Bias is bias, or is it ?

[Paul P]

Say you have a 6v6gt, cathode biased through a 470 ohm resistor
and that you measure 22v across it and 48ma through it. I take
it that the grid bias for the tube is then -22v ?

How can it be that to achieve the same 48ma through the cathode
using fixed bias, you would have to supply -51v ? Is the absence
of the cathode resistor sufficient to explain this ? With the tube's
resistance in the K ohms wouldn't the 470 ohms be insignificant ?

Paul P

[loudthud]

Voltages applied to a tube should be measured relative to the cathode. This is what the tube "sees". If an amp is converted from cathode bias to fixed bias, the tube sees increased plate and screen voltages that that is why it takes slightly more (negative) grid bias to make it idle at the same cathode current.

[Old Tele man]

...the effect of existance of cathode resistance ( Rk ) in power tubes is the same as for triodes, it raises both Zin and Zout, but not as dramatically as with triodes because of the much lower effective gain ( Av ) in power tubes as compared with triodes, ie:

rp' = rp + Rk*( mu + 1 )

where:
rp' = Plate dynamic resistance with Rk present
rp = Plate dynamic resistance, without Rk present
Rk = Cathode resistor
mu = Tube amplification factor (triode); effective circuit gain, Av, (tetrode, pentode).

...and, the much lower effective gain ( Av ) is calculated:

Av = dVp / dVg
Av = Zo' / (Rk + 1/gm)

where:
Av = Effective circuit voltage gain (v = voltage)
dVp = Change in plate voltage (output) due to signal
dVg = Change in signal voltage (input) to control grid
Zo' = Effective plate load resistance, with both Rk and reflected load Zoo/4 present
gm = Tube transconductance, amps-per-volt

...the astute tube-afficiandos will recognize that 1/gm is effectively the intrinsic cathode resistance of the tube itself, hence its summation "...(Rk + 1/gm)..." in the second effective voltage gain equation above.

...thus, cathode-bias is "sucking" current through both Rk and 1/gm, while fixed-bias is only dealing with 1/gm...why? because fixed-bias will have higher plate voltage (there's no Vk subtracting from the plate voltage) and, so, fixed-bias requires more negative bias voltage to "throttle-down" the tube to the same idle current (but totally different idle wattage!).

[Paul P]

Thanks loudthud and Old Tele man.

I'd completely missed the fact that the dissipation will be different even
though the currents are the same, so my calculations were sort of
meaningless.

Paul P

[Paul P]

The schematic for an AB763 Deluxe shows the 6v6 plates at 410v and the
bias set to -35v. So for a dissipation of 14w x 70% there should be about
24ma of current through one tube at idle.

In my amp (SE for the moment,and using fixed bias for this question) I have
378v on the plate and even with the bias set to -51v I get 48ma through
the tube (using a new Tung-Sol 6V6GT).

What could explain the difference ?

Paul P

[Satamax]

The schematic for an AB763 Deluxe shows the 6v6 plates at 410v and the
bias set to -35v. So for a dissipation of 14w x 70% there should be about
24ma of current through one tube at idle.

In my amp (SE for the moment,and using fixed bias for this question) I have
378v on the plate and even with the bias set to -51v I get 48ma through
the tube (using a new Tung-Sol 6V6GT).

What could explain the difference ?

Paul P

May be Tung sol 6V6GT isn't a 6V6GT but something else totaly diferent :D

http://cgi.ebay.fr/6P6S-6V6-6AY5-USSR-Tetrode-Tube_W0QQitemZ310014429863QQihZ021QQcategoryZ58174QQrdZ1 QQssPageNameZWD1VQQ_trksidZp1638.m118.l1247QQcmdZViewIte m

[Bruce / Mission Amps]

The schematic for an AB763 Deluxe shows the 6v6 plates at 410v and the
bias set to -35v. So for a dissipation of 14w x 70% there should be about
24ma of current through one tube at idle.

In my amp (SE for the moment,and using fixed bias for this question) I have
378v on the plate and even with the bias set to -51v I get 48ma through
the tube (using a new Tung-Sol 6V6GT).

What could explain the difference ?

Paul PAll tubes are different when manufactured and sometimes, radically. That is the reason we try to use matched pairs or quads when replacing power tubes. At least those "matched" sets are close to each other.
When power tubes are tested for presorting before distribution and ultimately, retailing, ... some power tubes off the line will naturally draw much more idle current or much less idle current then the next run.. if all of them were tested with the exact same grid and plate voltage parameters.
Why?
All I can say is, it is the opinion of myself and other techs I've talked to that these wild variations are the price of modern, mass production tube construction vs the sales staff making sure that all product off the line is sold somehow.
Compared to the Golden Days of vacuum tube assembly, I believe vacuum tube construction and QC standards have been greatly lowered in order to sell anything that works. They might call all these new tubes "within parameters" but those variable parameters are mostly made up to suit the manufacturer.
What does that mean? Some are terrific, some good, some are bad and all
the others are just mediocre with wildly variable idle current parameters.

Under a test set up of using a fixed plate to cathode voltage and the exact same fixed bias voltage, tubes that draw higher idle current are sometimes called soft and will be hot. Those that come in with less idle current are called stiff or cold.
As long as you know which is which and you can alter your bias voltage, you should be able to use most of them in some application or another.

