Tube pre oscillation

[ruffrecords]

I am testing the latest PCB layout of my Twin Line Amp design. These are more compact that the originals because the tubes are now on their sides so it will fit in a 35mm wide module. The PCB holds two identical amplifiers. Each uses a 6922 SRPP output stage and shares half a 12AX7 is the input stage. NFB extends to dc and there is a single coupling capacitor between stages so it is unconditionally stable at all gains. Closed loop gain is altered by a single gain resistor, With no resistor fitted the gain defaults to 6dB. The output of one stage drives a 2K4:600 output transformer.

Each preamp works fine on its own just like the originals. I then went on to test the two amps cascaded. The output of the first preamp is  connected to the second from which the output is taken. The problem I am getting is when cascaded like this, even at the default 6dB gains, the whole oscillates at 1.479MHz and I get a nice 20Vpp dose of this at the output.

There is a screened cable carrying the input signal across the board to the first amp input. There is another screened lead carrying the first amp output to the second amp input. The second amp output goes straight to the output transformer.

If I disconnect any of these cable at either end, the oscillation stops. I think it stops if I disconnect the transformer (need to re-check that).

So I need to fix this. The layout, especially around the tube bases is quite compact but don't think this is the problem because it goes if any of the cables is disconnected.

Any ideas?

Cheers

Ian

 

[JohnRoberts]

Perhaps look at where the cable capacitance (shield) is dumping current?

JR

 

[moamps]

-  check it with changed order of the stages,
- add a small miller cap to the first stage, or low pass somewhere
- test it with ecc85 (ground pin 9) instead of ecc83,
- do not split dual triodes in cascaded stages anymore  .

 

[ruffrecords]

-  check it with changed order of the stages,
- add a small miller cap to the first stage, or low pass somewhere
- test it with ecc85 (ground pin 9) instead of ecc83,
- do not split dual triodes in cascaded stages any more  .

Order of stages makes no difference
33pF across NFB resistor of one stage fixes it
Unfortunately ECC85 has only 6.3V heater but I use 12.6

I suspect the 1.479MHz is external. I had a problem similar to this some years ago that turned out to be interference from a local mobile phone mask. This frequency occurs at a small level at input with pre switched off.

Cheers

Ian

 

[walter]

Are the cable shields connected to ground at one end only, should be to prevent ground loops. Instead of ground, connect  the shield to the output of the stage it feeds, faze cancellation. Just a couple suggestion I learned about guitar amp voodoo, I still have a lot to learn about HF oscillations.

 

[Tubetec]

You could try a screening can on one of the valves , or maybe a metal screen between them.

 

[CJ]

put I/O jacks right next to each other.