Cathode bias in 4x6V6 output stage (caps, resistors?)

[SRT]

Hi all,

Just a quick one about the resistors and caps I'm about to order for rigging up a cathode bias in a 4 x 6V6 output stage (loosely based on a Fender Pro Reverb but 4 x 6V6 instead of 2 X 6L6, ~430V B+ at the output tube plates).

I've calculated out that I need a 470 to 500 ohm 5 watt cathode resistor per tube and am going to use a 100uf bypass cap.

Does anyone have any recommendations about sharing the cathode resistor between all 4 tubes? If I understand correctly, all that generally needs to be done is multiply the resistor value by the number of cathodes (thus giving me a 1.8k to 2.2k resistance). To me, this seems like the best way ahead from a layout standpoint rather than having 4 big resistors and their bypass caps.

Please feel free to chime in if I'm way off base or if you have some useful input!

--SRT

[pdf64]

No, for a shared cathode resistor divide the single tube value by the number of tubes, eg 500/4 = 125 ohm shared cathode resistor.
However, I prefer a resistor for each cathode. It allows tubes to settle at a bias voltage they're happy at, not being forced to a common bias voltage, and generally unmatched tubes settle at pretty much the same current level, ie the tubes tend to match themselves.
Plus it makes the amp more robust, eg with a shared resistor if a tube fails then the current drops as does the does voltage, resulting in the remaining good tubes running hotter.
Plus the individual bias resistors can be lower power types, 2 or 3 watt types can be used as only 1 watt is being dissipated at no-signal.
Plus the bypass cap value / frequency breakpoint can be calculated; with a shared value it's next to impossible as it's signal level dependant.

[SRT]

Many thanks! Reviewing my notes, you are most certainly correct (I mustn't post so late at night).

It's good to know about the wattage of the resistors though I've got a selection of 1% 5 watt Kiwames that should do the trick regardless. I'm not necessarily hurting for space (I'm using an old Fender Twin chassis) but Iwas hoping to avoid the clutter.

Thanks again!

--SRT

[Gingertube]

I have a 4 x 6V6 parallel push pull Guitar Amp I built. I have switchable fixed/cathode bias for each output pair plus pentode/triode switching for each pair.
I looked up my build notes.
In cathode bias I used a common 470 Ohms 5W for each push pull pair with no bypass caps
(that is, you pair one tube from each of the push pull sides, don't try to pair the tubes on one side).

Set to work notes:
HV Rail +345V
Pair #1 - volts generated across common 470 Ohms = 23.9V, Tube idle currents were 25.9mA and 25.3 mA
Pair #2 - volts generated across common 470 Ohms = 23.7V, Tube idle currents were 25.2mA and 25.3mA

So the tubes were idling at about 9 Watts dissipation - right on the 70% of max rating which is where I like to bias guitar amp outputs.

For a +430V Rail a common 470 Ohms would result in dissipation per tube approaching 12W. I think that is a bit high although I wouldn't hesitate to use that in a HiFi Amp.
You may wish to:
Use 560 Ohms common on each pair
Or use 1K to 1K2 on each tube

The tubes I used were vintage "Coke Bottle" shaped 6V6G from the Sydney Australia AWV Factory - probably mid to late 1950's tubes.

Output Tranny I used was a spare 50W Marshall, Raa 3K4 (which is a smidge low, would have been better with Raa = 4K3)

Is it any good? - well I think its drop dead gorgeous. You still get some of that 6V6 "grind" but it is smoother than a single pair.

BTW - I did not use bypass caps as I wanted to destinguish the sound between the fixed bias and the cathode bias switch settings. With bypass caps I found the sound in fixed bias and cathode bias was very much the same and there really was'nt much point in having the fixed/cathode bias switches at all. Now I find that one of the sounds I like is one pair fixed bias pentode mode and the other pair cathode bias triode mode.

Hope this is helpfull,
Cheers,
Ian

[jimboyogi]

After reading this last post from Gingertube about leaving out the Ck's. Just wanted to add that this only really works if you are using a shared Rk for each push-pull pair, or shared for all 4 tubes. If you are using a separate Rk for each tube then you will need to use a Ck for each or else you will lose too much amplification.

[MWJB]

SRT wrote: "I've calculated out that I need a 470 to 500 ohm 5 watt cathode resistor per tube and am going to use a 100uf bypass cap."

Not with 430vdc at the plates (to ground) you don't, you want ~800ohms per tube, I'd go for 390ohm per pair or 200ohms if shared by all 4 tubes.

Voltage accross the cathode resistors will be in the region of 30vdc...30*30=900.

900/800ohms = 1.125W a 3W resistor would seem to be adequate.
900/390ohms = 2.3W so a 5W resistor will be adequate.
900/200ohms = 4.5W so use a 10W resistor.

