Variac in Guitar amp

[mackie2]

I got out one of my old amps, that I never finished!!  I had installed a 3A Variac in the PT primary, so I could  Adj. the HV. I  installed a 7A  Fil Transformer (so the Fil Voltage never  changes).  I had breadboarded the amp with 6L6  pp output to check the action of adjusting  voltage --  to tone.  The Amp sounded good down to just  over 1/2 Voltage (of course Vol decreases proportionaly with headroom) .And, at lower voltages,  less treble.  Now that vvr has come along ,  I may just reclaim my parts for  another project.

Mackie2

[jjasilli]

Usually 90% supply power is considered the floor.  If you go lower I think you should measure your bias and current draw: @ idle & at max signal.  The point is that as voltage drops, current draw increases and might become excessive.  This could cause premature failure over time, or even sudden catastrophic failure or valuable components.

[tubesornothing]

jjasilli, the way he describes the way he is using his variac sounds similar to the way VVR and Power Scaling is used to scale the entire amp.  How is this different?  Or will those techniques cause the same problem you describe.

mackie2, speaking of which, if V1 is power scaled or in your case "variac'ed", I seem to recall there can be a problem with V1 grid bias and the guitar inert acing and a coupling cap needs to be inserted.  Anyways, something to look at.

[jjasilli]

With reasonably small fluctuations in supply voltage, tube bias will keep pace with the change in plate voltage.  This may not be true with major reductions in supply voltage. 

I don't really know how these things -- power scaling & vvr -- work.  I think they use use active or at least some sort of regulated circuitry to keep bias in concert with plate voltage.  Power scaling can also be applied to just the power amp, so you don't lose your preamp voltage.  OTOH loss of preamp voltage may be desirable for ealrier breakup.

[tubesornothing]

Interesting!  In the power scaling I built, there is a tracking regulator for the bias supply.  There is an optional sag circuit but this only applies after the tracking bias supply, so I think bias stays well regulated.

I have never scaled the whol amp - only the PI and power stage.  But I have heard many other scale the whole thing.  Different strokes.

[PRR]

> as voltage drops, current draw increases

No. Never.

> a tracking regulator for the bias supply.

Because the basic plan only drops the B+ voltage. Do that too far, while keeping bias fixed, the current falls to nothing, which sounds awful.

Mackie2's Variac drops BOTH B+ and bias. He can go a long way, just keeping them in the same ratio.

As we get down below a third of "normal" B+, tube parameter skew means the amp will get unhappy, probably tending to current-starve. KoK's fancier plan's offset the bias at low-low B+ to keep things happy down to very small power.

The Variac has elegance and is bomb-proof. But is heavy, especially since you need another PT to light the heaters. The siliconized plans are lighter and cheaper, and can be cleverer.

[mackie2]

PRR--

My experiment with the variac was several years ago--actually,  at first, I used my variable  power  supply with dc Fil, to gather a little info.  I was intrigued by the London Power Var Pwr when it hit the market.  That's when I breadboarded an amp with the variac. My amp then, was self biased.  I have junk box with variacs.  I use a 20 amp  Powerstat about everyday on my work bench.  The VVR wasn't out then.

Mackie2

[jjasilli]

JJ:   as voltage drops, current draw increases

PRR:  No. Never.

More transformer confusion.  If we use a tranny to step-down the voltage by 1/2, do we not double the current?  I was under the impression that power scaling uses some form of regulation to knock-down voltage, and burns away excess current as heat energy, thereby needing some form of heat-sinking.  Maybe I have this all wrong.

[FYL]

if V1 is power scaled or in your case "variac'ed", I seem to recall there can be a problem with V1 grid bias and the guitar inert acing and a coupling cap needs to be inserted.

Yup. There can be some DC appearing on the grid at very low B+.

[jjasilli]

JJ:   as voltage drops, current draw increases

PRR:  No. Never.

More transformer confusion.  If we use a tranny to step-down the voltage by 1/2, do we not double the current?  . . .

It dawns on me that the extra current might be felt by the variac, but not by the amp.

[PRR]

> use a tranny to step-down the voltage by 1/2, do we not double the current?

ONLY if you are "forced" to draw constant power.

Assume "full" is 300V 0.05A (a hot Champ).

Then "half" will be 150V at probably 0.025A. At least for self-bias, half the B+ will cause about half the current.

Note that half the voltage and half the current is a QUARTER the power.

Ignore some rectifier factors. "Full" is 300V 0.05A, and this is the transformer load. "Half" is 150V at 0.025A; if we get this from a perfect 300:150 transformer the current at the 300V point is 0.0125A.

PRR:  No. Never.

Before I get jumped: there are "negative resistances", in which current can go up when current goes down. They are very rare.

[PRR]

Here's a tube stage, 300V varied down to zero, biased three ways.

Self-biased

FIXED biased with an unvarying G1 voltage

G1 bias voltage proportional to plate and screen voltage



Current always falls.

For the FIXED-bias case (which could be VVR on the B+ but not on the C-), the current falls very quickly.

For the Self-bias case, current stays nearly proportional to voltage, with a small curve at the bottom. This is nearly true for the proportional bias case. This is very nearly Mackie2's case.

The V/I ratio is approximately a "good load impedance". We don't have an easy way to change load impedance, so we would like it to stay near ~~7K. The Self and Proportional cases do well down to quite low voltage. The VVR with FIXED bias goes way off-course.



The VVR/PowerScale gurus know this: some first-approximation correction of grid voltage is essential, or the amp just konks-out abruptly when the knob is turned off max. Simple Proportional works fine to perhaps 1/3rd of normal voltage. It gets a bit trickier as you pass 1/10th of normal. I'm sure the gurus of the technique have experimented for best-sound compensation at absurdly low plate voltage.