Direct Coupling with Zener Diodes

[jmaf]

So it is possible to measure a grid voltage lower than the cathode, and still have grid current flowing.

This is interesting, grid current flowing from where to where in that case? You mean AC current, from the peaks when the grid is slightly more positive?

[Steve Conner]

From grid to cathode as usual. The grid-cathode assembly is acting as a Thermionic converter - Wikipedia, the free encyclopedia and generating a tiny amount of electric power.

The statement I made above is in the RDH4 and not considered controversial in any way.

[jmaf]

see, jmaf, steve wanted to prove ME wrong!

:lol:

You are so charming, kg.

[jmaf]

The statement I made above is in the RDH4 and not considered controversial in any way.

Honest question. I did not controdivirsificate it by any means. On another subject, Steve, on your traces do you have any with the output of the CF's cathode?

[Steve Conner]

Then I reconnected the CF and scoped the CF output.
img_2933.jpg

Yes, the second screen shot shows the cathode voltage, which is what I meant by the output. On the scale of 100V/div, you can't see the difference between the grid and cathode voltages.

I'm getting more and more interested by the effects of grid current. With quite a few of my old tubes, I see them generate a fair grid current at idle, which makes the guitar volume pot scratchy.

The gain stage side of the Mullard I used for this experiment seemed to have something similar going on. I had been using the same breadboard to do experiments on blocking distortion, so the signal generator was AC coupled to the grid, which had a 1M DC leak, and I saw that just by turning the tube heater on, the grid biased itself ever so slightly negative.

I had never noticed this behaviour before, but maybe Merlin has something to say about it in the light of his grid current findings. Maybe I have duff tubes, and if I bought some new Russian or Chinese ones, they wouldn't do it. Or maybe they all do it and this little random-valued "bias battery" is what makes different tubes sound different.

[Merlinb]

Merlin, which book is it? The original preamps one? I've resisted the urge to buy it so far, but I really want to see that treatise on the effects of ECC83 grid current that was allegedly covered up by the AES.

It is my first book, but I suggest you wait for the second edition! You will like it better.
The AES paper is here BTW:
http://stashbox.org/34435/AES_Triode_NonLin.pdf

[Steve Conner]

Wow, thanks for the paper. Very interesting and very relevant to the discussion. (The 12AX7 grid current vs. voltage curve in the paper explains the negative grid voltage I saw on my test rig.)

I had never appreciated that the DC coupled CF circuit clipped by grid limiting on both half-cycles, and this sounded better than cutoff clipping. I've never used CFs in my amps because I felt that it was a waste of a gain stage, but maybe I will try it now.

[kg]

You are so charming, kg.

i was being serious!

as i'm sure you are, in turn.

[jmaf]

i was being serious!

as i'm sure you are, in turn.

Course I was being serious I noticed Conner was obviously on the intent of questioning you from the start. I hate them youngsters who come in knowing it all.

[jmaf]

Here are some interesting bits from Merlin's book, related to my earlier statements on this thread. I must say, a lot is getting a lot clearer for me....

1) The 5f6-a is quoted on Merlin's book(figure 5.16, page 127) : CF's grid is biased at -10V due to the difference in current between the upstream gain stage and the CF's higher current.

2) Merlin shows that grid current does not mean we have a positive grid. In fact, there's quiescent grid current and it is constant at 70uA.

3) Rarely does the 5f6-a CF get driven to cutoff

4) The grid current is constant and at 70uA which shows that the grid does not regularly swing positive? What does this mean exactly?

Source: Designing Tube Preamps, First Edition, Pages 126 to 131

[kg]

i think we have to keep separate the quiescent and the dynamic conditions.

ALL grids pull current, ALL the time. that's how grid leak biasing works. if there was no current, we wouldn't need a DC return path from grid to cathode.

at some Vgk that grid current will start rising exponentially, according to the three-halves law. depending on the tube type and the conditions of operation, that Vgk will vary, but will always be close to 0v. part of this is due to the velocity of electrons as they boil off the cathode, part is due to the inherent negative polarity of the space charge immediately proximate to the cathode which effectively creates its own slight relative bias.

i'm not sure where the idea that i ever suggested running with positive grid bias at idle comes from? while it can be useful for some high mu power triodes, and it certainly IS a valid mode of operation, it's not something typically seen because there are usually "better ways" of achieving a design goal.

during dynamic conditions, the primary issue with the direct coupled cathode follower is that at max positive excursion the voltage across the tube is greatly reduced. for a given plate current this means a MUCH less negative bias voltage--perhaps even positive. making matters worse, plate/cathode current in a CF is NOT constant with a simple resistive load, and in fact goes UP during positive excursions. so we have two contributing factors making our bias voltage "less negative": lower Vak, and higher Ik.

due to these factors, in my BAGA drivers i took great pains to reduce these two effects: 1) by using pentodes (el84s) run as pentode cathode followers--meaning the screen grids are AC coupled to the cathodes, and 2) by using constant current sinks in the cathodes, set to 15ma. this circuit topology allows the maximum positive signal excursion without running the followers into grid current.

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

[redelephant]

I've pretty much ignored CFs too, perhaps I'll join the bandwagon and try it out some more myself.
Thanks for that interesting paper Merlin, will enjoy it later this evening.
I got some ideas for a couple of floating high impedance inputs on my test rig out of this thread. I've wanted to be able to measure signals riding on top of other signals or other non-gnd referenced signals ever since I started building amps, but haven't given it much priority.
Such inputs will for instance enable measuring grid-cathode signals riding on a large signal swing at the plate, or measure over the grid stopper while the plate is swinging maximum.
Ah, now I need more time again So many projects - so little time.

[jmaf]

Just a quick note: I am unable to access the link provided by Merlin, says "Access Forbidden".

[J M Fahey]

1) The 5f6-a is quoted on Merlin's book(figure 5.16, page 127) : CF's grid is biased at -10V due to the difference in current between the upstream gain stage and the CF's higher current.

Are you sure?
I beg to differ.

The cathode follower triode does what any triode does, work as described in its datasheet.

Assuming +B=260V and Vk (referred to ground) of, say, 160V, plus a 100K cathode resistor,
1) that triode is "seeing" a Plate Voltage of 100V, since what we usually call "plate voltage" more precisely means "plate voltage referred to cathode voltage".
2) it's passing 1.6 mA (160V/100K)
And what does the datasheet show?
For 100 Plate Volts and 1.6 mA, bias voltage (relative to cathode) must be interpolated, but lies between 0V and -0.5V, I estimate it in around -0.3V.

-10V? No way José !!!!!!!!!!!!!!!!

Note 1: this is a "borrowed" datasheet, please ignore the red inscriptions which belong somewhere else.

Note 2:

due to the difference in current between the upstream gain stage and the CF's higher current.

What does it mean?
There is no relationship between the current passed through an earlier stage and the one passed by the CF.
Tubes are *voltage* controlled devices, .
Even if positive biased and passing grid current, the controlling parameter is still voltage. (Grid/cathode voltage, that is)
Please stop pushing that "balanced resistors" theory of yours.

[Merlinb]

Are you sure?

This occured to me too after it was posted. I *thought* that when I measured the voltages I measured the voltage across the previous anode resistor and subtracted it from the HT. However, I may have accidentally measured the CF grid voltage directly, which will have pulled its grid voltage down. That would explain the discrepency. In reality I suspect the grid is about 0.5V negative of the cathode.