power supply

For posts specifically relating to fusor design, construction, and operation.
Post Reply
guest

power supply

Post by guest »

Assuming I had a transformer that could cope with the high output voltages of up to 50 Kv (oil filled etc) could I just use a variable frequency drive to vary the output voltage?. What kind of switching element would be good in an application like this?.
guest

Re: power supply

Post by guest »

If you had a suitable line frequency transformer, a variable transformer feeding it would be the simplest and most bulletproof way of smoothly varying the output voltage. At line frequencies, microwave oven retifiers are probably the most available and cost-effective means of getting the DC, short of assembling enormous strings of 1N4007s. Finding a suitable oil capacitor for filtering the output will take some scrounging. You don't want to go overboard on the filter cap value, as the stored energy will be extremely dangerous. At 20kV and 10mA out, 50 Hz line frequency, with a full wave bridge rectifier, 0.1 uF will give you about 5% (1kV)of ripple. Stored energy will be 20 joules, which is enough to kill you. Accepting 10-20% of ripple will get you a safer capacitor value (25-50 nanofarads - still dangerous, though). The capacitor will also be a lot easier to find (and afford/carry). It is a good idea to insert a 10-20k resistor with proper holdoff rating in between the HV supply and the fusor to limit arcing damage. This has been discussed in the archives several times.
As for a suitable transformer, you might want to find some establishment that repairs X-ray apparatus (the telephone book is a good start) and see if you can talk them out of a junked tube head. The transformer inside will be generally good for 35-45 kV rectified output and will require oil immersion.
guest

Re: power supply

Post by guest »

Mark -
What Richard Hester has written is by far the easiest way to go. But, if you want to go to high frequency drives, you need to look for the biggest fly back transformer you can find. Phillips has a nice selection of cores if you want to start from scratch. But the bottom line is.. kV x mA = watts. The core has to handle the power you want to get out... so at 20 kV and say 10 - 20 mA, you are looking at 40 W. At 50 kV, same current you are up to 100 W.

The good thing at high frequency is the volts per turn figure, which at 50/60 Hz is only about 0.05 to .07 under load, rises to nearly 0.5 to even 1.0. thus a 20 kV output, good for nearly 28 kV pk in a simple full wave rectifier, comes out of a spool of 10k turns of number 30 wire. not too terribly huge. The drivers are typically pulse width modulator chips.. and there are cookbook apps notes on them..It is a way to go if your handy with electronics... and have some time.

Also the output caps are quite small, (few nF will do it), so stored energy is not to much of a hazard.


Dave Cooepr
wrba
Posts: 1
Joined: Fri Jun 29, 2001 5:50 am
Real name:

Re: power supply

Post by wrba »

Hi!

Just a small comment on the power requirements: 20 kV times 20 mA results already in 400 W power.
BTW: an ordinary TV flyback transformer can deliver at least (that's what I've tested) 50 W output power without overheating the core or the secondary winding and based on the core volume should be able to handle even more. However it won't deliver the 400 W for a full blown neutron producing fusor.

Juergen
User avatar
Richard Hull
Moderator
Posts: 14992
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: power supply

Post by Richard Hull »

I agree with juergen, A flyback is just not suitable for real fusion. A demo unit maybe, but a neutron producing device, no.

I have worked at 30KV and 10 ma recently and that is 300 watts. I am using the old 60hz linear supply mentioned at first which is controlled by a variac (variable auto transformer.)

For normal 50/60hz operation with the least hassle,
only an x-ray transofmrer or old used high pot transformer can give you a high enough voltage without using a voltage multiplier.

Voltage multipliers can allow the use of regular plate or potential transformers but have bad filtering habits and involve multiple capacitors and lots of deadly energy storage units in the chain inorder to achieve low ripple factors.

The ideal would be a custom wound 1kw high frequency core and drive electronics with about 20 kv output which could then be voltage doubled via conventional means. Such a transformer device would be very expensive indeed. The capacitor requirements would be greatly reduced and the supply much more efficient.

Most of us are more likely to have success casting about for an X-ray power supply or transformer and going for the inefficient linear system.

