Audio as a diagnostic helper

[Doug Coulter]

I've been going on about this for awhile, so I thought it'd be good to show people what I'm using. The idea here is to use each kind of thing for what it does best, and this includes the human as part of the overall system. With not much practice, your hearing can tell you a lot, and it's effortless, you can listen while looking at something else, for example. Feedback is more or less instant, and humans can hear correlations easily (stereo imaging) that would otherwise take a bunch of signal processing to detect numerically. So I have this in the loop on the way to the data acquisition stuff, for instant realtime feedback as I twiddle things on the fusor.

Cheapo audio amp, rack mount

This is just a cheapo Radio Shack audio amp, really the guts of a car stereo, but has the nice switching etc so I can use multiple sources and switch around. All I did was make it a rack mount, using some oval speakers filched from a dumpster TV set, and an old rack panel I had laying around. Just milled it out for things, and made a shelf to hold the little amp in there. I have "tape out" on the front panel, as well as a couple of inputs for things where I might not want to crawl under there to hook up for a quick test.

I've got the linear signal out from the neutron detectors going in there now, and you can tell quite a lot counters will miss. For one thing, the pulses aren't all the same height, they are taller for "double hits" as well as wider, so you get more bass when bursting is occurring -- where a counter would simply toss that information away. Not to mention, you can tell quickly if you're just picking up EMI or it's real signal when there is some arcing on the main HV.

The big amp above that is for driving things like X ray transformers, and is used with an arbitrary waveform generator to make interesting wave shapes -- you don't need 1.5 kW for listening!

[William A Washburn]

WE, as humans, function mostly on feedback. Imagine, for example, walking without the feedback signals from the inner-ear.
We are full of high-sensitivity feedback loops that allow our bodies to function. When we are in a different domain (e.g.fusor)
these are pretty much useless in this new domain.
However, what you have done, is to have allowed our auditory feedback loop to provide humans with this high-sensitivity sensor.
I am truly amazed that you can hear pulse height and pulse width signals while tuning rather than having to take your eyes off of
the tuning controls to look at the scope and other instruments.
Congratulations on building a device to allow us to use our senses in this totally different domain. You've done it again!

[Doug Coulter]

I am also recording this output on the computer next to it, and hope to get my 10 channel super fast hi rez soundcard going soon so I can do analysis on it all later.

One of the things this is showing me right now, especially in conjunction with the new preamps/power supplies for detectors, is just how variable the fusion rate is, even when all the meters are stable, all the counters are reading about the same -- all that stuff has too much time averaging in it to see what I'm "seeing" with my ears and no averaging. If what I'm seeing is correct (it is very repeatable anyway) then if I could just get the fusor to sit on the peak, instead of drifting all over, I'd be seeing nearly 10x more average output. But you can't even see that variation with most techniques -- easy with this one. I didn't think of this myself, of course, I simply added the idea of stereo and fairly high fidelity to the old geiger clicker.

During the short time over which this scope trace was generated, you can see (and I heard) considerable output variation, that a counter, even a one second period, would just average out.
And in fact, that's how I knew to "pull the trigger" on the scope snapshot without even looking down at it. That's the beauty of this.

Example from the scope

stereoneuts3.png (5.77 KiB) Viewed 2811 times

Top trace is 3He tube, next is B10 tube (lots less sensitive and farther away too) purple trace is NC, and green trace is a faraday probe in the tank -- showing that when we're running strong we have large net negative charge in there -- excess electrons. The upper two traces are somewhat corrupted by the audio amp, which is AC coupled, In both cases the real tube output is the positive going pulse, the negative undershoot is from the cap coupling and a protection diode in the amplifier. While not trivial to see in this picture, I've noticed that the onset of fusion after a dead spot has exceptionally high output. In another thread here, I've posted about "driving" that instead of just letting it happen randomly, with some success -- it's the road I'm now traveling down.

Here's a scope trace that shows that. Top is a neutron detector, bottom is volts on a secondary grid inside the tank, driven through a capacitor from a neon sign transformer. Note that it can't make positive voltage -- it's busy collecting electrons. Then when it goes negative again, we see a bunch of neutron output in bursts, that are matched by bursts on the NST waveform. Those kinda have to be large groups of positive charge forcing that high impedance up to positive voltages, even though the NST is trying to drive it negative at the time. Should also be obvious that 60 hz sine isn't the ideal drive for this, but it was easy, so I tried it.

Neutrons synced to drive waveform on 2nd grid

SyncPulses.jpg (25.69 KiB) Viewed 2811 times

There are a number of things in this waveform I can't yet easily explain, but it's encouraging. During this, the main supply current was in the .1 milliamp range, yet I was getting as many neutrons/second average as with 400w inputs. The top trace shows some corruption again due to the coupling caps in the audio amp and the resting DC value of that particular preamp, but you can get the idea.

FWIW, the things I think are interesting here and can't explain are:

1. Why does it go positive for a little while with the NST? Electron collection should start much faster than this sweep speed could even show, not a millisecond! Should be in the many mhz speed range, like any other vacuum diode.

2. Why did it take so long once the 2nd grid goes negative to start pulsing? It's reached ion source status a long time before that first burst.

3. Why are the burst times so short, but so long in between?

All these times are eons compared to computed transit times in these fields.

At any rate, I now have an arbitrary waveform generator to play with and that 1.5kw amp to drive transformers with said waveforms. You'd think the times should all be in the RF range, but we see what we see, so I'm simply getting ready for both. The wave generator goes to 125 mhz sample rate (it's basically a modified VGA card) and all the way down to millihertz.

If I can learn anything encouraging down at the audio range, I'll finish building that video amp I've been working on and move on up. The challenge there is more the output magnetics than anything else. If I had a clue what frequency I'd be working around -- trivial. But designing for octaves of bandwidth isn't by far so easy. I may have to just build a bunch in various bands.