The Dark Arts of Direct Coupled Cathode Follower Drivers!

[priesty]

Over the last few weeks I’ve been delving into the confusing world of direct coupled post phase inverter cathode follower drivers for push pull amps. The most obvious example of this in a commercially available amplifier would appear to be the big ol’ Ampeg SVT with 12BH7’s (and latterly 12AU7’s) fulfilling this role.

I’m rather perplexed to the way Ampeg set up the positive and negative supplies to this tube. Referring to my sketch below, it would appear that the potentiometer P1 varies the voltage at the grid of the CF triode, but how does it alter the voltage at the cathode of the driver (and hence the grid supply of the power amp)? It would also appear that with a -30V grid setting that the 12BH7 would be biased as cold as ICE! How can the power stage be properly biased with this set up? It obviously works, but to me it appears that the 12BH7 triode is in a voltage divider with a 47K resistor and a -150/+350 supply!

I’m perplexed and it’s bugging me a bit! I even bought TubeCAD, but it hasn’t really answered questions, just posed some more!

[octal]

Well, since no one more expert has bitten on this one, I'll take a stab. I wonder if those driver tube voltages could be wrong. Usually the cathode voltage on a DC follower sits quite close to the grid.

Here's an example of an amp with DC coupled cathode followers and voltages that make sense. Remember, the difference in voltage between the grid and cathode of the driver tube (cathode follower) determines its bias, not the absolute voltage per se.

http://www.triodeel.com/heathw6.gif

Nathan

[kg]

like nathan, at first it looked odd to me too.

but then i took a look at the datasheets...

100v across 47k gives a cathode current of around 2.1ma

a 12bh7 DOES need somewhere around -30v bias @ 450vak to keep plate current in that range. believe it or not!

that's what you get for using a triode for a follower... there are better choices.

http://ken-gilbert.com/images/pdf/BA...iverschema.pdf

^^ my design, which is admittedly a bit more complex, uses sv83 (or el84) as followers. the cathodes are loaded by constant current solid state devices. large value resistors may be used (and that's how i started off). also notice that the followers are connected in true pentode mode, which allows for much better performance. i see a bias voltage of around 5v on them.

the input valve/driver can be any higher current triode. in my case i use 6n1p.

edit: found an older schemo: the followers are triode connected, and use a resistive cathode load:

http://ken-gilbert.com/images/pdf/600w.PDF

[priesty]

Nathan & Ken - thanks for your insights into this subject! The voltages are straight from the Ampeg schematic, and Ken, I take your point about the need for the level of bias seen.

Is it the case that the more current the CF passes, the greater the voltage drop across the cathode resistor and hence the greater the output tube bias? Hence altering the bias of the CF alters bias of output tubes?

I'm considering using an ECC99 in a similar application, it can dissipate a "massive" 5watts per triode, so sits somewhere between the 12BH7 and Ken's SV83's!

Thanks again guys.

[mooreamps]

I’m rather perplexed to the way Ampeg set up the positive and negative supplies to this tube. Referring to my sketch below, it would appear that the potentiometer P1 varies the voltage at the grid of the CF triode, but how does it alter the voltage at the cathode of the driver (and hence the grid supply of the power amp)? It would also appear that with a -30V grid setting that the 12BH7 would be biased as cold as ICE! How can the power stage be properly biased with this set up? It obviously works, but to me it appears that the 12BH7 triode is in a voltage divider with a 47K resistor and a -150/+350 supply!

I’m perplexed and it’s bugging me a bit! I even bought TubeCAD, but it hasn’t really answered questions, just posed some more!

[ATTACH=CONFIG]12398[/ATTACH]

Well, of course I have some thoughts on this. But I'll make this as simple as I can.. I don't think it matters how it's set up ; with either one voltage rail ; or a split voltage rail. But, if the "Grid-To-Cathode" voltage is -30 volts, then I am thinking ya, that tube is biased into cut-off...

-g

[redelephant]

Looking at the data sheet for the 12BH7, http://www.tubebooks.org/tubedata/HB..._2/12BH7-A.PDF ,
a plate-cathode voltage of 400V and 2mA plate current, the grid voltage needs to be about -25 volts. That's not too far off the values on your schem. The point is to bias the CF so you get the -47volts at the output tube's grids. The 12BH7 will be biased so it passes 2.1mA at that point, and whatever grid voltage it needs is what it gets, either -20 or -30volts, it will be anywhere in that ballpark. So the important thing is that the potentiometer has a range that allows that. 2mA is not much for that tube, but at that point it has 400volts over it, and although it is more linear and perfectly within specs at 4-5mA, it is probably run conservatively for a reason. (Easier on the powersupply, less ripple, less heat, bla-bla-bla.)

