"universal" P-P driver board

[tubelab.com]

Many of you know that 2008 has been a difficult year for me. Layoffs at work, two very ill close relarives, yada yada, yada.....

I was thinking over the Christmas holidays that I had gone through all of 2008 WITHOUT BUILDING A SINGLE AMPLIFIER.
I started a 307A SE amp, but never finished it. I also came to the realization that EVERY amp that I have built since 2000 has been a SE HiFi amp. No guitar amps, no P-P amps, NADA. Talk about being in a rut. What am I going to do about it? Build an amp, a P-P amp, a complete amplifier, chassis, cabinet, stand alone, no external power supply, a real useful amp.

Well, my budget right now is nearly zero, so I got to use what I got, but I got a lot. Last year did not result in any completed amplifiers, but it did host a few experiments, P-P experiments too! The tube sale at AES resulted in some glowing tubes, and some seriously impressive power output numbers from some sweep tubes using conventional grid drive, screen drive, and cathode follower operation. It also resulted in a big box of cheap tubes to experiment on. I recently scored 8 X 6LR6 sweep tubes on Ebay for under $30. I feel the need to make em glow. I got a new power supply that produces 600 volts at 1.5 AMPS! No more power supply limitations. I have a small collection of P-P OPTs in sizes from 10 to 400 watts. Whats missing? A suitable driver for all of this of course.

I have been working on this driver circuit on and off over the past two years. There have been several changes, and new directions during that time. I have now realized that It may be possible to come up with a "universal" driver circuit that could satisfy most, if not all of those requirements. This driver circuit is not done yet, but I better start chronicling it now before I forget some of the important details. Time is limited now, so it may take a few days to catch up to where I am at now. The driver will be used to torture a few tubes, learn a lot of new things, and most importantly, build at least one amp. Hopefully more than one. I will test build (external supplies, no chassis) several kinds of amps based on some of my recent "experiments". I plan to post them all here.

"Tubelab" is known for SE HiFi amps, but long before I had a web site, I built other amps. Yes, solid state amps, tube, and hybrid, guitar amps, and HiFi amps. All of them were P-P except for some "Champ" clones. I built a few HiFi amps, mostly P-P leading up to the 300Beast, and then I got the SE fever.

Well, this P-P driver actually started about 10 years ago with my last P-P amp, the 300Beast. I threw this amp together in a hurry and used some rather cheap components but the sound it produces is amazing. It is the only amp that I have saved all of this time, but sadly, it has died. One of the reasons for this board design is to build a new Beast, so the board must be capable of driving push pull 300B tubes, and it must sound as good as the original Beast.

The Driver board used in the original 300Beast used a 5751 in LTP for the first stage. It used a pair of 6FQ7's wired in SRPP that are also cathode coupled as the "LTP" second stage. Kind of wierd, but it worked good. I never drew up a complete schematic though.

It's late, more tomorrow.

[kenpeter]

Can't even give us the smoking etch-a-sketch scribble version?
I'm having trouble imagining how the FQ7 LTP could be SRPP???

[tubelab.com]

OK, here is a schematic that I found on a hard drive that came from the time frame that the amp was built. This is NOT from the actual design (since it is a single channel), but was likely from a test board that I did during the amp development. I am not 100% sure that it matches the schematic of the 300Beast amp.

I designed this amp about 10 years ago. I didn't know about CCS chips then, so LTP's use "tails" made of resistors off of -300 volts.

The 300Beast is still repairable (the photoflash grade elecrtolytics in the supply are shorted) so I do not want to disturb it until I have built something that sounds as good. It is still my reference P-P amp.

[tubelab.com]

The 300Beast amp has served me well over the past 9 years. It has however been a bit unusual. When I made it, it was an experiment. It was my first foray into "expensive" audiophile tubes. The amp was built using cheap components including some "guitar amp" OPT's and "ceiling fan" coupling caps. It does sound very good, and I have yet to build another amp with the punch and dynamics that the 300Beast delivers. Early on I made several attempts to upgrade the Beast (UTC OPT's, WonderCaps, etc....). Every time I wound up putting it back the way it was when I built it. It has remained untouched for the past 6 years. I decided that I would not touch it until I built something better. So, it is finally time to build something better.

