The 813 Rig Troubleshooting Mystery

[K1JJ]

Now that the GFZ MOSFET audio driver is finished and working, for the last two days I've been testing, ironing out problems and generally going thru the new homebrew pair 813s plate modulated by a pair.  This is "Hollywood" as seen in a different thread. Everything is generally working except for one baffling problem...

It seems that the screen current is as dead as a doorknob.  Grid current rises way up past 40 mA under normal drive, but screen current, even with 350 volts on will not go over 5-10 mA.  Using 3KV, the pair of 813s will not go past 150 watts output.

How can this be? The screen supply is putting out plenty of voltage and the grid current is strapping and the plate current is 75 mA with a nice dip of the final tank. But the screen acts like it is blown out.

Walking Yaz today it suddenly dawned on me what was wrong. I made some simple measurements and found the problem. It was not bad parts.   I have yet to make the change, but without a doubt, the symptoms told the story.

I'll give it a day to see who can come up with the answer and may give some hints later...    It's one of those really dumb specific situations that may have happened and will probably happen to others too, so I want to highlight it.

BTW, hint: The SAME problem independently affected the two modulators too.

T

[Tom WA3KLR]

Suppressor grid pin 5 left floating?

[K1JJ]

Suppressor grid pin 5 left floating?

Wow!  So close but no cigar, Tom.   The exact root cause is not an easy one to figure out.

[KA2DZT]

No screen by-pass caps.   That's my guess.

[K1JJ]

No screen by-pass caps.   That's my guess.

Nope.  Good guess though.   Each screen has a 500 pF bypass cap.

Hint: The rig was basically working - but not very well. The screen meter showed more current when driven harder, but only about 1/10th of normal. It acted like the screens had big holes in them and were not pulling their loads.  

I just fixed the finals and working on the modulators now...  Power test later tonight.

T

[K1JJ]

Test:   700 watts out carrier at 2500 VDC.  Full screen current.   Screen problem solved.   Gobs of power now.

Now working on improving RF input network for 1:1 swr on all bands.


Finer points: My fixed protective bias seems a little too high.  There is also too much DC resistance in the screen circuit with the  41 H choke and RF choke.

I decided to try driving the modulator grids, screen and suppressors all tied together using the GFZ MOSFET board. I was going to ground the suppressor but something tells me to try it all together first.  The tone tests will tell which is best. No one seems to agree on this one.

T

[w9jsw]

Let me know on that. I have my mod suppressors tied to ground per the schematic. How are you driving the grid/screens on the modulator. GFZ suggested thru a 26 ohm R with a 300pf to ground. What are you using?

How much P-in for 700W carrier? What are you driving it with?

Do you have any diodes in the center tap of the fils on either pair?

John

[K8DI]

Your hint to Tom re. possibly unconnected suppressor grids. has me thinking maybe you swapped the control and screen grids, or suppressor and screen...”big holes” mirrors a comparison of the various grids....  The screen grid still would act like a grid just like it does in triode connected mode as you’re testing with the modulators, but the output would be down..

Ed

[Tom WA3KLR]

Now I'm wondering about low filament voltage or current meters and shunting, volt meters  not set up properly.

But now going along with Ed - screen and suppressor swapped. (Toob semantics - in a beam power tube, what's drawn as a suppressor grid is the beam-forming plates, only rarely shown as beam-forming plates on some tube schematics; beam-power 'pentodes' are actually tetrodes.)

[K1JJ]

Your hint to Tom re. possibly unconnected suppressor grids. has me thinking maybe you swapped the control and screen grids, or suppressor and screen...”big holes” mirrors a comparison of the various grids....  The screen grid still would act like a grid just like it does in triode connected mode as you’re testing with the modulators, but the output would be down..

Ed

BINGO , Ed, congrats!  You guys are good!

Yes, I had swapped the the screens and suppressors by mistake.  Hope you can follow this:  If you look at the datasheet, it shows a socket "bottom view" with the screen on the left and the suppressor on the right. Now, if you flip the picture top to bottom to get a top view, the screen and suppressor stay in the same relative positions left to right. But if you flip it over left to right, the screen and suppressor swap positions.  This  is just the way the world works and it fooled me.

