• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Parallel Push Pull Drivers

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm working on a 4 x KT88 Ultralinear Prototype and am having trouble with the driver stage.
I was using 6SN7 cathode followers to drive the output tubes with good sucess but wanted a current sourced diff amp driver. 6SN7 with 22K Anode loads and 8.5mA per triode doesn't cut the mustard. I lose way too much top end (-3dB dropped from >70kHz to 19kHz).
Should I try EL84s triode strapped in the diff amp to get some serious grunt or should I abandon the plan and look at SRPP on each side (after all they were designed to drive high capacitive circuits) or is there some better scheme I should look at.

Once I get this sorted I also want to apply it to a couple of 4 x EL34 Triode Mode Parallel Push Pull Monoblocks I have under construction. Anyone with experience or advice to offer feel free.

Output Trannies in both cases are Plitron VDV2100.
PS I also have one pair of VDV2100-CFB/H to try in the KT88 amp.

Thanks in advance
EC8010 - I removed the cathode follower because I was just a smidge short of open loop gain to apply adequate feedback to get the damping factor I wanted. (I'm using the CAE Currrent Sourced Cascode Diff Amp Front End). Also I'd been reading some reports of inferior sound from the cathode follower (Allen Wright). I also want to take maximum benefit of the VDV2100 Output Tranies ability to keep phase response accurate to greater than 50kHz. i.e. they have a -3dB frequency point 2 octaves greater than that. The EL84 idea was my "Brute Force and Ignorance" approach to that.

WayneM - Thanks for the offer of the schematic using 6U8A. I'm familiar with this circuit (or at least 6AN8 etc. variants of it) and it would have higher output impedance and hence less drive capability than the 6SN7 diff amp I've already tried.

Well, you shouldn't get much voltage loss through a CF, maybe 5% with a 6SN7. And it will greatly help you keep the frequency and phase response better at the high end, as you've observed by removing it. The CF has an advantage, too, of letting the previous stage drive a nice high-R load while letting the output tubes' grid leak to be reasonable low. And finally, the CF lets you cheat a bit into AB2...
Thanks SY,
I will try the current sourced diff amp and add another 6SN7 tube to provide cathode followers after it. The grid leaks are 100K each tube so load is 50K per side + the capacitance which I'm estimating at about 120pF for the Ultralinear KT88s and possibly a bit higher again for Triode strapped EL34s.

That 50K resistance is going to shunt the cathode follower cathode load resitance a fair bit (I was using 47K) and hence I'm not going to get true 100% feedback of a cathode follower - should I actively load the cathode followers?

I think a better approach is to direct couple the CF to the output stage grid (with a stopper resistor, of course). The cathode resistor is brought down to a negative rail and doubles as the grid leak. The grid leak of the CF is reurned to a voltage divider between the negative rail and ground and this voltage is used to set output tube grid voltage.
A bit of a rave

SY - Actually thats a MUCH better idea - eliminates any source of blocking from overloads and keeps low impedance drive of the O/P Tube Grids - also max bias at power up until driver tubes start to conduct
I then loose the ability to balance tube currents between the two tubes on each side - I can still balance the Push Pull sides which is essential for Toroidal O/P tranny - may have to have cathode follower for each O/P tube (c.f. the Norman Koren TENA) OR Cathode Bias the Output Tubes using individual current sources and muck about with cathode bypass and/or "ultrapath" capacitors. (Is "ultrapath" the right term? - I mean caps from O/P tranny CT to the cathodes of the O/P Tubes).

Actually typing this in - as if often the case - just answered my own question. In the original post I also stated that I wanted to try the VDV2100 CFB/H some time in the future. It has a single cathode feedback winding for each of the push and pull sides and about the ONLY way to accommodate it neatly is the individual cathode follower for each output tube.

Since this requires an extra tube section for each output tube I'll have a think about if there might be a better way to use an extra tube section to turn each output tube into a "compound" device, hopefully with lower rp as a side benefit to reduce the amount of feedback required elsewhere. Scott Frankland (I hope I got his name right) talked about this in one of his interviews but did'nt give any details.

Any comments/musings?

Sorry that this is a bit "shotgun at the barn door", all over the place musings, but I find all your comments very valuable in narrowing down possibilities.

If you use a matched quad of output tubes, you should be able to get away with a single CF on each side. And if you want a bit of adventure, you could use source followers; in my amps, I use IRF820, which work quite well.

My only caution (and it's probably redundant to you) is to put a diode from grid to cathode to protect the CF tube on start-up. Regarding CS, I get very nervous about using them on anything but a stage guaranteed to stay in class A and guaranteed never to overload. That restricts them to only the input stages.
Thanks for all your inputs.

My concern arose from trying to achieve (possibly unrealistic)150kHz to 200kHz drive for maximum phase accuracy.

I = C.dV/dt so for 120V pk to pk output tube grid swing at 200kHz (ie in 2.5 us from +ve peak to -ve peak) driving say 100pF you end up with 4.8mA pk to pk drive required. Thats going to need some serious grunt.
Once more into the fray -

The cathode follower has a problem. It will pull up OK (+ve half cycle) but when released (-ve half cylcle) it will just wander down according to the load resistor - that is, we get far different slew rates in the two directions. To properly handle (charge and discharge) the input capacitance of the parallel tubes we need a Push Pull ish driver. At the very least a current source load for the cathode follower to help equalise the slew rate in +ve and -ve directions OR and SRPP or a Mu Follower.

This was what I was trying to get at.

Anyone have any experience with these various methods of driving Parallel Push Pull Output Tubes OR even you SE folks trying to drive that huge Triode.

From my reading it appears that John Camille and mates preferred the mu follower, while Allen Wright preferred the "Super" Cathode Follwer (current source loaded and bootstrapped to keep Vak constant).

All ideas opinions greatfully received. I'm determined not to waste ANY of the Output Transformer bandwidth with inadequate drivers.

Well, it's worth seeing what the chosen idle current of a standard CF will do in terms of charging and discharging the input capacitance of the output tubes. If you want a certain bandwidth from the driver, then it's rather simple to determine what standing current will achieve this- the output tubes don't care whether that current comes through the cathode or from the resistor.
:) Hi everyone

intriguing discussion, i was thinkin about driving problems for 6550 too. I've found this interesting scheme:


If you like SS stuff, this author use an hybrid totem pole paraphase splitter to drive a pair of EL34 t.s. , and an optocoupler (TLP627) to control negative bias, monitoring catode current of the final stage.

I'm trying to understand how to apply this concept to this problem, but this circuit involve a lot of math... so difficult for me!

Any comment?
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.