HV Plate Chokes - Rolling your own - Some Thoughts and Questions

[K1JJ]

Hola!

I'm in the processing of winding up a bunch of HV plate chokes for the various linear and plate modulated rigs here.

A friend said that in the old days, some of the guys would make their chokes as much as 1000 uH ! for 160-10M. They wanted to make sure that there was absolutely no HV supply influence on the RF tank circuitry.  That is a BIG choke.

A web article mentioned that a standard 100 uH choke can be brought up to 1200 uH using a ferrite rod inside. It said it didn't get warm. I once tried it and the ferrite got so hot on 160M, it snapped...

Today, I wound an air choke that is 1 5/8" diameter, 105 turns, 7" long using #22 insulated wire. It calculates out to be about 95 uH. This is about 2,200 ohms inductive reactance at 3.8 mhz according to the web calculator. 

What is the guideline used  for plate impedance vs: choke impedance? I always thought that a 10:1 ratio was used for most circuit isolation jobs, but in this case the tube plate impedance will be about 2K and the choke about 2K.  There would seem to be a big interaction potential here between RF tank and HV supply. 

This amp is going to be the RF negative feedback 75M monobander I talked about before, so I'm not worried about a bigger choke having capacitance between turns or other problems on the higher bands...  I want this amp to be be as close to perfect as possible, since I'm shooting for -55db 3rd order figures, etc.

I've used similar 100uH chokes in the past that "seemed" to work FB on 160/75M, but should I be winding something much bigger?  My buddy, who is also building a 4CX-3000 linear wound up a choke that was 20,000 ohms!  That must be close to 950mH !

What am I missing here?

T

[WA1GFZ]

Tom,
I never had any problems with a B&W800 in my 4CX3000A I think it is about 90 uH
I would think 200 UH enough for 160 m.  gfz

[K1JJ]

Frank,

I don't understand  how a 2K inductive reactance choke connected to ground [HV supply] from the tube's RF plate can be invisible in a 2K tube plate impedance circuit??

T

[wavebourn]

By the way,
did anybody try ballast chokes from luminiscent lamps?
It seems to me they should work well...

[WA1GFZ]

Gee Tom,
That is interesting since I come up with the same number. A cap in series to ground say 1000 pf has about 50 ohms of reactance. I suppose it would be a real problem if the final tuning was ever way off or the swr went very high. I have never seen any signs of heating either. Inductive reactance not resistive.
I did blow the wire off a B&W 800  in my old 4X1 once.

The pi network is presenting a lower Z load compared to the RF choke which may be why it hangs in. Also it is resonant.
I suppose it is time to open the west coast handbook.... I thought it should be higher L for 75  m.

[John K5PRO]

Tom
Since you are building a monobander, you can work out a more perfect choke, and get more than 2K ohms of inductive reactance. If you get it very near self resonant, you can see >10K ohms of reactance, which is nearly transparent for a 2K pi network. You can either measure it with a device like Vector Impedance meter or a grid dipper, in the general layout and proximity to chassis that will be used. More the resonance to just above the top frequency, so that it is inductive and you should be good. The best chokes are long and thin, not fat coils like you would use in a Pi or other tank. The Q of the choke self-resonance is lower with something like a B&W800. Something about 5/8 - 3/4 inch diameter and long enough to have close wound enamed magnet wire, and get a high Z is good. I once did a study of various solenoid RF chokes for a high power industrial oscillator at 27-28 MHz ISM band. The long choke won for having slightly larger bandwidth of good operation. Oh, and I used Nichrome wire for the choke, which really de-Q'ed it. Only problem was that it glowed slightly from the 5-10 Amps DC plate current.

[nu2b]

Hey TomJJ,

The 90 uhy (2K ohm) choke will be just about parallel resonant on 3.8 mhz with an additional parallel cap
of about 20 pf supplied by the tube, wiring, etc.

With a 2K load Z, this would make a Q of approximately 1 for the choke circulating current.
This is in parallel with (and won't affect much) the Q of about 10-12 reflected up by the pi-net.

(90uhy) With 1.62 dia and 7 in length and 103 turns of #18 wire the DC resistance is about 0.29 ohms and the ac (skin effect) series  resistance is about 5.25 ohms (Qu is approx 409)

The Vpp at load is approx 6Kvpp=2100Vrms.
The choke circulating current =V/reactance= 2100/2000=1.05 amp
You can calculate the pdiss from the Series Rac since 1.05 amp is circulating in the Choke.
So I*I*R=5.25*1.05*1.05= 5.8 watts.

Under 100% mod this would go to about 23 watts.

I would recommend going with 200uh min if going to 160Meters.
(I did this quickly-so maybe somebody else can double ccheck)
Regards,
BobbyT

[K1JJ]

Thanks for the analysis, Bobby T!

