AB2 power supply rails?

[JazzGuitarGimp]

Hi All,

I seem to remember a post - and I think it was here though an exhaustive search session has produced no results - where someone said that in order for class AB2 driver circuits to work properly, the B+ rail needs to be 3 times the B- rail. I can't remember why this is, and I cannot remember if that advice pertains to valve drivers or MOSFET drivers.

I'd like to experiment with AB2 in my latest build, and I will be using MOSFET drivers DC coupled to the output grids. I don't think it will be possible to push the grids of the 6V6's any more than 25V above the cathodes (at GND). I have a 50V bias tap on my transformer, that will produce plus and minus 70V unloaded and probably +60V under load. Since I am using a MOSET driver, and I only need to get up to +25V, shouldn't I be okay with a 60V supply? I could easily do a voltage doubler on the plus rail to get about 100V under load. But I'd rather not go there if it's not necessary.

All comments welcome.

Thanks everyone,
Lou

[pdf64]

Sounds like Gingertube.
Whilst not specifically for AB2, he outlined such an arrangement to me over on MEF a while ago, see http://music-electronics-forum.com/t32435/#post302238

[JazzGuitarGimp]

Thanks, pdf - that was an interesting read.

I wasn't planning on using a CCS for the MOSFET, but rather, a 24K source resistor. This would put the idle current at (-70V - 35V) / 24,000, about 5mA. My impression is, since the 6V6 grid presents a nearly infinite load impedance at voltages below GND, biasing the driver stage at 5mA should be sufficient to keep the driver stage stable. Then when the waveform transitions to the positive going half cycle, the real work will be done by the positive rail, and the 24K source resistor will pose an insignificant additional load to the source. Of course, I could be all wet here. But what I took away from the article you posted is that you need to keep at least 20V across the MOSFET. I think a positive rail of 60V will easily accomplish this. I expect to see no more than, probably, +15V on the 6V6 grids.

I was planning on using a 100R break-out resistor right at the MODFET source (for stability), then on to a grid stop resistor right at the 6V6 socket, though I have no idea what value is appropriate for this resistor. It seems to me that tube data sheets don't really talk about control grid max dissipation. Also, since I have no experience in class AB2 at this point, I have no idea how much current the control grid may pull at various voltage levels above 0V. I'll need to figure out a way to measure this under AC signal conditions. But even if the control grid pulls 50mA at +25V, that's only 1.25 watts of power. However, Steve Conner's point about burning out the control grids is a bit alarming, and serves as a poignant reminder that I am treading in waters I know absolutely nothing about.

Many thanks,
Lou

[Stevem]

What's your reason for wanting to go AB2 with 6V6s?

[JazzGuitarGimp]

What's your reason for wanting to go AB2 with 6V6s?

Hi Steve, there are a number of reasons:

- I've been interested in AB2 for a number of years now.
- I am poor, and so not able to do many builds - and I am currently in the midst of a 4x6V6 build. This amp also runs a pair of 5881's nicely, and although I've not tried it yet, it should work nicely with a pair of 6L6GC, KT66, and EL34. So I think it's a goid platform for AB2 experimentation.
- I'm curious how much extra output power can be gained by AB2. I am currently getting 40W from the quartet of 6V6's.
- I have heard the distortion in the output stage is more musical with AB2.

Are you thinking 6V6 is not a good choice for AB2?

Cheers,
Lou

[David Root]

The "theoretical" max Po in AB (1 or 2) is 2Pa, so 4x12=48W or 4x14=56W.

Jim Kelly claimed 60W from his FACS amp back in the early '80s. 485Vp, 460Vs. It runs in AB1, no drivers, and obviously pushes the s**t out of the 6V6s.

Haven't seen any 6V6 datasheets showing AB2 but the GE 6L6GC datasheet indicates for a 60% increase in signal from AB1, a 160% increase in power output in AB2, at the same plate/screen voltages and OT primary impedance. This makes sense as if AB2 signal voltage is 1.6 x AB1 signal voltage, then at the same Zpri, Vsq'd/R power is 1.6 x1.6=2.56, close enuff to 2.6.

I built a 4x6L6GC Steel String Singer in AB2 a while ago but haven't measured its power. It has a real beefy sound and one of my customers described it as "a serious gigging amp", which is probably an understatement.

[stephenl]

https://tubeamparchive.com/viewtopic.ph ... ght=driver

[martin manning]

In Class B, running the load line through 200% Pa max and assuming that the anode voltage can be pulled all the way down to zero would produce 4x Pa max in output power on the load line. Of course anode voltage won't go less than 40-50V, and in AB1 or AB2 the idle current is not zero, so the output power (again on the load line) is more like 3x Pa max.

In the AB1 Deluxe Reverb, Fender uses 400V Va and 400V Vg2, and 6k6 Zpri, which runs the load line through 200% Pa max. Load line power is about 32W. A similar load line (7k2) will work in AB2 if the screen voltage is reduced to 250, and Vg1 is driven to +15V. You can see in GE's 1955 6V6GT data sheet curves for 250V Vg2, Vg1 -20 to +15V, where Ig1 is shown for +5 to +15V. There you can see a value of 12.5 mA at the +15V Vg1 knee.

[roberto]

Has anyone continued these tests?

I've played with a pair of 6P6S (russian equivalent of the 6V6GT) in AB2 with 340 B+ and a 8k OT with 23% ultralinear taps.
The SS driver is supplied at +40V, that is +15V maximum for g1 plus 25V to keep the driver input capacitance constant and -80V, that is the expected bias point of -25V minus 40V of swing (that is -25V - 15V, enough with CCS as load on the driver, even if the CCS is not strictly needed) minus 15V of margin. Of course it is directly coupled to 6P6S' g1.