Richard Hull
Progress may have been a good thing once, but it just went on too long. - Yogi Berra
Fusion is the energy of the future....and it always will be
The more complex the idea put forward by the poor amateur, the more likely it will never see embodiment
DaveC
Posts: 2346
Joined: Sat Jun 30, 2001 1:13 am
Real name:

Re: power supply

Post by DaveC »

Sorry about the "mere" factor of 10, in the power figures!! Late night fog. Everyone is right, of course.

Spellman DC supplies in the 80 kV, 150W to 600W ranges are high frequency multistage multiplier -types with extremely good regulation and ripple.... these have remarkably small driver transformers, a bit larger than the typical flyback.

A 0.5" sq core will handle around 50W at about 22 kHz, depending on the actual material. So a 1kW capacity would have not more than a 5 sq inch cross section, ~ 2. 25 " sq. Respectable, but still very small compared to a 1 KW 60 HZ core.

HF supplies, do not usually run as the conventional flyback in a TV set, where the rapid collapse of field is needed for beam retrace. A common mode of operation is in a quasi-resonant mode, in which a PWM driver is set for an optimal pulse width, near 50% at a frequency close to resonance in the secondary. This reduces magnetizing current drive. Voltage control is achieved by varying the DC voltage supplied to the drive FETs (typically HEXFETs) using a second PWM DC supply. This has the same effect as varying the amplitude of the input AC to a conventional transformer. The second PWM can be regulated with feedback from the HV divider to maintain constant HV output under various load conditions.

Dave Cooper
guest

Re: power supply

Post by guest »

Hmm I can get huge C section ferrites made by Neosid for use in welders, the material is supposed to be good to 200 Khz. They are stacked side to side normally to get an E section shape but I'd have to muck around with something like an 833B push pull stage........that X-ray supply idea of Richard Hester's is sounding a lot less painful all the while.
guest

Re: power supply

Post by guest »

Check out a web page from Marco Denicolai. He built a multi kW switching supply at 20 kV to drive a tesla coil for his licentiate thesis. Nothing in the design is hard to get or particularly expensive. I suspect that if a number of folks were to get together, he would make the PCB layout (or even PCBs) available, which would greatly simplify things. http://www.saunalahti.fi/dncmrc/ is the web site... Look under Project THOR, power supply design.
guest

Re: power supply

Post by guest »

It makes sense to do whatever works most painlessly for you.

Where good 50/60 Hz High Voltage transformers are available at the right price, it is hard to argue for anything else. They are by far the simplest approach...but size and weight are factors.

Referring to the big " C " cores, I believe you probably need just two. The primary winding will consist of no more than about 20 or 30 turns. The secondary will have about one -two turns per volt of fully loaded output. (The 833B's, pardon my ignorance here, but those sound like transmitting tubes. You will need a "few" other parts there to make them into a switching pwr supply.)

I think some simple HEXFET 's (or equivalently the IGBTs) are what you need. One product line of HEXFET , International Rectifier's , includes the IRF740 and IRF360.

The pulsewidth modulator chips, I am familiar with (was made by Unitrode, now made by TI) are 14 pin DIP. They're used in many types of X-ray and other HV supplies. The Unitrode series has a whole raft of types. I have used the U3524, and the U3825. They are good for just about any frequency you could want. The package contains oscillator and feedback controls, and the output can directly drive many FETs without anything else!. The U3825 can handle about 1/2 amp output, unaided!

You can get nearly all the circuitry you need from the application/data sheets for these pieces. An 8 pin DIP FET driver, whose number I can't remember, will boost the drive current output to 2 amps, more than enough to control the largest FET.

The HEXFET designs are nearly bullet proof with built snubbing diodes, 400 - 600 volt source to drain voltages, and... they operate like a vacuum tube in control characteristics. So,.other than not having 6 kv plate voltage capabilities, if you like those transmitter tubes, you 'll love these little things. The IRF740 is a TO3 - tab pkg, the IRF360 is a bit larger, also Tab pkg. Forward resistances (the on resistance) is 0.5 ohm and 0.25 ohms , respectively.

The whole driver will run off power supplies of about 12 vDC or higher! Of course for a kW of ouput, that's a lot of input current and mains voltage driver is a better idea. I am less familiar with the IGBT's (insulated gate, bipolar transistors) which are typically used in higher power motor controllers. These are a bit pricey, (the HEXFETs are a few bucks each) but in either case you usually only need two for a push pull driver).

