Help me understand - two types of negative feedback

[BluegrassDan]

Here's my perpetually evolving tube preamp again. This time, trying to add in 9dB negative feedback is causing some issues. Can you please help me understand why?

Version 1 - From between R7 and R8, voltage divider from cathode, to input transformer (-) secondary. This arrangement is causing a significant high frequency lift upwards of 70kHz and getting really steep over 100kHz. Oscillation at high gains. Tidying up the layout and using shielded cable helped quite a bit, but it's still there. Any thoughts as to why?

Version 2 - From between C6 4.7uF cap and 10k potentiometer to input grid, after a 4.7uF cap. This arrangement produces a high frequency rolloff starting around 10k and down about -3dB at 40k.

My goals are to learn what is causing these phenomena (impedence changes, phase anomalies up top?) and, once that is figured, to add some soft of stepped NF switch to control gain reducation of V1.

Here are schematics. Gotta run to Walgreen's real quick, then I'll come back and show you the frequency sweeps.

Thanks, guys!

 

[BluegrassDan]

I'll add that I've experimented with three different input transformers that I know the response of under normal open loop situations.

 

[ruffrecords]

There is definitely something wrong with the second schematic because the output of the cathode follower is connected directly to the grid of the first triode.

Cheers

Ian

 

[BluegrassDan]

Ian, is that not an acceptable way to inject negative feedback? It seems to work in practice, but I'm curious.

 

[BluegrassDan]

MY BAD! I forgot to include the resistor in the schematic. Been experimenting with all sorts of values to give up to 20dB of NF.

 

[5v333]

2nd looks abit like a virtual ground circuit. perhaps try with a series input resistor (equal to what you want the transformer to see)

 

[abbey road d enfer]

Version 1 - From between R7 and R8, voltage divider from cathode, to input transformer (-) secondary. This arrangement is causing a significant high frequency lift upwards of 70kHz and getting really steep over 100kHz. Oscillation at high gains.

At a certain frequency, the xfmr combines with its own stray capacitance to form an oscillator. You should notice that the oscillation changes with loading the primary.
Taming this oscillation is tough, since it involves damping that's gonna affect negatively the noise performance.

Version 2 - From between C6 4.7uF cap and 10k potentiometer to input grid, after a 4.7uF cap. This arrangement produces a high frequency rolloff starting around 10k and down about -3dB at 40k.

As 5v333 suggests, it's a VE stage, so the FB ratio depends very much on the source impedance. What is the Open Loop response?
Beware that the more NFB you apply the more you decrease the input impedance, which is not good for performance, particularly dynamic and ribbon mics.
TBH, I would use none of these circuits as a front end.

 

[Winston OBoogie]

Bode plot of number one would help you to stabilise that circuit.

 

[BluegrassDan]

Just updated the schematic for V2.

I also have tried it without the cap between secondary and grid. Makes for a flatter low frequency response.

 

[abbey road d enfer]

Just updated the schematic for V2.

I also have tried it without the cap between secondary and grid. Makes for a flatter low frequency response.

The cap acts as a HP filter against the low-ish input Z of the VE stage (wet finger says about 5k).
Do you mean flatter as in less-attenuated?

 

[BluegrassDan]

Yes. I did notice the value of that cap affected the HP filter point.

 

[abbey road d enfer]

Yes. I did notice the value of that cap affected the HP filter point.

IMO you should abandon this avenue. Voltage-to-current NFB decreases the input impedance. You will not benefit from the free gain of the xfmr.
You should pursue with Plan A.

 

[Winston OBoogie]

On plan/option one: That only 9dB of feedback causes oscillation says to me that you need to look at the phase shifts in the input transformer.

 

[5v333]

i wonder if you wouldnt throw out almost as much gain just from removing the bypass cap...

cathode follower already has alot of NF

 

[BluegrassDan]

It's not oscillating now that I've moved things around physically. But it does lead me to believe that it is overly sensitive and could be on the verge.

 

[BluegrassDan]

Here are a couple sweeps. Max gain, no attenuation, feeding 50mA with 150 ohm input load.

 

[BluegrassDan]

There's one more version. In this sweep, I have added a cathode bypass cap to the NEXT stage, V2 (not shown on schematic). This creates evem more bump up top.

 

[abbey road d enfer]

There's one more version. In this sweep, I have added a cathode bypass cap to the NEXT stage, V2 (not shown on schematic). This creates evem more bump up top.

That's the typical resonance of an undamped xfmr. Good case for a Zobel at the secondary.

 

[solkatten]

Maybe you could implement the feedback this way, simular to the V72A preamp.. Have the feedback derived from the CF to the cold end of the trafos secondary. The 3k3 resistor can be of larger value...

 

[abbey road d enfer]

Maybe you could implement the feedback this way, simular to the V72A preamp.. Have the feedback derived from the CF to the cold end of the trafos secondary. The 3k3 resistor can be of larger value...

What would be the difference? NFB is applied via a voltage divider in both cases.