Use 100v rated bypass cap(s) and keep them away from the cathode resistors.

[SRT]

How did you calculate that out? We're obviously using different equations (though admittedly I'm going by some info gleaned from a local old-guy whose memory may be failing). FWIW, I'm using JJ 6V6S tubes.

[kleuck]

The Valve Wizard -Push-Pull

[SRT]

The Valve Wizard -Push-Pull

While this helps understand tube curves and where the bias point should be, it doesn't help me choose a value for the cathode resistors (unless I'm being immensely dense and just not seeing it in the text) because the equation I'm using obviously ends up with different values than MWJB.

Anyone care to chime in?

[Gingertube]

SRT,
Using that article text (EL84 example) it states that bias will be approx -12V.
You can see that from the graph. The Green spot is the chosen idle point. Extrapolarting between the -10 and -15V grid voltage curves gets you that -12V value. Anode current at that point is 30mA.
Cathode current will be Anode Current + Screen Current but you can normally ignore the screen current contributuion as it is much less than the anode current.
THEN
From Ohms Law Rk = 12/0.03 = 416 Ohms. That would be per tube.
So for individual tube cathode bias use say 390 Ohms or 2 x 820 Ohms in parallel (for 410 Ohms). Power will be 12 x 0.03 = 0.36 Watts (at idle) use 2W resistors
Therefore you would want 416/2 = 208 Ohms per pair, use say 220 Ohms. Double the power so use a 5W resistor
This gives you the method, just adapt this to 6V6 curves.
Hope this helps
Cheers,
Ian

[MWJB]

SRT I used a little known method called "practical experience" ;-) Actual measurements taken from an amp: "Biased to 35mA @ 395v (424v B+, less 29v @ cathode, Rk = 330ohms)". This gave 13.85W idle dissipation, if your plate voltage is higher, go for the 390ohm per pair, I'd aim to keep idle current over the 30mA mark. If after measuring, your plate voltage is 420vdc or lower, then 330ohms per pair will do. 270ohms per pair up to 380vdc from plate to ground. Measure, don't assume or calculate, your calculator has not batched the power tubes by their differing idle characteristics.

At 430vdc on the 6V6 plates, with a 470-500ohm cathode resistor per tube, your 6V6s will be dissipating over 20W each...they can handle that in a SE Champ, but not in a P-P amp.

DO NOT under any circumstances use 208/220ohms per pair for the 6V6s at 430vdc from plate to ground, the amp will just get very hot & eat tubes, probably won't last a set without redplating.

[kleuck]

First, you must tell the OT's primary impedance

[MWJB]

SRT kicked off the thread with: "(loosely based on a Fender Pro Reverb but 4 x 6V6 instead of 2 X 6L6, ~430V B+ at the output tube plates)."

So given the lack of any more specific information we may assume a primary Z of 3.6K to 4.2K (or 7.2K to 8.4K per tube). Even if it were 5K or 14K per tube, the tubes will only take so much dc current before fading/redplating due to excess heat.

Even so, idle dissipation is more directly linked to idle plate voltage, under 30mA at 430v from plate to ground the tubes may start to sound thin & sterile, at more than 35mA per tube you will exceed 100% plate dissipation, they may take 16W or so, but why push them if we still have to choose a cathode resistor?

Primary Z is primarily determined by the sound you are looking for, not by idle current which is to some degree a moveable feast (you can't predict idle current to any accurate degree, individual tubes vary), idle current is fine tuned based on actual operating conditions found in the amp.

[kleuck]

Even so, idle dissipation is more directly linked to idle plate voltage

You're right, but let's calculate.
We want a class AB bias, so let's go for a hot 80% idle power = 11,2 watts (assuming 100% is 14 watts).
Assuming the plate to cathode voltage is about 400 volts, we find a 28 mA plate idle current.
With the JJ 6V6S datasheet, this leads to a bias voltage of -35 volts (approx) so plate to cathode voltage should actually be 400 volts, and G2 current will be about 4 mA, so we need 35 volts with a total idle current of 32 mA >> 1K for one tube (500 ohms for a pair, and 250 for a single cathode resistor)

[MWJB]

We don't really want an idle current of 28mA (parts/tube tolerances could see us drop well below 30mA which would be tonally undesirable)...we want a little over 30mA, we can bias hotter than 80%. G2 idle current will be negligible in a design based on a Pro Reverb...<1mA?

Calculating gets you so far, then you have to measure & assess what you actually have. Interestingly, I know a guy who primarily builds 4x6V6 amps that run ~430vdc plate to ground/30v accross the cathode resistor. He actually uses 390ohms per pair, his amps are reliable and well regarded.