The AM Forum

I am looking for the permeability for the "classic" ferrite material used in the balun choke providing the filament feed for the 811A...etc.. In lieu of the rod, I was considering a toroid structure to reduce the size of the final unit and I have large toroids on hand; ui in the 800 range. The classic BW FC-xx like FC-25a or the Lafayette MS-333 antenna rod don't provide a tag to the actual ui of the ferrite. I believe the L on these fellows is about 40 uH or so and of course I need to handle the current 4-10 A. Thanks!

Alan

1/2" rods of 61 material are typically used for filament chokes. The permeability is about 125. Amidon has rods and even a kit which includes the wire.

https://www.amidoncorp.com/items/10 Rods
https://www.amidoncorp.com/items/61 Kit
http://www.fair-rite.com/newfair/materials61.htm Specs

The reason you don't see toroids used for this purpose is because the filament current can saturate a toriodal core quicker than a rod, reducing its effectiveness at RF.

bifilar choke does not saturate due to current. Fields cancel with both leads on the rod. Type 61 will have the lowest loss above 40 meters.

Interesting comment on the saturation. So, an AIR GAP is required and it begs the question, if the torus is cut and an air gap introduced would that work? The choke need not be a balun and evidence is 8877 amplifier designs seen show independent rods for each filamnet feed. So... the eye opener is the comment on torus vs. rod and lack of saturation in the rod format. Thanks again!

A quick research look and indeed the air - gap will slice the B-H curve and reduce the ability for the core to saturate. Thanks on the heads up on rod- vs toroid!

""Introducing an air gap in the magnetic
flux path sheers the hysteresis loop so that it requires more magnetizing force to saturate the
core. The more air gap introduced into the flux path the lower the permeability (ratio of B/H).
Note that the saturation flux density is unchanged even though a gapped core requires more
magnetizing force before reaching saturation... ""

Reference: Specify Saturation Properties of Ferrite Cores to Prevent Field Failure
by
George G. Orenchak
General Manager
TSC Ferrite International
Abstract

Again it will not saturate due to heater current because the fields cancel
You only need to worry and calculate for DC cathode current.

RF parts has the chokes already wound at a good price.

I think you're right. I stand corrected. :-)

http://lists.contesting.com/_amps/2006-12/msg00130.html

So, is the use of rods for filament chokes just design tradition?

ok, well this is getting confusing. Lets focus on the 811A. No cathode. In lieu of the ROD, can we use a toroid. If I use 2- toroids of suitable size, each wound with #12 or 14 G wire and I obtain L~ 40 uH for each unit, will that work? A single 811A is just 4 A filament current. I think the goal in this thread is to filter out the folklore. Must the 811A filament feed be via a ROD?

the heater is the cathode in an 811. The center tap of the heater transformer secondary is the DC cathode connection to the tube. Since the choke is bifilar wound one lead is seeing reverse current of the other bifilar winding so fields cancel.
The only current that will effect core performance is the DC offset due to tube DC current and the AC peak drive voltage.
The 1/2 inch rod donated by Lou McCoy was used in my first dual 811 linear back around '75 now sits in my 4CX3000A linear.

So, therefore, a torus bifilar wound works?

A filament choke is there to provide a high impedance between the RF driven filament and the filament transformer at the other end. Loss is what you want and 61 mix is no good unless its for CB or 6M.

For a 4 holer use the 7 1/2"  x 1/2" 33 mix or the 4" for 2 tubes.

The FC-25A and others are 33 mix.

I have both sizes here and plenty of enamel wire up to #10 (for big tubes), PM for pricing, etc.


Carl

I haven't seen it mentioned yet, but Bill/ HG suggested a great idea years ago for a large current bifilar choke.  

Take the two straight fil leads and hold them together. Slip about 12" of large ferrite beads over both wires and secure. No turns required. That's it.

For larger directly heated tubes, (3CX-3000A7, etc) the fil currents can get big and using heavy wire on feriite rods is not easy. By slipping beads over, it's easy and the fil path is short.  I once measured the inductance of one of these chokes on 160M and it was more than enuff inductance.  I don't recall the permabilty/type material of the beads I used, but use something suitable for the frequencies involved. The beads were the large 1" diameter X 1.5" long types.