The trick being used today is to wind the primary with two windings in opposite directions. The outer ends are connected together and go to + VDC. The other ends go to ground through the FETs. Schematically it looks like a center tapped winding, but allows a single power supply to give fully alternating drive voltages. These are the solid state equivalent to the old style "vibrator" used in car radios in the 50's. (Those ran, interestingly enough, around 400 HZ).

If you have the time and interest, build a simple low voltage transf/driver.. and get the hang of the circuitry..which remarkably forgiving and simple. Then just add windings to get the output you need.

I know this is oversimplifying it.. but with modern plastics like LEXAN for spools, you can forget about G10 machining, and with a litle handiwork... produce some nice gear of your own design.

Sorry for the length of this post... didn't intend to give a course on switch- mode pwr supplies in one note.

Dave Cooper
guest

Re: power supply

Post by guest »

Hi Dave, the 833B's are a transmitting tube good for outputs in the kilowatt region, the main reasons i'd use them is that I have a few already, they are relatively cheap and very rugged compared to silicon devices and I would hopefully (not 100% sure on this) be able to run them off three phase which is 415 volts over here. I forget the exact ripple figure for rectified three phase but it almost doesn't need filtering caps at all from what I remember. So ideally using valves (tubes) it would run off three phase with no step down transformer and give a bullet proof output on a smaller (than 50-60 Hz transformers) hand wound transformer
DaveC
Posts: 2346
Joined: Sat Jun 30, 2001 1:13 am
Real name:

Re: power supply

Post by DaveC »

Mark - thats an interesting approach. The ripple frequency for a full wave rectifier is 2x mains frequency per phase...so for 3 phase, you would have a 300 to 360 Hz ripple which is much easier to filter. The unfiltered ripple voltage for three phase is only about 13.4% of peak voltage (100 - 86.6) rather than 100% of peak, so you gain there as well.
With 415 V (rms ? ) input you have 586 volts DC available for primary which means the turns ratio is quite modest for an output of 20 or 30 kV.

Tubes also are nice for direct line connection in that the higher plate resistance means more protection against shorts and overloads.

With the transformer cores you mentioned, you could build the supply as a low frequency Tesla coil, and let it free run. The cores would keep most of the EM fields inside, so noise would not be too serious. You would no doubt need a good shielded enclosure, for safety, anyway.

The only issue on frequency that comes to mind is the diodes for HV rectification. They typically like frequencies below about 50 kHz for the fast recovery ones. (Microwave oven diodes are mains frequency types. But the 10kV fast recovery diodes are only a couple bucks each and they are good for 20 -30 mA)



The only issue
guest

Re: power supply

Post by guest »

I spoke today to a radio ham friend who is a bit doubtful about using rectified 415 Vac through 833B tubes, he tells me they are happier with around 2000 volts to run off which would require an input step up transformer and would start to make things pretty complicated. I scored three steel cabinets from the rubbish dump today, they are 8 foot tall modular units in grey powder coat and lined with refractory composite insulating material about 3/4 inch thick throughout. Apparently a local electrical engineering firm went bankrupt and these items weren't bought at auction and so were dumped along with a spool half full of co-ax cable about an inch in diameter (looks like foam dielectric). I bought them (brand new) along with a 5 Kw three phase transformer for $50 Australian which is a bit over $25 US bucks!!!. Anyway whatever power supply I end up with, I now have a cabinet for it and maybe some cable to connect it up with.
DaveC
Posts: 2346
Joined: Sat Jun 30, 2001 1:13 am
Real name:

Re: power supply

Post by DaveC »

The cabinet find sounds great. I would be careful with the coax, to find out if it is good for HV applications. Foam dielectric cable is usually used for low loss characteristics. It is typically NOT suitable for HV work. For HV work, you need a solid insulation (like PE or PTFE) is you are going above a few hundred volts. It's fine for other thing , of course. Just watch the applied voltage levels.

Regarding the tubes preferring 2kV for plate voltage, that would mean your plate current is limited to about 500 mA, so at 580 or something VDC you would only get about 1/4 to 1/3 the power. Also the transconductance will be lower at the lower plate voltage, meaning maximum plate current could be lower still.

So to follow your original idea about 3phase 415, you would need to consider a pretty stout voltage quadrupler circuit. Not impossible, but a bit more involved.

Dave C.
Post Reply

Return to “Fusor Construction & Operation (& FAQs)”