Finding the Right Transformer for Fusion!

[Jeff Robertson]

After extensive testing on my homemade 24 kV power supply, playing with plasmas and fiddling with the variac and pressure valves, I've decided to scrap my current power supply and start fresh. My power supply was run off of two 12 kV neon sign transformers in series, but there was a lot of internal circuitry which was intended to protect the supply from surges, unexpected open circuits, etc. Long story short, these transformers were built to be "dummy proof" and used for neon signs and nothing else, so when it came time to really start pumping up the voltage and playing with the fusor they just weren't able to do what I wanted.

So here I am, looking for a transformer again. I'm making this thread not only for myself (as I know a lot of this information already exists in the search function), but also to consolidate a lot of information about transformers that has been floating around this forum.

Results will vary, but I think most people will agree that 20 kV or so is just enough to barely limp over the fusion line (or rather, generate enough neutrons that amateur equipment can detect them). As Richard Hull has stated in a couple FAQs, fusion can still occur below this point but it is much more difficult to detect it. This threshold of 20 kV is a very awkward number to the amateur fusioneer. The most common "step up" transformer designed for very high voltages, which runs off a standard wall outlet (~110 Vac, 60 Hz), is a neon sign transformer. These transformers are cheap and one can find tons of them on ebay at any given time, ranging between prices of $10-100. What's frustrating about them is that they generally output anywhere between 3-15 kV, not suitable for fusion. Every once in a while you will find a 18-20 kV NST (I have never seen higher, do higher ones exist?), which I suppose would be barely enough to cross your fingers and hope for fusion, but these are a lot less common. Because of this, many people consider neon sign transformers only adequate for demo fusors, and not viable for a real fusion-producing fusor.

Next are flyback transformers. My knowledge on these transformers is limited, but I believe they run off of much higher frequencies than are provided by the mains electricity (I've seen numbers ranging from 5-50 kHz). Creating the circuitry to up the frequency of the current is a mess I never bothered learning, and I see it as an unnecessary step. So I believe the general consensus is that flyback transformers aren't too appealing in fusor power supplies, but someone with more knowledge on them might be able to say more.

Microwave oven transformers (or MOTs) are another type of step up transformer which are commonly seen on ebay. These transformers are even wimpier than NSTs, as they are generally capable of only a few kVs. Completely acceptable for sustaining plasma in a good vacuum, but not even close to enough for fusion.

X-ray transformers are yet another type of step up transformer which the amateur experimenter may have access to. From what I've seen, these are the most ideal choice for a fusion-producing fusor. These suckers are capable of pumping out anywhere from 30 kV up to 100+ kV (just from the numbers I've seen). Unfortunately, my patient lurking on ebay has turned up fruitless, as these transformers also carry a much higher pricetag. I've heard of people snagging them for as low as $100, but the only ones I see are going for several thousand dollars. I will continue to be patient and frequently check ebay, but as more time passes I'm seeking alternatives to my transformer dilemma.

So this is where I'm at. As far as I can tell, my only two options are finding a 20 kV NST, which I've found to be fairly rare, or crossing my fingers for a good deal on an x-ray transformer. What are your guys' thoughts on this? Have I overlooked an alternative option?

Thanks for reading, and I anticipate the feedback. Until then I'll continue researching, bargain hunting, and holding my breathe while I search for the perfect transformer.

Jeff

[Rich Feldman]

Here's some feedback, Jeff.
1. I applaud your effort to consolidate some information about transformers here.
Not sure it isn't already consolidated at least by type, findable with more searching and reading.

2. You talked about having used two 12 kV NST's in series, and mentioned their internal circuitry. Were they mains frequency coil-and-core transformers with protection added, or were they lightweight electronic switcher types? Your answer will lead to more questions about HV isolation and rectification.

3. Your discussion about the 20 kV threshold, and available NST's, doesn't even try to qualify the kV values as RMS or peak. Do you understand that the nominal voltage (which means, literally, the voltage in the name) of an XRT or system is the peak voltage with a normal load? While the nominal voltage of a NST is its RMS voltage with no load? The simple HV metering described in our FAQs responds to _average_ DC voltage at the feedthrough.

A 15 kV NST, as generally configured here with grounded center tap and diodes from each hot terminal, should deliver between 10 and 11 kV peak with no load. Used with a 4-diode bridge and HV isolation of the case and primary, should be about 21 kV peak with no load. Either way, 120 peaks per second. I have seen the two-diode configuration trip SGFP circuits immediately.

[Chris Bradley]

This is an 'armchair' idea for someone else to try as I neither have, nor need, an NST to try this out with.