All of my homebrew linears employ this method these days.

This will work just as well for smaller amps, but for a smaller linear, the standard rod-wound or commercial "FC-XX" series chokes are fine. Just thought I'd mention this in case someone was looking into a bigger fil requirement.


T

I used to sell the 12" x 3/4" 33 rod for the serious QRO years ago but Fair-Rite quit making it.

You can also use one choke for each tube when dealing with a pair of 4X1's or similar.

Luckily a lot of big tubes (YC-156, 3CX5000, and up) these days are cathode type which minimizes the problem.

Coming from no less an amplifier guru than KM1H, this took the wind out of me.

I've always used 61 material for directly-heated filament chokes in the few linear amplifiers I've built or modified, from 160 through 10 meters. 61 Mix is also the choice of Amidon Associates.

But coming from Carl, what can I say? :-)

Is anyone gonna answer this question. He's asked it like three time now.

I'd say yes, since I've seen it done. Anyone else?

heheh - looks like we blew that one.

Yes, I've seen several manufacturers using toroidal bilfilar chokes. I think my Henry amp 3-500Z's uses one. There's no reason why most any configuration like this will not work as long it is of enough inductance.

The only caution I would suggest it to keep it a single layer with no turns overlapped back on itself to prevent cancellation or stray inter-winding capacitance effects. Just use the normal toroid winding procedures and you shud be FB.

Also, wind BOTH wires on the same toroid(s). I have seen hum produced by using separate leads and ferrite.  Frank/GFZ, I am referring to your problem with the 4X3 amp years ago. Is this correct?

Hope this answers what you axed about.

T

My answer to the question is, forget about making any type of toroidal filament choke.

It may very well be that you can do this, but of all the filament chokes I've ever seen throughout the years in numerous publications, 99% of them have been wound on a rod.

I just came down from my library after poring over many different directly-heated cathode filament amplifier designs, and without exception, ALL of them used rods, and not toroids.

If you want to experiment and build a toroidal filament choke, go ahead. But you are in uncharted (read: undocumented) territory.

Henry used toroidal fil chokes in a few thousand amps. It's pretty well documented.

Yes, but Henry Radio had issues with them.

http://www.google.com/#hl=en&xhr=t&q=henry+toroidal+filament+choke

Interesting.  I didn't read all the links, but this sounds familiar....

"This is KX5JT.  You guys may remember a year ago after acquiring a
Henry 2K4 amplifier that I pressed into A.M. service, I had to replace
the improperly designed toroidal filament choke.  It was NOT bifilar
wound.  A bifilar wound replacement took care of the AC hum on the
carrier."

I will have to look at my own Henry next time it's apart to see if it is bifilar wound. It has no hum on the carrier, but worth a check anyway.  As I said in a previous post, Frank/GFZ solved his hum problem too once he wound both fil leads closely coupled on the same toroidal core, in the same direction.

I can't see why a toroid as a ferrite medium would not work correctly once wound this way - but open to opinions...

T

Well, I appreciate all the feedback. Items like this are particularly interesting and the question is mainly prompted by what I have on hand. If I could find a good reason not to fabricate the choke in torus fashion then I would. However, I could not. That said, I am happy to take the risk and see "what lesson is learned".

Thanks!

That's documentation.  ;)

Yup, I had two short rods side by side and the amp could hum a nice tune.
When I swapped over to bifilar the problem went away.
Tom, I think you used type 43 beads on that big choke you built..

Type 43 is good down to 1 MHz. Type 31 covers down to 500kHz. Either one should work for this.
I'd be interested in trying a fil. choke wound on a tooide in typical common mode fashion ie... opposing directions.

I suggest doing a bit of winding, measuring, and calculating Mike and determine the reactance on 160 & 80. I back up calculations with a SA or VNA to make sure nothing funny is going on.

As far as Amidons charts or whatever there is a huge difference between using a ferrite mix in a reasonant circuit and as a choke.

