8000 Upgrade

[K9ACT]

Now that both of my homebrew rigs are working well, it's time to "fix" one one them.

The 811 rig will be my backup till I get the 8000 back on the air.

The new project is to make my "big rig" bigger.  After a day of quality time in the shop, it is now two 8000's in parallel.

The plan is to run it at 1600V and 350 ma for a cake walk at the legal limit.

After making the necessary changes, it is acting very strange so I never got to turning on the HV.

It resonates where it should and neutralizes as it should but it draws more plate current than grid current.  I can not drive the grid to more that 5 or 10 ma without the plate current meter pegging.  This is with nothing on but fils and bias supply.

I checked all the connections and wiring changes and see nothing wrong.

What could cause this?

Schematic is at http://schmidling.com/2x8000_rf.emf

Thanks,

js

[N3DRB The Derb]

have you got 2X the bias voltage on the tubes? You doubled everything. You are drawing plate current with no plate voltage?

still unsure if thats whut u mean? you got 2 tubes now. hould need 2x the bias voltage to cut the toobs off. Someone confirm please, I have lost a lot of knowledge over the last 4 years. Stuff I say may not B true.

[K9ACT]

Bias voltage has been doubled or at least the resistor values have been halved.

Just to reiterate, this anomaly shows up on the plate current meter but there is no HV on. I assume it has to be coming from the bias supply.

js

[The Slab Bacon]

Bias voltage has been doubled or at least the resistor values have been halved.

If you are using grid leak biass instead of fixed biass, you have cut your biass in half by halving the grid leak resistors. If you are using some fixed biass you may be ok. Also make sure that the cathode (filament center tap) and grid leak resistor have a good DC path to ground.

                                                                                the Slab Bacon

[Bacon, WA3WDR]

It must be some sort of screwup with the metering shunts.  With no plate DC, the plate meter should not move much from grid drive, no matter what.

[The Slab Bacon]

Well, I've had a good nights sleep now and a very good & big breakfast. Unfortunately the only thing I remember about my friend's HB rig with the same problem is that it was parallel 810s modulated by triode connected 813s. There's definitely only 2 sources involved, the grid bias supply and the RF grid drive.

Mack

Parallel 810s mod by P/P 813s?? I would have done it the other way around!!

                                                                                          the Slab Bacon

[Bacon, WA3WDR]

I had darn good luck with 810s modulating a pair of 813s.

[K9ACT]

Well, I've had a good nights sleep now and a very good & big breakfast. Unfortunately the only thing I remember about my friend's HB rig with the same problem is that it was parallel 810s modulated by triode connected 813s. There's definitely only 2 sources involved, the grid bias supply and the RF grid drive.

Mack

I suspect the reason you can not remember is for the same reason I don't now why it went away.  After checking everything over again and putting it back in the rack, it acted normally.

Not very satisfying but I guess the case is closed.


>Parallel 810s mod by P/P 813s?? I would have done it the other way around!!

Ah... but then you have to deal with icky screen grids.  Triodes are nice.

Sounds like my rig as there is not much difference between an 810 and 8000.

js

[k4kyv]

After making the necessary changes, it is acting very strange so I never got to turning on the HV.

It resonates where it should and neutralizes as it should but it draws more plate current than grid current.  I can not drive the grid to more that 5 or 10 ma without the plate current meter pegging.  This is with nothing on but fils and bias supply.

I checked all the connections and wiring changes and see nothing wrong.

What could cause this?

Schematic is at http://schmidling.com/2x8000_rf.emf

I'm not sure about the plate meter pegging, but it is normal for a triode tube to show some plate current with grid drive applied, but with no plate voltage.  I ran into this problem and it gave me fits until I figured out what was going on, the first time I attempted to build a high power triode final back in 1962, using a single 304-TL.

This is the reason the older handbooks recommend that you physically disconnect the +HV lead from the final when neutralising.

With rf applied to the grid, and the rectification resulting in DC grid current, the grid is being driven positive on rf peaks so that it acts like the plate of a diode rectifier tube.  The positive grid attracts electrons from the cathode.  Because of the bias on the grid due to the grid leak and any fixed bias, the grid still reads negative with a voltmeter, but it is being instantaneously driven positive whenever the rf voltage exceeds the negative bias voltage.

