I just did this recently.? I bought a used Dahl? choke (4H,? 3 amps CCS, and 15 kv rated). Weighs 127 lbs and uses the same hypersil core as the Dahl 10 kva plate xfmrs...which also weigh 127 lbs.? Bought it across town from a ham, who got it from the? old? satellite uplink station, just north of me The date code is 1991, and it's in pristine shape.? They had a total of 19 of em.?
It's called a? C-L-C? filter setup.? ? Basically? ?you take a regular? 'high C' standard B+? power supply, and add a choke in series with the B+...followed by a cap. In software, that's called adding a 'LC' section.? ?Mine is setup for 60 hz single phase. Modelling it all in software 1st (Duncan's PSUD-2), which does NOT handle 3 phase supplies, ( but will handle 50/60/400 hz single phase)... there are some? serious concerns, and caveats.? (You would probably need Spice, or LT-Spice, to handle 3 phase).
In software, it all goes to hell, and fast, IF the added cap after the choke ( let's call it? 'C2')? is not at least the same value as the main filter cap before the choke.? It should really be 50-100% bigger than the main? 'C1' filter cap.?
If the C2 is smaller than the C1 cap, what happens is you get this severe? yo-yoing... down-up-down-up-down-up output voltage waveform, until it finally settles down after 200-600 msecs, depending on values.? For something like say FM broadcast, non issue.? For? CW, and esp SSB, it's a real mess.? ?The 1st cap? (C1 cap) still gets hammered with the high AC ripple current, same as before the LC section is added.? ?The 2nd cap has extremely little rms current flowing into it.?
With any flashover, you still need a fast, HV fuse in series with the? final B+ output, followed by a HD glitch resistor.? ?( I had that before the LC section added).? ?I got the choke for minimal cost, and so it's just a good experiment I always wanted to try out.... more to verify the software than any practical use.? End result is, it's gross overkill for ripple reduction.? ?Mine has a 0db cut off of just 1.7 hz...and a 3 db cut off of only ( I think) 2.5 hz.? Ripple reduction at 120 hz (2F on? single phase) is -129 db.? ?My L4B with it's 25 uf filter cap (8 x 200uf caps in series) is only -30 db... and 2.9% ripple.... which is blatantly obvious on any RF monitor scope, drives me nuts.??
The added LC section gets better. The harmonics of 120 hz, (240-360-480 hz etc)? ?get killed at an even greater rate.? The added LC filter attenuates above it's cut off freq at a rate of? 12 db per octave.? ? I verified all this through measurements. The software is dead on.??
Ok, here's what you don't want to do,? and that's wire the choke in series with the B-.? ? Reason is, any V drop in the B- is just extra, yo-yoing extra, unwanted bias on the GG triode.? ?And that is in addition to the normal regulated bias on the tube.?
The V drop across the choke is simply the current through the choke? x DC resistance of the choke.? ?My 4H choke is 9.56 ohms when cold, and 10 ohms when warm.? I tested it with my small lab supply, and stuffed 3 amps? CCS through it for hrs...doesn't even get warm, stays at ambient room temp.? It does heat up with 6 amps through it.... and right where the gapped cores are.? ( it also becomes the mother of all electromagnets).? Crank the current up..and u will reach a point of thermal runaway.? ?With normal 2.6 to 3.0 amps of plate current through the 10 ohm? choke, = 25-30 vdc V drop.?
I tested a 2nd choke, a new? Hammond 1H choke, which is rated at 1000 ma CCS....and has a 5.75 ohm resistance.? ?I tested it with 800 ma though it, stone cold.? The Hammond is only rated for 1 kv ( 1.5 kv hi-pot test)..so it has to be installed in series with the? B+....AND well insulated from the chassis.? Treat it like a string of 450 vdc lytics.
?
Same deal on the big, 127 lb dahl choke.? ?It's wired in series with the B+, but well insulated from the chassis.? I used 4 x of those red...? HV glastic insulators, that come with threaded? 3/8" inserts in each end.? Choke sits on top of a pair of upside down aluminum U channel's....which are in turn? insulated? ( via 4 x glastics) from the? 1.5" thick plywood? base installed at the bottom of the rack.? ?Insulated from chassis, the V stand off is a non issue.? ?You will still get the normal V drop across the choke, but at least it's not adding bias to the cathode.?
On any single phase (60 hz) FWB / FWD? B+ supply,? the raw DC ( before any filter cap)? has? 67% ripple...and ripple freq is 120 hz ( 2F).
?On 3 phase, the raw DC? has? just 5.2% ripple.? ?On 3 phase, the ripple freq is 6F (300 hz in the UK, 360 hz in NA).? ?With 3 phase, your raw DC is not much worse vs the 2.9% ripple, after the 25 uf filter in my drake amps.?
IE: caps get charged up (or topped up as folks in the UK say) 100/120 times per second on single phase......and? 300/360 times per second on 3 phase.? ?That's the HUGE advantage of 3 phase power.?
?
Before you try the choke, measure it's DC resistance 1st..then model it in software.? ?And make sure it's not a swinging choke.? See if you can measure the choke on an LCR meter.? (peter told me eons ago, that with typ 3 phase B+ supplies,? a small filter cap was used, followed by the series choke...and no C2 cap).? The ripple is then down to zip.? If it works in software, it will work? in? the real world.? ?If your DC resistance on that 10H choke is high, like (as I suspect) in the 40-60 ohm range, any normal V drop across it, will be in addition to the V drop across any glitch resistor.?
?
With the typ series string of resistors used for the multiplier string for the B+? HV meter,? you can put the? DVM across the last resistor, or 2nd or 3rd to last resistor,? and measure the AC ripple component...with a full bore load applied.?
For my setup, for safety purposes...and to protect the choke,? I installed a HB adjustable spark gap across the choke terminals. Made from solid brass? 1/4-20 hardware, mounted on a pair of 3/4" x 3/4" x 1/6" thick aluminum angles.? Angles are mounted directly to the threaded studs on the choke.? Solid brass acorn nuts used where the gap resides.? Acorn nuts are semi domed, and ideal for this application.? If they get trashed, they are easily replaced.? ?Acorn nuts face each other.? Solid brass lock nuts are used to lock the gap into place....done with feeler gauges, and also a hi pot tester.? I also installed a surplus SPST Kilovac ceramic vac relay across the choke terminals.? I can shunt out the choke at any time, for testing purposes.?
Typ, with any B+ to grid arc, or say B+ to chassis arc, the 50 ohm glitch resistor will? LIMIT the fault current to just.... B+ / 50 ohms.? Meanwhile, the fast? HV fuse before the glitch resistor assy? will open off in < 1-2 msecs every time.? The C2 cap will discharge 1st...followed by the C1 main cap. When the C1 cap discharges, all that fault current flows through the choke... with a resulting MASSIVE V spike across the choke.? Chokes impede current ( both AC + DC) by dumping voltage.? I'm 95% convinced this is why chokes fail shorted to the frame in Henry radio amplifiers that use a resonant choke input.? You can also simulate the fault current in software...it's an eye opener.? This is a problem when chokes are wired in series with the B+.....and are not insulated from the chassis and / or no spark gap used.? ?The spark gap across the choke trick was used by Harris, and several others in the past.? ?I just re-invented the wheel.?
On software, you can see the normal negative / positive swing on the choke with normal plate current drawn, either CCS cxr, or up / down like CW / SSB.? ?Typ it's not very much, < 100 vdc.? ? The spark gap across the choke is dirt cheap to implement.?
?
Jim? ?VE7RF