Just because it works for you means little if your dumping part of the drive into the choke bypass caps and filament transformer. OTOH it could be a good thing for those who like to overdrive their amps :o

So besides liking to follow me around what is a slopbucket man doing on this forum?

Carl

Type 43, ui:850,  2 inch toroid is what I have in hand. Will try that first.

A neat closing line on this is:

In GOD we trust, all others bring data.  :)

The problem....sometimes....when using a lower mu ferrite is that you have to add many more turns to get the same inductance as a higher mu mix. In some cases it works OK. Other times all the extra interwinding capacitance can/may create nasty resonances or otherwise degrade the choking effect. When its just a choke and not a tuned circuit I like to use as few turns as possible and as mentioned earlier back it up with the SA or VNA to look for the nasties. A GDO can find the resonances.

I didnt read much about the Henry problem but Id suspect that a small toroid (about FT-140 size) used with 30A of filament current (2x 3-400Z/500Z) plus trying to stop RF is not a good approach. They do tend to be drive hogs when 160M models are involved. It might be fine with a pair of 811/572B, 813, or similar. The last filament choke toroid amp I worked on was a Johnson Thunderbolt II which became the Tempo 2000; 80-10 with 3-400Z's. It had its problems when it arrived but that circuit worked fine at 1200W out with an overkill 100W drive.

Speaking of not enough choking impedance, the Dentron Clipperton L and its similar family is another one that suffers on 160.

I also have mix 77 in a 2 inch size and this ui is 2000. However, the ui is frequency dependent and I'll need to measure both the 77 and 43 for the same length wire and see what the XL does with frequency. I have a VZM which is hand for these measurements. Thanks.

Well, I'm not following anyone around.  :) Amfone.net seems to be a pretty special place. I just started reading some of the posts last month, and I see there's some good people here.

You might very well find me on AM someday. Since reading this forum, I'm again toying with the idea of two paralleled 833-C's on the lower HF bands, perhaps plate or cathode modulated by a single 833-C.

Yes, my Henry 2k4 had an original toroid filament choke.  On AM, I had experienced AC hum on the carrier.  I dug and poked and prodded and someone finally admitted that during a certain period of time, the toroid choke was wound improperly and some 2K4's were out there with the mistake.  On sideband, it wasn't much noticed.  On AM it became clear.

Those particular amplifiers had each filament lead wound SEPERATELY, one on each side of the toroid, not wound TOGETHER in a bifilar fashion.  I had one.

There is nothing inherently wrong with a filament choke on a toroid but it needs to be wound BIFILAR fashion, just like if it were on a rod.

John KX5JT

When constructing chokes from 61 material be especially wary of resonances. Unlike other ferrite materials, inductors wound on 61 will often have Q's approaching 300. Inductors wound on other ferrite materials are lucky to achieve 1/5 of that - some considerably less.   

Jay,
I always load the choke with around 2K ohms of resistance. I've seen HF amplifiers take off in the VHF range due to a high Q resonance in the choke.
I do not know what material is in my Lou McCoy donated rod.

Thanks for the added data. I'll measure the 77 material and 43 material from 500 kHz-110 MHz and report what I find. The material set has ui which are frequency dependent and I'll be able to find the real and imanginary parts of the Z over that range. Ideally there will be no resonance loops! It will be bifilar (pairs of wire) and will probably be space wound between each turn about 1-2 x the diam of the wire.

Alan

I had to use #7 to handle 43 amps so couldn't afford spacing turns.

# 7 WoW. Here is a link to a very interesting article:

http://www.highfrequencyelectronics.com/ see this monthe issue it is free to download

The small signal response of ferrites and the author explores the saturation affect.

here are the results of one xmfr. This should work well from 160 M- 30 MHz. Even at 110 MHz the resulting Z is ~ 2 pF.

The 77 mix is good stuff; I use the FT240-77 to solve most noise problems from household devices.