But some of the electrons attracted from the cathode by the instantaneously positive grid are accelerated past the grid and end up bombarding the plate.  If there is a DC path to ground from the plate, these electrons produce measurable current as the electron flow returns to ground.  This rectified rf appears as DC plate current on the meter.

As long as this rectified  rf finds a return path to ground, it will excite the resonant tank circuit with rf.  You can verify this by holding a small neon lamp near the tank coil.  The current will be sporadic, and will vary as you adjust the grid tuning, plate tuning or neutralisation setting.  This rf in the tank coil makes it impossible to adjust neutralisation if you are attempting to do this by nulling out the rf at the output circuit while maintaining grid drive at the input.

Physically disconnecting the HV plate lead opens up the DC path to ground that exists through the power supply rectifiers and plate transformer.  That will eliminate the extraneous current flow through the plate, and will eliminate the induced rf from the tank circuit.  You can then neutralise in the normal fashion, by adjusting the neutralising capacitor for minimum rf voltage indication at the output, while keeping the grid and plate tank circuits in resonance.

When the HV lead is reconnected and +HV applied, the final should behave in normal fashion.

[K9ACT]

This is the reason the older handbooks recommend that you physically disconnect the +HV lead from the final when neutralising.

I happen to have one of those older handbooks, 1949 West Coast version and it says exactly that but gives no explanation or reason.  None of my ARRL books mention this starting with 1957.  I did not take it literally or seriously and it was never a problem with the single 8000 or the 811 so I considered it quirk.

I have just done as you suggested last night on the air, viz., dummy load on output, disconnect HV and it neutralized cleanly with a sig gen in and then with the Kenwood driving full grid drive.

When I connect the the HV lead it still pulls some plate current but it is greatly reduced so I guess  I will fire it up later at low voltage and see what happens.

I also removed 10 more turns from both sides of the coil and the cap is not almost fully messhed at 3700 and a bit less at 3885.

Thanks for the help,

js

[K9ACT]

The good news is that the "mystery" plate current with no HV turned out to be an open shunt in the meter.  Not hard to pin a 1 ma meter.

More good news is that I fired it up and ran it for two hours on the Noon Time Forum and it was unanimously voted working well.

The bad news is that for the extra 200 watts input, I am only getting about 25 watts more output.  Efficiency went from 72% to 54%.

Both tubes were tested in the single tube config and produced the same output.
The DC grid voltage is -320 at 40 ma drive as prescribed.  Plate voltage is 1600 at .350 ma output is about 300w.

So, where is all the power going?

js

[k4kyv]

Both tubes were tested in the single tube config and produced the same output.
The DC grid voltage is -320 at 40 ma drive as prescribed.  Plate voltage is 1600 at .350 ma output is about 300w.

Is that 40 ma total, or 40 ma per tube?

With HF-300's or 8000's, I always ran about 37 ma per tube, but I always ran them at 2000 volts on the plate.

[k4kyv]

The -320v grid bias is a little more than the spec sheet value of -300v for 1600vdc on the plates?

I always run my grid bias about 50% higher that what the spec sheets call for, by using higher than recommended grid leak resistance while running the grid current at the recommended value.  This takes more driving power, but preserves modulation linearity on the positive peaks by causing the tube run "super class-C", that is, to conduct over a smaller number of degrees of the rf cycle.  In other words the tube pulses at a smaller conduction angle  in class-C.  This should increase the efficiency as well.

[W2PFY]

 Well the first thing I would do is un-ground the center tap of the grid coil and feed your bias isolated with an RFC. It should be bypassed with a cap to ground from the cold end of the RFC. The second thing I would do is get rid of that .001 cap going to the grids and feed the grids direct from the hot end on the coil. You may have to adjust N/C. Never seen an input circuit like that.

Hope it helps and would like to hear other comments about this change.

[K9ACT]

Hope it helps and would like to hear other comments about this change.

Me too.

Here are my data points.

With a single tube it worked like a charm for the past 10 months.  This design is basically by my elmer, an engineer for whom I have the utmost respect.  Everytime I ask him why he is doing it some way, he has an iron clad explanation.

I built my 811 rig as you suggested and it also works like a charm.  The design is basically out of the handbooks with minor variations.

Before I change anything, I would really like to hear some opinions.

js