LF Parasitic oscillation?

[W8ACR]

I finally got around to tuning up my homebrew 254W rig on 75 meters, and I have a problem. When I apply plate voltage without grid drive, I get about 150mA of plate current despite having adequate cutoff bias. This only happens with the 80 meter grid coil in place. and with the grid capacitor tuned near resonance. If I detune the grid capacitor enough, the plate current drops to zero. Despite this phenomenon, the final amplifier seems to tune properly when grid drive is applied.

The final is properly neutralized, and it behaves perfectly on 40 meters. I do have a 1mH RF choke in the bias supply lead and I know that this sometimes will lead to LF parasitic oscillations. The B+ plate choke is a B&W 800. I do have VHF parasitic suppressors in the grid leads, but not the plate leads.

Any words of wisdom?

Thanks, Ron

[K5UJ]

I think Rodger WQ9E had a rig with a LF parasitic once--maybe he'll give some advice.  I wish I could remember the details--I recall he had a hard time figuring it out but he did find the problem eventually.   Don't hold me to this but I think it was in his Valiant. 

Rob

[KM1H]

Ive had those critters several times in the dark past thanks to following ARRL schematics.

Finally a wise ham suggested either tossing that choke and replacing it with a 1K resistor or just placing a resistor across it.

Carl

[k4kyv]

I was told long time ago that when building a push-pull triode final, never to use an rf choke from the mid-tap of the grid coil to the grid resistor and bias supply. Instead, just run the grid resistor directly to the rf cold spot on the coil, and connect the by-pass capacitor to ground on the opposite side of the resistor. If not the whole grid resistor, insert at least several hundred ohms of resistance between the coil and the by-pass capacitor. Also, if you use a split stator variable for the grid tuning capacitor, ground the frame directly to chassis; don't return it to ground through a capacitor and connect the frame to the grid bias lead. Directly grounding the frame (rotor) does require more spacing between the plates of the grid capacitor, but it eliminates a lot of self oscillation problems. Those precautions cured the frustrating self-oscillation problem I had the first time I tried to build a push-pull RF final.

The same principle should work with single ended as well.

[KA2DZT]

I finally got around to tuning up my homebrew 254W rig on 75 meters, and I have a problem. When I apply plate voltage without grid drive, I get about 150mA of plate current despite having adequate cutoff bias. This only happens with the 80 meter grid coil in place. and with the grid capacitor tuned near resonance. If I detune the grid capacitor enough, the plate current drops to zero. Despite this phenomenon, the final amplifier seems to tune properly when grid drive is applied.

The final is properly neutralized, and it behaves perfectly on 40 meters. I do have a 1mH RF choke in the bias supply lead and I know that this sometimes will lead to LF parasitic oscillations. The B+ plate choke is a B&W 800. I do have VHF parasitic suppressors in the grid leads, but not the plate leads.

Any words of wisdom?

Thanks, Ron

Do you have a bypass cap on the cold end of the grid choke??  Between the choke and the bias supply.  I never put VHF suppressors in grid leads, always put one in the plate lead near the plate ternimal.  Also, usually you should use a 2.5mH choke in the grid lead.  Do you have any grid bias resistance or is all your grid bias coming from the bias supply??  How is your grid current meter inserted in the grid circuit??

Not sure what your circuit layout is, maybe you can post a schematic.

You say that the rig is properly neutralized but you sure have some type of  oscillation occurring,  may not be low freq.  First thing I would do is get rid of the VHF suppressor in the grid lead.

Try to post a drawing if you can.

Fred

[W8ACR]

OK Fred,

Here is a schematic of the pertinent parts of the circuit. Now that I think about it, I'm prety sure the RFC in the bias line is 2.5mH. I forgot to put on the schematic that grid current is 90mA. Also there is a 500pf bypass to ground on the B+ side of the B&W 800.

Thanks, Ron

[w3jn]

Replace the 2.5mH choke with a 1K/10 watt or so, as Carl said.  It's not a great idea to put a choke in series with another hi-Q coil like that JCL, you're begging for multiple resonances.

Look at some old West Coast handbooks for the right way to do it...