A couple of years ago, SY mentioned some extreme power that he was getting from a single pair of sweep tubes using screen drive. I had read about some commercial screen driven amps (Berning) but never tried it. I had just developed this breadboarding system, so I screwed together a screen drive amp using a pair of $3 sweep tubes, and yes some serious power flowed forth. Screen drive requires some serious drive voltage, hundreds of volts P-P and the driver must source screen current which could be tens of milliamps. The current is not an issue with PowerDrive, but the voltage requirements are not trivial.

I threw together this complete amp breadboard in one Saturday afternoon. I proceeded to torture it for a day or two, and then took it apart in order to build something else. This guy had no problem making 80 watts, and did so without a hint of distress. I knew that I wanted to revisit screen drive some day, so I actually drew a schematic of my creation this time.

[tubelab.com]

More info about this amp can be found at the bottom of this page:

http://www.tubelab.com/6AV5.htm

The schematic is included here.

OK, the new driver board needs to be able to deliver the voltage and current necessary for screen drive.

[tubelab.com]

A question I got via email back when I posted the screen drive schematic was "What happened to the SRPP, don't you like them anymore"? Well The SRPP was in vogue when I did the 300Beast, and it seemed to work well in that design. I have learned a lot in the last ten years, especially in the judicious application of silicon in tube amplifiers, so I no longer believe the SRPP is of use here. Why? PowerDrive.

First we need to analyze what an SRPP is. See the first diagram on the left. The bottom triode is just a typical common cathode amplifier stage. The top tube is a cathode follower that also functions like a CCS. How? The top tube's grid is connected to the plate of the bottom tube through a grid stopper resistor, which can be ignored for analytical purposes. The plate of the top tube is connected to B+, sounds like a cathode follower to me. It would be if the cathode had a resistor to ground.

The cathode of the follower has a buffered and almost identical copy of the signal on the bottom tube's plate. The "magic" happens when we connect that resistor to the bottom tube's plate instead of ground. Now we have a resistor with an almost identical AC voltage on each end. This would mean that the current through the resistor does not change with signal, which is the definition of a CCS. Now the cathode follower is not perfect, so the CCS isn't either. The "Kimmel" or Mu follower improves on the SRPP by using a pentode for the top tube because a pentode makes a better follower than a triode does.

We can seperate the follower from the CCS as shown in the middle diagram. In theory the CCS can be made better by removing the follower duty. It could be implemented with a tube or silicon components. A silicon based CCS will have a lower minimum operating voltage, which is important when we need a bunch of output voltage swing. The follower likewise could be implemented with a tube or a mosfet. A mosfet has a much higher Gm than any tube, and will make a far better follower than a tube if a few guidelines are followed.

When I was exploring the limits of "cranked to 11" guitar amps I learned about blocking distortion. I learned that a heavy enough overload could actually silence the amplifier for a fraction of a second under the wrong conditions. Guitar players called this effect "farting out", and it is generally an unwanted phenomenon, unless you are Neil Young. The short story is that the capacitor and bias resistor on the output tubes grid are to blame, and must be eliminated. See the PowerDrive page for the long story.

http://www.tubelab.com/powerdrive.htm

OK, the circuit on the right can be made out of the middle circuit just by rearanging a few components. This is what I call PowerDrive. In my testing it delivers more output voltage, nearly unlimited output current, and far lower distortion than a SRPP or Mu follower, so long SRPP.

[smoking-amp]

There is another option that is half way between SRPP and Mu follower. Take a look at the IXGH6N170 above the 3C24 in this link:
http://www.diyaudio.com/forums/showt...73#post1707173

This has been referred to as accurate complementary drive or "anti-triode" in previous discussions. The idea being that the top device produces current variation which if summed with the bottom device's variation, would be a constant. (the bottom 120R resistor actually needs to be a little bigger than the top 120R when using finite gain devices on top, to allow for some AC gate drive signal generation) The accurate complementarity is intended to preserve the bottom single ended gain tube's characteristic sound.

The anti-triode....scheme should work well for a non-grid current tube drive since both devices operate with near constant current then.

Also, take a look at this scheme for avoiding blocking distortion:
http://www.diyaudio.com/forums/showt...75#post1689375

Normally Rk1 would = Rk2 to symmetrically neutralize the blocking effect at large signal, but some interesting results may be possible by making Rk2 smaller, so that grid bias will be decreased at some signal level up until g1 conduction restores neutrality by symmetry. This could be used to decrease the grid bias for "efficient class A" operation with increasing signal amplitude. (so it operates with a low idle current without sound signal present, and then increases the idle current to remain in class A with an increasing signal)

Finally, take a look at the 6GF5 for a HV driver tube. 770 V B+ rating, 9 Watts diss (in HO mode). This tube is just a 6GE5 with a smaller plate and stuck in a smaller bottle. The screen voltage won't take that much B+ though in triode, so some kind of "Schaded" feedback to g1 or g2 would be needed for lowering output Z.

