Cathode follower question.

[Diabolical Artificer]

I'm looking at designing a driver for an AB2 OP stage - two 807's. I've knocked up the KF as per the STC 807 datasheet, so far it's working ok - ish. The datasheet says for AB2 conditions the drivers should be able to drive the grids @ 200mw, 4% THD, this with the 807 grids at -25v.

First question - my Vgk voltage is 18-20v, so in cutoff! But the stage still works, why, how? 2) when drawing a loadline for a stage with split rails, in this case +300v & -80v does one draw it as from 380v? 3)the V as measured at the KF cathode is -67v, this is wrong if we want to bias the 807 @ -25v; the dynamic characteristics graph for an 807 g2 300v had the grid lines finishing at about -45v, so what gives?

Lastly, I calculated that the C input for this stage would be about 120p (C x gain [gain is 10 EG 25v in, 250v out, g1 driven to +5v, OP 62W)], I'm using a capacitance decade box as load , there is no attenuation of OP from driver @ 700p, it starts to roll off above 800p, THD is less than 4% but is reduced if I increase the negative rail.

Andy.

[kalee20]

Well,

1) the CF tails are taken to -78V, the 6SN7 grids are to 0V, so the 6SN7's can't be cut off!

With Vk = -67V, then this should give Vgk = +67V but you say it is 18-20V, something wrong there!

I have looked at 6SN7 data, with 300V on anode. cut-off is indeed at Vg = -18V.

2) no, the 807's are running between 0V and +300V so the loadline would be for 300V. If that's what you meant.

3) Seems strange! The 807's should indeed be cut off at -67V... Can you rely on your measurements?

Input capacitance... 250pF seems a lot for a tetrode, but I'd agree more-or-less with the formula.

A bit more accurately:

Cin = Cgk + (Gain+1)xCga.

Increasing the negative rail would reduce distortion in the CF, because it's biasing the 6SN7's 'on' a bit harder, where they are more linear. The 807's, on the other hand, I'd suspect, won't be though.

[GMB]

the CF tails are taken to -78V, the 6SN7 grids are to 0V, so the 6SN7's can't be cut off!

Is he talking about the follower that drives the 807's? There is a potential divider (easily missed) so the grid is not -78V. I would have said it was near to cut off, but not fully.

It might help to know all the various voltages relative to ground.
Also, when a circuit does weird/impossible things it is usually oscillating (in my experience), often at unexpectedly high frequerncy - so worth a check.

[PJL]

I would have guessed he is referring to the 807. The 807 grids will be the 6sn7gt grid voltage (-75/2) + 6sn7gt grid to cathode volts. The 6sn7gt autobias on the 10K in the cathode and a rough guess of 5mA for the anode current gives us -25V on the 6sn7gt cathode and -12.5V for the 6sn7gt grid to cathode that fits fairly well with the datasheet.

Class AB2 it may be, but there is not much 'A' as it's mostly 'B2'. Good for getting the most out of the 807's but not so good for the ears.

[G6Tanuki]

As others have noted, when looking at the phase-splitter/driver you need to measure voltages WRT the -78V supply, and when measuring the voltages associated with the 807s you need to reference to ground.

That design-ethos is quite familiar to me, having used it in a number of AM modulator and guitar-amp designs over the years. I always found the big problem being that once you started to drive the output-pair into significant grid-current (as you need to do for 'GOES UP TO ELEVEN!!' loudness) the driver often struggled; a problem solved by using the centre-tapped primary of a small push-pull output-transformer as the 'load' for the cathode-follower stage. This provided a nioce low DC-resistance-to-ground for the grids of the 807s.

[kalee20]

Originally Posted by GMB

Is he talking about the follower that drives the 807's? There is a potential divider (easily missed)...

You're right, I did miss it!

[Radio Wrangler]

I don't see a potential divider for the DC level of the 6SN7 grids. Unless they're into grid current level of drive, they sit at 0vDC quiescent.

The cathode resistors will try to pull the cathodes negative but the 6SN7 will turn on and fight this. The cathodes will sit with a DC potential suited to the current through the valve. Draw a 10k Ohm line on a set of 5SN7 characteristics to find the quiescent point. The cathode will be a few volts POSITIVE with the valve passing 60v/10k = 6mA

With 300v on the anode and 6mA flowing Vgk is -11.5v from the curves so the cathode wants to stabilise at +11.5v

So this circuit forward biases the 807 grids because the cathodes are at 0v DC

Those 807s are far from cut off, they're working hard and getting into grid current without even a signal.

Of course the 6SN7 has an output impedance of 1/gm ohms so the 807 grid current will ull down that 11.5v somewhat and the dear old 807s won't run quite as much current.

David

[ms660]

807 Ik = 47mA with no drive.

807 Ik = 125mA at max output.

-ve 78V bias supply....I = 14mA....constant.

Lawrence.