[KM1H]

The grid suppressor is a very good idea, oft times just a 10-20 Ohm carbon resistor is required to keep a VHF parasitic from starting saving the requirement for one in the plate lead.

It took the ARRL decades to get away from that stupid idea of a RFC in the grid bias lead, they must have finally read a West Coast/W6SAI Handbook. It pays to have several reference works to consult.

Carl

[KA2DZT]

The grid suppressor is a very good idea, oft times just a 10-20 Ohm carbon resistor is required to keep a VHF parasitic from starting saving the requirement for one in the plate lead.

It took the ARRL decades to get away from that stupid idea of a RFC in the grid bias lead, they must have finally read a West Coast/W6SAI Handbook. It pays to have several reference works to consult.

Carl

I always used a grid choke, maybe out of habit.  The difference is I never used that split-cap type grid tank circuit.  That circuit works like a balanced bridge.  The bridge balance changes when the tube is in operation.  Under different conditions the center of the coil is probably not at a low potential (if I understand the circuit correctly).  This may be creating the unstable condition, especially with the choke.  The West Coast Handbook doesn't recommend that circuit be used above 7mHz and below that freq be used only with low capacity tubes (OP is using two tubes, twice the capacitance).

I always use a single ended grid tank coil that I wind close spaced (hopefully to keep the Q low). The hot end is connected directly to the grid. The cold end is bypassed with a 250-300pfd cap.  The feedback voltage is injected at that point.  The NC and the 250-300pfd cap create a voltage divider for the feedback voltage.  From that point I use the 2.5mH choke which has a larger bypass cap on the low end.  From that point you can add you grid resistor, bias and meter to complete the grid circuit.  I link couple power to the grid coil at the cold end.

All this is fairly standard circuit design, nothing novel about it.  In fact, I never even record schematics of the RF sections of my xmtrs.  I figure that anyone who needs a drawing to figure out the circuit shouldn't be working on the xmtr to begin with.  I do record all the PS's, relay control circuits, audio and other circuits that are not usual.

Ron,

At this point, I guess you can leave in the VHF suppressors as suggested by Carl, especially since there isn't any in the plate leads.  Try replacing the 2.5mH choke with a resistor as suggested by Don, Carl, and W3JN.

The thing I would avoid is using a wire wound resistor.  Use a few carbon resistors to make up the suggested 1K ohm at some wattage.  Put a .01uf bypass cap at the low end of the resistor.  Putting the resistor and the bypass cap on the coil essentially puts a load on the coil at that point.  This lowers the Q.  You can lower the Q of any coil by putting a resistor across it.

In fact, maybe putting about a 10K carbon resistor across the whole grid coil may solve the whole problem (you'll need some wattage and more drive power).  Trying different ideas is what home brewing is all about.

Good luck and welcome to the real world of home brewing.

Keep us posted on your progress,

Fred

[W7TFO]

Frank Jones was in the habit of using a light bulb in the 50- to 100Watt class as a swamping load across the secondary of input coils like yours.

73DG

[W8ACR]

I think I figgerd it out. First, I changed the RFC in the grid bias line to a 1000 ohm WW resistor. No joy. Then I put a probe from my freq counter into the tank coil and got a reading of 57 MHz. So I assume this was a VHF parasitic. I rearranged (shortened) the leads from the final plate to the neut cap and also to the output cap. I was able to shorten them by several inches. I also changed the leads to copper strap rather than wire. Voila! no more runaway plate current! Will reassemble everything and tune up tomorrow, and give another update then.

Ron

[KA2DZT]

Ron,

Glad you're making progress,  Use carbon resistors for the 1K.  WW resistors can act like coils and may cause other problems.

We hadn't looked at your output circuit and layout but good work cleaning up your connecting wires in the tank circuit.  I always use heavy silver plated copper braid in the output circuit.

Good layout of components and wiring is just as important as the circuit design.

FB on using your freq counter to pickup what freq the oscillation was on.  Before freq counters one would have to use an absorption wave meter that had a little light bulb for an indicator.

Let us know if you've got it solved.