Based on this info and other stuff I read, I wound four chokes that are about 291 uH each using #22 insulated wire.     204 turns on a 1.9" diameter form, 12" long.

That's about 7K impedance at 3.8mhz.  Should be overkill.

Thanks again for the info.

73,
T

[W4AEE]

One reply eluded to your problem.  Certainly, isolation of RF energy from ground by virtue of the choke's reactance value is important.  But you can't just wind a bunch of wire on a form, trying to maximize inductance and expect it to work.  The key is to avoid self-resonances (chiefly parallel resonances) in the choke.  Any choke has not only inductance but also distributed capacitance between turns and sections or pi's, (if present).  These form parallel resonant circuits (little tank circuits, if you willl) within the choke.  The property of a parallel resonant circuit is that it has maximum impedance at the resonant freq.  When that happens, RF circulating current inside the parallel combination is maximum.  That means voltages built up across the capacitive and inductive reactances will be maximum.  If the choke has one or more self-resonances near the transmitter operating freq....ZAP!  Use a grid-dip meter and alter the choke in number of turns, distribution of turns, spacing of turns, dia. and lenght of form, etc. so that all reasonances are far from the operating freq. (best if above it). 

73,
Mike Carroll, W4AEE

[K1JJ]

Hi Mike, and welcome to the board!

You are correct, of course.

Self resonance is a killer for plate chokes.  Though, measuring them out of the circuit will give readings that are different than when connected in the circuit with all the effects of the tube's capacitance, stray capacitance and additional lead lengths added in, etc.

So, it may sound a little JS, but I don't bother measuring them anymore before use. I look at it this way: To wind a choke that happens to fall within a ham band for self resonance is a harder thing to do than have it not resonant in a ham band. The hams bands are narrow ranges compared to the whole spectrum. Probability is on our side. So I just wind it and fire it up and wait for the fireworks.     

Over perhaps 35 years and winding maybe 30 big HV chokes, I can remember only two that blew up. They imploded and melted the wire - had to peel it off the form in chunks. No other damage to parts. I just cleaned it up and wound a new set of windings, this time with more or less turns and everything worked FB on all bands. I usually check the choke for warmth after running it hard on each band with various tune/load settings, trying to break things.. Most of my amps were used on 160-10 including the WARC bands.  But this time around I'm building a 75M monobander, a 40-10M amp and a separate 6M amp.  Makes things a little easier and more optimized.   

Probability has a tendency to even out over time, so guess I'm due for another big bang - it's half the fun...

73,
T

[WA1GFZ]

Tom it would be easy to check the choke in the circuit.. Just feed your signal generator into the top of the choke througe a resistor equal to plate Z. Then monitor the top of the choke with your scope. Resonance is when you see a suck out on the scope as you sweep the generator across the bands. (HV off)
BTW if you wind chokes with teflon covered wire be very careful. Burning teflon makes poison gas. fosgene (sp) I think.

[K1JJ]

That's a great idea, Franz.  I will do it when the amps are ready. Seems I saw that technique once on the Amps reflector, but forgot about it.

I ran out of Teflon wire here recently, so have been using regular plasdick insulated wire for the chokes. Considering most are wound with wimpy insulated enamel wire, the plasdick stuff will be FB.

T

[Rob K2CU]

HI Tom,

There are a few simple things to remember. Capacitors are series resonant circuits at self resonance and inductors are parallel resonant.  This means that a cap will look like a low impedance at self resonance ans an inductor a high impedance. IN either case, there will be high circuating currents, just as in the PI output circuit.  Below self resonance, the inductance is doominate and above self resonance, the stray/interwinding capacitance will dominate. The general rule of thumb is to have the self resonance well above the operating frequency. IN your typical shunt fed PI output RF amplifier, the reactance of the RF choke will be in parallel with the transformed parallel RLC circuit of the load and PI netwrok.  The purpose of the PI netweork is to ransform the 50 OHm load to a 2000 Ohm load for the plate in your example.  IF you target a Q of 10, then, at resonance,  the total plate capacitive reactance would be roughly -200j Ohms, and the transformed inductive reactance would be +200J Ohms. AT 3.8 MC, the equivalernt inductor would be about 8 uH. So, the 90 uH choke could be combined to result in  7.3 uH.  This would make an effective inductive reactance of 174j Ohms., upping the plate circuit Q to over 11.  The plate tuning cap would end up being less so that the total plate capcitve reactance is -174j Ohms, for resonance. 

But, what the heck! we use adjustable caps and often roller inductors, and how often is the load truely 50 Ohms resistive anyway, not to mention the true value of the plate resistance of the tube.