[kalee20]

Originally Posted by Radio Wrangler

I don't see a potential divider for the DC level of the 6SN7 grids

It's there! 22kΩ from -78V and another 22kΩ to 0V.

So the 6SN7 grids sit at -39V.

If the CF cathodes are at -67V (as per original post) then Vgk = +28V, which can't be right!

Is there any chance of reposting the circuit diagram, but with voltages marked on, relative to 0V?

[Radio Wrangler]

How did I not see that? the horizontal 22k! and GMB pointed it out as 'easily missed'!

So that drops the 6SN7 grids to -39v

say 39 +10 (estimate) volts across 10k gives 5mA per triode reasonable

807 grid sees -39 +10(estimate) volts = -29v on the grid.

So with 500v on the anode and -29 on the grid, the 807 look biased for about 30mA each and it looks like class AB.

It looks like it ought to work, though is the 10k cathode pull down and 6SN7 quiescent current a bit light? The RC timeconstant comes out around a megahertz if wiring C is negligible. That ought to be out of feedbach stability zone.

Sorry, all

David

[Diabolical Artificer]

Seems I've caused lots of confusion...to clarify, there's no 807, and there was no voltage amplification stage, I'm talking about the cathode follower in isolation but with a cap decade box as load.

So, what we have is a 6SN7, grid, 470k, anode straight to 300v, cathode, 3 resistors, a 10k, and 2 x 22k, which as mentioned form a potential divider, -78v between the 10k and 1st 22k, 470k grid leak connected between the 22k's, therefore the grid must see 78v/2 = -39v. the grid has to see a negative voltage, otherwise there is no way we can bias the cathode negative.

When I posted I was getting around -69v on the cathode, confirmed because I measured a 6v drop across the 10k, therefore the valve is drawing 600uA, so is in cutoff, but still works, why? How? Vgk was definately 18 - 20v approx.

I then had a think and rejigged the circuit. Instead of taking the 470k grid leak to the junction of the two 22k's, I got a 100k pot and put this on the wiper, one end of the pot to ground, tother end to -78v, I then moved the -78v to between the 22k's so as to act as a bit of a CCS, see attached schematic (dashed lines original circuit as per STC datasheet, solid lines mark 2 circuit). now I have more control and can get the cathode to sit at -25v, with Vgk at 15v, better, but still near cutoff, again, the KF still works.

Normally, one would center bias a AC coupled KF, so Vgk should be in the region of 6 - 10v (8v say), see load line (in blue) on attached graph point B = Va 325v (300v HT + 25v Vk) Ia = 0mA) , this means the grid should be at - 33v (if we want the cathode at -25v), with the 100k pot, I can't get this to happen. The cathode follows the grid, with a 10 turn pot, one end of wiper travel = Vgk 20v, the other end Vgk = 15v. I've noticed this in other cathode followers when I've used a CCS in the cathode, there is some action independent of Ik that is keeping the cathode and grid apart by more than we want.

I'm on the edge of sussing this out, a bell is ringing in my head but it's in a sea chest under 500 fathoms of water. I hope I'm making sense here, the best I can do at present is to state that "Vgk is big, because of the following action of a cathode follower", it's something to do with one of the 3 constants, gm or mu or something.

Lastly, I can't figure out why the valve is working, I've had over 10v RMS out with little distortion but the negative part of the sinewave should be clipped....hold on, no it shouldn't because the cathode is sitting at -25v, or should it? AGGGHH, it's bending my mind trying to visualise what's happending. Times like these I really have to take my hat off to the lads at the beginning of electronics for their powers of thought, still in wigs and stockings yet able to intute things like why a lyden jar holds a bigger charge if it's not insulated. Kudos.

Apologies if this still doesn't make sense, it's hard to clearly articulate.

[ms660]

Just an observation, that load line.....if that's supposed to be the cathode load line it looks like it's Va on the X axis not Vk.

Lawrence.

[kalee20]

Hi Andy, there's something very wrong.

Your third diagram, with the voltages: Grid -39V, cathode -67V, you put 39-67=28?? Which I can't disagree with. But look! It'd be grid 28 volts POSITIVE to cathode, not negative! No way would it be cut-off.

And, big grid current would be drawn, which isn't possible via the 470kΩ grid resistor.

[Chris55000]

Hi!

At the risk of upsetting the applecart, may I ask where the circuit came from please?

It's very Mullard–esque the way it's drawn, and I've got a chassis I could build the first parts (output excepted) on to, so I might give it a try myself and see what the voltages come out at!

Chris Williams

[joebog1]

It's here Chris.

http://people.physics.anu.edu.au/~dx...807/807STC.pdf

Joe.

[Diabolical Artificer]

Hi Lawrence, right, but if we take the Va = 325v and call that zero volts then count to the left, that would give us out Vk, therefore our bias point is a bit off.

"But look! It'd be grid 28 volts POSITIVE to cathode" Your right, there is something off, however that -39v is what we calculated as being on the grid, not measured, this is a fault on my part not helped by then skipping on and changing the circuit by the inclusion of a pot.

