The AM Forum

I would like to use #43 cores to make a filament choke.  Simple yeah?   I want to run 4 ea #8 THHN inside the cores to lite 2 tubes.  Will it work like that or should I make a filament choke for each tube?  Also how snug should the wires be in the core and how much core length is enough?  I'm working on a pair of 4-1000.   6 KV  Thanks   Steve

A single pair of #8 would suffice.

The impedance goes up with the square of the turns.

I've used two #31 cores and run 6 or 7 turns through it on a pair of 4-1000s.

Ive also used a string of beads and #6.  However, this wasn't nearly as much impedance nor as good across HF.

#8 is rated for 40A in conduit.  Quite a bit more inn free air.

--Shane
WP2ASS / ex KD6VXI

Ive got my beads and cores mixed up....   I want to use #43 beads.  I have a filament transformer that has two separate secondaries. each with its own ctr tap.  I figured 2 #8 to each tube.  4 #8 will fit inside some beads available.  make a stack of beads..will that work with all 4 filaments leads inside?  I could probably cut the wire size down to #10

I was experimenting a while back with #43 ferrite beads (3/4 od)
I stacked 6 on a 1/2 inch fiberglass rod and put epoxy in the seams.
36 turns of a double winding of awg 10 magnet wire gave me about 80uH of inductance, I think it was. I wound the windings one at a time on a 5/8 wooden dowel a then (screwed the double winding on to the stack of ferrite beads.
I drilled the fiberglass at each end and ran the awg10 wires through, locking it all together. A coating of epoxy on the top of it mage it rigid and physically stable.
The effective permeability of the solenoid on the 43 fabricated rod was 49, by the way.
Used about 6.5-7.0 feet of wire for each winding.
73, Mike K9MB

I have been having fits trying to get a Swan Mark 1 linear to work on 160 meters.  It uses two 3-400Z tubes.  The filament choke has a 3rd winding for the center tap on the filament transformer.  I am using the Ameritron filament choke for the other two windings.   So far have tried a Torrid care and beads but no luck.

Be interested in how yours runs out.
Bill KO4NR

First, you don't need #8 or #10.  That's just over engineering it.  #12 THHN is rated at more than you need for one 4-1000.

That said, why not parallel the transformer rmer windings, if they match up enough and use a si gle piece of #8.  One #8 is good for better than 50 Amps in free chassis wiring.

Since you have the beads, run what you bring.  But 43 isn't very good on the lower bands. It's impedance starts to drop off.  That said, I do get OK power out with my 16 x bipolar device with type 43 transformers on 75.

The issue some people have with not returning the center tap through the choke is hum.  Honestly, I don't know if you will or not, seeing that you have separate filament circuits.  Possibly enough coupling theough the feed caps?


Bill, your center tap doesn't go to a transformer winding, unless you bulldozed the original swan trafo.  Those use a 10 volt trafo and then ground the junction of the two tubes on TX for cathode return.  No center tap on the original mark 1.

Without enough reactance with that Fubar system you instantly start bleeding drive power to ground.  Henry does the same thing.  So did Kenwood.  So did Tokyo Hi Power.
 All Fubar, IMHO.   Now you need a tertiary winding on the filament choke.  For what?  To save the winding of a center tap?


I'd be interested to see your drive voltage on the filament on say 80 meters vs 160.  A grid current meter would also tell you what you need to know, IE, is the grid getting enough drive or being shunted to ground via the filament circuit.  Or is the antenna tuner input network Fubar and causing grief.

This is where a tube exciter would be great.

Or an MFJ259 and a resistor.


--Shane
WP2ASS / ex KD6VXI

I take that back, RE the filament xformer.

It does have a 10 vac xformer.   And a center tap.  Not exactly like Henry, but close enough.


The rest still stands.  Still Fubar.


--Shane
WP2ASS / ex KD6VXI

I am out of town now.  Will check everything when I get back next week.

I did get 40 and 80 meters working ok.  1KW out on both.  Grid current was a little over 150ma.
Bill

That's good!  That means we have something to go on, that the amp DOES work....!

