Designing for AB2

[Steve Conner]

I propose a Tube Tech-Argue-A-Thon to celebrate the return of KG! Here's my contribution:

I was never convinced about the benefits of driving audio pentode and beam tetrode tubes into AB2. They weren't designed for it. I always got the impression that you would burn out the control grids (which seem to have a pathetically small dissipation rating in these tubes) before you ever got much extra power out of them. You could probably get the same power boost more easily by raising the screen voltage and staying in AB1.

I got a shock when I first saw that pair of back-to-back diodes clamping the input signal to the Ampeg SVT's power section. I thought at first it was designed to be heavily overdriven right into grid current. But I guess someone at Ampeg might have come to a similar conclusion to what I mentioned above.

However, I think an advantage of the cathode follower circuit is that it might stop the coupling capacitors from pumping up due to the output stage grid current and causing the amp to do what I think of as "farting out". I guess it can do that even if it can't drive the output tubes into grid current. The zener trick described earlier achieves the same thing.

I say if you're itching to drive something into grid current, you should use a hefty triode that's designed for it. Discuss

[frost]

Hello everyone,

Awesome subject for a thread. I'm just curious to see what people's opinions are of the three major techniques meantioned above: 1) direct coupled power tube grids 2) clamping zeners on power tubes grids 3) replacing LTP PI with interstage transformer and tube driver tone wise?

I'm considering trying one of the above on an EL84 amp to get around the "farting out" issue that seems to plague these little guys. Thanks,

-Heiko

[Arthur B.]

how do you propose to "stifferin the cathode?"

That's supposed to be "stiffer in the cathode". I fixed it now.

umm..

no.

cathode (or self) biasing REQUIRES that a resistor be used.

whether or not you are injecting a bias voltage at the cathode does not matter. if i put a positive voltage at the cathode, via a battery, or a zener, then it is NOT cathode biased any longer... it is FIXED bias. whether you put a -ve voltage on the grid, or a +ve voltage on the cathode, the tube doesn't care.

Too true, so what's the difference between fixed and cathode biasing?

you MUST have a pure resistance across which degeneration at DC occurs. that is the absolutely key, critical point. it is precisely the reason why max g1 circuit resistance values are HIGHER for cathode biased tubes vs. fixed bias--the tendency for positive grid current, and subsequent runaway is decreased via the DC degeneration occuring across the cathode resistance.

Your point about degenerative negative feedback is irrelevant. To know such behavior, one would have to perform small signal AC analysis. I mean DC is "flatline". How would you know the behavior around the bias point if it never wiggles around the bias point? Also, don't forget that one can insert DC servo loops in both fixed and cathode biased amplifiers. Negative Feedback isn't exclusive to cathode biased amplifiers.

Arthur,
Funny you should use this term... it's exactly the impression I get every time I read one of your "corrections".

Ray

My comment was not meant to be condescending. I'm sorry if I came across that way. I was just surprised that you were surprised that it took so much power to push the grids into conduction.

I propose a Tube Tech-Argue-A-Thon to celebrate the return of KG! Here's my contribution:

I was never convinced about the benefits of driving audio pentode and beam tetrode tubes into AB2. They weren't designed for it. I always got the impression that you would burn out the control grids (which seem to have a pathetically small dissipation rating in these tubes) before you ever got much extra power out of them. You could probably get the same power boost more easily by raising the screen voltage and staying in AB1.

Actually that's not true. Many power tubes had recommended values for AB2 operation. 6l6GC comes to mind, as well as the KT88/6550. And AB2 (for audio amplifiers) isn't about quantity of power in as much as it is about quality of power.

However, I think an advantage of the cathode follower circuit is that it might stop the coupling capacitors from pumping up due to the output stage grid current and causing the amp to do what I think of as "farting out". I guess it can do that even if it can't drive the output tubes into grid current. The zener trick described earlier achieves the same thing.

I say if you're itching to drive something into grid current, you should use a hefty triode that's designed for it. Discuss

Actually the usual practice is to have the power triode driven by a more conventional power tube, since power triodes have little gain.

[Ray Ivers]

My comment was not meant to be condescending. I'm sorry if I came across that way. I was just surprised that you were surprised that it took so much power to push the grids into conduction.

A half-hearted "apology", followed by more condescension - or perhaps you simply can't see it as such?

Ray

[kg]

That's supposed to be "stiffer in the cathode". I fixed it now.

whether you fixed it or not, you're still flat out wrong.

self bias cannot be used with class ab2 or class b operation.

(p. 547)

i don't care how "stiff" you make the cathode. if it's cathode/self biased, then it won't be pulling g1 current.