The reason why I used 8k is that the power is almost the same than with 6k6, but the "class A" range is extended, and the tube is always working below its maximum plate dissipation, so cooler and so safer. Power is around twice than what would be in AB1.

Having the PA in UL like come DR.Z amps, the amp need a different voicing on the preamp, because you don't have the effect of the presence control to make the highs "shine".

[Colossal]

I've just completed the bass amp running 4x6550s in U/L we've been speaking about using Martin's 12AU7 AB2 driver. It worked very well and the circuit came right up with no problems. The AB2 rail is +/170VDC rectified. Plate voltage was 468, screens 465V running into a Dynaco A-451 2k2 OT and tubes drawing 48mA at idle. I will report further on the amp over in the thread shortly.

[roberto]

Great! I'm happy to know it works well! Please update thre thread, I'm curious to know how it sounds and the details!

470V x 48mA it's 22.5W so 64% of 35W max plate dissipation. Have you tried increasing a bit and see how it sounds?
I ask you because with 4k4 per pair you are slightly exceeding plate dissipation in a small part of the loadline, so a bit more juice can be taken from the tubes without issues on tubes, if you can get any sonic improvement.

Looking at the curves of the 6550 in that configuration, I'd say that you can go at +5 on g1, but to my hears the direct coupled driver gives more "authority" to the power amp even in AB1. Here below the loadline of your amp (one pair of 6550 is shown).

6550 470V 4.4k 40%UL.png

How many Watts do you get from it? Around 150 clean and 200 distorted?
Tube drivers need more voltage to work correctly (I've used 12AT7 with EL34s and 6L6GC), more heater current, more heat, etc...
That's why I use SS now: cleaner, simple to supply, needs less space, etc... But I am quite prone to blend vacuum and sand, and not everyone likes this approach.

[Colossal]

Great! I'm happy to know it works well! Please update thre thread, I'm curious to know how it sounds and the details!

470V x 48mA it's 22.5W so 64% of 35W max plate dissipation. Have you tried increasing a bit and see how it sounds?
I ask you because with 4k4 per pair you are slightly exceeding plate dissipation in a small part of the loadline, so a bit more juice can be taken from the tubes without issues on tubes, if you can get any sonic improvement.

Looking at the curves of the 6550 in that configuration, I'd say that you can go at +5 on g1, but to my hears the direct coupled driver gives more "authority" to the power amp even in AB1. Here below the loadline of your amp (one pair of 6550 is shown).

6550 470V 4.4k 40%UL.png

How many Watts do you get from it? Around 150 clean and 200 distorted?
Tube drivers need more voltage to work correctly (I've used 12AT7 with EL34s and 6L6GC), more heater current, more heat, etc...
That's why I use SS now: cleaner, simple to supply, needs less space, etc... But I am quite prone to blend vacuum and sand, and not everyone likes this approach.

Hi Roberto,

Thank you for the loadline, that is very useful! I don't want to derail this thread, but I ran into a mishap with the amp but unrelated to the AB2 driver, so the amp is down for the moment. A very small oversight on my part when I first powered up the amp ended up stressing the PT's secondary so I am awaiting the arrival of a replacement. I had the amp operating for about two hours and (with a guitar) it was sounding very wet and lush; just a huge sound actually. Yes, I absolutely noted the physical punch of the amp which I attribute to the driver. I have not tried to increase the current, however. I will put the amp through its paces once I install the PT. It should be here soon. The Thiele TL606 cabs are done and loaded with Eminence CB15 drivers. I will discuss more at length in the other thread. I am pleased with the way it came out though, so far.

Cheers,
Dave

[Stevem]

Sorry this is the wrong schematic I posted!
I will have the right one up later.

[roberto]

Re-reading the thread, I would like to correct some statements that could be misleading:

The "theoretical" max Po in AB (1 or 2) is 2Pa

No it isn't, you can reach way more without issues, it highly depends on the tube.

GE 6L6GC datasheet indicates for a 60% increase in signal from AB1, a 160% increase in power output in AB2, at the same plate/screen voltages and OT primary impedance. This makes sense as if AB2 signal voltage is 1.6 x AB1 signal voltage, then at the same Zpri, Vsq'd/R power is 1.6 x1.6=2.56, close enuff to 2.6.

This comparison is unfair and has not to be taken as a reference for AB1 vs AB2 comparison. The system was simply optimised for AB2 and not for AB1. The same test, with the system optimised for AB1, would have shown the same driving signal and the same output power in AB2 compared to AB1. Is then AB2 useless? No.

A more fair comparison should have been keeping the same plate and screen current, and changing the Raa to fit the two working points.
I add here below two loadlines to better explain the point:

This is 450V B+ with screens at 300V and an Raa of 7k, optimized for class AB1 with around 40Wrms

6L6GC 450Va 300Vg2 7kRaa.png

This is 450V B+ with screens at 300V and an Raa of 3.5k, optimized for class AB2 with around 65Wrms at g1=+15V and around 25Wrms in AB1.

6L6GC 450Va 300Vg2 3.5kRaa.png

[roberto]

Then of course there is another approach: lowering the screens to fit the knee of the curve to the loadline, like in the example below.
You end up with a 6L6GC in PP supplied with 450V on plates and 180V on screens, that needs only around -8V to be biased at 50 mA, and will swing between +15 and -31V to get full power. This is helpful with SS drivers, because you can supply it with a specular +-50V obtained by a simple winding and two diodes instead of an asymmetrical one.

6L6GC 450Va 160Vg2 7kRaa.png

What I haven't seen yet is a plot of the ellipsoid loadline in guitar amps.