Searching for gonzo mode

[Doug Coulter]

I thought this might be a better place to discuss the various attempts at replication....since that's really been the main priority, and info has been leaking all over the board about it (I may use this first post to put in links to all that later on). For now...just a starter post.

This is continuing discussions from various places. One of them is: viewtopic.php?f=15&t=1130&p=6673#p6673

Found this while looking for that paper on weird effects of inductors with long length of wire (this isn't it but it another example of that relaxation mode)
viewtopic.php?f=40&t=744&p=4593&hilit=inductor#p4593

[Doug Coulter]

Back in the other thread, I mentioned that with what I think the capacity is, and with the inductance measurments with the MFJ meter on both inductors, the big original 5mH one ought to resonate around 250khz or thereabouts. Was I ever wrong (again) - try around 426khz, but wait, there's more - see bump way up there. For giggles, I mounted it back up as if to use it (but not power supply on the input) with no gas present and ran the VNA over it, getting these two interesting plots...nowhere near 250k.

From the miniVNA pro wide sweep

It's kind of interesting, and from my POV, quite a coincidence that that other bump is where I happen to be working now. This is the first I've seen of it with this inductor. I saw, and proved out for myself, some interesting work on the sort-of spiral antenna effect when you have 1/4 wave of wire in a coil and so on (and I think I put the paper up here somewhere, or can find it again), but despite this being a zillion turns of wire, it's gotta be no where near 1/4 wave of 2.7 or so mhz?
Now I'm going to have to count it (or at least TPI and do an estimate of wire length).

On a narrower sweep it looks like this:

Narrow sweep, otherwise the same conditions

The marker here is 428khz or so - it's hard to hit on the head.

This is all looking into the power supply end of the inductor, being series resonant with the capacity of the tank and feedthrough.

One issue I have is that under no or light load conditions, the circulating current makes even my 115uH inductor - wound on pyrex of #16 wire - hot enough to smoke. This little stuff wouldn't last a couple seconds with real drive...hmmm. Under load (like when there's a lot of gas lit off by the RF alone and it's a near-short), no problems. However, that appears not to be a useful mode for making fusion. We need volts for that.

I note that with a different driver arrangement and (I'll have to go check this number) 66 or 76 khz. is where I saw some bunching, and I'll have to go and watch my own video about all that to verify the conditions then. That is, FWIW, a big reason for this board and my youtube channel. I try lots of stuff. I forget...while I do backups, sometimes things crash in hard to recover ways...
And that's the theory behind this thread, or part of it - lets do some more backup and condense the info and discussion that's actually meaningful to the search here.

As an aside - in the other measurements up around 2.6 mhz - the lower the gas pressure the more delay shown between the main grid signal (my HV probe) and the faraday probes, which are basically just feedthroughs terminating in a little wire inside the tank. Less gas, more delay - and it's a strong function with gas pressure, which seems to be a strong function for every other thing. Time to pay more attention to that methinks. When gonzo mode happened, we were down at "barely can do this at all" gas pressure of .016 mbar indicated. Usually that means the little spellman that makes up to 40kv at up to 2.5 ma is current limiting around 15 or 20kv. An unmeasurable (by me) change in gas pressure below that and we're out of business - can't keep it going and have to deal with hysterisis even at 40kv on the ion grid out in the big space (better for Paschen's law) to re-light, then edge down the gas....the voltage curve is very near-vertical there.

I am also thinking at this moment that when we are running in "normal fusor mode" that we actually have very small fraction of the gas ionized, from what I've seen here. When we manage to get the 2.6 mhz AC-only thing lit off - it's far lower impedance than would be indicated by the DC meters - we're seeing 100's of ohms. With just DC (it's lower with ion grid on, other things equal) it's more like 750k ohms. Just an observation. While we get weak rays in the middle of a fog of light in RF drive only, we get only the rays in DC mode in the poisser.

For anyone wondering about that funny glow on the glass (white or blue) in the youtube movies, it appears to be electrons, we tried to measure that with the original pinhole cam and a probe as well. It's a lot of them. In continuous running it will heat that stainless steel screen in front of the glass to orange hot. You're only seeing what made it through a grounded screen....

[Doug Coulter]

What started this lab down this particular rathole was the observation - very repeatable - that just about anything at all the knocked the normal stable fusor off its dynamic equilibrium increased both output and Q. Thus began a search for the best way to do that on purpose - most of the original measurements of this had micro-arcs in early versions of a feedthrough design that wasn't quite up to the task, as the stimulus.

