

The original output transistors are 2SC1116A (10MHz) and 2SA747A (15MHz). The replacements are 2SD424 (5MHz) and 2SB554 (6MHz).

However, the replacements for the original 2SB528 (70MHz) and 2SD358 (ft unknown, probably around 70MHz) were replaced with 2SA794A (120MHz) and 2SC1567A (120MHz). Would that be enough to do it?

I dug around, and I have a couple of old driver boards with good 2SB528's and 2SD358's, which I'm gonna pull out and put in this amp. But before that, I'm going to check R42, and maybe even replace C11 with a .1µf cap. Couldn't hurt....

[EchoWars]

Well, the 2SB528's and 2SD358's were put back in, and no difference in the output was observed. R42 (the 10 ohm Zobel resistor) is fine. Bummer.

I'll see if I can follow the signal from input to output and determine where the ringing starts.

[EchoWars]

Originally posted by johnnyx
Could it be the output inductor ringing with soe load capacitance? If the frequency of the ringing is fairly low then this is probably it. Instability from the change of components would ring at a few megahertz.

Forgot to reply...the ringing is there with either no load, or with an 8-ohm dummy load. There should be no oddball capacitances with enough magnitude to account for the ringing.

[Bill Fitzpatrick]

Originally posted by EchoWars

I'll see if I can follow the signal from input to output and determine where the ringing starts.

Unless I'm mistaken, you'll see ringing all along the signal path unless you lift the feedback. And if you lift the feedback, the input signal should be considerably reduced.

[EchoWars]

Originally posted by Bill Fitzpatrick


Unless I'm mistaken, you'll see ringing all along the signal path unless you lift the feedback. And if you lift the feedback, the input signal should be considerably reduced.

And that's pretty much what I see...about the only place I don't see ringing is on the base of the input.

I did calculate the frequency of the ringing...it's about 290KHz...