Let's say you have your NST (or MOT) demo fusor and want to 'upgrade'. One thing that I think might be a fruitful experiment is to obtain one of those variable speed motor drive inverters, the ones that put out 0 to a few hundred volts of 3 phase at 0 to 400Hz. If you now take one of those 400Hz outputs - or maybe if it can't work unbalanced you run it through a 3-to-1 phase transformer - and you use that 400Hz input into your NST, I'd lay odds that your ironed core transformer will still work usefully at those frequencies but you've just multiplied up its power capacity by several times.

Not sure what the upper useful frequency would be - that's the experiment. With the variable-speed inverter you'd be able to tune it to optimum. If you can find a way to drive an NST (or MOT) with a higher frequency then I think you'd find you'll be getting higher power performance from it.

That might just be enough 'edge' to get to the voltage and power ratings you're looking for.... if someone's got the parts, it'd be a straightforward enough test to see what max. frequency an ironed-core NST works at.

Incidentally, some time ago I acquired a couple of oil-burner ignition modules. These were Danfoss 'EBI' units. They take in 180 to 250V and convert it to 10-15 kV (I have one that is nominally rated 12kV and one 15kV), and are rated 45mA short circuit current at 50% duty cycle. The EBI units looked particularly useful as they are electronic in that they are fully isolated from ground, and output 20kHz, so easy for feeding into a voltage doubler, or taller stack. (They also come in other 'earth-tapped flavours'.)

The following URL should provide a link to all such transformers *currently* listed worldwide on ebay:

http://www.ebay.co.uk/dsc/i.html?sacat= ... _PrefLoc=2

and I attach a datasheet...

Again, I cannot say whether oil burner transformers are a promising route to your fusor psu, but is a suggestion for you to contemplate.

[Jerry Biehler]

You can't run a transformer off a VFD.

[Tyler Christensen]

Any reason why not? Pretty sure you can, as far as I know a VFD is just an inverter which you can certainly run a transformer off of. If you were going to DIY the circuit, you would just build an inverter anyways without any motor control systems.

[Rich Feldman]

Like Tyler, I see no problem running a transformer (or a light bulb) from a VFD.

But I think a NST won't get as much of a boost from higher-than-mains freqency as Chris hopes. Here's why:

1) Yes indeed, you can increase the drive voltage in direct proportion to the frequency, before running into core saturation. But the higher frequency does not increase the voltage limit due to secondary insulation. At a given flux density, the core loss (core heating) from hysteresis goes up in proportion to f. The core loss due to eddy currents goes up faster than f, IIRC.

2) the current-limiting feature in an NST is leakage inductance. So to first order: for a given primary voltage, the short-circuit current is inversely proportional to frequency. I bet you could increase the maximum output power by REDUCING the frequency and holding voltage constant. Until the transformer overheats.

Real experiments are better than thought experiments, of course. Who has a VFD?

[Chris Bradley]

Rich Feldman wrote:
> 1) Yes indeed, you can increase the drive voltage in direct proportion to the frequency, before running into core saturation. But the higher frequency does not increase the voltage limit due to secondary insulation. At a given flux density, the core loss (core heating) from hysteresis goes up in proportion to f. The core loss due to eddy currents goes up faster than f, IIRC.
I think it is only right to point out that my idea is likely to be heavily dependent on a particular transformer! It may not be generally applicable. But I think 400Hz is quite a modest increase in operating frequency.

Just recently I was experimenting with 20kHz transformers. I decided to give a few iron-cored ones a go at that frequency, just to see what happened. Two of three did very little, but one (presumably with core metal more like an audio transformer's) was quite happy up to 20kHz and a little beyond. I put 60W into it without any particularly obvious overheating after a few minutes, despite being 6VA rated at 50Hz.

I wasn't thinking that it is necessary to increase the Volts to get extra power through it. It is simply that each cycle is pumping flux more often into the secondary (if the core can work at that frequency), thus it draws more current. I'm sure leakage current on the secondary will bite at some point, but would the question not simply be whether you can beat leakage current by driving it harder? We're working in a world of lossy mechanisms, and it doesn't matter much if you lose a few 100W so long as it doesn't cause excess overheating.

Remember; magnetic flux is Volt.seconds. So if you reduce the time and keep the volts the same, you reduce the flux. As the core losses increase exponentially with flux, you should end up with lower core losses for a higher AC frequency at the same voltage input (- I think!?!). This should therefore be a double-advantage - you can both feed in slightly higher voltage (dependent on the power losses in the core, wrt flux) and it will draw higher current due to the increased rate of change of flux. Not entirely sure about all that, but I think that's the way it is.