All the other tube elements (fil and grid) were correct because they were originally at the top or bottom and didn't flip when I flipped  it side to side..  The reason I ran into trouble is because I needed to work from both the top view and bottom view due to my broadcast style build. Working just from the bottom of a chassis is easier to visualize rather than both.


Anyway, the screen voltage was feeding the "lossy" and web full of holes suppressor. Whereas the real screen was grounded. Thus, the "screen" current, while still working, barely attracted electrons to the suppressor while the real screen was grounded and barely worked too.


So the problem was human error. The funny part was that I was using an old 813 socket that someone before me had marked the tube elements in pencil. I noticed that he had the screen and suppressor swapped backwards and I smugly laughed that he probably never got the rig running right. But I was the real fool and should have double checked after that warning.

Look at page 7 on this 813 datasheet and you will see how this could happen...  bottom view and top view can be tricky depending on how we orient the picture.  
https://frank.pocnet.net/sheets/079/8/813.pdf

So, I wired both the modulators and the finals the same way.  My reference penciled in labels were wrong, so I kept measiring the wrong points thbinking they were corret. Never trust anything when troubleshooting.


John, I have some more data to answer your questions later today.  As a teaser, I ran the HV up to 3KV last night and tuned eveything for maximum carrier output. I was seeing 1300 watts carrier into the Bird wattmeter. I doubt it could be modulated cleanly at that level but it took it with no problems. I plan to run it around 1/2 power or maybe 500 watts for best cleanliness, but that point will have to be determined. My 500 watt mod iron is the limiting factor and it would blow at 3KV I think.  I like to push a rig to the limit in testing and then back way off for reguar use.

I am using the arc "snubber" quencher circuit with a vcuum relay and 3K power resistor across the mod xfmr sccondary. No unkey arc overs at all even with no sequencing. The mod iron magnetic collapse goes right into the power resistor. That is one FB circuit.

I also got rid of the broadbanded RF input circuit because it showed 2:1 on some bands. What worked better for perfect 1:1 on any band... is an L network.  50 ohms >  L series > 350 pF cap to ground > into grids with 1K non-inductive to ground swamping resistor.  Very stable and versitile.

T

[K1JJ]

Question:

https://frank.pocnet.net/sheets/079/8/813.pdf

I'm trying to determine what to do with the suppressor with my 813 modulators. The data sheet above says to ground it to the filament center tap. That is about 24" away from the tube.  Should I bypass it for audio at the pin socket and run it to the fil CT?

I plan to drive both the grid and screens for audio but not sure about the suppressor. I don't remember if I grounded it to chassis or drove it with the grid and screens in the past. Whatever I did, it tested quite good and never popped a tube.  The difference between a direct chassis ground and the fil CT is about 14 VAC, the fil voltage, so it seems to be a fragile voltage rating.

Wouldn't the suppressor be across the filament xfmr voltage and modulate the suppressor with 60 Hz?

Also, diodes for biasing in the cathode would put more voltage potential on the cathode/suppressor area.  I have used diode biasing for years with 813s with no problems.

It appears the voltage rating between the cathode and suppressor is quite low because the data sheet always shows a zero rating for the suppressor.  So maybe it is wise to ground it. I noticed that it DID have an effect on tube gain when I put screen voltage on it by mistake. It did not blow out with 400 volts, so I just don't know...

Any opinions on this?

T

[K8DI]

I don't have an opinion on this but point out a couple ideas/informational items...

https://frank.pocnet.net/sheets/049/8/803.pdf links to the 803 tube and contains data for grid, suppressor, plate pentode, and plate tetrode class C amplifiers, one of the more complete datasheets I've found.

Suppressor modulated, to get 53 watts of RF, with 2kV plate and 80mA plate, takes 150v audio on the suppressor, along with a  substantial negative bias.  It's so-called efficiency modulation. same term is applied to grid modulation, which, while accurate, is kind of a misnomer -- not efficient modulation is more like it...

It seems not unreasonable that the 60Hz ac from the filament-suppressor connection would end up being a minor part of the modulating going on with a plate modulated rig. Note also on this data sheet the maximum voltages for the suppressor -- that would give me some confidence that the wiring error wouldn't have hurt the 813's either.

Ed

[K1JJ]

Thanks for the additional thoughts on the suppressor, Ed.    The 803 certainly looks like it will handle suppressor voltage from the specs.  