Now to get 160 working



--Shane
WP2ASS / ex KD6VXI

The filament choke 3rd winding is suspect.  When I get home I plan on trying a 20uh Torrid Core choke, FT-240-31.  I can measure the inductance with my VNA.  The Ameritech filament choke is connected to the other two taps.  The drive is going
somewhere other than the tube’s cathodes, I believe.

I should have started a new thread.  I apologize KL7DF.

Bill KO4NR.

The entire point of filament chokes is so that the cathode is not loaded with a low parallel load impedance.
The ferrite used may be less important than the total impedance of the choke in the same way that a balun choke prevents current from flowing on the braid of a feedline.
#61 material will make a more reactive load and #31 has a high loss resistance, so a choke on #31 will have a high resistive component.
If the choke appears as a high impedance, then little or no rf current will flow through it at the low load impedances seen in the cathode circuit(s).
A toroid is not a magic wand. The solenoid I described utilizing stacked cheap ferrite beads will result in a high impedance load in parallel with the cathodes.
I know that guys often run wire through the centers of 20+ ferrite beads to get a moderate amount of impedance, but it takes a lot of beads and space, so unless you have a very high filament current, winding a solenoid of #10  for each tube is adequate for the 21 amps needed and you get separate cathode returns so you can apply separate cathode biases to balance the quiescent cathode currents that can be measured- a good thing.
Of course, you can wind a pair of awg8 wires too and tie the cathodes together, but you lose the ability to monitor and balance the separate tubes and awg10 is quite adequate, as Shane said. In my choke, it only took a few feet of #10, so the resistance was low.
As I stated, the calculated effective permeability of a stack of #43 beads is about 49 for calculation purposes. The 800 permeability figure holds true when winings go through the core.
Core loss means little as long as the total impedance is very high at the operating frequency, as in the case of antenna balun chokes. 73, Mike K9MB

Core loss means little as long as the total impedance is very high at the operating frequency, as in the case of antenna balun chokes. 73, Mike K9MB
   

What would be a high enough impedance at 29.1 Mhz?   

I've always worked to a minimum of 10 times whatever the cathode impedance is.

For a pair of 1000s 600 ohms or more.


--Shane
WP2ASS / ex KD6VXI

A few cathode choking questions for grounded grids.  Please explain like I’m five….

Why is a separate choke needed? That big iron filament transformer has a lot of inductance, even if most of it is balanced out by the opposing windings.

Why does the center tap (vs. one end) get grounded? We like symmetry, but does it matter to the return HV DC at all?

Instead of choking the lines to the filament, why not choke the single line to ground (regardless of center or outside of winding)?

Trying to learn the why behind the “always done that way” aspect here….

Ed

1.  Because if you don't choke the cathode off, on a cathode driven amplifier, the cathode circuit (including fils xformer) will suck drive.

They have sp vial transformers made for exactly what you ate talking about.  Very low C from pri to sec.  Those you love u can use a Gil choke on the primary side.  Which is how I do anything bigger than a 3000.

2.  It depends on how the tube is made.  If the cathode is a separate pin, you don't need one, although it's still good practice.  And even then you'll still need some type of choke.  The cathode needs to return to DC to turn the tube on, but you don't want the rf sent to ground.....!

3.  Depending on how the amplifier is made, some due that.  Look at the Swan amp.  It's a trifiliar.

But basically, it's this.  The C from filament transformer to chassis is usually enough to throw the tuned input off and divert some, if not most of, the drive to ground.

Rather than resonate the tranaformer, choke it off.

