driving a pair of 811s

[Ed/KB1HYS]

I want to drive a pair of 811s with zero bias with out using a driver tranny.  I would be using a paraphase type driver but I don't think that will provide enough drive to the grids of the 811's.  I also don't know about the load the grids would provide to the phase splitter.  Some one must have tried this before. Will I get enough drive from a phase splitter?  I was figuring on using a an 6SN7 as the phasor,(60 volts but no current) with the remains of the audio circuit probably being sandy state.

Just looking for thoughts..

So grid drive requirements for 811s zero bias?

[K1JJ]

Hi Ed,

Consider this excellent generic FET cathode audio driver for your 811 modulator. No audio transformer required. It was designed by Tom, WA3KLR. It runs only 5 watts and can fit entirely on a RS perf board. (Not including the 12V supply)  <1 volt input drives the 813s to full power ++.  The 813's also need a fixed 50 v on the screen/grid as positive bias. They are TRIODE connected.  For 811A's,  just eliminate the screen connection and all shud work FB.

I just did a proof on my 813 X 813's yesterday and am amazed at how clean this system is. It also has provisions for negative feedback.

Much simpler and cleaner than a tube audio driver could possibly be.

Highly recommended.

Later -

Tom, K1JJ

[Opcom]

Peak grid current must be supplied to avoid distortion and it depends on other conditions like B+ supply volts and power output requirements.

FETs are certainly a good choice and will supply the drive, no disputing that. I prefer tubes so I'll only comment on them.

From the tables, the peak drive voltage can be obtained, and from this, the peak grid current can be estimated. When the tube is driven hard, the plate voltage will become low and at that time the grid current will especially increase. So there is no simple answer but it could be calculated.

I hope some experts will comment on these next paragraphs and show me my errors:

]]]]]]]]]]]]]]]]]]]]]]]]]]]

One thing that is confusing the plate to plate load impedance. For this explanation of my opinion I will use the CCS 1250V table with zero bias and 12400 ohms plate to plate to plate load.
I found out that 12400 Ohms plate to plate does not mean that each class-B tube is presented with 3100 Ohms load, even though common sense says that the center tap on a 12400 ohm winding would present 1/2 the voltage and therefore 1/4 the impedance. This is assumed out by looking at the same conditions for a single tube as a class B RF amplifier. The RF amp takes a 5700 Ohm load, roughly half the push-pull figure (RCA TT-4 handbook).

So from there it is to calculate the minimum plate voltage and from there the grid current at that moment of peak drive voltage. Since 1250V is supplied to the plate, 6200 ohms is in series with the plate in question, and 260mA DC is flowing through the tube during its half cycle, the resistance of the tube is 4807 Ohms on average (or RMS?) during its half cycle. Since the table assumes a sine wave, the peak minimum value of the tube would be 0.707 of the average value, or 3399 Ohms.

So there is a voltage divider consisting of the 6200 Ohm load and the 3399 Ohm tube resistance. With 1250V on the plate supply, the minimum tube plate voltage should be 1250 * (3399/(6200+3399)) or 442V. The peak grid to grid voltage is 145 volts. Each grid has a positive peak of 72.5V.

With 442V on the plate, a 72.5V grid voltage would cause a grid current of 55mA to be drawn.

Another way to figure this is to look at the peak g-g voltage and compare to the driver power. If the peak grid voltage is 72.5V, the RMS voltage is 51.25V. If the drive is 3.8 watts, the one tube uses 1.9 watts RMS. During the tube's half cycle, the RMS grid current is 37 mA and the peak value would be 52.4 mA

So the driver must be able to supply 55mA to the grid without distortion. Good practice requires the driver to be able to provide substantially more, to account for irregularities. If it were me I would try for a driver that could do 100mA peak.

To drive with no transformer, I'd want to use a cathode follower. I'd be tempted to use a 6080 or 6AS7. It would be possible to use a pair of low voltage beam tubes like the 6AU5 and 6AV5 and they are cheap because they are seldom used. Any tube used in regulated DC power supplies as a pass tube would do. Other seldom-used tubes like the 815, 829B, and 832 could be considered because they are cheap and can be had by the shoebox full, as well as sockets for them. An advantage of a beam tube or tetrode for a driver is that the gain of the stage can be tweaked with the screen grid voltage. The screen voltage is also a good place for a little feedback.

