Modulation transformer arcing

[K4NYW]

I have a theory about a problem I encountered and seem to have solved - but I invite comments about whether my thinking is screwball or I just got lucky or what - this is not like anything I have run across or attempted to solve before.

I am restoring an AN/FRT-17 transmitter (4-400 modulated by 4-125's) which I had completely disassembled for cleaning.
When I got it back together I saw HV arcing on the scope at both pos & neg modulation peaks.
Shows up on both modulator plates, modulated B+, and modulated RF. Fastest rise times appear on modulator plates. Shows up at the same p-p modulation voltage whether I run in TUNE mode (2kv) or OPERATE (2.6kv). Can't get above 30% modulation or so before this happens.
So I gradually tried to isolate it.
> Listened - mod xfmr buzzes a little when arcing appears on scope. Is this a direct or secondary symptom?
> Turned out the lights - can't see any blue glows anywhere.
> Cleaned all the terminals/insulators (again) - no change
> Disconnected leads to shorting relay for CW operation - no change
> Replaced modulator plate to mod xfmr wiring - no change.
> Disconnected PA screen winding on mod xfmr - no change.
> Disconnected PA plate/screen and connected 10K 300w resistive load to mod xfmr
   AHA - no arcing with B+ disconnected from load, BUT the arcing returns if B+ is connected.
> So I tried switching the mod xfmr secondary terminals
  > And got no arcing! See scope photo of modulated B+ (1 KV / division)
http://www.navy-radio.com/xmtrs/frt17-8/DSC02244.JPG
And once I hooked the PA back up I got some real AM
http://www.navy-radio.com/xmtrs/frt17-8/frt17-am-131208.JPG

So here is my hypothesis -
Let the primary  terminals be:
A plate1
B 3kv
C plate2
and the secondary terminals be:
D 3kv
E modulated B+
 
Now suppose that the A winding end is physically closest to the E winding end and that phase relationships are such that when A is at min (~0v), E is at max (~6kv) (und so, ve get der sparken). 1/2 cycle later A is at max and E is at min and the sparks will fly again. Seems to describe the symptoms.
 
Now suppose we swap connections to DE - now A-E will be 3kv max instread of 6kv max. C-E will now be 6kv max but they are farthest away from each other. Whaddya think????

[KA2DZT]

Have you considered that you may have a bad mod xfmr?  Connecting the xfmr as it was originally designed to be connected, the rig should work.  I have to re-read you post. 

What type of mod xfmr is in the rig??  Open frame, potted or what??  Potted xfmrs can be a problem,  it may have developed carbon paths through the potting to the frame or between terminals.

Fred

[w4bfs]

It is possible that this is a parasitic oscillation .... need to try some experiments to determine if this is so ... sniff with a neon bulb on a stick ( insulated of course )  .... the scope display is not enuff info to tell

[KA2DZT]

I re-read your post.  The class C Terminal of the mod xfmr takes the biggest abuse.  The B+ end never sees more that the B+, not so at the other end of the secondary winding.

You stated that you reversed the secondary connections and you had no arcing.  This tells me that there is a problem on the original Class C terminal.  Could be a bad insulator or if it's a potted xfmr, there may be a carbon path through the potting to the casing.

Maybe you can post a pic of the xfmr, might help.

Fred

[K4NYW]

Thanks for the feedback, guys - some responses -

This origin of this problem is mysterious to me since the transmitter was working at W3FJJ's before I transported, disassembled, cleaned, and re-assembled it. I initially re-wired the mod xfmr as it was and as the manual's schematic & wiring diagram show it. I can't imagine what I might have done in the process to induce this internal arcover - perhaps something related to cleaning (windex) or soldering...or perhaps due to removing the protective coating of mouse pee & poop....
If you want to see the whole transmitter project as I stripped it down and reassembled it -
http://www.navy-radio.com/xmtrs/frt17-2.htm

This is a steel encased military mod transformer, I presume oil-filled. I have no idea how the thing is actually built inside - The parts list has a Federal Tel. drawing number. HV terminals are typical 1" ceramic standoffs with solder lugs. Transformer (before cleaning) is here
http://www.navy-radio.com/xmtrs/frt17-2/DSC01136.JPG
A photo taken while I was experimenting with temporary connections is here
http://www.navy-radio.com/xmtrs/frt17-8/DSC02241.JPG

I am really doubting any arc to casing/ground since the arc shows up on negative peaks as well as positive.

Yes my theory is that there was breakdown inside the mod xfmr between one end of the primary and one end of the secondary. And my fix moved the "hot" end of the secondary away from the mod plate lead that was out of phase with it. The schematic shows an electrostatic shield between primary and secondary but again I have no idea of internal construction.

The scope doesn't show any problem until I reach some p-p maximum and then it definitely looks like arcing to me (hash) rather than parasitic oscillation (on my 300MHz scope). Sorry, I should have taken a photo, but I was so happy to have gotten rid of the problem that I didn't take any until afterwards.

Also I'd think parasitics would change when I substituted different length wires from mod tubes to mod xfmr (this is typical military and the mod tubes are in a pull out drawer, the mod xfmr mounted to the main frame and wires are in a harness, maybe 6'-7' long). Also I'd think parasitics wouldn't appear/disappear when I run the resistive test load with and without B+ connected, but it definitely does. And I wouldn't think parasitic behavior would be exactly the same with a substitute resistive load or with the PA as load.