--Shane
WP2ASS / ex KD6VXI

As Shane said, the minimum is 10 times the input impedance. A pair of 4xKs will look like 50 ohms or so.
Impedance is reactance and loss resistance in ferrite products. Above self resonance, the choke will be capacitive and as frequency rises, the reactance goes down. Loss resistance can stay high enough or rise, however, so the cathode has a large shunt impedance anyway.
If the impedance-resistance is too low, however, it can cause heating in the ferrite, so i like to see thousands of ohms.
I think that you can feed a small signal through the choke and put a series resister in series and observe the rf voltage drops using a good scope. If a large drop is across the choke compared to the resistor, it can be seen as effective at a given frequency. Use non inductive (carbon comp or equivalent) resistors.
If you have a good network analyzer, you can see much more, like polarities of impedances- inductive or capacitive, etc…
You could also see if the ferrite is getting hot as a primitive method…😉😂73, Mike

That’s how I tested my 10 bead filament choke for the 3rd/center tap on the Swan Mark 1.
10 beads used
Voltage at Signal Generator side of a 1k ohm resistor is .5vdc and on the choke side of the resistor it was .25vdc.  12 beads was .3vdc so looks like I need a minimum of 15 beads to achieve a 90% voltage drop across the choke.  Used my oscilloscope to measure the voltage.

Should also use a 60 ohm resistor for two 3-400Z tubes.
Bill

I got a Nano VNA and did some filament choke testing.  I am building a 2 hole GG 4x1 amp.  The design frequency is 29.1 Mhz.
I have never been able to get real good efficiency on 10 meters from a 4x1 in grounded grid in a 5 band setup.   Hence the monobander.  I'm trying to tailor the components for this band with as few compromises as possible.  I don't want the filament choke or the plate choke sucking power.  Close spaced layout, short connections, etc.  
The values  for this filament  choke should be + -    10uhy , 500 ohm Resistance ,  Low series "C", and not resonant near 29.1 Mhz.
  Choke construction is K1JJ style using #10 THHN thru the hole in the beads.  Beads are Fair rite #2643102402 43 mix  Test wire was 14 in long.... measurements made with Nano VNA H4.  Refl mode,  cheesy clip leads on 3 inch coax to SMA connector.  all measurements taken at 29.1 MHZ
I started with 4 beads and added one at a time until I got to 8 beads.
  
# beads                       4        5         6        7            8
resistance  ohms       506       650     780    878       1k
reactance x100         172       130        62       -50     -48
inductance  uhy        9.4        11.4       13.5     15.1   17.1
series C  10nf/          -32pf     -47pf      130pf   108pf  113pf
resonance  Mhz        45.7       36.9       ?          27.9    28.1

I don't understand the series "c" measurements  except to note that the C goes up as more beads are added.

As more beads are added, the R goes up, the reactance goes down , the C goes up , and the resonance goes down.

   I'm going with 4 beads for  29.1 Mhz..    I am a newbie with the Nano,  The numbers seem to be OK.. I invite your comments.  Hope this is useful.
 
Steve

UPDATE....   I got the amplifier going .  Each pair of filament leads (#10 THHN) is run thru 4 #43 beads that are taped together. Chokes are very close to the tube pins.
 I'm getting decent power out of a couple old coffee stained pulls and there are no ill effects from insufficient filament chokage.
I'll report again as the voltage goes up........   Steve

Pic

Hi Steve,

Good show on getting it working on 10M!   The fil chokes look FB. They should work well.  Small and sweet, good for higher freqs.

Is there enough socket gap for an ample supply of air?  It's hard to tell from the pic.  Use real 4X1 chimneys, of course.

Be sure to add in a parallel tuned cathode circuit. Use a quality tuning cap. And also use quality ceramic "door knob" shaped Russian caps and fat copper strap througout.  Be sure the tank circuit is very tight and short.   Insure that C1 + the 4X1s total plate-to-cathode capacitance is correct for the tank impedance you want.   Sometimes C1 has too much minimum C and the Q is way too high for good efficiency.  I had this problem with my YC-156 amp.  There are ways around this if it's a problem.

When finished, post some pics of the amp in operation with plate color.   You'll be king of 10M when that honey is running!

** Not sure why this post came out in duplicate..  **

T

4-1000A Rico Suave II thread:  http://amfone.net/Amforum/index.php?topic=34543.0

Third pic:  160-20M fil choke.  #6 wire for low 43 A  voltage drop, ten cores on outside chassis and 10 cores underneath chassis at tube pins.

