Has anyone dabbled in AB2 ?

[JazzGuitarGimp]

Hello All,

I am curious about class AB2 lately. Has anyone here experimented with it? If so, can you tell me:

1. What output tubes you used?
2. How far above the cathodes were you able to get the grids?
3. How does it sound?

Many thanks,
Lou

[JazzGuitarGimp]

Or how about this question: Are there any famous designs from yesteryear that employed class AB2 topology?

Thanks!
Lou

[es345]

e.g. Fender 400PS

[David Root]

Very commonly done using direct coupled cathode drivers in the '50s with higher powered audio amps. McIntosh is a fine example, although they used other refinements to, like C-core OTs. Also used in higher powered bass amps, as already noted. A modern example is the Fender Custom Shop Rumble Bass, see Schematic Heaven for schematic. IIRC uses a 12AT7 driver.

Dumble Steel String Singers use it too. I built an SSS#001 using a 12BZ7 driver, with 1.5 to 2V grid to cathode. Works fine, very clean, very loud, extended bass. 4XMesa 415STR 7581As.

[pdf64]

The Rumble Bass has 47k grid stoppers on the 6550s; I don't see how that can operate in AB2, even though the 12BH7 driver stage looks like a contender?
http://www.webphix.com/schematic%20heav ... e_bass.pdf
Pete

[es345]

Have a look here.

http://music-electronics-forum.com/t28617/

[JazzGuitarGimp]

Hi Guys,

Thanks for your replies. While I have worked in the electronics industry for over 30 years, and in spite of having played through valve amps over the last forty years, I've just become interested in valve amp design over the last four or so years - and certainly, there is still much to learn. But it has become a real passion for me.

It sounds like class AB2 yields more clean headroom, which leads me to believe more output power. This thinking leads me to a few more questions:

1. In class AB2, do you still bias in the 65% to 85% range, or do we need to be cautious about biasing as high as 85%, considering we'll get more power out of the amp?

2. Is it possible to push so much current into the the control grids that damage to the grids is the result? If so, how do you tell how far you can push a given valve? I am thinking a pair of KT66's with direct-coupled 12AU7 CF's.

3. I saved the most important question for last: In class AB2, when the grids are pushed far enough to get into power-stage clipping, does the distortion sound any different than the same output stage in class AB1?

BTW, Hans, your GU50 design is off the hook! Thanks for sharing.

Many thanks for all your help!
Lou

[es345]

1. In class AB2, do you still bias in the 65% to 85% range, or do we need to be cautious about biasing as high as 85%, considering we'll get more power out of the amp?

A little bit background: Using AB2 does not change the behaviour up to the point of G1 getting positive. Going positive the plate voltage will continue to go down. The result is that the peak to peak voltage at the OT is higher than in AB1 resulting in more power and consequently in an increased cathode current. So you need to check if you are still inside the spec of the allowed cathode current
Regarding plate dissipation: The maximum plate dissiption is at about 2/3 of the drive you need to get the maximum peak to peak voltage - and not at the maximum output power.
Coming to your question: AB2 has no impact on biasing regarding plate dissipation. Keep control of the cathode current when you go into power stage clipping. Myself I prefer cold biasing.

2. Is it possible to push so much current into the the control grids that damage to the grids is the result? If so, how do you tell how far you can push a given valve? I am thinking a pair of KT66's with direct-coupled 12AU7 CF's.

Taken into account the known allowed G1 power dissipation of the old comparable transmission tubes I would assume ~0,5 max power dissipation for G1 at a KT66. Now you can calculate and test. From my experience I have limited the max positive voltage at G1 to ~25V.

3. I saved the most important question for last: In class AB2, when the grids are pushed far enough to get into power-stage clipping, does the distortion sound any different than the same output stage in class AB1?

My prediction: the tubes will clip very sharp, the sound might be harsh. The best is to do a smooth clipping at G1 - thats what I have done.

Hans- Georg

[JazzGuitarGimp]

Thank you, Hans. I appreciate you taking the time and sharing your knowledge.

Lou

[pdf64]

I think this Gibson http://schems.com/manu/gibson/ga20rvt.pdf may have been designed to operate in AB2, with the transformer coupled 12AU7 driver and no grid stoppers.
My guess is that plate saturation in AB2 may be smoother than grid clipping in AB1 (which is pretty abrupt); I can't think that it would be sharper.
Pete

[es345]

...a little bit of background. AB2 is known to get more output power.

If you go to AB2, the impedance of the source driving G2 is shaping the overdrive behaviour and is defining the maximum power you can get. Have a look to the pictures I have taken when I have analysed this at a 200W KT120 UL Amp I have built. It is working in AB2 with MOSFet follwer as drivers, the driver impedance is very low. I would estimate less than 10 Ohm. I have scoped the ouput with 2 different gridstopper values.
What you can see is that there is a tradeoff between maximum clean power and soft clipping. If you want to maximize the clean power you will get a very hard clipping.

Looking to the schematic of the Gibson GA20 I expect a much higher driver impedance than 45 Ohm, perhaps 1KOhm or so. This will result in a more smooth clipping but without significant increased power compared to AB1.

[pdf64]

How interesting, thanks.
I didn't expect the plate current to hit a 'brick wall' max like that.
What does the driver waveform at the power tube grid look like - stay unclipped?
Pete

[JazzGuitarGimp]

Hans,

Are you getting 200W out of a pair of KT120's, or a quad?

Thanks,
Lou

[JazzGuitarGimp]

Here is a schematic of the power amp section. Keep in mind that this design hasn't left my schematic capture program, so it is unproven. I have 100R build-out resistors at the cathodes of the 12AU7 and 330R grid-stoppers at the KT66's. I have not yet calculated the output impedance of the cathode followers, so don't know as of yet what that adds to the 430-ohms I have built-in. But even if it were zero, and I wanted to get +10V to the grids, the current would be limited to 23mA by the 430R - and I have a feeling 23mA won't be enough to get the grids to 10V. I also suspect that the bigger limiting factor here is the Zo of the CF's. My goal really isn't to get a significant increase in output power - the amp will easily produce 50W, and that is plenty. I am curious about the difference in the sound of power stage distortion. This particular amp is for a good friend, and while I am more than intrigued to explore AB2, I am starting to think it makes sense to just pull the CF's out and get on with this build. But here is a question, I am assuming that it's just good "design practice" to add the build-out and grid-stopper resistors, though I know I could get more current to the grids if I left them out. Would most designers leave them out? Any obvious indications of instability without them? I have a feeling the 12AU7 Zo would give a nicely rounded clipping, yes?

BTW, this amp is 5E3-inspired, hence the split-load PI.

Many thanks,
Lou

[es345]

Lou,

Some comments

if you decide to go fixed bias and AB2 you don't need R13 and C5, instead 10 Ohm to allow measurement of the cathode current

I have sketched an AB2 circuit below, the main difference.
- put the bias control to the grid of the T12AU7
- change to bias and bias balance, then it is much easier to change from "cold" to "warm" biasing
- now something new coming into my mind: put a stero pot to allow adjustment of the gridstop resistance "on the fly" and consequently the overdrive behaviour.

Hans- Georg