4-125a amp for fusor ion trap

[Doug Coulter]

I don't need exact value here, it's more or less a low ESR/L bypass. But I can't seem to find doorknob caps with anything like those values. All the HV ones are in the hundred pf range...and all the big capacities are in the couple kv range (surplus sales etc). As far as searches here in the US go, so far.

Do you know if those glassmike plastic caps are any good? We did find one for "only" $139 + ship.
We might have up to an amp RF there (optimistically, it's probably going to be a lot less). I'm just trying to keep the Spellman safe here and not waste too much power in heating capacitors. A little reactance will tune out just dandy if it isn't lossy. In this scheme, at least for now, we just get "close" in frequency and tune the rest (like they did with cyclotrons and for the same reasons - everything else is easier to tune).

This place is one of our better sources in the USA: https://www.surplussales.com/home.html#Capacitors
Amazon is showing some ceramic disk caps with the right ratings(!) but I'm betting it's those lazy chinese leaving out a decimal point in the kv rating, as they're too small to not just arc between the terminals at 50k..https://www.amazon.com/50000V-2200PF-Vo ... B075SRXR5J
Unless those are really 5" in diameter or more and that lead wire is #8 or bigger...kinda doubt it.


I'll keep you in mind about those NOS tubes. These seem new so far - the crumbling brown paper they were labeled with said so! (and they measure good and reasonably matched). I'm trying to take as much of Bob's advice as I can and keep from frying them (or anything else).

Don't you love it when you see new printed on something decades old?

[Bob Reite]

I love NOS stuff!

Anyway hvstuff.com has some 50 Kv 4800 pF ceramic doorknob capacitors $69.00 each as well as other vaiues at that voltage. They also have a 60kV 3000 pf polystyrene capacitor for just $35.00. While I never bought capacitors from this outfit, I have bought HV diodes from them and they lived up to their specs. Only downside is that the parts shipped from China and you have to wait a month to get them.

The ceramic disk capacitors that you saw on Amazon have to be operated in oil or encapsulated in epoxy. In air, they will only work at 1/2 rated voltage.

[Doug Coulter]

Adds to cart....http://hvstuff.com/doorknob-capacitor-h ... 0kv-4800pf
You betcha, and many thanks Bob - we didn't find that one ourselves. Should handle what I need there just fine. I'm just trying to have something fairly close to RF ground there to drive from, and not back feed RF into the Spellman ballast too bigtime. (it's 8k + 3k pf IIRC what CliffS told me)

I'm surprised if those caps on Amazon would do even if encapsulated. Maybe I should flag Bill to get them next order.

My only other option was we own a ton of .25uf (!) at 6kv caps, but by the time you hook 10 or so in series, damn, that's a lotta joules to dump in a failure...ow.
Not to mention kinda big....

Finally some weather not trying to kill me for a day or two. I may get some other work in today. For starters, a few pieces of shielding for this thing. I once scored a bunch of double sided PCB material about 10 mils thick around 2 feet on a side - makes great shields.

I'm working on a plastic table in the normally heated building...my guess is that it's metal frame is picking up lots of this and taking it to all sorts of places I don't want it to go - like my scope, the computer, all those microprocessors I normally use that station for...I'm going to run a fat ground wire to the outlet box underneath and tie those metal frame pipes and keyboard tray together....that was scary when the keyboard lit up, the scope changed settings...and so forth.I guess for only a couple MHz it should be OK with a 2' wire that goes to the same box the power is coming from.

[Bob Reite]

Yeah you should be able to get away with the ground that you describe on 160 meters (which is where you are at. Maybe I should ask for a QSL card if I can identity your signal LOL). It was only when i was on 20 meters and higher at 1000 watts that I started having problems with "RF in the shack" and had to go to 2" wide copper strap as short as possible to a ground ring around the building.

[Doug Coulter]

We'll see, I did it at any rate. Tied all the tubes in the table frame together (drilled and tapped holes...) as well as the pull out keyboard tray - a pretty big piece of isolated metal mounted under-table and where I noticed the backlights on the keyboard coming on during a lower power test...It's a mess, but I'm working there because heating more than one building at all times is too much for me..."Spring is on the way". But for now, I'm working here, right next to the heater.

Messy space with tons of uP's and semis to fry laying around...

Gee, I hope you can't hear me, but maybe we should do a test on purpose when I get to that point. Normally you can't radiate this stuff that well without a long antenna...but when we step it up to insane volts, well, maybe there's some conductor with some current in it to radiate with. I've seen things as silly as bumper mount 160m antennas that worked according to their users - at one frequency (super high Q loaded). Similar situation here till I can get it all back inside my faraday cage, just a shorter whip.

And hopefully I land somewhere in this band or even a bit lower - if I jam the AM band talk shows a little, so be it. That's still up in the air as I measure and try to find a place where the voltages I can generate move the particles enough to cycle through the grid and out to a decent radius.