I just got back from WB9DNZ's QTH where he wound no. 10 enamel on a form to make a choke using that  7 1/2"  x 1/2" 33 mix rod.   I provided the wire, rod and the coil form, which was a bit less in diameter than the rod.  I figured the wire would "spring back" so I thought we would need a form just a tad narrower than the rod.  I kept the tension on the wire while he did the winding--I had the easy job.   I am hoping it will be enough for a pair 4x1 filaments but I may have to have another one. 

Here's what we discovered:  The no. 10 doesn't have anywhere near the spring back we expected.  The coil was real tight and it was therefore a b**ch getting the rod into the coil.   That is one hell of a filament choke.

The inadequate grid choke protects the inadequate plate tank on 160.

Doesn't Don have a 4-holer out in back of the schoolhouse?

Here is a cool site:  http://www.stormwise.com/page26.htm

Mike WU2D

There is no grid choke in a Clipperton L and the tank circuit is an easy fix. With a bit of work the L can run 1200W out 160-10 with 100W of drive and the tubes showing no color.

Duh. I meant to say filament choke.
Too much vacation.

No such thing.

Agreed! I'd like to have two vacations a year: six months each. ;-)

Im on a permanent vacation

For completeness, here is a pix of the filament xmfr ready for insertion. The Delron block is 3x3 inch. The 77 material is 2.5 inch diameter toroid not 2. There are 28 turns and I spaced them wider then I originally stated. This is not heavy gauge, 16AWG, however, I am at 4A, not 40A, so a BIG difference. Which leads me to believe that the ROD became popular because it would be a real chore to put # 10 on a torus!

Finally, the resonance is NOT that of the winding, but is induced by the measurement itself. Turns out, the voltmeter probe has about 2 pF of C in its clip on version. So to get a real measure... go to a GDO! If I have time, I will do that and see where this unit is actually resonant.

I have a collection of ferrite core toriods that I have been able to use successfully as filament chokes.
I am aware of the various grades of ferrites and their frequencies of operation.

Powdered iron like wise is used in frequency bands indicated by their coloring.
I would like to know if I can use powdered iron cores and also when and were I can use them.
What are the general rules covering their use, is it some thing to do with the RF power they can handle?

Good questions! For starters, see this months issue of High Frequency Electronics, I posted the link a few notes earlier in this thread. There is an excellent article in there that should address your requests... its about 12 pages long so quite in depth.

This should keep you out of trouble for a few days ;)

http://www.micrometals.com/

And this for the ferrites

http://www.fair-rite.com/newfair/index.htm

For the uknown materials you need to wind some turns, measure the L and then attack the formulas to determine what its useful for. A lot of surplus powdered iron cores come out of switching supplies and other LF applications.

Carl

You guys will probably think im crazy but hey thats nothing new so here we go. I have not used ferrite for winding bifiller chokes in years.The choke in my main homebrew 3-813 AM LEENYAR amp is wound around an old broken ratchet with the head sawed off.I save all the made in China broken ratchets now days.Ferrite is too expensive. I have also wound chokes on rebarb,I even used a motercycle axal one time with success. I have never had one of these contraptions to not work.I have never worked 160 so I dont know if one of these chokes would work on that band but they work fine 10 through 80 meters.You guys probably have a good supply of ferrite and dont want to try this but if you ever get in a situation where you need it and dont have it dont be afraid to try other materials.They work !                                              73s Mike

Cool!  :D If I get a hold of a broken wrench I'll put one together and give it a try. Can you imagine sitting in front of a 80K$ 40 GHz VNA, measuring a choke wound on a broken ratchet tool at 7 MHz. That would be quite FUN. Seriously, if the permeability of your broken tool is not unreasonable, than the inductive reactance present may be more than sufficient and it simply works. 

I've heard everything now.

I think I'll go wind a toroid using a stack of big steel washers as the core. ;)

I wonder how much #6 I can get on a crowbar. Then I can try an old axle shaft.