POInts to remember: the choke is part of the output matching network and will have some circulating currents in it. Also, the bypass cap at the bottom should be low loss. I would try for inductance of 5 to 10 times the equivalent parallel L, while trying for the hsghest self resonance, hopefully 5 to 10 times F operation. It is a trade off.  Minimize winding resistance losses by using good size enamel wire or better yet, Litz wire.  I like your use of plastic insulation covered wire, preferably the high temp stuff. It increases the windiingi to winding space and reduces capacitance. Easiest way to measure self resonancet is with a dipper.

[K1JJ]

Thanks, Bob.

That's the best description of plate chokes and how they relate to the plate tank circuit I've read.  It takes several reads to sink in fully... :-)   I think G might want to put that as an article under "East Coast Sound".

When using the grid dipper, I take it you can couple it into the choke with it in the circuit - with the tube fil and voltages off, of course, correct?

And, what do you think of Frank's suggestion in a previous post for using a signal generator, looking for suck out - with the choke in the circuit?  Any preference to either technique?

73 and thanks again, OM.

T

[Rob K2CU]

HI Tom,
 
I would dip the choke out of the circuit and with a dipper. you can just disconnect one end. You want the trickey self resonance, not that from being in a circuit.  As to FC idea, the self resonance is parallel resonance for a choke, and you would get a peak, not a null.  I would be concerned about a scope probe's typical 10 pF affecting readiing, and compare the value i get with direct connection to that from a single loop at end of probe with loose  magnetic coupling. Another way top check the validity of any method would be to get severral chokes out on the bench and use the GDO method on each, then connect up with genreator and scope and try these other two methods and compare results. You don't need real accuracy, just ballpark.

rob

[K1JJ]

HI Tom,
 
I would dip the choke out of the circuit and with a dipper. you can just disconnect one end. You want the trickey self resonance, not that from being in a circuit. 
rob

Hmmmm.. that's interesting.   I think I've read on  Measure's site, or maybe it was the Amps reflector, that it must be in the circuit. But you say not. What about the additional tube to ground capacitance and the tank circuit stray?  If a scope probe will add 10pf, won't these other components add at least as much or more?

And yes, with Frank's method, if it is parallel resonace you are looking for, the scope will not suck out.  I can see that..

73,
T

[WA1GFZ]

Rob,
Sorry I couldn't talk today. I was in the middle of a big S$$t sandwitch.
Consider this, in the days before the pi network a parallel tuned circuit was used in the plate circuit not a series tuned circuit. I still think you need to avoid series resonance but it was a long crappy day...................fc

P.S. At 6:10 I just wanted home so didn't get a chance to read your article Rob.   

[nu2b]

Hi Guys,

Two quick comments:

1) The series resistance of the choke can be measured indirectly
by series tuning the 90uh choke with about 20pf and terminating
with 50 ohms. At series resonance the MFJunk meter should read
about 55ohms. e.g. the 5 ohm difference is the choke Rac.

2) The analysis I gave a few days ago can also be worked
bass-ackwards to test the rf choke at the ratings involved.
Use a low power transmitter to force the expected current through
the choke. In the case mentioned, this is about 1 amp rms. Just
use a Ranger at about 50 watts.

Series tune the choke with a vacuum variable of about 20pf
terminated in a 50 ohm, 50 watt dummy load (50 vrms  142vpp).
This is very high series Q so be very careful!!! The series
current of 1 amp will create 6000vpp at the junction of the tuning
cap


Note: Change the numbers appropriately for other chokes. The
measurement problem will increase  at higher uhy values where it
only requires a few pf to series resonate. Be sure to reduce the
current appropriately.
If you can't series resonate, the choke is already close to
parallel self resonance. Try at a lower frequency.

Also, you can find the srf of the choke with the setup above.
Just drive it with a noise bridge and search on your receiver for the null.

Regards,
BobbyT

[nu2b]

Sorry Folks,

I wasn't clear on the noise bridge measurement for
SRF. This could be used if no griddipper is around.The
noise bridge is just used as a broadband noise source
at the input to the series resonant circuit. The SRF
null is measured across the 50 ohm output termination
on the series resonant circuit (not the normal noise
bridge receiver output connector).

Just feed the receiver with a short cable
Tee-connected across the 50 ohm term. You should be
able to estimate SRF within =/- 50 kc or so. 

Please see the attached figs relating to a typical
measurement. The circuit shows the test setup. The 5pf
cap (guesstimate) represents the choke self capacitance.

The second figure shows the voltage across the output
50 ohm termination versus freq. Theres a peak at 3.750
mhz and a null at the SRF of 7.502 mhz.

BobbyT

[nu2b]

Circuit

[nu2b]

Freq Sweep