Got an overshoot problem.

[Shoog]

Hi all,
I was listening to my new Tabor amp clone today and it suddenly dawned on me that there was a hard grainy quality to some of the midrange on complex passages. I hadn't taken a close look at things on the scope so took it to the bench and did some square wave tests.
Turns out I have a sever overshoot problem, in the order of 2x the square wave, but very little ringing. Its like a big nasty spike which dies straight away. I am using mains toroidals as outputs and my speakers are 4ohm. I checked my input transformer and there is no ringing there. I think the plate to plate feedback maybe masking some ringing.
Where do I start with damping this nasty artifact.

Shoog

[Tom Bavis]

If you want to avoid global feedback, you my want to try a Zobel network across the primary - try a series RC network, maybe -3 dB at 40 kHz, and see what it does.

[ilimzn]

Would be nice to see the exact schematic you are using, but here's an idea:

What you may be seeing is overshoot due to transformer stray inductance. In general, when applying square waves to the input of an amp, it is advisable to somewhat limit their slew rate (or, if you will, the bandwidth of the amp). The stray inductance of the transformer is that part of the inductance which is not couplet primary to secondary. In a manner of speaking, it is in series with the mutual inductance of the transformer's windings, so if your rate of signal change is sufficient, inductive kickabc even from this relatively small inductance (sompared to the total primary inductance) will create huge spikes, because for that part of the waveform the stray inductance effectively uncouples the amp load! The usual method of correcting this is incorporating a snubber network across the primary of the OPT.
Another possibility is sort of the reverse of the above - a situation where various capacitances which end up shunting your feedback signal, break the feedback loop at HF creating a peak in the HF response. This, however, is not possible to know without a schematic.

[Shoog]

I think I am experiencing a tank ringing because toroidals have tiny leakage inductance but high interwinding capacitance.
I remembered that previously when I used the same transformers I got a ringing/rising response when they were wired in one direction, but a falling response beyond 50kHz when wired the other way.
Since this looks very similar, I have tried switching the leads around. Initial impressions on the listening test seem quite promising.

Shoog

[oshifis]

Is there a compensating capacitor parallel with the feedback resistor? It is for compensating the effects described by ilimzn above. You can find the optimal value by experimenting.

[Jeb-D.]

If it only happens with a speaker as the load and not with a resistive load, add a Zobel on the secondary.

Unfortunatly, with the type of feedback the tabor uses, it's hard to tell if the conventional compensation capacitor in parallel with Rfb will be effective.

You may be right in assuming the iron.