You have to use multiple crowbars or axles to break up the eddy currents, Carl. ;-)

Ferrite rods are cheap, find a 2nd hand BC Band radio and remove the ferrite antenna rod.  I used one in my GG linear and no problems at all. I think you can get them for a buck or two at yard sales or flea markets (not the Ham ones, the other kind) or some times junk radios can be had for Free just carry them off.  Strip some 12 or 14 gauge romex scrap (save the bare copper piece, and wrap as many bifilar turns as will fit. tape with crappy black electrical tape. Done. It'll carry a load of current too.  Works FB.

The toroid choke measured and posted in prior posts was re measured via a GDO. There were no anti-resonant frequencies found. So indeed, the resonance in the early plot was due to the measurement itself.

The measurement was done in the same manner that plate chokes are measured, the total choke shorted start-to-finish winding ends and parallel resonance searched.

Its very hard to get a good reading on a toroid with a GDO, even a high level tube job such as the Millen and Measurements 59 can have problems especially with ferrites.

An SS version is mostly useless.

Before you buy, make sure there's a loop-stick antenna in there of sufficient length.  The older AA5 receivers up to around 1955 with the big octal tubes usually had a flat coil of wire glued to the inside back panel for an antenna--no ferrite at all.

Very well. I thought about that, so what I will try is to deliberatly resonant a ferrite structure, I will add C across the winding. Since the torus is self shielding, I coupled the GDO coil well within the structure. And again I will try. As a dress rehearsel of the effort, I took some plate chokes that I had built, they looked great physically, but awful electrically. Again the GDO was used here. But they are solenoid, not toroid.

Plate chokes should be dipped in circuit, it doesnt take much of an enviroment change to move an inband resonace well away and vice versa.

I looked at this issue in a little more detail. I measured some material 73 and 61 with ui in the range of 850-2500. If you look at the impedance of these particular elements as ferrite "chokes" they have a significant real part and therefore the loss is high and the Q is small. Yes, there is no GDO dip. However, in the case of the toroid I posted, the real part was VERY small. The angle of the Z nearly 90 degrees. I added 22 pF which would resonate this unit near 3 MHz, indeed the GDO dips. Have to tune with care, but it is there. So... in many cases if the ferrite is chosen to be used as a lossy element with no resonant condition and a large reactance, NO DIP. After I removed the 22 pF, again a search showed no resonance present. So I guess the up shot of all this, is IT DEPENDS.  :)

Yup, thats what makes homebrew and restoring boatanchors fun. Sure as hell beats using a black box and a PC...BORING

Sorry to resurrect an old thread . . . but having searched the internet this is the only decent information I have found on this topic!

I have a Yaesu FL2000B Linear, that is basically the same circuit as the Heathkit SB200 (which itself I believe was a copy of a Collins linear?) . . also the FL2100B and KW1000 all use the same circuit, ie two grounded-grid 572B valves and a voltage doubler PSU.

The unit I got had been cannibalised for spares (by the Yaesu dealer I was working for while at University) so I rebuilt it just to cover 160m & 80m, by winding my own PA coil and not bothering with any input matching circuits (with a 5000pf coupling capacitor straight to the filaments it's about the right impedance . . certainly OK for a valve PA to match into.

The only problem I had was there seemed to be insufficient inductance for 160m in the bifilar filament choke feeding the filaments. (presumably the later FL2100Z which covers 160 has a bigger choke?) . . . but I got round this by taping an equal length of ferrite rod to the existing choke - this was a normal ferrite rod from an old portable radio.

This all worked fine . . . 600W output on 160 & 80 for 1200 W input, about what you'd expect.

HOWEVER . . . . and here's my question coming up!  Having used this linear for over 30 years I have always found that if there is heavy use (like in a contest) the power gradually goes down . . . . not just the output power, but the input current too.  If you let it cool down for a couple of minutes, the power comes back up again . . . but once it has started getting this hot, it very quickly starts going down again.

Now I had always presumed this was the valves getting too hot . . . . and maybe there was secondary emission from the anodes?  But to be honest, the anodes can glow red quite quickly ANYWAY, and this problem DOESN'T occur . . . not until the rig has been on for a while and the whole thing is getting hot.

But what ALSO occurs when this is happening is the dial lights glow BRIGHTER in time with speaking or CW transmission.

