Understanding "grid drive" to drive, 2, 4, 6 power tubes

[nsubulysses]

I don't knwo why I didn't think of this before but there is an off the shelf 6 power tube output transformer, Hammond 1650WA 280W output transformer made for 6 or 8 power tube. to my surprise the primary is 1900 ohms. I thought it would be lower.

[Jime]

Sorry I can't take on repair work. I would recommend starting a thread if you have some questions and me or other people way more knowledgable than me will answer.

Ok, thanks Nsubulysses.

[nickb]

I don't knwo why I didn't think of this before but there is an off the shelf 6 power tube output transformer, Hammond 1650WA 280W output transformer made for 6 or 8 power tube. to my surprise the primary is 1900 ohms. I thought it would be lower.

That would suggest a supply of about 800V.

[nsubulysses]

I can see some but not all screen grids glow when I crank the amp whether it's a high voltage 580V B+ version with 1K resistor in place of choke or if it is the "normal" version with 485V B+ and 500R dropping resistor in place of choke.

I am just guessing, but wouldn't a 485V B+ with 500R rather than a choke be easier on the screens than your typical Marshall that has same power supply voltage and a choke.

I chose 500R dropping resistor in the 4 power tube 100W version based on my crude screen dissipation measurements to roughly match the 4 power tube 144W with 1K screen grid.
That makes me assume that both versions are equally hard on the tubes, but probably less hard on the tubes than a typical amp with a choke. Totally guessing. Guess I should get a JCM800 and put it on the bench and measure. There is one near so maybe I will.

I have seen PV and other brands use 400R dropping resistor in place of choke for similar B+s that are in the high 400V range, but actually to me 500R sounded better than 400R. 400R had a tiny bit more power but the tone was a little more ragged

[Helmholtz]

The choke version may be easier on the screens if used with a lower capacitance screen filter cap. The better efficience of a LC vs a RC filter would allow for this.

Generally, higher value separate screen resistors are much more effective in reducing screen dissipation peaks than increasing the screen voltage dropper resistor in the power supply chain.

I prefer the lower B+ version because it doesn't exceed the screen voltage limit. I don't believe that tubes only die from excessive power (even though this may be the more common reason) and not from overvoltage(s). There must be a reason for the voltage limits and manufacturers certainly would prefer to advertise higher voltage limits if possible.

[nsubulysses]

Pretty interesting to study and compare.

JCM800 has only 470V B+. It has a choke and I guess someone changed screen grids to 2.2K? I thought they were usually 1K with JCM800

anyway

voltage over screen grid resistor -- voltage at screen -- mA -- W

JCM800 w/ real choke and 2.2K screen grid resistor, Svetlana EL34

At clipping
37V -- 380V -- 17mA -- 6.4W

Full power
47V -- 352V -- 21mA -- 7.5W

JCM800 real choke, 2.2K screen grid resistor JJ 6CA7

at clipping
39V -- 382V -- 17mA -- 6.7W

full power
51V -- 351V -- 23mA -- 8.1W



My 4 power tube 485V B+ amp

1K dropping resistor, 1K screen

at clipping
21V -- 350V -- 21mA -- 7.35W

full power
30V -- 303V -- 30mA -- 9.1W

w/ real choke (hammond 193H), 2K screen grid resistor

at clipping
38V -- 377V -- 19mA -- 7.2W

Full power
56V -- 331V -- 28mA -- 9.3W

It seems I can continue buying a $4 resistor rather than a $40 choke. JCM800 does slightly better with choke and 2.2K screen grid than my 1K "choke" and 1K screen grid but I would think that can probably be chalked up to my "JCM800" 290HX PT from Hammond yeilds about 15V more B+ than some random JCM800 from mid 80s that's laying around here. Also JCM800 screen supply node is 50uF and mine is 110uF

[Gingertube]

I am concerned that guys above have been talking about 500K and 700K maximum grid leak resistance values for the 6CA7.
Those are indeed the values for CATHODE BIAS.
For fixed bias the values are 1/2 to 1/3 of those cathode bias values.
These days, for multiple output tubes, I used the standard PI followed by individual MOSFET source followers, one for each output tube.
Direct couple the output tube grid to the mosfet source (just a grid stop in betweeen), use a good low value (47K or 100K) source resistor, apply individual bias adjustment to the gate of the source follower. For HiFi I "spiff" that up by using current source loads (ring of two transistors) on the source followers.
Cheers,
Ian

[Helmholtz]

The advantage of a lower value screen filter cap won't show in an unrealistic steady state full power situation where screen supply voltage is lowest anyway. But it will cause screen voltage drop faster in a dynamic situation where screen dissipation can be considerably higher.

