Interpreting datasheets for class AB2

[cdemike]

I've been trying to learn more about output stage operating points and am trying to better understand class AB2 in particular, especially when it comes to designing a power supply that can support that kind of operation. RCA's datasheets are very detailed in giving information about some tubes' characteristics when drawing grid current, but I'm finding that some datasheets imply class AB2 operation without giving details in terms of the amount of grid current drawn, grid resistance, etc. For example, the RCA 6L6GC (https://www.tubezone.net/pdf/6l6gcrca.pdf) and Marconi KT66 datasheets (https://frank.pocnet.net/sheets/179/k/KT66.pdf) explicitly describes AB2 operation, whereas the Mullard EL34 datasheet (https://frank.pocnet.net/sheets/129/e/EL34.pdf, page D2) only implies it, describing a fixed bias push-pull pair operating with a B+ of 400 and a bias voltage of -36V. Under those conditions, they input a maximum signal of 50V RMS, i.e., 70.7 V(peak). Compare with the Marconi KT66 datasheet, which describes Vg1 = -30 and a peak input signal = 38V. So I'm thinking I either am misinterpreting the datasheet or EL34s can be run in AB2, even though I'm having trouble finding circuits designed to run EL34s under those conditions and am actually finding some instances of people saying EL34s weren't designed to operate drawing grid current. It does give me pause to see only one half of the waveform described in the Marconi datasheet, so I'm thinking there might be convention differences between the datasheets. Moreover, it would seem like a strange coincidence that the V(peak) on the Mullard EL34 datashset is just about double the bias voltage, i.e., could really be describing AB1 operation.

If that is a misunderstanding on my end and the EL34 datasheet does really only describe AB1 operation, it would seem the tubes in common use in guitar amplifiers that can be run AB2 would be 6L6 and derivatives like KT66. I'm surprised to find that I can't find other examples, so it's possible I'm not thinking of some. The practical question I'm trying to answer is how to approach the power supply such that the driver and bias supply are capable of providing enough current to support grid current. I'm also trying to figure out design considerations when choosing drivers' cathode resistor values, which seem to function as grid leak resistors analogues, for example, in a Dumble SSS, especially since that might impact choice of driver tube and corresponding operating point.

As an aside, it seems like it might be an interesting experiment to see what an amp might sound like by using an asymmetrical driver tube like 12DW7 to impart some even-order harmonic content, since the "12AX7" side could be set up to draw grid current at certain signal amplitudes like it does as a directly-coupled cathode follower due to its high plate resistance.

[bepone]

EL34 is not good candidate for AB2 class.. more are transmitter tubes for that.. grid g1 must be sturdy and dissipate some power (1-2W).

Check 2E22 datasheet you can find some nice data there..

G1 on out. tube is pulling from 10-50 mA from the driver tube in AB2 usually - so driver tube must deliver that (driver must be biased hard) - better to use mosfet driver, but from lower voltage suply (+80, -80v supply) to avoid strong dissipation.

Read also for A2 driver here:
http://www.pmillett.com/A2_buffer.html

[Stevem]

Take a peek at a Fender 400 PS schematic.

[pdf64]

The massive benefit of modern pentodes was that they could provide high power output without the added complexity of grid current '2' operation.

For output stage conditions with -36V bias to operate in AB2, the max power input signal would need to be >72Vp-p.
As it's not, we can infer AB1 operation.

[cdemike]

For output stage conditions with -36V bias to operate in AB2, the max power input signal would need to be >72Vp-p.
As it's not, we can infer AB1 operation.

Thanks, that clarifies it. So it was a misunderstanding on my end, and there were two different conventions being used between the Marconi datasheet and the Mullard datasheet.

Take a peek at a Fender 400 PS schematic.

Really interesting design. Cool to see a comparatively modern design using an interstage transformer as a phase splitter. Also interesting to see a power supply that looks so different than a lot of other designs. Am I seeing right that this has a half-wave rectifier for the main B+ rail? This power supply is especially wild after learning how much current G1 can draw in AB2. That PT must be capable of some mammoth current. The MOSFET driver thing makes sense, especially given the wider rail-to-rail swing capability. Definitely having fun learning all the different ways of approaching the same operating condition with the phase splitting transformer on the one hand and drivers on the other. It seems to me to be much more approachable to design using the transformer vs. drivers in terms of having fewer operating points to worry about. Is there a specific advantage to using drivers, besides the ability to further shape the sound viz. operating points?

What is it specifically about the pentode that removes the "need" for AB2 operation? My understanding was that beam tetrodes were invented as a workaround for the Phillips/Mullard patent on pentodes, so if that were the case wouldn't the same obsolescence apply to beam tetrodes like 6L6, despite their use in AB2 applications? Sorry if I'm missing something.

[pdf64]

...

Take a peek at a Fender 400 PS schematic.