Too true, so what's the difference between fixed and cathode biasing?

i just illustrated it to you.

cathode/self bias will exhibit degeneration at DC. as cathode current rises, cathode voltage will fall, and vice versa.

Your point about degenerative negative feedback is irrelevant. To know such behavior, one would have to perform small signal AC analysis. I mean DC is "flatline". How would you know the behavior around the bias point if it never wiggles around the bias point? Also, don't forget that one can insert DC servo loops in both fixed and cathode biased amplifiers. Negative Feedback isn't exclusive to cathode biased amplifiers.

in that case, IF the characteristics of the "black box" surrounding the tube under test EXACTLY duplicated the characteristics of a truly cathode biased tube, then you're right. the tube doesn't care. if it looks EXACTLY like cathode bias, then it IS cathode bias.

however, this is not where the discussion began. it began with your refutation of my assertion that a zener between cathode and ground no longer constitutes cathode bias. it does not, and you are wrong if you think it does.

furthermore, i challenge you to create a servo-based fixed bias amplifier that EXACTLY mimics a cathode biased amplifier. possibility does not indicate probability, and such a circuit would have a absolutely ludicrous parts count when compared to the TWO parts needed for cathode bias (or ONE if you don't bypass the Rk). this circuit would be far more complicated than a zener hanging off the cathode, which is what got us into this conversation (and still does not constitute cathode bias, despite your arguements).

lastly, if you DO manage to create a working model of a "fixed bias" tube stage that EXACTLY mimics the behavior of a self/cathode biased stage, look carefully at what you have just done: you have created a stage which has degeneration at DC, which is how i have defined self/cathode bias in the first place (not accidentally, i might add).

ken

[Satamax]

no, it does not. grid current and cathode bias do not get along due to the degeneration across the Rk

OK Ken, thanks a lot!

[Satamax]

cathode/self bias will exhibit degeneration at DC. as cathode current rises, cathode voltage will fall, and vice versa.

a zener between cathode and ground no longer constitutes cathode bias

Hi Ken, about the first phrase, huh, i thought from Ohms law, that as cathode curent raises, voltage drop across the cathode resistor augments
so cathode voltage gets higher above the ground. No?

And about the zener in the cathode circuit, i understand that it's not "usual" cathode bias, but cmon it's in the cathode circuit, how would you call that then? fixthode bias

Bye.

Max.

[kg]

Hi Ken, about the first phrase, huh, i thought from Ohms law, that as cathode curent raises, voltage drop across the cathode resistor augments
so cathode voltage gets higher above the ground. No?

yeah, that sounds a bit weird. you've got the idea though.. as Ik rises, the voltage drop across the Rk will increase, and that means that the ground-referenced g1 will appear "more negative" wrt the cathode.

And about the zener in the cathode circuit, i understand that it's not "usual" cathode bias, but cmon it's in the cathode circuit, how would you call that then? fixthode bias

i understand the difficulties with the terms. unfortunately "cathode biased," in just about every reference i've come across, is used interchangably with "self biased."

you do undertand the difference, though? look at it from purely the tube's perspective.. if an increase in Ik means an increase in Vgk (ie degeneration at dc), then it's self biased. if it does not, then it is either grid biased or fixed biased. looking at it this way should keep things simpler to understand.

[Steve Conner]

I bet you I can create a cathode biased amp that does draw grid current. All I need to do is disconnect the plate and screen voltages, and the only thing that can possibly happen is for the grid-cathode to act as a diode and conduct positive half-cycles of grid current through the cathode resistor to ground.

Now that is an extreme example, but I bet if you drive the tube hard enough, you can make the same thing happen with the plate and screen voltages present, since these voltages will sag down on peaks when the tube is conducting strongly. So I don't see cathode bias and AB2 being mutually exclusive.

I'm still going to stand by my position: The main selling point of beam tetrodes and pentodes was that they cranked out loads of power efficiently and with lowish distortion, without needing to run any grid current. That made it very easy to design drivers for them at audio frequencies. A tetrode or pentode tube that needed grid current, and the complex driver circuit that goes with it, to achieve its full rated output would just be missing the point.