Since it's already wired up as grid-screen-suppressor driven, I will begin my audio tests in this config.  I did some RF tests tonight after making some general changes and the rig is now very stable and dependable for RF.  I did an efficiency test at 1KW AM carrier out, (Bird watts) 2800VDC  with an input of 1092 watts. This is 91% efficiency which is obviously not correct. The Bird is probably reading a wee bit high due to a little swr.   But I am reading 2800V at 390 mA.  The tubes show no signs of color so I won't complain.  I am running 35 mA of grid current and 75 mA of screen current. (2 tubes)   And yes, I did subtract the screen and grid current from the cathode current to make the calculation.   The tank coil runs cold. I'm running the recommended -175V of class C negative grid bias.  25V fixed and 150V grid leak.  The plate dip is quite sharp due to the Q=18 or probably higher. Chuck advised a high Q, so I'm giving it a try. So far Hollywood  seems to like it.  I have tested just 75M so far.

Anyway, I did have some instability with both the grid and screen suplies. The fixed protective grid suppply was charging up from the RF drive. The screen voltage was dropping too much with RF drive. Bottom line is I was bleeding both supplies too lightly. I added enuff bleeder power resistance to bleed at the same rate as the requirement. IE, for the protective grid supply, I am now drawing 35Ma of bleeder current and 35 mA of grid current = 70 mA total. This solved both problems and the supplies are now stable.  I think this is an important thing to be aware of.


I changed my RF input swampling resistor to 1K. It's a 20 watt Glowbar and makes a perfect match for the L network input.  1:1 on any band, any freq.



So now that the RF and power supplies are OK, tmw I'll hook up the GFZ MOSFET audio driver and put some audio tones thru the rig. I'll need to hook up the NFB (negative feedback)  and see what we have for freq response and just how clean this rig is out of the box.  I have all the tube voltage and current parameters working by the datasheet, so the audio will tell the story.  If it works out well, I might even put it on the air, though I am doubtful things will go as smoothly as that... :-)

I was thinking tonight that this may become my favorite rig for a while. It's rare that everything comes together so well without major problems.  The audio testing and troubleshooting coming up is really the hard part, so I won't celebrate just yet.

* I'm anxious to get back to the new 4D32 plate modulated summertime rig. Got all the parts on the bench and the chassis and panel are just beginning to get drilled and blasted.  Pictures will be posted once I make some progress.  I need a little rig to add to the harem.

T

[K1JJ]

Let me know on that. I have my mod suppressors tied to ground per the schematic. How are you driving the grid/screens on the modulator. GFZ suggested thru a 26 ohm R with a 300pf to ground. What are you using?

How much P-in for 700W carrier? What are you driving it with?

Do you have any diodes in the center tap of the fils on either pair?

John

John,

I am using 39 pF cap bypasses at the modulator grid tube socket pins.... no resistors. If it becomes unstable I will add some filtering as Frank suggests.  I have plate parasitic suppressors for now.

I am using an FT-1000D for carrier drive, about 15 watts.  You will need less drive cuz I am using a 1K swamping resistor. It's all I had non-inductive but makes the rig rock stable.

For now, I'm not using cathode diodes on either the finals or the modulator.  I'm finding that the modulators are idling about right between 2500 and 2800 VDC with the grid/screen and suppressor all grounded to the chassis. I can later set the GFZ board bias if need be to idle them anywhere.

The plate power efficiency seems superb. The rig just wants to run and put out power. A pair of 813s X 813s really are a miracle rig if set up right with plenty of plate voltage a clean audio driver and BC mod iron..  This is my third and best 813 build ever and I like it!

T

[Tom WA3KLR]

Third time is the charm.  I hope to work you soon OM.

[K1JJ]

This one has to go into the books as a real BA-buster.  I spent the good part of the day testing the audio system and 813 modulators.   The first time I fired up the rig with the modulators on, the audio was extremely distorted.  Probably the worst I've ever seen.  There was just a carrier with no peaks at all, just a negative collapsing carrier as the tones were applied. There were audio spikes and some parasitics mixed in.  The modulators were drawing huge current - pinning a 500 mA meter as the carrier collapsed and the grid meter jumped up.  Cut me some slack!

I worked around it and checked all the schematic wiring and fixed some other related audio stuff, but it persisted.  Nothing I could do or check could make it better.  The GFZ SS audio driver was putting flawless audio into the 813 screen/grids as designed. It was happening after the input stage.