]]]]]]]]]]]]]]]]]]]]]]]]]

I hope some experts will comment on the above paragraphs and show me my errors.

[WBear2GCR]

Sorry Patrick, I didn't read through ur description, so can't comment yet.

Ed, as we talked out on the air earlier today, the trick is to get the voltage swing from the preceeding stage, and use a cathode follower to drive the grids of the next tube. Some prefer the resistor to ground at the cathode, I prefer the choke method.

If you get online and find a copy of the Altec 1570B amplifier schematic, you'll see an example of the exact circuit you need.

To drive the 811 in class B2 or AB2 you need to be able to handle some power when driving the grids positive, thus the requirement for DC coupling. The tube manual shows the drive requirement, iirc something like 2watts. You want some margin, so a tube like the 6W6 is ideal, as is a 6L6, strapped for triode is best. Anything will work in that slot, as it is a follower. HLR mentioned the 6AS7 or 6080 in that position, but to me that is far more tube than is needed for the job, and wastes a relatively rare and very useful tube (very low plate Z) on a job that doesn't need a tube with low plate Z. Also they take more filament juice. At least that's my thinking on it.

Typical plate Z iron for two 811s is about 8-10kohms on the full primary, iirc.
And it does depend somewhat upon the plate voltage that you pick... for the proper match - if you want or need that.

               my 2 cents worth...

      _-_-bear

[KE6DF]

Attached is a 6bg6 cathode follower driver used in a Gates broadcast transmitter to drive 810s.

6bg6 is a beefed up 6l6 with a plate cap used as a sweep tube. Sort of an octal 807.

Fairly cheap.

[K4TLJ]

6BX7 or 6BL7 dual triodes are another possibility as cathode follower drivers.

[Ed/KB1HYS]

Good information all!! 

Bear, I grabbed that altec schematic off a web page and it does look like just the ticket, especially since it is a circuit specific to driving 811's, it looks alot like I was thinking. It was the part after the 6sn7 inverter that I was fuzzy on.

By the way, you have a great sounding signal there OM, absolutely first rate.

[Opcom]

I'll agree 100% on the Altec driver. It will provide enough drive even when the choke is jumpered and the Altec's HV raised to 1250V, and the output to 250W. Not that the amp will run long at that level before the Klixon pops.

But I don't like the choke. Nothing wrong with it, just a preference, if you are eliminating iron, then eliminate it!

Bear -  you should please mention what to use for the cathode follower driver choke since the Altec one is not usually available and I think you found a good substitute?

[WBear2GCR]

Yeah, give away all my secrets! 

Use any 10watt class PP output iron, the primary windings.
Works fine.

You could use a CT power tranny, or choke, but the inductance has to be equal or greater than the stock choke and the DCR should be similar + 100%/- 20% tolerance on the DCR. The trick with the choke is to try to get the two halves to match in DCR and inductance within reason, say 5%, for best balance. But actually, that isn't terribly critical for modulator service, is more so for hi-fi service, and/or if you are not using any feedback at all.

If you copy the Altec exactly, you'll need to add a small negative bias supply to the CT of the choke to drop the grid lower IF you raise the B+ enough to make the plates start to get red in the face. Also be sure NOT to use the cap that bypasses the feedback resistor in the circuit, it screws up the output waveform in order to get a "flat" response. I prefer a square wave that looks like a square wave, not some thing with excessive ringing on peaks... sounds way better without the ringing to the human ear.

Use film (mylar or polypropylene) coupling caps, you'll do best with them.
Bypass ur electrolytics with film caps of 10% or so value and that will help too... those in the signal path. That includes the one on the filter of the bias supply, if you add one. It's in the signal path.

Anyhow that's the whole deal there...

Oh, the good part of the choke thing is that since you want it to have rotten HF response (increasing impedance with frequency) you can use a rather poor output iron for the choke, whereas you can't use poor output iron as output iron! 

Caveat, I have notice that for hi-fi apps, a decent although inexpensive outpoot iron in that slot appears to sound substantially better than the stock choke. I do not know if it is because it has better leakage inductance and less interwinding capacitance, or if it was just because it had a bigger core, and didn't saturate on peaks as a result. The latter would indicate that a suitable HV power tranny with the requisite windings would do just fine - keep in mind these are relatively small bits-o-iron, 7-10 watt class here.

Dat's eet, mon!

              _-_-bear