Going lower in frequency seems to compensate somewhat for the fact that the ions don't see the field we think we're putting on there - some of it they neutralize, more or less. So at a lower F they have more time to move, since they're going slower than the simple math would predict. On the flip side, I don't want slower, I want faster, generally speaking as I want them to be coming in fast enough to fuse if they do hit. A whole lot of balls in the air for this tradeoff, and the math seems to be ill defined at best - all I have is a direction, not a slope (or even a shape, I'm using an assumption of a line for the moment). The math I'm using now - which did hit it on the head at my low pressure limit or very close - small perturbations around those numbers just made it worse - doesn't account for inter ion interactions, and things get lousy quick with pressure going up to the range we'd rather run in - I think, as a bit higher pressure is where the DC Q is highest.

Which is itself an assumption. So what's really going on is "test some stuff and adapt based on what I learn". That sweep tube amp could probablty have been somewhat tweaked - it was close to enough running well below what it should have been able to do, but at the limit of what it DID do - around 160w output.

This guy should be loafing by comparison - and can hit class C instead of B, but of course I still need to do some serious protection (of the screens) around what would happen in the main plate supply isn't working, and in fact, just building a decent main plate supply in the first place - that's most of what's on my plate now. Somehow I don't think that just powering down the screen supply when the main supply is powered down is enough. What if it is powered up, but fails or is somehow disconnected? I've seen those failure modes in other things. So, a little skull sweat to be applied...

[Doug Coulter]

What does anyone think about using something like this PTC in the screen supply?
https://www.digikey.com/product-detail/ ... ND/1679670
These are 50ma hold, 100 ma trip, 200v continuous, 600v peak (we have 350v).
The data sheet for a tube in class C shows typical screen current at 55 ma at 2kv plate volts (less as plate volts go up) per tube.

These would smoke if tripped and ignored - which is fine, they're cheap. A neon and R across them would light up if tripped, and of course the RF output would pretty much stop at that point.
Maybe one for both tubes (which are in parallel). The supply transformer itself wouldn't do that kind of current for very long, I'd bet, and the 6146 is already using 20-40 ma from it.

Trying to keep it simple. A latching cutout or a fuse that low current are either complex or a little scarce out there.

[Doug Coulter]

FWIW, the intel NUC you can't quite see in the picture above, just behind the amp, went up in smoke, won't get into bios anymore. I'm copying the disk which seemed to have survived now. Looking for a used one about the same generation so I don't have to buy new ram and disk too....this one's ddr3l and sata, much newer and they need ddr4 and nvme. I'm hoping all the pies and odroids on the table and shelf lived...
And this is why you use tubes on fusor stuff - they eat lightning and don't even burp loudly.

[Bob Reite]

The new LDMOS RF devices are pretty rugged, but they still can't take the beating that a 3CX3000A7 tube can. I'm thinking the last time that the 103.5 I take care of took a hit. Actually I had that place pretty well protected and the UPS bought at Office Max gave it's life to protect the STL receiver connected to it. The Poly Phasor lighting arrester on the antenna input did it's job. The tube output transmitter didn't even notice.

Oh. Back to power supply design. One thing I've seen done is a full wave bridge rectifier with a center tapped transformer. But unlike the configuration used in audio circuits to get equal plus and minus voltages, the minus output of the bridge is grounded. Then the center tap will give half the supply voltage. Still may not be low enough. I still work on transmitters that use a 5CX1500A. They almost always have separate supplies for fixed grid bias, screen and plate. Sounds expensive, but in their day the 5CX1500A was the cheapest way to get 20 db of gain, which was needed since exciters of the day could only put out 20 watts at best.

I'd still just use a 240 ohm 1/2 watt screen resistor for a "fuse".

[Doug Coulter]

240 ohm 1/2w hits ratings at 45 ma. Are you thinking one for each tube, or for both?

We probably don't hit the 35 kiloamps or so of lightning here, but hard to say - the very low series impedance for the source of those arcs would let us get there....
As John pointed out, things like coax and air capacitors have really good characteristics and do hold some charge here. The pulse might be short, but the 2nd breakdown of things is easy to reach too.

Looks so far like what fried was the wifi card in the nuc - the antenna is in the top plate and is apparently what saw the "hit" - which might have been just plain capacitive coupling....thank heavens it boots again now that I took that out - and now I should add some deliberate shielding there I guess. These compact little things aren't easy to get apart to modify, but in this case, it might be worth it. Even used ones sometimes are north of $1k. (saw one for $2300 on ebay, but I think that was a stupid-tax collector trying to score). And that's for the 4th generation (Intel is on 8 now, moving to 9).

Oh, we did get those caps from Amazon. The pix don't do them justice - they're .7" thick....OK, getting some encapsulant going....now I believe the ratings.

[Bob Reite]

240 ohm 1/2w hits ratings at 45 ma. Are you thinking one for each tube, or for both?

One for each tube.