I convinced myself that this was a blown mod xfmr. I had used it in the last two 813 rigs and the last time I had popped a wire at the terminal, but it was repairable -  but I could not be sure.   I then disconnected the RF final, put a 2K 200 watt resistor in place and ran some tones thru. They were better, but still pretty nasty looking.  I started to think of where the heck was I going to get a BC mod xfmr or maybe I shud convert the 813s into a series modulated rig..

The RF portion of the rig acted flawless, but there was something seriously wrong that was limiting the positive peaks and seemed to be a big load on the audio system.

Man, was I happy when I found the problem!  This is a surprising cause but once found the audio leaped right into good form and the positive peaks stood high and proud. The high end looks great though I don't have the NFB hooked in yet.  However, the low end starts dropping off at about  50Hz, probably due to the temporary 18H of mod reactor chokes I have. I need to get some more choke iron to bring it down below 30 Hz. The NFB will help clean up the extreme low's waveform a lot.

Anyway, this problem was not due to bad parts or operating parameters...  

Hint: I could spend 10 seconds with any opened plate modulated rig and duplicate this problem and then fix it in 10 seconds.


Can anyone guess what caused it??

T

[WR6J]

The high plate current on the mod tubes sounds like a biasing problem or maybe a floating grid. Was the control grid floating? Or maybe the suppressor?

Richard

[K8DI]

Mod transformer wired to wrong side of Heising choke??

Ed

[Tom WA3KLR]

Only 10 seconds?  Plate cap on one modulator tube not connecting, the one which drives the PA voltage positive. But high current?

[K1JJ]

All good guesses!   In fact many of those suggestions I checked at first.  The GFZ SS board provided good bias at the grids, plate caps were tight and had continuity - and the mod xfmr was eventually wired right. No change in severe distortion.

I hate to admit it, but one of the things I found was the modulators were STILL  wired with the screen and suppressors swapped. Thought I fixed it.  The upside-down diagram fooled me a second time!   And, one of the plate leads to the modulators was swapped with the modulated HV lead. (The mod xfmr is mounted remote, so the leads are bundled)  I checked it several times before but somehow missed it. That is a sign of getting old I guess... :-)  Still, no change.

Anyway after all of this troubleshooting it became the process of elimination -  a clearer idea to what is was not -  but the severe distortion was still there. I did this at 1500V to be sure I didn't break anything else. At 2KV the results were the same, just nastier looking.

So, what was the problem?  The ending is a head slapper for sure. 

T

[Tom WA3KLR]

One modulator tube not installed?

[KD6VXI]

One modulator tube plate cap not connected or bad connection....?

--Shane
KD6VXI

[K1JJ]

One modulator tube not installed?

HA!  With a big viewing window all tubes present or accounted for... and the AC cord was even plugged in.  

Nope, not a loose connection. It was something more subtle.   It was something else that wud cause tremendous distortion, huge modulator current, next to no carrier modulation, heavy audio loading, spikes, etc...  

Another Hint: EVERY big plate modulated rig should have one, (and 95% do) but it is NOT required to produce a big, clean signal.  

T

[WD5JKO]

The spark Gap across the modulation transformer secondary distance set to zero!

[K1JJ]

The spark Gap across the modulation transformer secondary distance set to zero!

BINGO!  Right on the nose, Jim!

You've solved a few of these so far. I'd hire you in a second to troubleshoot my stuff....

Yes, it was the spark gap on the modulation transformer secondary. The gaps were pushed together (out of sight) creating a dead short across the winding.   I had them set correctly when I first installed the iron in the cabinet, but somehow they got pushed out of alignment.  I had to use a mirror to look at them.  It all makes sense now.

It was like putting a  clip lead jumper across the winding terminals. Dead short, big current, no audio peaks. 

I just finished some more testing at 2KV. After working with the finer adjustments it's putting out a pretty nice waveform from 30 Hz up to about 9 KHz so far. I am running out of GFZ driver board voltage swing and may need more than + - 275V when running full power.    Also, if I dial in more than about -4 DB of negative audio feedback I get some oscillation at the higher freqs, so plan on some cap bypassing.

But all in all, the major problems are now ironed out and what's left is the finer refining details to make it consistent and sounding very FB. The spearheading is over. Time to call in the mop-up crew.

T