6V6 Output Transformer NFB resistor mod / value

[systemtruck]

Hi all,
I’m modifying a Hewlett Packard 201C, which is an old tube oscillator unit with a nice strong PP output stage with a 600 ohm output impedance. Going to delete the whole osc section.
Utilizing just the output stage, by cutting the osc signal and inserting a line input, it already sounds pretty great going into a 600:8 line to voice coil transformer and into a speaker.
There is a negative feedback pathway with just one resistor of 4700 ohms. It’s not a divider, just a resistor in line from transformer output secondary (or maybe it’s a tertiary) and into the cathode of preliminary stage.

The cathode is otherwise not connected to anything else. Thus it is not a voltage divider setup.

I’d like to experiment with making this unit flexible on the fly….. louder and dirtier to use as a guitar amp, but also to retain a somewhat cleaner sound for HiFi playback. Appropriate preamp stages assumed, to be built once I settle this output stuff.

I followed the suggestion of another tinkerer and snipped the feedback line and grounded the now loose end of the NFB resistor so that the cathode now has 4700 ohms to ground. This increased the gain greatly and allowed for excellent distortion/drive/breakup of the output stage where as I wasn’t able to distort it at all before. This is a great guitar amp version. There is some noticeable hum though, as to be expected with all this gain I suppose.

Is the NFB transformer winding bad to leave disconnected? I’m unsure if it is a tertiary or if it is actually just part of the secondary. Schematic makes it look like it’s just the bottom half of a center tapped secondary.

1. Would it be better to handle that now disconnected lead somehow? Ground it?
2. Would it be better to never snip this NFB line and instead put in a “variable” resistor? I’m unsure what the current going through that line is and if any pot can handle it.
3. If I can put a pot in as a variable resistor, how high of a value will effectively create the same extra gain as snipping the line? And is that perhaps better for noise / hum than snipping the NFB line outright?

Ideally I’d be able to have a pot active at all times that was simply an NFB resistor. Great for experimentation immediacy too. Adjust for guitar amp, and adjust other way for clean lower gain. But if current is too intense, I could create a switch with a few beefy resistors.

 

[systemtruck]

Just adding a follow up here..
I inserted a 100K pot in line with the 4700 NFB resistor…i had thought that turning it up in value above 4700 would decrease NFB and thus increase total gain, but it is the opposite… increasing value reduces gain. I clearly don’t understand NFB in this context haha.

 

[abbey road d enfer]

Just adding a follow up here..
I inserted a 100K pot in line with the 4700 NFB resistor…

What do you mean "in line"? Have you connected it as a rheostat in series, or like a potentiometer with one end connected to ground?
A schematic wouold help.

 

[ruffrecords]

Just adding a follow up here..
I inserted a 100K pot in line with the 4700 NFB resistor…i had thought that turning it up in value above 4700 would decrease NFB and thus increase total gain, but it is the opposite… increasing value reduces gain. I clearly don’t understand NFB in this context haha.

The problem with that is that the 4700 sets the dc bias conditions of the first tube so it should not be changed. The way this particular NFB works is that the transformer winding is effectively in series with the cathode. You cannot change that by altering the 4700. You need to alter the voltage across the transformer winding but this is not really practicable with this circuit.

Cheers

Ian

 

[abbey road d enfer]

I even wonder how the amp can work with 100k in the cathode...

 

[emrr]

You can try a pot across that portion of the secondary with wiper connected to the 4k7 for less NFB. Need to consider the Z of the sec, I’d shoot for bridging load, also considering it shouldn’t change the Z to ground substantially. Prob 50-100 ohm range. Turned all the way down there’s no NFB. You can also put a small end stop resistance at the bottom of the pot to give a minimum NFB value.

——edit——
Was reading this as an 8 ohm output, see it’s likely 600 total sec. What is the DCR of each half of the sec? Agree prob 1-2k range pot.

 

[systemtruck]

Yeah I quickly realized that I was totally destroying the bias by increasing that resistor value. Still getting acquainted with tube design here.
Considering that the 4700 cathode bias resistor can be connected to ground to retain about the same bias and that the tube stage / bias performs pretty well (I’ve tested this), I had thought about implementing a variable resistor between bottom of transformer secondary and the cathode pin of tube, so “above” the 4700 to ground. I believe this would give me the same voltage divider NFB design that many amps seem to have. Maybe put like a 2k minimum in series, but a 20k variable resistor, so that nfb value is variable from 2K-22k, and bias resistor remains 4700.