Don

[Michael Koster]

Originally posted by tubelab.com
A question I got via email back when I posted the screen drive schematic was "What happened to the SRPP, don't you like them anymore"? Well The SRPP was in vogue when I did the 300Beast, and it seemed to work well in that design. I have learned a lot in the last ten years, especially in the judicious application of silicon in tube amplifiers, so I no longer believe the SRPP is of use here. Why? PowerDrive.

First we need to analyze what an SRPP is. See the first diagram on the left. The bottom triode is just a typical common cathode amplifier stage. The top tube is a cathode follower that also functions like a CCS. How? The top tube's grid is connected to the plate of the bottom tube through a grid stopper resistor, which can be ignored for analytical purposes. The plate of the top tube is connected to B+, sounds like a cathode follower to me. It would be if the cathode had a resistor to ground.

The cathode of the follower has a buffered and almost identical copy of the signal on the bottom tube's plate. The "magic" happens when we connect that resistor to the bottom tube's plate instead of ground. Now we have a resistor with an almost identical AC voltage on each end. This would mean that the current through the resistor does not change with signal, which is the definition of a CCS. Now the cathode follower is not perfect, so the CCS isn't either. The "Kimmel" or Mu follower improves on the SRPP by using a pentode for the top tube because a pentode makes a better follower than a triode does.

We can seperate the follower from the CCS as shown in the middle diagram. In theory the CCS can be made better by removing the follower duty. It could be implemented with a tube or silicon components. A silicon based CCS will have a lower minimum operating voltage, which is important when we need a bunch of output voltage swing. The follower likewise could be implemented with a tube or a mosfet. A mosfet has a much higher Gm than any tube, and will make a far better follower than a tube if a few guidelines are followed.

When I was exploring the limits of "cranked to 11" guitar amps I learned about blocking distortion. I learned that a heavy enough overload could actually silence the amplifier for a fraction of a second under the wrong conditions. Guitar players called this effect "farting out", and it is generally an unwanted phenomenon, unless you are Neil Young. The short story is that the capacitor and bias resistor on the output tubes grid are to blame, and must be eliminated. See the PowerDrive page for the long story.

http://www.tubelab.com/powerdrive.htm

OK, the circuit on the right can be made out of the middle circuit just by rearanging a few components. This is what I call PowerDrive. In my testing it delivers more output voltage, nearly unlimited output current, and far lower distortion than a SRPP or Mu follower, so long SRPP.

Hi George,

I'd like to check another idea in with you. Here's what I'm calling
stacked power drive in honor of your power drive circuit.

It's part gyrator, part follower, and has a fixed bias reference
like the power drive circuit.

I think this may be better than the series balanced drive Don
mentions above if the goal is to drive an A2.

The drive is asymmetric, as is the grid load when driving A2.

The output at the MOSFET source should be well isolated from the
driver tube, allowing the driver to operate on a flat load line.

The grid current loop is local to the MOSFET and grid drive supply.
The grid drive supply can be stacked on the power tube cathode.

It provides stable fixed bias like power drive, but presents a very
low DCR load to the driver anode (1/Gfs) + R like a choke.

I'm putting together a breadboard using this to load my
801-driving-3C24 test circuit. I'm sure it isn't new...

Cheers,

Michael

[kenpeter]

I fixed one of Zobsky's amps with a circuit like that.
Abusing a 10M45 I think it was... Original problem:
the grid of his 12B4 kept forward conducting, and
his 12AT7 just couldn't drive into A2 without help.

His 10M45 was already there as a current sink.
I just moved the tapping upward to the top end
of the current set resistor.

The issue here is that your stacked Mu Follower
drive voltage is too high for direct G1 drive. But
might be OK for G2.

Some kind of capacitive coupling to deal with this
discrepancy before the actual superdrive, if it truly
going to be "universal". Because we skipped this
step, we never got Zob deep into A2. But even
with simple capacitive coupling, it could certainly
handle transients.

Leaning toward a triode + stacked IGBT anti-triode,
driving an IGBT superdrive, driving the output tube.
Less complicated than it sounds...

[kenpeter]

Hey MJK,

IXGH6N170 looked good on paper, how did it work in real life?
You actually managed to get ahold some of them, right?