Fred

[KM1H]

I think I figgerd it out. First, I changed the RFC in the grid bias line to a 1000 ohm WW resistor.

A 2-3W MOX resistor should be fine, the WW will be inductive even though its made with nichrome. The grid bias isnt drawing much power in that amp. When a lot of grid current is involved a 1-2K across a RFC will swamp it enough to tame.

No joy. Then I put a probe from my freq counter into the tank coil and got a reading of 57 MHz. So I assume this was a VHF parasitic.

Either that or it wasnt completely neutralized and you had a TPTG oscillator. Ive used a 10-20K swamping resistor across the secondary especially with parallel tubes when I have drive to burn, want somewhat broadband tuning, plus it helps the neutralization to cooperate.

I rearranged (shortened) the leads from the final plate to the neut cap and also to the output cap. I was able to shorten them by several inches. I also changed the leads to copper strap rather than wire. Voila! no more runaway plate current! Will reassemble everything and tune up tomorrow, and give another update then.

That is a huge step in the right direction, wide strap and short leads are the only way to go.

I get extra fussy when building amps since they have to work 160-10 or 80-10 for most customers and for myself. Plus i do a lot with 6M also where things can get real touchy. After 50+ years Ive put most of my bad habits behind me

Carl

[W8ACR]

OK, it seems that the problem is solved. I tuned up the rig on both 40 and 75 meters tonight with no problems noted on either band. With no grid drive but plate voltage applied,  I was unable to cause any plate current to flow at any settings. Here are the changes that I made.
1. I changed the grid bias line RFC to a 1000 ohm WW resistor. The 1959 west coast handbook actually recommends a WW resistor, and at least in my case, it seems to work OK.
2. I removed one of the finals so that the rig now only runs one tube rather than two in parallel. I was previously tuning it up to 180 mA with two tubes, now I tune it up to 160mA, so I haven't lost much power. I have about 290 watts input and about 210 watts output with one tube. I may go back to two tubes in parallel eventually. I have a total of five good 254W's. I don't know if my overall tube life would be better by running one tube at a time full throttle, or two in parallel at reduced settings. I was able to get 400 watts of carrier with two tubes full bore. I am seeing small spots of purple glow inside the glass envelope with power applied. Pretty, but I assume this indicates gas that shouldn't be there.
3. I had to change the grid leak resistor since I am now only running one tube, and I reset the neutralization capacitor. Again, the circuit seemed to neutralize properly with no problems.
4. I changed the output coupling capacitor from a 500pf/10KV large molded cap to a high quality 1500pf/15KV ceramic doorknob unit.
5. I significantly shortened the leads from the tube plate to the RF choke, and from the RF choke to the output coupling cap. I probably saved 5 to 6 inches of lead length total.
6. I changed all the plate leads from heavy wire to copper strap.

Ron

[W7TFO]

Hey Ron,

Blue glow indicates gas only if it within the space between the fil & plate.  Blue on the envelope inner is just ions, no problem.

BTW, the xfmr arrived fine!

73DG

[KA2DZT]

Ron,

I'll have to check the WC Handbook about the use of WW resistors in the grid circuit.  Once the grid return circuit goes past the first bypass cap it doesn't matter what type of grid leak resistor is use.  Not so sure if the resistor is coming from the coil.  But, if it's working, leave it as is.

FB on all the other changes, it's working.

Is your rig going to be a linear amp or plate modulated AM??

Fred

[W8ACR]

Hi Fred,

Thanks for the comments. This is a classic plate modulated AM rig, class C final amp, class B modulator. Everything is homebrew except RF exciter, which is a Viking I. Speech amp, modulator, and both power supplies are all homebrew. The RF plate supply is capable of either 1800VDC or 2200VDC @ 275mA - full wave center tap using 3B28 rectifiers. Modulator is modified Heising using 811A's, S22 mod transformer and a 50H Dahl audio reactor. Plate supply for the 811A's is capable of 1400VDC@300mA - bridge rectifier with SS diodes. This rig is an evolving project. I first built it as a single band 160M rig, and had it on the air last winter on multiple occasions. Got very good audio reports. I tried to modify the RF deck to cover 160-40, but couldn't quite do it with the components that I had on hand. I had it on the air on 40 meters a few times and again got very good reports. After some additional changes, I was able to successfully add 75 meter coverage, but then I ran into the parasitic problem that started this thread.