SO . . . I am now thinking that the problem is the BIFILAR CHOKE starting to not work properly . . . thereby reducing the drive . . and also why the dial lights glow brighter, due to the RF getting into the PSU.

I have also realised that the PA current of the driver rig goes down too - as would happen if you started feeding a lower and lower impedance. (which obviously would be happening if the problem IS the filament choke)

SO . . I figure that the problem must be the FERRITE in the choke . . and also my additional ferrite rod . . DON'T like getting hot! And when they reach a certain critical temperature, they STOP having their normal property of increasing the inductance.

I can find no info about this anywhere . . . but would you agree with my conclusions chaps?

If that IS the problem, I don't know whether the ferrite is getting hot from the RF from the driver rig, or whether it's just the whole Linear Amp case getting hot causing the problem.

Wonder if anyone else has had a similar problem . . . and what you suggest?

If RF plus DC current is present on the Ferrite rod and the rod is constructed with high perm material, its material losses could generate significant heat. The heat gets elevated to the point that you can actually smell it as in ceramic heater disks. If you get the rod hot enough (above the Curie temp), its perm drops and the problem gets worse, analogous to thermal runaway! Not suggesting that is your problem, however, if you believe that is an issue, I suggest you first measure the rod inductance with NO current. Then heat the rod, either pass the IV similar to its operation in the amplifier and even add additional heat, say a hot air blower gun. And measure the L value as the unit heats. You did not say what Ferrite material the rod is using. That would be helpful.   

this was a normal ferrite rod from an old portable radio.

Ah, I missed this. The perm used in these AM radios can be quite HIGH, say 1500-2000. Their magnetization loss could be significant and aggravated at elevated temperature. However, in the end, this is all speculation and a measurement is in order.

Sadly I have no test gear to measure the inductance !

But given what you have said about Ferrite not liking getting hot, I would have thought my diagnosis (from the various symptoms) makes sense?

If this is what is happening, it's not just my ADDITIONAL ferrite rod that is failing to increase the inductance . . . . as before I added it (ie with just the original ferrite wound choke on its own) I could still drive the amp up to 800W input on 160M (ie 400W out)

When this starts happening, the input and output drops right down . . ending up with less than 100W out!

By the way, I failed to mention that this has happened over the years with different sets of valves (I've replaced them every 6/7 years) . . . I originally had spare Cetron ones, but recently fitted some Chinese 572Bs, which are SUPERB copies of the originals ! (behave just like them)

No problem. In the 1973 ARRL Handbook, the classic B&W like Filament choke is noted on page 155. It is a 30-A choke using No. 10 AWG, 28 turns on a 1/2 inch diameter rod, 7 inches long. Similar is sold and advertized at 7.5 inches in length. In any case the same is stated at 40 uH of inductance and at that value of XL is 250 ohms of reactance at 1 MHz. That value seems reasonable for GG although perhaps off by a factor of 2 for the balun connection. In any case, NO where does anyone state the perm value. So, for grins, lets reverse engineer this thing and see what perm results. Stay tuned.

I calculated backwards the permeability required to obtain the inductance per the ARRL Handbook and the dimensions used in the 73 edition, Fig. 6-22. I find that a perm of 125  gives good agreement with 10 gauge wire, 28 T on the Ferrite form size they suggest for an Inductance of 40-50 uH. This would suggest Material -61 would be a good choice. If the choke temp is kept under 300 degrees C  :-\, you should be fine. Welcome any comments. If anyone knows the exact material type used in these Filament chokes in the past, that would be great.

And if you look at all the prior posts as I should have, indeed -61 was indicated by several folks in prior notes.

Hope this helps,

Finally reading this thread, I imagine that an additional reason rods are typically used is the isolation
that they afford, having the RF voltage at one end only. There is very low capacitance from that end to the cold end of the rod. In a toroidal winding, one has to take care that the hot and cold ends of the bifilar winding are not adjacent or overlapping, as others have said. Otherwise, the effectiveness, esp at higher end of the HF frequencies, diminishes. We are trying to bring the filament AC power into the RF circuit, at the cathode of a GG tube, without loading the RF circuit, and without allowing RF to get back into the AC circuit. Toroids can certainly be used, but in high Z circuits, strays from the end of windings can limit effectiveness.

