6SN7 input stage unexpected frequency response

[gerrittube]

I'm testing a Williamson style input stage with a 6SN7 and the bode plot shows a rolloff of more than 10 dB when the frequency goes up.
Of course I tested the signal generator too and it's flat within 1dB up to 100 kHz.

I measured at the top of the first anode, using a 1 uF capacitor into a 10M probe. The capacitor is there to get only AC for the probe.
The next tube is another 6SN7 with 220K from grid to ground.

What could be wrong? What else should I measure to get a better understanding of what's going on here?

Regards, Gerrit

 

[jan.didden]

Doesn't look like HF rolloff, more like a LF boost due to pos fb.
But hard to say without seeing the rest of the circuit.

Jan

 

[gerrittube]

Hi Jan,

There is no feedback anywhere in the circuit.
What could cause positive feedback when it's not part an intended part of the circuit.

Regards, Gerrit

 

[Merlinb]

What does your power supply look like? Do you have enough decoupling?

 

[gerrittube]

Hi Merlin,

My PSU is a zener regulated MOSFET source follower with approx. 400 VDC output.
I will check tomorrow or perhaps this evening with more decoupling.
As you can see both the input tube and the phase inverter are on the same power rail.

Regards, Gerrit

 

[jan.didden]

Hi Jan,

There is no feedback anywhere in the circuit.
What could cause positive feedback when it's not part an intended part of the circuit.

Regards, Gerrit

I don't know because I only see part of the circuit.

Jan

 

[gerrittube]

Jan,

There is no feedback into the part of the circuit that is shown.
So how could it be influenced by feedback? I don’t get it right now.
There is no feedback added in the entire circuit.

Regards, Gerrit

 

[Koonw]

Are u sure voltage U1b cathode is same as U1b grid (or U1a plate)? As shown, it's zero bias and so will not work properly esp when drive voltage is greater than the bias.

 

[gerrittube]

Hi Koonw,

I will check this tomorrow.
Could you share your ltSpice file with me? I would like to see how you create the frequency response diagram.

Regards, Gerrit

 

[Koonw]

Sure, here it is.

 

[daqvin_carter]

Maybe replace LED1 with an appropriately sized cathode resistor and see if the problem goes away,

 

[gerrittube]

I already tried using a single cathode resistor (without the LED); I tried with several values without any difference to the frequency response.

Regards, Gerrit

 

[rongon]

I can't view the .asc because I'm not at my computer with LTspice on it, but...
What is the decoupling network from the output stage to this cathodyne stage?

My PSU is a zener regulated MOSFET source follower with approx. 400 VDC output.

Is there one plate supply feed common to this cathodyne stage and the driven push-pull output stage?
Is this a push-pull power amp? What about the time constants formed by bypass cap(s) on cathode bias resistor(s)?

 

[daqvin_carter]

Maybe make sure R2 is really 1K? Sort of an odd frequency turnover that I would not expect with that schematic.

 

[rayma]

Looks like the B+ to the first stage has a rather large value of dropping resistor, and lacks enough decoupling
to avoid a boost at lower frequencies. If you run the first stage directly with a lab supply, the boost will go away.

 

[gerrittube]

Thanks for all input, this is very much appreciated.

1. R2 is absolutely 1K, but I will measure it again to be 1000% sure.
2. I have a lab supply 0 - 600 VDC @ 2Amp. I will try this too tomorrow (and with extra decoupling).
3. I can try a different anode resistor value, but according to Koonw’s ltSpice simulation this should be OK.

Regards, Gerrit

 

[gerrittube]

Rongon,

The 400 VDC regulated supply is created from my 600+ VDC raw DC voltage. In order to keep the MOSFET in the 400 VDC regulater a bit cooler I use a few high wattage series resistors. They are just getting warm, not hot. The same goes for the MOSFET.
I use another regulated circuit to feed the final tubes with approx. 600 VDC.

Yes, it’s a Push Pull amplifier. The pain however is (only) in the input stage. I don’t measure a significant difference in frequency response between (1) input stage as described in post 1 and (2) the output measured over my 4 Ohm dummy load.

Regards, Gerrit

 

[rayma]

Don't use any decoupling with the lab supply. Just connect it directly to the first stage instead of what you have.
If there's no low frequency rise, the decoupling is the problem.

What exactly is the decoupling circuit used? Seems like it must be well over 100k in series with the B+..

 

[gerrittube]

Hi Rayma,

Right now there is no decoupling between the input triode and the phase inverter triode. Both are connected to the same power rail.

Regards, Gerrit

 

[rayma]

Ok, but it could still cause that LF rise even if the first two stages share the decoupling network.
Just run both stages with the lab supply instead of what you have. The rise will be gone.

Can you post the supply schematic?