[mooreamps]

OK, fair enough.. But if that tube is capable of so much current, then why bias it so cold ? Why not let it run at 10 mills or 20 mills or so if that helps it drive a load better ?

[redelephant]

First: 'mills' is a dimension, not current, 'mA' is correct...I've heard people use 'mills' a few times, at Edcor for instance, so is it some term tought at some technical school or something? Is it the 'oldschool' term for mA?
Anyhow...
The tube can be perfectly happy at low idle currents, a CF is very linear driving a light load where the currents don't vary much. The load only becomes difficult when the signal swings so high that the output tube's grids go positive in relation to the cathodes. At that point the grid will draw current from the cathode, and the directly coupled CF happily starts to conduct more. Now the amp is into AB2 and at (or near) max power output so the CF will be farely seldom drawing that grid current.

As the CF goes up, it conducts more. Just b/c it idles very lowly, doesn't mean it cannot suddenly conduct shitloads of electrons.

Why they let it idle so coldly? Like I said in earlier post, I dunno, but perhaps just being economical....
The max plate dissipation of that tube is 3.5W, so you don't want to bias any more than some 8mA thru it. As a compromise, I'd go smack in the middle f the original design, and the max rating, and set it at 4-5mA.

[kg]

as far as why so cold, i think redpachyderm hit it.

followers DO kind of "wake up" at higher cathode currents... transconductance goes up with Ik, and transconductance is the primary factor in cathode impedance (which is what you want to minimize in a follower). so the hotter follower is a stronger one.

that said, you have to consider when this kind of follower will be asked to do work... it's when the grids swing high. once they get close to Vk on the output tubes, then they'll start to intercept a lot of electrons and conduct a lot. the electrons will "appear" at the follower's cathode, where they can be quickly dispatched through the plate impedance and off to the power supply. essentially you get cathode current that does NOT flow through the cathode resistor.

the long and short of it is that they just don't need to be run hot to retain good performance sourcing grid current to the finals, and retain decent linearity while doing so. power supply requirements, heat in the chassis, tube life--all benefit from the cooler operating point.

[Steve Conner]

essentially you get cathode current that does NOT flow through the cathode resistor.

Yes. The followers are essentially being run in Class-AB, as opposed to pure A.

[mooreamps]

Well, it seems to me after reading some of the earlier comments about using cathode followers to source more current to charge the capacitance for tone stacks and power tube grids ; it just seems to me this would be a good tube to do that with..

and if so, why not "bias" it to run at 10 mills instead of a just a couple ?


It seems like an interesting tube.. For awhile I've been considering going to a more active style tone stack. I even drew up a switchable midrange notch boost for a metalhead amplifier.. I was considering using something like an EL-84 in triode mode to drive the tonestack ; given that tube could sink more current than a 12ax7 or a 12au7.. But, I checked one of my suppliers, and I can source this tube..

-g

[redelephant]

and if so, why not "bias" it to run at 10 mills instead of a just a couple ?
-g

There u go with those 'mills' again...
U can bias it any way u wish as long as plate dissipation is below 3.5watts. Blackstar uses that tube (12bh7) as the powertube in their 5watt amps.

[priesty]

There u go with those 'mills' again...
U can bias it any way u wish as long as plate dissipation is below 3.5watts. Blackstar uses that tube (12bh7) as the powertube in their 5watt amps.

OR use JJ ECC99 with plate dissipation up to 5watts!

Thanks everyone for this discussion, its getting some good ideas and discussion out there!

[defaced]

There u go with those 'mills' again...
U can bias it any way u wish as long as plate dissipation is below 3.5watts. Blackstar uses that tube (12bh7) as the powertube in their 5watt amps.

The irony.

[mooreamps]

U can bias it any way u wish as long as plate dissipation is below 3.5watts. Blackstar uses that tube (12bh7) as the powertube in their 5watt amps.

Still, just not the answer we were looking for... The question was : why not run your cathode follower to source 10 ma instead of 2 ma ?

{of course I already know the answer to this question. I'm just seeing who else is willing to chime in on this....}



-g