As an example:
I buy 6V6s in large qtys of say, 100 (usually 25 pc quad packs) for my tweed 6V6 amp kits.
The power tube quads (or pairs) will have a number on the pack, meaning every tube in each pack will have matching numbers.
The number on the boxes from my vendor represents their testing setup for me using a -45vdc bias voltage, 425vdc on the plates and 415vdc on the screens.
I have found those numbers are the absolute best for all my cathode biased 6V6 applications using a NOS 5Y3GT rectifier with my custom wound power and output transformers, yet the above described voltages have nothing to do with the actual cathode biased 6V6 amp they are being used in!
This number grouping was empirically derived by me over years of experimenting with those power tubes, a specific rectifier and my transformers.
However, my vendor still makes a mistake here and there by sending me random matched pairs or quads and I have received 6V6s from this vendor with QC number as high as 48 and as low 14!
Although those tubes still work in some other applications, that kind of a spread would be totally unacceptable to me and the majority of them will get sent back because the won't work right in the amp line I spec'd them for.

On a side note, most good techs DO NOT use an "idle current number", such as 30ma for any BF Deluxe Reverb.... as seen many times on the Net (although mostly in the past).
But instead, the tech will do the math using measured idle current vs measured plate to cathode voltage to see where the tube is in idle power.
With respect to the 6V6GTs you mentioned, a power rating using 70% of 14 watts.

The modern made power tubes from Russia, China and some Slovakians just don't seem to be very robust to me, so, I think a slightly lower dissipation rating should be used... unless retubing is no financial concern to the owner/user.
In push pull amps, (PP), I think 70-80% does sound good to most players but many modern made power tubes don't last as long as brand new old stock USA made power tubes so you have to be a little more careful with high idle current power tubes, especially if the plate and screen voltages are at high modern levels.
I think you might find that many pro shops, who offer warranties on parts and what they do, will set the power tube idle current around 50% to 60% most of the time in order to insure a longer tube life and avoid come backs.
Yes, I will set power tubes higher if my customer requests it but I also will warn them of lower tube life and not offer any warranties on power tubes if they want something over 70%.

[loudthud]

Something is not right here. I should have paid more attention the first time I read this. If a 6V6 is idling at 24mA in cathode bias I would expect it to go up to maybe 27mA with the same bias voltage in fixed bias mode. That -51V tells me something else is going on. Either there is a problem on the 6V6's grid like leakage through the coupling cap or the tube starts to oscillate in fixed bias mode. Another wild possibility is that the grid leak resistor is too high a value and the tube is putting out grid current.

Measure the DC voltage across the grid leak resistor in fixed bias mode, it should be zero. Is all 51 volts making it to pin 5? Check the AC volts across the speaker with no signal. You should have a little hum, less than 0.1V but nothing like 5V if there is an (ultrasonic) oscillation present. Do you have resistors on the grid and screen? They can prevent oscillations. That's why they call them "stoppers".:)

378V @ 48ma is 18W. That's about right for 2 tubes, not one.

[Paul P]

Thank you, loudthud ! You've put an end to two days worth of headscratching
on my part.

The bias supply was supplying -55v to the grid leak resistor. I measured the
bias at pin 5 and found -27v. Then I measured across the 220K grid leak and
found 27v. Pondered a bit and voila ! A stupid thing, really : I have a dual
bias supply for when I add a second power tube. I also wired a Y between
the coupling capacitor and the grids of my two power tube sockets for when
I try to run two tubes in parallel. Well, the other tube socket was set to
cathode bias, so it was grounding its 220k grid stopper. So from my -55v
bias, through the 220k grid stopper to the other 220k grid stopper to ground.
Divide by two.

So I rebiased aiming for 10 watts dissipation, which with 390 on the plate
meant 25ma, which I achieved with 29v of fixed bias (now there's a pun..).

Bruce, thanks also for taking the time to provide more information. I always
appreciate stuff like that from those of you that have so much more experience
than I do.

My amp sounds great now. I notice that on a loud chord the current through
the tube goes from 25ma to up to 45ma or so. I presume this is ok ?

Paul P

[Old Tele man]

...excellent job of troubleshooting loudthud, because I too "didn't read well" (obviously) when I spouted off my posting.

[Paul P]

Hey, Old Tele man, I think you answered my initial question, which I asked
in the abstract, very well. And I learned a lot from your answer.

It seems that my idea of driving two output tubes from the same signal,
using a "Y" off the coupling cap, won't work because the individual biases can
talk to each other through the grid leak resistors and the "Y".

So what would be the proper way to drive two tubes in parallel from the
same signal ? Should I use two coupling capacitors ? Would doing that
halve the signal level ?

Paul P

[Old Tele man]

...my answer: do it just like Leo Fender did with his Twin and Showman amps which used "paralled push-pull" tubes for 4 x 6L6 operation.

...it "worked" fine for him, CBS and the current FMIC...so why re-invent the wheel?

[Paul P]

The thing is Fender's tubes all shared the same bias supply. I have two
supplies so that I can bias my output tubes separately. With push-pull
this is not a problem since the tubes each receive their own signal but with
the tubes in parallel, using a "Y" connection also connects the two bias
supplies together, something I'd like to avoid if possible.

Paul P

[Old Tele man]

...hmmmmm, sounds like you purposely want to be able to operate: (a) with two different types of output tubes, or (b) with two rather unmatched output tubes.

...sounds like two separate isolating/coupling capacitors, as you suggested, might be one simple answer...sorta like a "reverse" version of how the NORMAL and VIBRATO channel signals are "summed" just before being fed into the phase inverter circuit.

...I'll search my books and magazines to see what's been written about doing this (odds are it certainly has).