Really interesting design. Cool to see a comparatively modern design using an interstage transformer as a phase splitter.
...
Am I seeing right that this has a half-wave rectifier for the main B+ rail?
...
What is it specifically about the pentode that removes the "need" for AB2 operation? My understanding was that beam tetrodes were invented as a workaround for the Phillips/Mullard patent on pentodes, so if that were the case wouldn't the same obsolescence apply to beam tetrodes like 6L6, despite their use in AB2 applications? Sorry if I'm missing something.

Linear operation in the positive grid area pretty much rules out capacitive coupling, which leaves transformer or DC coupling (eg cathode or source follower to drive the output valve control grids) as the only options.

The Marconi KT66 info contradicts this, as it notes that the driver stage is capacitively coupled. I suppose they were able to jiggle the conditions to accommodate the resulting bias shift somehow?
Generally, with capacitive coupling, when signal levels are high enough for the top of the upper lobes of the waveform to move the grid positive, the signal effectively rectifies and pushes the bias more negative, causing crossover distortion and eventually, blocking distortion.
That effect might be mitigated a little by a hotter idle, but it's not usually feasible to facilitate AB2.

The split rail HT supply of the 400PS uses a voltage doubler for the anode feed, and a high power potential divider off that for the screen grids etc. https://el34world.com/charts/Schematics ... _400ps.pdf
I wonder why a standard split rail supply wasn't used (ie with a CT on the winding)? As per the 160PS or the 140 (not that the 140 power amp uses both HT rails simultaneously).
https://el34world.com/charts/Schematics ... _ps160.pdf
https://el34world.com/charts/Schematics ... _schem.pdf

As per most valve manufacturers in the golden age, I use 'pentode' to cover beam power and suppressor grid output valves. The 'beam tetrode' term seems to have become obsolete by the 1950s.
eg TungSol describe the 6550 as a pentode at the top and then qualify it as a beam pentode in the description https://frank.pocnet.net/sheets/127/6/6550.pdf

My meaning in the previous post was that output pentodes were embraced by designers as being a significant technological improvement over output triodes. My understanding is that a key reason for this was whilst output triodes almost inevitably had to be opererated in the positive grid region to realise good power output, pentodes didn't. Hence amplifier designs and construction could become somewhat simpler and their operation more reliable.

What benefit to you perceive AB2 operation might offer?

[Helmholtz]

The Marconi KT66 info contradicts this, as it notes that the driver stage is capacitively coupled.

Mabe you misread?
I'm only seeing "cathode coupled".
As you said, A2 rules out capacitor coupling, because A2 requires a net DC current drive not possible with caps.

[pdf64]

The Marconi KT66 info contradicts this, as it notes that the driver stage is capacitively coupled.

Mabe you misread?
I'm only seeing "cathode coupled".
As you said, A2 rules out capacitor coupling, because A2 requires a net DC current drive not possible with caps.

Yikes, yes I misread, apologies. I noticed the bias voltage magnitude increasing from -25V at idle to -30V at full output, and my brain must've done the rest.

[mhuss]

The 400 PS (and 300 PS, I actually own one of these - anyone want to buy one, lol?) use a voltage doubler scheme to get 700 VDC out of a 245 VAC PT winding. So its actually one step 'worse' than a normal half-wave rectifier circuit. They make up for this by using massive "soup can" 450 or 650uF stacked filter caps and a sturdy filter choke before the output tube screens and downstream.

[Ten Over]

The 400 PS (and 300 PS, I actually own one of these - anyone want to buy one, lol?)

No thanks, I already have one of each. I bought the 300PS new when they first came out. After the second set of overly expensive 6550's blew along with all the collateral damage, I shelved the thing.

[Ten Over]

So its actually one step 'worse' than a normal half-wave rectifier circuit. They make up for this by using massive "soup can" 450 or 650uF stacked filter caps and a sturdy filter choke before the output tube screens and downstream.

I wonder how it is worse than a half-wave rectified circuit. The voltage doubler plate supply is full-wave rectified and those big computer grade capacitors are for this part of the supply. The big computer grade capacitors are stacked in the schematic, but they are not functioning like the stacked capacitors in other Fender power supplies. The screen/pre-amp supply is half-wave rectified, but that circuit isn't pulling anywhere near the current that the plates are pulling. The choke is the same one used on the 100W Fender models as well as others. Those amps were dead quiet at idle.

[martin manning]

Music Man used that same arrangement.

[mhuss]

The 400 PS (and 300 PS, I actually own one of these - anyone want to buy one, lol?)

No thanks, I already have one of each. I bought the 300PS new when they first came out. After the second set of overly expensive 6550's blew along with all the collateral damage, I shelved the thing.

I've been running a 'matched set' of JJ KT88s in mine without issues.

[Ten Over]

I've been running a 'matched set' of JJ KT88s in mine without issues.

Thanks for the tip.

[cdemike]

Really appreciate everyone’s responses! Thanks!