I appreciate your replies chaps

Well looking at the original bifilar choke in my FL2000B, it looks like it's wound on half inch ferrite rod with very thick enameled wire . . . . but it's only about 4 inches long, so only has about 20 turns!

That's only 10 turns per winding, so I guess from the ARRL design that means it's only about 15 uH . . . which on 160 would only mean a reactance of about 150 ohms. (On 80m I guess 300 ohms, which is obviously enough)

The FL2100 and FL2100B are internally the same . . . but the later FL2100Z (which covers 160m) PRESUMABLY has a bigger choke . . . . but the only online photo I have found of the valve base compartment shows NO choke there!  Certainly there is no space for a bigger choke there . . . but I can't see it under the chassis either! (I really can't see where they have put a bigger choke, so if anyone knows, please tell me!)

Anyway, as previously stated, just taping an equal length of ferrite rod to my original choke made it work OK on 160m, so presumably doubled its inductance . . . as per my original post my issue is I think the overheating. Presumably that's because it's relying on a LOT of ferrite to up the inductance? (what do you think?)

If that's the case, I guess the best fix would be to make a proper longer choke with twice as many turns on a SINGLE ferrite rod. But the problem is I can't see WHERE I could fit a longer choke in the Linear!

The only other solution I guess would be to drill holes in the back panel opposite the choke and fit a fan on it (I already place a mains operated fan on top of the Linear to improve cooling if I'm operating for any length of time)

Something's heating up but what, that's the question.  What band or bands does this happen on?  

The usual suspects are your caps in the pi tank circuit on the output--the fixed value load caps.  Other possibilities are the blocking cap and plate choke bypass cap.  

Because of the dial light thing I think your problem is in the power supply, especially if you have one of these deals with a single transformer with with windings for all your voltages.  I'd first try getting the amp out of the cabinet, firing it up and running it in the condition that causes the problem and put a six inch fan that moves a lot of air, 150 to 200 CFM right on the power supply iron and see if the problem goes away.   If you have an IR thermometer measure the temp. rise on the transformer(s).

If the problem is only on some band like 160 then the load fixed caps and fil. choke maybe but these don't explain the higher voltage on the lamps to me.  You can try putting some snap on material 31 beads on the leads from the fil. choke to the sockets and see if that changes anything.

I really only actually use it on 160 . . . the new (bigger) PA coil I fitted is only tapped for 80, and as there are no longer any input switching/tuning circuits I figured it might be unstable on the higher bands. I used to use it on 80 occasionally . . I never noticed this effect, but that MAY have been due to not operating for over an hour.

When I first noticed this happening (YEARS ago!) I measured the PSU voltages and they were fine . . and they still are. As I say, I originally just put it down to the valves overheating, but now realise it ISN'T.

I also wondered about the output capacitors, but if any of them were failing, the plate current would surely go UP (as the pi tank isn't fed or going off tune) . . but the current goes DOWN, as if the DRIVE is going down. And even when this effect is happening, the pi-tank still tunes in the same place.

More recently I've noticed that my FT101E's PA current ALSO goes down when this happens, which would happen if it was feeding a very low impedance - such as the bifilar choke having no reactance! (there's nothing wrong with that rig, if you by-pass the linear it's immediately fine)

If the bifilar choke is ending up with very little reactance when it's hot, surely the RF will be trying to get back in the PSU? (despite the decoupling at the bottom of the choke) . . . that's why I figure the dial lights start to glow slightly brighter as I key. (this only happens when this overheating effect occurs)

Thanks for the feedback John. If the stray C coupling is an issue it would show up as a reduction in the SRF and the same case occurs in the rod (solenoid). Allowing the start-finish ends of the torus to come to close is NOT desirable and it certainly would lower the SRF. I have used the toroid in Filament choke, no issue, as have others. I think the real issue of ROD-vs-TORUS is PRACTICAL. Have you tried winding #10 onto a toroid? Unless you have a huge one, its tough to get the bending radius. However, the 1/2 inch rod is quite straightforward! My motivation for looking at the material property was to address G3YRO query as he indicated broadcast band type rod material was used. The material I have used in those applications had quite larger perm values than desired for a fil. choke. Anyway, it looks like he is on track.