[Helmholtz]

I am concerned that guys above have been talking about 500K and 700K maximum grid leak resistance values for the 6CA7.
Those are indeed the values for CATHODE BIAS.
For fixed bias the values are 1/2 to 1/3 of those cathode bias values.

Spec says 700k max. for class A/AB and 500k max. for class B. At least class B circuits are not typically cathode biased.

[Gingertube]

Yes - I checked the linked datasheet above. The Class B data was certainly fixed bias but Class B infers very low idle current which in turn infer very low negative/reverse grid current and low bias disturbance.

I would be suggesting 1/2 that (250K max) for Class AB fixed bias if idling at not more than say 70% of max anode dissipation.

If (Class A or Class AB) fixed bias idle current is set for more than 70% of max anode dissipation I would be suggesting 1/3 of that (say 150K max).

A reminder about grid currents if interested: It is negative or reverse grid current we are concerned about in this instance.
https://music-electronics-forum.com/...ad.php?t=32978

Cheers,
Ian

[Helmholtz]

Here is an idea for a simple and accurate method to determine screen dissipation in steady state. It only requires a simple DMM:

The sum of screen dissipation and screen resistor dissipation must equal the DC power delivered by the screen supply. So screen dissipation Ps = Pdc - Pr. Furthermore screen resistor dissipation Pr is the sum of its AC and its DC dissipation components: Pr = Prac + Prdc.
It follows that screen dissipation Ps = Pdc – Prdc – Prac.

So, one only needs to measure screen supply voltage and AC and DC components of the voltage across the screen resistor.
If the value of the screen resistor is well known, accuracy only depends on the meter.

[Mike Sulzer]

Yes - I checked the linked datasheet above. The Class B data was certainly fixed bias but Class B infers very low idle current which in turn infer very low negative/reverse grid current and low bias disturbance.

I would be suggesting 1/2 that (250K max) for Class AB fixed bias if idling at not more than say 70% of max anode dissipation.

If (Class A or Class AB) fixed bias idle current is set for more than 70% of max anode dissipation I would be suggesting 1/3 of that (say 150K max).

A reminder about grid currents if interested: It is negative or reverse grid current we are concerned about in this instance.
https://music-electronics-forum.com/...ad.php?t=32978

Cheers,
Ian

But class B requires very accurate bias because a small drift in the off direction introduces a high level of crossover distortion. My inference from this is that the upper limit on the resistor is about the same for AB and B.

[Helmholtz]

I think what the manufacturer had in mind with class AB and B can be deduced from the typical operation examples in the datasheet. These include "class B" operation at 30mA and even 35mA idle current. Class AB example is for cathode bias.

[nickb]

Here is an idea for a simple and accurate method to determine screen dissipation in steady state. It only requires a simple DMM:

The sum of screen dissipation and screen resistor dissipation must equal the DC power delivered by the screen supply. So screen dissipation Ps = Pdc - Pr. Furthermore screen resistor dissipation Pr is the sum of its AC and its DC dissipation components: Pr = Prac + Prdc.
It follows that screen dissipation Ps = Pdc – Prdc – Prac.

So, one only needs to measure screen supply voltage and AC and DC components of the voltage across the screen resistor.
If the value of the screen resistor is well known, accuracy only depends on the meter.

Nice one

One snag to watch for is that for many DMMs I think you will need a bit of low pass filtering ( I mentioned this earlier) to find the DC current. They often perform poorly on a DC range when AC is present.

[Helmholtz]

Nice one

One snag to watch for is that for many DMMs I think you will need a bit of low pass filtering ( I mentioned this earlier) to find the DC current. They often perform poorly on a DC range when AC is present.

Thanks.

Good additional advice. Seems I am not familiar with non-pro DMMs. But I think analog meters would do as well if used within their frequency range.