[kg]

So I don't see cathode bias and AB2 being mutually exclusive.

well, you'll just have to disagree with fritz, then.

notice that the RDH does NOT say anything about a class A2 case. the reason why is because it's not about grid current.

cathode biasing falls apart when Ik(avg) becomes far greater than Ik(idle). this is exactly what happens in a AB2 case. biasing the tube cold enough for AB means an Rk that has too much resistance to be able to contend with the massive swings in Ik without severe degeneration. when you reduce Rk so that there isn't a huge voltage drop across it when Vg1(pk) occurs, you fall out of ab2 and into a2--IF your tube can handle the heat!

now, is there an absolute point at which the tube is cathode biased, happily running in a hot ab1 case (ie, vox ac30), and the thing just stops working as soon as Vgk becomes positive? no. but what you will see are increasingly large positive variations of Vk. is it possible to pull a little grid current, remain in ab, and have cathode bias? the laws of physics don't prevent it. as in your extreme example, pull the grid to 100v above the cathode and it WILL draw current.

BUT are you operating with any sort of efficiency? if the cathode ends up rising 60v as you pull the grid to 100v, what have you gained? also keep in mind that any lack of symmetry in terms of plate current swings will tend to be cancelled out in a PP OPT. since we're talking about ab operation i'll assume we're talking push pull. if the plate current doesn't fall as fast as it rises on the "other side" then the swings will be averaged out.

if you're going through the trouble of adding current sourcing capacity to your driver stage, then you might as well take steps to avoid throwing out all of your hard work... switch to fixed bias and get rid of as much cathode impedance as you possibly can, and take every step you can to retain symmetry between your push and pull halves.

fritz comes from the standpoint of retaining as much fidelity as possible, which is not always the goal of MI amps. as a result some of the "impossibilities" are relaxed. still it's important to know why they were there in the first place. the achillie's heel of self bias has always been the tendency for Ik(idle) to be less than Ik(avg), and this is ALWAYS the case due to the characteristic of the gm curve. the greater the disparity between Ik(idle) vs. Ik(avg), the worse your self-biased stage will perform.

I'm still going to stand by my position: The main selling point of beam tetrodes and pentodes was that they cranked out loads of power efficiently and with lowish distortion, without needing to run any grid current. That made it very easy to design drivers for them at audio frequencies. A tetrode or pentode tube that needed grid current, and the complex driver circuit that goes with it, to achieve its full rated output would just be missing the point.

i agree that the much higher sensitivity and efficiency of BTs and pentodes freed designers from a lot of the constraints of high power triodes: with low mu triodes you could get away without grid current, but you needed gargantuan voltage swings, and with high mu it was just the inverse. with bts and pents if you want more power you just add more in parallel, with only a slight penalty in terms of driver requirements.

[Ray Ivers]

Ken,

I recently did some 'symmetrical-clipping' experiments using a single cathode-biased 12AX7 stage and driving the grid real hard using a MOSFET, and I found myself continually reducing the Rk value and having it work better and better; less tail-chasing against the Vk rise, as you pointed out. I concluded I'd probably have to go fixed-bias/direct-coupled to really do it right, so now it's on the shelf awaiting some free time.

The results were promising, though - the waveform I was getting could have come from a push-pull output stage with very poorly matched tubes. I was getting visible overdrive with a lot of audible distortion, and what distortion I was getting I liked a lot better than regular preamp-tube clipping.

Ray

[Ray Ivers]

Heiko,

My advice would be to try the Zener thing first - it's just so much easier to retrofit than the other two, and if you like the sound of your amp now, it will reduce the things you don't like without affecting the things you do, which IMO is the coolest thing about it.

As far as taste-testing the three methods you mentioned (there are others as well), using an identical-except-for-drive-method circuit - sorry, I haven't done this. I can tell you I have yet to hear a well-executed AB2-drive circuit I didn't like the sound of.

Ray

[frost]

Ray,

Thanks for the info. I appreciate it. I think I'm going to try the zener trick on an existing amp first and go from there due to the ease of installation. Thanks again,

Heiko

[Ray Ivers]

Heiko,

Cool - we've got a few posters trying it now, and FWIW I have a real good feeling about the outcome, especially considering BobW's initial report.

Ray

[Smitty]

Ray,

The more you talk about Driver Transformers the more I want to try one in my amp.

Let's say I'm going to use your choke input mod and that I'll have about 310V at the primary. 12AU7 paralleled running about 6ma each side.

I'm not looking for more power. I want to drive the grids just far enough positive to round off the hard clipping I'm getting with a conventional LTP.

I also want to eliminate a some/most of the bias shift by lowering my bias supply impedance. A split secondary would be nice. I like dual bias pots.

I'm in Super Reverb land now and want to end up somewhere in between the Musicmaster Bass and the 300PS.

Is there an existing design spec I can send to Heyboer?

Oh yeah. Would the Zener mod work with a Driver Transformer. There is still a DC component to the AC during clipping, right?

Thanks.

Smitty