Dual bias vs single

[daz]

I dunno guys, maybe it was placebo. But i haven't had a chance yet to test it at volume where it seemed to make a difference for more than a few seconds literally so i'm not even sure. Should be able to crank tonite tho.

[Chuck H]

For those of you who are wondering about SGM's post... I think his implication was that there would be inevitable phase cancellations because of the shared node of the grid bias resistors. This ignores ideally VERY low impedance of the bias supply. But SGM did mention in another post that increasing the bias supply cap value could improve bottom end. So...

I surmise that SGM has either built or dealt with circuits that weren't idealized in this area. And, accurate to his observation it is necessary to use a capacitance value on the bias supplies dropping resistor that presents a low impedance to well below 50Hz. Since any reasonably designed bias supply does this, and most bias supplies are reasonably designed, SGM's observations on this can be ignored most of the time.

[soundguruman]

Unless you change your bias filter cap to 100uF, which seems to be popular.
It causes the low frequencies to distort...and boom...

But as I said, people who don't understand (or don't care) ignore it.
People who are trying to preserve sound quality pay attention to it.

Further, your test with O scope is not making you appreciate it.
Test must be a guitar.

So yeah, you actually do have some control over low frequency with the bias filter cap.
And this seems to happen more often in Fender amps.
But it would apply to Marshall.

Ever wonder why Marshall is using 8uF in Plexi bias circuit? That may be why.
Why does fender use 70 instead of 100uF?
You can really hear a difference, if you can hear...

[Chuck H]

Certainly frequencies are affected by that cap value. I agree with you on that. My point is that in a reasonably designed circuit it should be utterly insignificant. BUT... Looking over the Fender BF bias supply design I decided to run some tests on the impedances with a simple SPICE type program. It looks like some models, when adjusted for higher bias voltages to accommodate modern tubes and wall voltages, could indeed result in the chosen cap value being too small to fully decouple the AC at that node! And I didn't expect that. Considering the relatively low impedance of the overall circuit, even discounting the function of the capacitor, I would still think the a affect would be small, but I'd have to listen to it to know how small. I didn't check the typical Marshall circuit but looking at it I would assume it's not a problem. I can see now that under certain circumstances with some of the BF amps it actually could be. FWIW you could always modify the resistances in the Fender amps. Keeping the divider ratio the same and increasing the resistance would effectively lower the knee of the capacitor without causing the possible problems that come with the inrush and charge time of using a really large capacitance. Just thinking out loud.

[daz]

Ok, i skimmed the last couple posts so maybe i'm missing something. But 100uf bias caps? Heres a story for ya....last time i fully rebuilt this board i accidentally used 100uf bias caps and i could not get the thing to bias up right. i forget the exact issue, but i think the bias was roaming all over the place. Something like that. Anyways it was all F'd up. I realized my mistake and put 10's in it like on the previous board and all was well. Sometimes people will question if i really noticed what i said i did...i guaran-damn-tee you that happened. 100UF bias caps did NOT work right at all.

[uneumann]

I didn't quite get it how dual bias could sound different than single one either.

I didn't either, at first, but I find a day or two of thinking about something almost always leads to some new insights.
I'm just wondering if perhaps any difference in tone is caused by exactly what the the two-bias to one-bias change actually does. It changes the bias voltages on the tubes. Assuming the two-bias initial condition has different bias voltages on each tube, at least one (or both) tubes will have a different bias when driven with a single bias voltage. Admittedly, that could be a small difference, but no one has mentioned it here and it's at least as likely to cause a tone change as the value of a bias supply bypass cap. I'm thinking that the test conditions should be normalized... test the same amp with the dual bias controls set to exactly the same voltage (not tube currents). Then switch to a single bias source (keep the same voltage). If the tone change is still there, at least my bias theory can be eliminated.

[elipsey]

The way I explain it to myself (which might not be the best way) is by remembering the effect of changing the value of the bias resistors in a single bias amp. Their values can reduced, to reduce blocking distortion for example, at the cost of reducing the driving voltage to the output tube grids, and introducing some phase cancellation which I think attenuates the signal differently at different frequencies, and also introducing some compression. I perceive a difference in tone when I change these values. At the limit of that range, imagine using bias resistors of infinite resistance. Of course this would starve the grids of bias, but it seems to me like this is essentially how the output behaves in a dual bias PI, because the two outputs are not coupled through the bias resistors.

I'm definitely not an expert here, but that's how I imagine that they could sound different.

[Mike Sulzer]

Any chance the differences in sound single vs. double bias is from differences in grid current during heavy overdrive?

[g1]

Further to what uneumann was saying about the interaction, this circuit is even more isolated than the usual "dual control" bias circuits as it has separate rectifiers, I'm not sure I've even seen that before.
Perhaps there is some small amount of loading effect with single bias that is avoided in this circuit.

[g1]

Any chance the differences in sound single vs. double bias is from differences in grid current during heavy overdrive?

Not to discount this idea, but in Daz's case I don't think there is any chance of overdriving the power amp?
Daz: we are still talking about low volumes here?

[daz]

Not to discount this idea, but in Daz's case I don't think there is any chance of overdriving the power amp?
Daz: we are still talking about low volumes here?

Yes, pretty low. Well before any output OD.

As to the dual diodes to supply each side, i'm not even sure where i got that idea or if i came to it myself. But i will try using one to power both sides.

[GainFreak]

But 100uf bias caps? Heres a story for ya....last time i fully rebuilt this board i accidentally used 100uf bias caps and i could not get the thing to bias up right.

Thinking back now I don't remember building an amp with a non adjustable single bias point. A single non adjustable bias point always seemed to be a very bad idea. Maybe I got that from my Hi-Fi amps. So almost every guitar amp I built so far had at least two separate bias trimpots (some of them 4), most of the time both caps were 100uf without any biasing problems.
Now I have several experiments in mind which I will conduct maybe next week to see (hear) any difference between single/dual and small/high value caps.
Somehow after all those years of reading about and building tube amps this is the first time I see this subject of "tone shaping via biasing circuit" brought up. Will have to hear that with my own years.

[Chuck H]

I probably don't need to mention this, but if you look at some of the vintage circuits it seems possible that the caps may not decouple and filter enough in the LF. In some simulations I ran it seems possible. Many Fenders and Marshalls have impedances between 250 and 350 ohms at 60Hz. Now, that's pretty low compared to the size of the biasing resistors typically used so I don't know how much trouble it actually causes. But it's a lot less decoupling and filtering than I would have on any other power supply. Any ripple is probably filtered by the PP circuit. Kudos for jumping in. I'll look forward to your report.

[loudthud]

I simulated a bias circuit once that showed what might be going on here. Daz's circuit does not use a bias tap, it uses a big resistor (220K) from the HV winding. In my simulation the cap (might have been 100uF, I don't remember) hadn't charged up to it's final value after 60 seconds.

[Chuck H]

I mentioned this above WRT increasing resistances in the divider rather than using larger caps. This is a good addition to that. Large feed resistance + large capacitance + virtually no current = shifty bias it seems.