Stubborn Noise/Oscillation - Another Call for Help

[LinuksGuru]

I have removed output tubes. Here are oscillograms from the one 6N6P LTP plate.

Signal is very small, almost on the edge of oscilloscope sensitivity, made with 2 frequency resolution. Kill me but it looks like 86 KHz oscillation modulated with 50 Hz hum.

[avp1]

Originally Posted by LinuksGuru

AVP1, are these mods to reduce phase shift between input/output signal and make GNFB more stable? Is there any formula in order to exactly determine these values of R and C?

Large transformer have quite high stray capacitance, so actually increasing it by 1000pF may lower resonant frequency of primary winding of the output transformer. Or am I wrong here?

There is a way to calculate them, but you need to measure gain and phase shift without feedback for that.

But in your case NFB is not a problem. You have oscillation without it too. So it either ground or power filtering or physical layout problem. You need to address it first before working on GNFB.

[avp1]

Originally Posted by LinuksGuru

I have removed output tubes. Here are oscillograms from the one 6N6P LTP plate.

Signal is very small, almost on the edge of oscilloscope sensitivity, made with 2 frequency resolution. Kill me but it looks like 86 KHz oscillation modulated with 50 Hz hum.

Do you see this 20 mv (P-P) hum with input shorted? If so, you power filtering is not adequate. Are you sure that ALL your electrolytic capacitors are functional?

[LinuksGuru]

Originally Posted by avp1

Do you see this 20 mv (P-P) hum with input shorted? If so, you power filtering is not adequate. Are you sure that ALL your electrolytic capacitors are functional?

Yes, input is shorted. Right now I use CLC filter 200uF + 4H choke + 100 uF. May be on of the cap is defective, I can't exclude that. I also suspect this might be due to long wires from power supply (it is mounted separately for a while).

But at the moment oscillation at LTP 6N6P is a problem #1.

As Geek from DIYTube.com suggested, 6N6P is prone to oscillation which is very difficult to get rid of. I'm going to buy 12BH7 (original schematic used them), and 6CG7 (very linear triodes). So I will try to use these tubes instead of 6N6P.

[LinuksGuru]

Originally Posted by avp1

There is a way to calculate them, but you need to measure gain and phase shift without feedback for that.

Input (w/o GNFB) = 200mV, output 10V. Gain: 20*LOG(10/0.2) = 33.9 dB

I'm certainly need to refresh my math skills back to my student days.

Here are oscillograms of 20 KHz input/output. Can you please point to the reference how to calculate these RC values?

[avp1]

Originally Posted by LinuksGuru

Input (w/o GNFB) = 200mV, output 10V. Gain: 20*LOG(10/0.2) = 33.9 dB

I'm certainly need to refresh my math skills back to my student days.

Here are oscillograms of 20 KHz input/output. Can you please point to the reference how to calculate these RC values?

One point measurement won't help. You need to cover all frequency range from 20 Hz to 500 kHz. An example: one amplifier I measured recently had problem with stability at 280 kHz.

[LinuksGuru]

Originally Posted by avp1

One point measurement won't help. You need to cover all frequency range from 20 Hz to 500 kHz. An example: one amplifier I measured recently had problem with stability at 280 kHz.

What frequency point are required (e.g. 20 KHz, 50 KHz, 100 KHz)?
At 207 KHz input and output signals are completely out of phase.

This amp have quite large output transformer (4.5 Kg, or 9.92 lbs), so it is not meant for so high frequencies as 500 KHz, because transformer of that size at 500 KHz will act like a capacitor.