811 project

[K9ACT]

I have my 811 amp up and running in the CW mode and it works quite well but has one quirk that I don’t know what if anything to do about.

When set up according to “typical operation“….. I get 115W out at 150ma and 1000vdc. This is with 50 ma on the grid and -55 dcv on the grid derived from a combination of -28v fixed bias and 500 ohm grid leak.

That’s the good news.  The problem is that when I turn off the HV, the grid current rises to almost 100ma.

From a practical standpoint, it’s easy enough to just leave the HV on but if this indicates a serious problem somewhere, I need to know what it is and I have no clue.

Any ideas?

js

[Bacon, WA3WDR]

Is that with the RF drive still on?  It is best to drive a modulated stage (especially a triode)  with a soft source like a tetrode or pentode, to regulate the grid current.  Of course, normally you would turn the drive off when you drop the plate voltage. I assume that you left it on while testing.

A grid draws more current with low plate voltage because the plate isn't sucking on the emitted electrons as hard, so they go to the grid instead, which means more grid current.  This is true of screen grids, too; if you cut plate voltage with the screen still powered and drive on, you can blow the screen.  Of course the 811 is a triode, but pretty much the same issue holds for the control grid in that case.

[K9ACT]

Is that with the RF drive still on?

Roger.  The RF drive produces the 100ma grid current if the HV is not on.

 >It is best to drive a modulated stage (especially a triode)  with a soft source like a tetrode or pentode, to regulate the grid current.

Not sure what you mean but it is driven by an 807.

>Of course, normally you would turn the drive off when you drop the plate voltage. I assume that you left it on while testing.

Right but also, it is my MO to adjust the grid drive before turning on the HV.  With my 8000, the grid current does not change at all when I turn on the HV.

>A grid draws more current with low plate voltage because the plate isn't sucking on the emitted electrons as hard, so they go to the grid instead, which means more grid current.

I understand that but a factor of two?

js

[Bacon, WA3WDR]

2:1 does seem to be a lot.  A few thoughts:

The 500 ohm grid leak and -28V fixed bias is probably making the grid current of the 811 more sensitive to the presence of its plate supply voltage and load tuning.  With an 811, you could get away with zero fixed bias at 1KV, and it shouldn't need more than about -6V to cut it off pretty well.  Maybe -10V to be sure.  This would allow the grid leak resistor to increase to about 900 ohms to 1.2K.  That would help, I think.

About that "soft" RF driver business - I was sloppy to use that term, I was talking about the plate resistance of the RF driver amp.  With an RF driver having high plate resistance, the driver output will behave as though it is coming through considerable series resistance, and this will normally reduce variations in 811 grid current when its plate voltage is turned off.  You want high plate resistance in this RF driver for this reason, and also because it levels the grid drive during modulation too.  Triodes modulate more linearly if their RF driver has high plate resistance, because of that effect.

A couple of problems are possible, though - if the 811 is not neutralized, high driver plate resistance will cause a reduction of drive because of grid-plate capacitance when the 811 plate is powered and swinging over a thousand volts of RF.  This will be worse with AM because you will get less drive on modulation peaks, which is when you would want more drive, if possible.  You want to neutralize the 811.

Also, the 807 plate resistance will be lower if the 807 screen supply is powered from its plate supply through a simple series resistor.  A resistive divider providing screen voltage would be better, and a lower fixed-voltage supply or some zener screen voltage regulator would be best.

I assume that there is no significant resistance between the filament center tap and ground.  An ohm or two to monitor current is OK, but a higher resistance would cause a loss of grid bias when the plate voltage is off.  From what you described, I don't think this is an issue in your case, it's just a thought.

Another thought - I am assuming that the 811 is not set up grounded grid.  For plate modulation I think it should be grounded cathode (filament).  GG might not like a high-resistance driver either.

That's a cool looking telescope in your avatar.

   Bacon, WA3WDR  (the old Bacon)

[K9ACT]

Thanks for the ideas.  I have a gut feeling that it is somehow connected with neutralization or coupling.

The usual test for neutralization (no change in grid current when plate cap is turned) says it is neutralized but the coupling between grid and plate seems rather excessive.

If I put a sig gen on the input and a scope on the output, I get the same level out as I start with when both stages are resonant at the sig gen freq.

What really surprised me as I get the same results when I remove the tube.

Turns out, disconnecting the NC makes it all go away.  This may be what it should do but it is disconcerting to me.

js

[WD8BIL]

Bacon has a good question. Is this grounded grid or common cathode ?

If it's GG used as a linear amp loose the neut cap and have at it !

[K9ACT]

I think I solved the leaping grid current problem or at least know what causes it.

I back engineered my changes and found that changing from link coupling to PI output on the exciter is the culprit.  Why, I haven't a clue.

I used the Kenwood to excite it and got only a nominal change with/without HV.

I then connected the input of the 811 to a two turn loop over the output tank and got about the same results.

Any ideas?  The schematic can be found via http://schmidling.com. radio.htm

js