Just found the FL2100Z choke - it's twice as long as in the non-160m versions, and mounted horizontally across the top of the valve base compartment.

Yes I meant to comment on that suggestion . . . . Do you really get enough inductance that way?  How long would the leads need to be?

Maybe the answer in my case would to REMOVE the additional ferrite rod I strapped on, so it's just the original Bifilar Choke . . but then add ferrite beads on the wires to the filaments to increase the inductance sufficiently for 160M ?

This is the new filament choke for a customers Clipperton L that I wound a few days ago, the rod is a 7.5" x 1/2" Type 33 rod that I shortened a bit to fit, I like to use 5-10X the reactance of the input impedance of the tube(s) used. In this case the original choke is 24uH and the new one is 55uH. After tank circuit changes, 100W drive, the Pout on 160 went from 900 to 1200W and holds there to 10M where its 1100W using 4 new Shunguan tubes I bought directly from China. Stable as a rock and no tube color even with a few tests I did with 150W drive from the TS-950SD.

This is a very late model Clipperton L with a PC board holding the tube sockets, input networks and filament choke, Ive only seen a few others. It also has a filament winding CT with a transistor and 1W Zener, no hum reports even on AM. They also used what looks like an AL-80/SB-1000 plate choke and a 5000pf bypass. The circuit is similar to the next edition 4 X 572B Dentron amp, the Clipperton QRO which came out just as Dentron went away. Parts have 1982 dates.

Carl

what awg wire did you use on the fil. choke?

I suggested 31 beads because that material type is fairly easy to get.  It is actually not the best for 160 m. but will work okay.  To really do it right you'd get a material type (F? J? can't remember now) in some smaller beads, undo the leads to the tube pins, slip on the beads (as many as you can fit) and solder the leads back to the sockets.   The choke Carl made is probably a better way to go.  that 7.5 inch 1/2 inch diam. material 31 rod is available from Amidon; maybe other sources too.

#14, there is 16A of current.


There are no 31 rods, they are 61 and 33 only. I have both here in 4" and 7 1/2" way cheaper than Amidon.

Sorry, I must have been thinking of the material 33 rod.  The 7.5 inch 1/2 inch diameter 33 rod is $20 now from Amidon!  I have one or two of them, one has no. 10 enamel wound on it to make a fil. choke for a pair of 4-1000As but no. 10 may not handle the current for a pair.   That's an experiment waiting to happen.

The material I was thinking of for beads on 160 was 75/J.  Gyro, if you want to try adding beads to your choke leads try to get 75s in a size that will fit on the wire.  A bigger choke like Carl's is probably a better multi-band solution.

Relating back to the post by GYRO that started this recent exchange, a rig for only 160m/80m might not even need a filament choke! Instead use a filament transformer with a split bobbin such that the filament winding is separate from the primary. This is not a good 160-10m option, but for top band and 80m operation this should work. The split bobbin type transformer will also be coupled less tightly; a good thing to limit the initial turn on current surge. I wonder if any suitable transformer such as discussed is readily available?

Edit: There would still be a need for a filament CT. RFC to ground, but it would only need to pass cathode current. One of those old single Pi chokes from an old WW2 tuning unit (~1mH) would do. A few ohms DC resistance would be fine.

Jim
WD5JKO

Finding the proper old iron will likely be harder than spending the same or less on a rod and wire. My only attempt at using one of those transformers and no choke for a pair of GG 813's in the early 60's didnt turn out well. The articles of the day specified only a few useable part # that had sufficiently low C.

Where a generic split bobbin one may work somewhat is in combination with a rod choke of less than needed reactance/inductance and then modify the input network to compensate. Expect higher drive power for comparable output using a proper choke.

Carl