Chris, the circuit is from the "STC 807 application report", I'd attach here but it's fairly big and most of it isn't pertinant to our thread. I built the KF on it's own as non of the other circuit (apart from the 807) should affect our DC conditions.

I've since jumped on to version #3 in which I ripped out the cathode resistors and used a CCS instead, still using a 100k pot to adjust Vg. This circuit works better, however the grid and cathode are still apart from each other by over 10v, that is until I adjusted the CCS to sink more current, then Vgk got to 9v, Vk sat at -25v and current through the valve was 12mA ( I used a 100r "sense" resistor from the CCS to the cathode.

I'll go back and re-build the circuit/s and post less ambiguous results as I've confused the issue, but essentially there are two unanswered questions. 1) Why did the cathode not sit at -25v (but at -67v) in the original STC circuit? 2) Why does the grid and cathode sit apart at around 20v? I have an idea about this but first things first.

Andy.

[Radio Wrangler]

Andy, whenever you give grid-cathode voltage differences, please indicate the polarity. I think this has been adding to the confusion. I find myself making assumptions about it, and assumptions are dangerous in problems that seem to defy expectations.

There is something very wrong going on here. The bias conditions the cathode follower should establish equilibrium at can be estimated reasonably well, and the reported values are quite different. Perhaps there is some difference between the circuit diagram and the built circuit, or perhaps strong oscillation is creating an apparent error in DC conditions?

David

[Diabolical Artificer]

Noted David. Just knocked the circuit up again, see schematic 1, this works as it should. Now this is how I had the circuit, see schematic 2. I was using a 100k pot to adjust -V, but the 100k pot is in parallel with the two 22k's, hence the wrong DC conditions. this answers Q1, operator error. It should be as schematic 1

One thing of note Vg isn't -39v as predicted, I've measured some voltage drop across the 470k before, therefore there must be a bit of grid current, 2/470k = 4uA, negligable.

So, q 1 answered, sorry for the fo par, I'll try to put Q2 in terms that are more clear by proper experimentation.

Right, see schematic three. With the 100k pot fully CCW I get Vg 0v & Vk +12.3v. With the pot fully CW, I get Vg -77v & Vk -61v. In the first example Vgk is -12.3v, in the second Vgk is -16v. My question is why the big Vgk? Why not Vgk = -2v? or -4v? I've tried this with various circuit tweaks, like using a CCS, Vgk is still larger than wished for. the only thing that does make Vgk smaller is increasing Ik.

A.

[kalee20]

Parasitic oscillation has been mentioned as a possibility. However, the 6SN7 is an octal double triode, which I would regard as comparatively docile.

Cathode followers can oscillate, yes - why not connect 10kΩ grid stoppers directly on the grid pins? While unlikely to have an effect, it's quick and easy to do, just so the possibility can be discounted.

[tri-comp]

Hi Andy,

We tried this a long time ago
Let's get it working this time.
And for others; the sound-quality is not B2/public address but good quality Hi-Fi with a potent bass-punch offered when entering AB2/grid-current.
First of all, remove the 807's from the sockets.
When setting up the quiescent state of the cathode-follower, and getting it wrong, there's no reason to abuse the output valves.
Besides, when you do get it right, the 807's will be almost cut-off, drawing no grid-current and therefore having no influence at all on the workings of the cathode-follower.
You may simply drop them back in and enjoy the music w/o further adjustments when you get the CF right.
The working of my amplifier, schematic attached, is quite similar to the one in the STC data-sheet.
For now, remove the CCS and replace it with a 10K-Ohm resistor between CCS pins 2 & 3.
My cathode-follower is somewhat beefier than a 6SN7 but I imagine the latter will be fine.
I did put a grid-stopper (1K-Ohm) in the circuit but that was just in case.
I actually didn't try the circuit out w/o this stopper.
As it doesn't do any harm I suggest to leave it in.
Now you'll have to get the DC output state of the cathode-follower in range to suit the 807's.
You do this by getting the negative grid voltage adjustment range of the cathode-follower right.
You could undoubtedly calculate this value but it's far quicker to get out a couple of 100K-Ohm trim-pot's (10-turn if you must) and connect them as shown.
Now get a 3rd. pot., possibly also 100K or higher (to 470K) and put it as a substitute for the 100K-Ohm resistor from the 2 x 2K7 resistors to ground.
(These two resistors may be shorted and left out entirely.)
Power-on and set the 6SN7 cathode-voltages to your desire as to where you want to bias the 807's.
Imho it's that simple.
The reason to bring in a CCS as a replacement to the 10K-Ohm resistor is to raise the impedance of the cathode-load
and so to achieve a better frequency response of the circuit.
That, however is a secondary priority. Get the CF working w/o the CCS and feel comfortable with how it reacts to whatever modifications you do.
And when that's nailed try the CCS out.
Good luck,

Best,

/Torben