Bias settings with higher plate voltages

[alerich]

I think I have asked this before (or some variant of this question) but I am still fuzzy on the answer. Maybe I should rephrase the question. My amp is a Sovtek Mig60 (see schematic). It is EL34 based and I have Tung Sol EL34B tubes in it currently. The plate voltage is 640VDC. It has 4 ohm (3.9 actually) sense resistors in each EL34 cathode. The schematic instructs you to set the bias for 80mv across each sense resistor. That yields 20ma of current through each tube. 640VDC x .020A equals 12.8 watts static dissipation per tube at idle. Very conservative. I really dig the tone as it is but I would like to experiment a little with different bias settings.

My question is this: Is there any reason to over bias (bias cold) a tube in a guitar amp simply because the plate voltages run higher than normal? Or is this 80mv suggestion on the part of Sovtek an attempt to keep you from pushing the amp too hard lest it fail? The iron in these amps was notorious for taking a dump when you pushed them hard but this amp has had the dodgy iron replaced with Mercury iron.

[phsyconoodler]

Some of the reason is that too much current causes lots of issues beyond just crappy iron.The relationship between voltage and current is a funny one.
  Pressure is voltage and flow is current,so increase the pressure and the valve needs to have the flow reduced so it doesn't blow a gasket.But the pressure on modern tubes is kind of a touchy thing cause they just ain't built the way they used to be.
  And the other question is,do they really sound better with more current?Maybe,maybe not.
I built a 2204 and it sounded killer good at 21ma with 500v on the plates.If I upped the current it didn't sound as good to my ears.The distortion is also a consideration.As the amp drives the power tubes and they start to distort,they kick up the current radically with a true pentode like an EL34.
  So add tons of voltage(pressure) and lots of current(flow) and drive the snot out of them with preamp signal and you have a recipe for diasaster if the conditions are right.
  There is a reason why most amps these days don't do both high voltage and high current.Longevity is one.
A Musicman amp with 725v on 6LGC's is another prime example.Too much current and the failures are catastrophic. You need to regulate the screen voltage so the screens don't melt down too.
  Lots of reasons NOT to get crazy with current in a high voltage situation.

[PRR]

Go 100, 120mV on the 3.9r resistors.

150mV would put you right at the conservative 25W rating.

You don't want to leave it this high because the grid resistors are over-spec (220K where tube specs say 100K), and modern tubes are not all as good as the golden-age bottles. When the tube gets hot, that grid leaks. The large grid resistor takes the grid more positive, increasing current, and it can run-away.

If 100mV-120mV does not sound different tan 80mV, put it back at 80mV.

[alerich]

I tried different settings between 80mv and about 110mv. I couldn't hear a significant improvement in my tone and if anything it got a little mushier and compressed around 100mv so I set it back to 80mv per the schematic. That should make them last a little longer. Thanks for the input!

[HotBluePlates]

My question is this: Is there any reason to over bias (bias cold) a tube in a guitar amp simply because the plate voltages run higher than normal?

We forget that output tubes provide voltage amplification to their plate (because we typically focus on plate current and how that interacts with load impedance to cause the plate voltage variation).

If you bias the tubes to idle at less current, you're applying more-negative voltage.

If you have more negative voltage, the tube can accept a larger input signal before the grid is momentarily driven to 0v on the positive input signal peak.

If the tube can accept a larger signal swing when driven to 0v on the grid, there is a greater difference in plate current between the zero signal condition and maximum signal condition.

This should yield more power output; bigger voltage swing x bigger current swing = more power, assuming the amp is designed to have the proper power supply and load impedance to allow this all to occur.

We forget that an output stage is designed in its entirety to function in a specific way, generally to hit some power goal at a given amount of distortion. Since we usually don't mind gaining/losing a few watts of output and care more about the nature of the output stage distortion, we've all ignored the fact that the designer specified a particular bias voltage to yield the designed-performance. The catch is they had to assume the tubes all had some arbitrary characteristic (plate curves or dynamic curves) which they could use to design around.

[PRR]

> a greater difference in plate current between the zero signal condition and maximum signal condition.

No.

The two idle currents cancel out. The no-signal signal current is zero.

The peak current is (generally) the zero-grid-voltage current of the one ON tube.

The -difference- between zero (cancelled) signal at idle and the peak current does not change with idle bias.

(There is a pathological case where idle current is set higher than half of peak current. Over-rich class A. In this case the "off" tube never goes near off, and that current fights the "on" tube, reducing max signal output. But we never find serious guitar amps anywhere near this point.)

> tube can accept a larger input signal

And for differences like 40V versus 45V, the player can barely tell the difference, will adapt his knobs or fingers to get _to_ the distortion-point more-or-less to-taste.

[HotBluePlates]

I see...

So idling cooler doesn't do anything for an already built amp, because we're not raising the supply voltage and changing the load to actually get more output power.

And I guess there's a (maybe wide) range of values were idling hotter/cooler has no real impact; as you say, the player picks harder or turns up the volume control.

And so above that range, tubes want to redplate; below that range the output stage tends to crossover distortion because a side turns off too early.

Is that more accurate to what's happening?