Well, I managed to run some yesterday, with the goal of checking the thing out - get a baseline just in case, with only DC but with all the RF junk inline - just not powered on. I had some micro-arcs, and a couple big baps - I never did make it all the way up to 50kv in several runs (I was trying to change things around the HV to stop those ticks and baps with some limited success).
And here we are again. This time at least I was taking nearly all the data I could take, though I was too engaged in other stuff to adjust the scope settings to investigate further. No worries, it's still set up that way and we have sunshine (if not warmth) so I may get some more info today. An odd thing is that it was able to stay "lit" at lower gas pressure than it normally will - down around 1.4e-2 mbar (indicated, real is about half that according to the pk251 dox) - and the curve is very steep there. I noted much higher Q than normal at this low pressure, even in the absence of the perturbations - several times higher.
But before getting into those plots, lets look at a frame from the initial run from the big scope. Nothing really changed from the time I was trying to measure transit times, except I slowed down the sweep a bunch to see the hornyak detector waveform as a relatively narrow pulse. In retrospect, I should have sped it back up and adjusted the faraday probe gain up again - I had it way down for transit time measurements - interesting that with the RF on the main grid, I had to turn the gain way down, and with ions present, the faraday probes got LOUD - which is why I didn't bother to say anything about capacity coupled feedthrough there - it was clearly swamped by the delayed signal due to stuff flying around in there.
But here we are, kind of full circle:

Should have twiddled more knobs, but those arcs were a little distracting - I was trying to even find which spot they were happening at.

Now, these little disturbances and ringing on the various inputs other than the yellow neutron detector (the hornyak, so it's fast-onset) were happening lots of times with no neutrons detected (it's not a very high quantum efficiency detector FWIW- so there's added doubt) but it was very noticeable that, triggering on neutrons detected, you saw this almost every trigger. It's unfortunate that the purple trace happened to land at the same spot as the yellow one during "resting" or we'd know more. It's interesting that the fast arc-spike usually happens just after a neutron detection as well, and nope, don't have a clue why.
FWIW, this detector clipping on the negative (true) part of it's waveform means more than one neutron hit it in a burst.

Then we have some Q measurements of interest. Those normally take a video to really show as you want to rotate the plot around to see all the interesting info on it. But of some note is that this is perhaps 4x what we normally had when running out highest output runs earlier in the game - but those were at 50 kv and 12 or so ma - this is more like 35-40 kv and only a couple of ma...lower output - a mere 1-2 million neutrons per second, but wow, almost no power input (or grid heating).
(More to come, I'm troubleshooting a computer network issue and need a boot)

Here's some more info and semi-useful blabbing about it:
https://youtu.be/EZVWrSrz6h4


Hoping to get more today and this time, adjust the scope etc, doggone it. I think with only a mediocre amount of hassle I can get rid of the micro and macro arcing, but maybe not right away since it's yielding some interesting info. Don't want to do that for long - it's scary around some real expensive and otherwise hard to replace equipment.

[Doug Coulter]

A capture of the tank innards. In the foreground is the ion grid, out in the middle of the big area (14" diameter, 26" tall). In the back 6" sidearm, roughly ending at the boundary to the big space, is the main grid.
The sparkles are X rays, hard to shield and let enough light through, and of course, a raspberry pi camera hunts around looking to reproduce the spectrum of D, and mostly fails - but we know that.

A little pi cam looking in from the top of the tank.

The neutron detector hornyak button is right against that sidearm, around a foot from the ion grid. I'm pretty sure what we're counting is from the main grid, due to inverse square, and due to the fact that it more or less stops when we turn off main grid power. Here, though brighter, the ion grid is at around 15kv and 2.5 ma,
the main grid is closer to 40kv at 3 ma. Being in the larger space makes Paschen's law kick in better for the ion grid, but for making neutrons, it seems that
less density - less scattering outside the attempt at a focus - is better. We don't want our ions to hit until they're moving fast at one another. The lower pressure seems to contribute to that in a couple of ways - one is there's just less to hit when the mean free path is longer, and the other is that it's more likely to generate copious new ions further from the main grid, at a better place in the electrical potential well, so they get accelerated more.

Remember, visible light is a loss here - it's a sign of ions recombining into neutrals at low energy, which is a fail. It may be happening "near the action" but it's not the action we're after at all. Those X rays, which the camera can't focus are a little more indicative, but neutrons are the real sign it's working.

Oh, the green glow in the lower left is a piece of a quartz stab-in heater, and the blue is from some ZnS:Ag I spilled in there years ago. It's hard to clean out a tank!
It's mostly electrons making the ZnS glow, dunno about the quartz for sure.

[Doug Coulter]

As promised, here's a measurement of the nominal 50k 225W cartridge ballast I was normally using. It's pretty horrible..

This one has pointy terminals on the side too - bad news at HV.

I like this style best no pointy stuf, and at 1.1", easy to fit a split plumbing itting to. I have a female socket in my feed through with setscrews for this.

A port in a storm. Only terminals.

All pretty close to the same. I'm using a Digilent Analog discovery 2 to do these measurements. It's OK when you can double check another way that it's not telling you a lie. Here I used a 10k precision series R.

The test setup. R held off the bench a little, but no go - the parasitic C is built in.

Open circuit at DC, a brown cartridge one they do go bad, and I hadn't tested this with the ohmmeter first....in use, it would probably have kinda worked, just arcing across the break.

Well, it's 50k *somewhere*...This one's actually open circuit on my ohmmeter, I'll go get another one.