Another question, along a similar line: Why can't we—at least for smaller amps—simply use commercially available axial lead RF chokes in the filament leads, rather than bifilar types? We would need to pay attention to the manufacturer's specs as to resonant points, of course, as well as the current rating. Bifilar chokes would exhibit strong common-mode choking, I would think, but other than that, if the impedance at the operating frequency was the same and there were no resonant gremlins, what's the advantage of a bifilar choke, other than the builder being able to design them for big honkin' currents?

A bifiliar choke, on a filament tube, prevents hum.

You really need a teifiliar on things like 8877s, gs35b tubes, etc.  The cathode style tubes.

You want currents balanced.  Individual chokes don't do that.

That said, they have been used.  Then people wonder why they have complaints of hum on their signal, etc.


Then they buy or wind a proper choke.

--Shane
WP2ASS / ex KD6VXI

I used the core of an old tv fly back for the choke in a dual 813 amp.  Works fb

Don't forget the coaxial choke, a length copper tubing with a solid center conductor insulated by teflon or whatever, slipped inside. It's then wound into a solenoid. Ferrite rods can be put inside the coil to give it more inductance if needed. Henry used them on their 3CX3000 industrial amplifiers.

Small axial RF chokes in pairs are frequently or historically used, with small indirectly heated tubes in VHF-UHF RF receive circuits like rf preamps and tuners where hum from imbalance isn't a problem. It's for isolation of the filament from other stages in the receiver, stability etc. 1 million tube type TV sets can't be wrong. But that's low power stuff.

Henry used those on monoband amplifiers, usually at 13.5 or 27.12 mhz.

Those are somewhat tuned.  No Bueno for multiband.  Problem becomes interwinding C causing resonances as you go up in frequency. Because one winding is inside the other, you can't bandswitch it. (it would be more like bandpass, but I digress).

Chuck Asamoto did a lot of cool stuff at Henry......  But he also has put out a ton of questionable amplifiers (like pallets full shipped to the middle east, that didn't work on arrival.  Uncle Sam was ZERO too happy about that).

Those fils chokes work great over a limited bandwidth.  Then they either don't have enough inductance or they have a series resonance.

--Shane
WP2ASS / ex KD6VXI

That's why stuff ferrite in for broadband.  There's also one around here somewhere with taps for several bands on the outer conductor, picked up at a fest. Never have needed it but might be interesting to try out.

It's a dual solenoid inductor.  Balanced feed for filament.

How does bandswitching half of it work?

Like I said.  Asamoto was pretty smart.

He also did some really FUBAR stud at both ASI and some of the Henry stuff.

--Shane
WP2ASS / ex KD6VXI

  Seems pretty much the consensus, so I'll run with it. I've talked to Frank, WA1GFZ, and he says the same thing. It still puzzles me a bit, though: the equal currents thing...I'd hate to have the ghost of Mr. Kirchhoff rattling its chains in my hallways over my presumption that equal currents don't flow into and out of a circuit element. And in audio applications, like a pair of 811's, for example, you've got 4 amps of 60hz AC flowing through the filaments of each tube—directly heated cathodes—and yet that 60hz hum doesn't find its way into the audio. Not disagreeing with the established wisdom at all; just scratching my head, which is probably why I don't have much hair up there.
  In any event, I'll be winding myself a couple of bifilars for my 2X 4CX400 amp ( http://amfone.net/Amforum/index.php?topic=48198.0 ). My original thought was to use 6 amp axial lead RF chokes, because they're small and it's easy to keep the leads short, but I'll stick with the conventional wisdom. Just waiting on a couple of ferrite rods to show up.

The taps for bandswitching are on the outer conductor, like they would be on a normal type of copper tubing plate coil. I would presume a cap to ground from the chosen tap or a pi setup of some sort but I'm not qualified to make that guess.
The center conductor just goes its merry way all the way through unmolested. Can't say how well or badly it may work.