Cheers,
Nick K4NYW

[N2DTS]

You might try insulating the mod transformer from ground, I do that with all my mod iron.
At those voltages, you need to make sure all the wiring that handles the modulated voltage is good and not close to anything else.
With 3KV, expect spikes up to 15 KV.

I run 2kv on my 813 rig, and when I had a negitive cycle loading deck in line, I was blowing out 17KV diodes...


What a beast! Way too complex, way too many tubes, but a bunch plus of very good parts.
What a nice cleanup job.

Bet that thing cost a pretty penny when it was new...

[AB2EZ]

Nick

I agree with your diagnosis, based on the experiments you performed, and the results you obtained.

The next question in my mind is whether this "fix" will hold.

The arc path may have formed just due to the long term effects of the high field between the primary and the secondary windings at one location where the windings are close together, and where the Faraday shield between them is missing. But, more likely, it resulted from an unusually stressful event... such as a sudden disruption of the plate current in the RF stage. With this type of traditional modulator, a disruption of the transformer's secondary current cannot be gracefully balanced by a corresponding rapid change in the transformer's primary current... because the modulator tubes' plate currents are insensitive to the modulator tubes' plate voltages.

If you can avoid those kinds of stressful events, your "fix" might last indefinitely.

Stu

[w4bfs]

I would not be comfortable until I disconnected and hi-pot tested to tranny working voltage .... it just may be about to quit

Stu .... this goes back to our previous discussions about how difficult it is to find GOOD mod iron .... I came up with a way to use the toroids without violating insulation methods but will require multiple units ...if you missed it, I will start a new thread

[K4NYW]

Thanks guys - I don't have any illusions that this might not be a problem that will recur - but I'm hoping it'll let me move along with getting this beast back on the air. I'll try to avoid dropping drive to the PA in the middle of a transmission, etc..
But if you hear me sizzling some day, you'll know what's happening. Then I'll go stand by the exit ramp with my tin cup and cardboard sign saying "Transformer needs rewinding. Please help."
Cheers,
Nick K4NYW

[KA0HCP]

Nick, Welcome Aboard!

It's great to see you here.  I enjoy and use your website for reference often.

73 Bill, USNR-ret.

[WD5JKO]

Also I'd think parasitics would change when I substituted different length wires from mod tubes to mod xfmr (this is typical military and the mod tubes are in a pull out drawer, the mod xfmr mounted to the main frame and wires are in a harness, maybe 6'-7' long). Also I'd think parasitics wouldn't appear/disappear when I run the resistive test load with and without B+ connected, but it definitely does. And I wouldn't think parasitic behavior would be exactly the same with a substitute resistive load or with the PA as load.

  I agree on that. Still, I was looking at several pictures on your web site that you didn't link to, and I see the little 6" pieces of wire from the 4-125 plates to a nearby ceramic insulator. This would be an ideal spot to add a parasitic suppressor.

  With my Gonset G-76, there are 15 ohm carbon resistors (2 watts IIRC) at each plate of the 6DQ6 modulator tubes. Without them resistors, those sweep tubes will "sing" on an audio peak..like a blocking oscillator that cuts in at a certain threshold, and then drops out at a lower threshold. I am not saying your setup is taking off, but there is a lot of gain there, and I bet the phase margin is small such that if it isn't "singing" on the occasional peak, then it might be awful close.

Jim
Wd5JKO

[W3FJJ]

Nick, Sorry to hear you are having trouble with my old transmitter.
I never had any problem with it modulating 100 percent, but I do remember
a few occasions where I had arc over problems, keyed up with no antenna, low rf drive,etc
 so can't say
mod xfrm never seen any stress. I hope changing the windings hangs in there and look forward  hearing you on the air with it..  Great job with the restoration!

Chuck

[K4NYW]

Hi Chuck - Yes, I'm sure it is due to something I did along the way, because I know it sounded great from your QTH. I did re-do the PA tank circuit back to the original configuration and have got the speech amp back in line - both of which I am sure aren't operational improvements, just back to the original design. I'm pretty close to putting it on the air - only took a year! It has been a great learning experience for me and lots of fun - thanks!
Nick K4NYW

[N2DTS]

This makes a lot of sense to me, some phase thing with the audio, because any time I had something arc, it started arcing and blew the breaker, it did not sizzle on voice peaks.
Once any arc starts at those voltages, it blows fuses/breakers in most cases unless its in something like a vacuum cap.
Once you ionize the air, it keeps going.....

Also I'd think parasitics would change when I substituted different length wires from mod tubes to mod xfmr (this is typical military and the mod tubes are in a pull out drawer, the mod xfmr mounted to the main frame and wires are in a harness, maybe 6'-7' long). Also I'd think parasitics wouldn't appear/disappear when I run the resistive test load with and without B+ connected, but it definitely does. And I wouldn't think parasitic behavior would be exactly the same with a substitute resistive load or with the PA as load.

  I agree on that. Still, I was looking at several pictures on your web site that you didn't link to, and I see the little 6" pieces of wire from the 4-125 plates to a nearby ceramic insulator. This would be an ideal spot to add a parasitic suppressor.

  With my Gonset G-76, there are 15 ohm carbon resistors (2 watts IIRC) at each plate of the 6DQ6 modulator tubes. Without them resistors, those sweep tubes will "sing" on an audio peak..like a blocking oscillator that cuts in at a certain threshold, and then drops out at a lower threshold. I am not saying your setup is taking off, but there is a lot of gain there, and I bet the phase margin is small such that if it isn't "singing" on the occasional peak, then it might be awful close.

Jim
Wd5JKO