So now I have a usable RF deck for 75 and 40 meters. As for 160 meters, I'm goint to build a separate, single band RF deck with a single 8000 tube running about 400 watts input. I will use the same modulator and power supplies, and simply switch RF decks back and forth. I rarely get the urge to work 20 meters or higher, and if I do, I can simply use the Viking I barefoot. Hopefully will have all this stuff done before winter so that I can actually get on the air occasionally! But I must say, this project has really been fun. I've learned a lot, and it's very gratifying to say "rig here is homebrew".

Ron

[KM1H]

1. I changed the grid bias line RFC to a 1000 ohm WW resistor. The 1959 west coast handbook actually recommends a WW resistor, and at least in my case, it seems to work OK.

Neither handbook should be considered the last word on any subject except maybe Ohms Law

In the circuit shown the inductance of a WW may or may not have any effect, particularly on a low frequency; sometimes we get lucky. The bypass takes care of any residual RF flowing thru it but the inductance could still cause an unwanted resonance. In a perfect world the attachment point on the coil has zero RF voltage. In fact a bypass cap ahead of the choke or resistor should theoretically have no effect.

2. I removed one of the finals so that the rig now only runs one tube rather than two in parallel. I was previously tuning it up to 180 mA with two tubes, now I tune it up to 160mA, so I haven't lost much power. I have about 290 watts input and about 210 watts output with one tube. I may go back to two tubes in parallel eventually. I have a total of five good 254W's. I don't know if my overall tube life would be better by running one tube at a time full throttle, or two in parallel at reduced settings. I was able to get 400 watts of carrier with two tubes full bore. I am seeing small spots of purple glow inside the glass envelope with power applied. Pretty, but I assume this indicates gas that shouldn't be there.

Letting tubes loaf always has a good effect as long as running the anode at color isnt required as part of the ongoing gettering process. I dont know the H&K spec but try running the anode a bright red to dull orange for 10-15 minutes without RF by changing the bias; a bit of air flow should be used. With some particularly gassy 3-500Z's Ive had to cook them for several hours and almost a day with a few 4-1000A's. I probably could have bought good tubes considering what the electricity cost
Filament life isnt dependent upon how its run but filament electron emission is. In your case I take it that you are drive limited?

Carl

[W8ACR]

I had to do it. I couldn't help myself. I put that second final tube back into the circuit. 200W of carrier just wasn't enough when I knew I could do almost 400. So I put it back in there, changed the grid leak resistor, reset the neutralization, and hit the HV, and she's behaving beautifully. No parasitic. 380 watts of carrier on both 40 and 75 meters, modulating very nicely on the scope. So now I gotta get a cable out to that 40 meter vertical, and mebbe I will be able to talk to you east coast guys.

Ron

[KA2DZT]

Ron,

Leave that second tube in the rig, go for it.

Glad everything is working.  And remember, you figured out your problem yourself.  Most of the help we gave you didn't seem to be the cause of the oscillation.  Of course, in our defense,  if we could have seen the rig close up, we would have seen the too-long wires and suggested shortening and replacing with branded copper strap.

So we're taking partial credit.

FB on trying to talk to us east coast guys.  After a few QSO's with the east coast crowd you'll probably want to take up a different hobby

Fred

[W8ACR]

OH Fred,

You guys can't be all that bad . I used to be one of you. Lived in Plattsburgh, NY for 6 years active duty USAF. Was on 160 meter AM frequently. Member #188 of Grey Hair net. Checked in frequently to Old military net on the T368. Fond memories of QSO's with N2KSZ(SK), KW1I, KK1K, W3PWW, WA3PUN, K4KYV, W1CKI, K1DEU, WA2PJP, W2VJZ, AJ1G, WA1HLR, WA3YXN, WA3VJB, K1JJ, Ashtabula Bill, and many others. Some of us just wised up and went west - probably raised the IQ of both regions.

73, Ron