Does that seem like a reasonable approach?

The secondary is intended to be 600 ohm.

I’m curious about EMRR’s idea though.. putting a low impedance pot at secondary to feed the 4700 as it currently is in schematic.

The schematic is in bottom of first post, fyi.

 

[abbey road d enfer]

The schematic is in bottom of first post, fyi.

I meant a schematic of how you connected the pot...

 

[systemtruck]

Here are the two options illustrated…
The first one seems to structurally replicate a lot of guitar amplifier NFB designs, but with an added switch to kill NFB.
The second one seems like it could work too, although i suppose the cathode bias will be affected around 10% -ish by moving the pot around.
I’m wondering if there are any benefits of one vs the other, for AC heater hum. With the NFB line completely broken there is a little bit of hum. However, there’s also a ton more gain. So maybe it’s just that the hum is already there but everything is louder. Haven’t run a proper test yet to measure the change.

 

[abbey road d enfer]

Here are the two options illustrated…
The first one seems to structurally replicate a lot of guitar amplifier NFB designs, but with an added switch to kill NFB.
The second one seems like it could work too, although i suppose the cathode bias will be affected around 10% -ish by moving the pot around.

I would tend to prefer the second option; it would require a low value pot (about 1-2kohm).

So maybe it’s just that the hum is already there but everything is louder.

Very likely.

 

[ruffrecords]

I would prefer the top option because it avoids a heavy load on the transformer. However, it will not work as it stands because the NFB network still affects the dc conditions of the first stage. So you need to introduce a series capacitor, preferable a large value on so as to avoid low frequency instability.

Cheers

ian

 

[systemtruck]

I would prefer the top option because it avoids a heavy load on the transformer. However, it will not work as it stands because the NFB network still affects the dc conditions of the first stage. So you need to introduce a series capacitor, preferable a large value on so as to avoid low frequency instability.

Cheers

ian

That’s interesting. I’ll throw a large electrolytic in series when I try this one out. ( I guess the stock NFB path design doesn’t require this. )

Is this circuit mod something that an average pot can handle? I mean current.. like a typical 1/4 W rated pot is ok or will it get fried? I am not confident about how much current is drawn in a tube NFB path between cathode and transformer secondary.

 

[ruffrecords]

That’s interesting. I’ll throw a large electrolytic in series when I try this one out. ( I guess the stock NFB path design doesn’t require this. )

Is this circuit mod something that an average pot can handle? I mean current.. like a typical 1/4 W rated pot is ok or will it get fried? I am not confident about how much current is drawn in a tube NFB path between cathode and transformer secondary.

The stock NFB parth works down to dc and carries the quiescent current of the first tube. In the proposed change, the quiescent current flows down the 4700 directly to ground. In your scheme the pot and transformer winding were in parallel with this thereby creating a second dc path to ground and upsetting the dc conditions of the first tube. The series capacitor simply blocks this dc route so that the first stage quiescent conditions are determined solely by the 4700 resistor. The ac condition are now determined by the added cap, the pot and the transformer. Since this is an ac path, the signal levels and currents will be small so a1/4W pot will be fine.

To recap. 4700 goes from cathode straight to ground. Cap goes from cathode to pot to transformer. Use an electrolytic cap with the positive connected to the cathode. 47uF 25V should be a good starting point.

Cheers

Ian

 

[systemtruck]

I implemented the one just discussed, with cap and resistor and pot, and it works great. Thanks for the help! This little amp sounds absolutely fantastic, but I’d like to address small bit of 60Hz hum so I’ll be looking into building a DC heater unit.
One other question though.. in that 6SN7 1st stage, there is no cap filtering out DC before it hits the grid of 2nd stage phase inverter. Is that 95V acting as a fixed bias for that 2nd stage?

 

[ruffrecords]

One other question though.. in that 6SN7 1st stage, there is no cap filtering out DC before it hits the grid of 2nd stage phase inverter. Is that 95V acting as a fixed bias for that 2nd stage?

Yes, it is dc coupled to the next stage. It has the advantage it removes a capacitor from the loop and so aids stability.

Cheers

Ian