Switcher supplies?

This forum is for specialized infomation important to the construction and safe operation of the high voltage electrical supplies and related circuitry needed for fusor operation.
Post Reply
Richard Hester
Posts: 1519
Joined: Thu Jul 05, 2001 12:07 am
Real name:

Re: Switcher supplies?

Post by Richard Hester »

The big problem will be getting the 2 mil silicon steel required to operate at 10kHz. Ferrite would probably be more available. Once that descision is made, there is no barrier to running 20-30kHz. I'd rather not mess with the tubes, myself.
User avatar
Richard Hull
Moderator
Posts: 14992
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: Switcher supplies?

Post by Richard Hull »

The iron losses would be horrific in a 10khz tube driven unit using a 60hz silicon steel transformer. I don't mind wasting a little power for amateur work, but I would like to get some bang for my buck! You really would have to do the thing in powdered iron or ferrite.

You need not make a super HV transformer system if working at or near 20khz for multipliers work nicely at this frequency with only small storage caps in each stage to achieve a very high voltage rating with decent current capabilities.

Rchard 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
guest

Re: Switcher supplies?

Post by guest »

I have made some 200W audio transfomers using conventional 4% silicon steel (Lycore 150) which is why that approach suggested itself to me. With interleaved primaries and secondaries (not easily possible in HV)I was getting a sinewave loss of 3dB at around 50 KHz and a squarewave with a 1 db loss and a small overshoot at around 30KHz. 2 Kilowatt HV transformers would be another matter but
I think conventional silicon steel should be up to the task to at least 10 KHz but even if you ran a transformer at say 2KHz the size of it not to mention filtering and voltage multiplying should be much less of a headache than at 50-60 Hz. The big pain would be implementing the driving circuit,
would some kind of triggered tube work here at all?.

All in all though I wonder if it's worth the hardship, after all If I just make a 50 Hz transformer good for say 50 KV (I can get a lot of engineering advice from the pros for this) and just hook up the output to a bridge rectifier and the input to a 2.5 KVA variac I already have, I'm then ready to go without all the complexity. There's a local company that makes CNC wound GOSS C cores to almost any size that can be easily removed and fitted, I could get a large one and play around with that.
Richard Hester
Posts: 1519
Joined: Thu Jul 05, 2001 12:07 am
Real name:

Re: Switcher supplies?

Post by Richard Hester »

Silicon steel will work fine at 10kHz, as long as you have 2mil (1mil = 0.001 in) thick laminations or use a cut core made with 2mil thick tape. Ordinary 60Hz laminations are 12-16 mils thick, and will heat up nicely at 10kHz, especially if the flux density is pushed near the limit. The big problem is finding 2 mil laminations or cut cores. The cut cores are made to order, and any overruns are greedily snapped up by bigger fish than ourselves. It is difficult to find a distributor that will sell less than a truck load of laminations. Hamfests might be a possible source, though most of the stuff you find will be suited for 50-400 Hz operation. A nice ferrite U-U core will do the trick for high voltage, high frequency operation. These can be bought readily from distributors like Elna Ferrite lLabs or Lodestone Pacific.
BTW, powdered iron is far better suited for inductors than transformers when operating at DC to about 400kHz or so, as the low permeability would result in unacceptably high excitation current for a transformer. This changes at frequencies above a few Mhz, and there are grades of powdered iron that could be used as transformers for low level RF. I am contemplating using Micrometals grade 2 powdered iron in a current transformer to measure the current pulse in a copper halide laser. Since the current pulse FWHM is not likely to be longer than 100nsec, exciting current is not likely to be a big problem. However, excessive inductance will bring the resonant frequency of the current transformer down to unacceptable levels. Powdered iron (or even a phenolic toroid) fills the bill nicely.
User avatar
Richard Hull
Moderator
Posts: 14992
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: Switcher supplies?

Post by Richard Hull »

The general point to be made was that you can't run a 60hz transformer at 10khz which is what I thought the original idea was. Secondly that at 10khz, you WILL be "rolling your own" if you want a real HV secondary with current capability.

You might not need a physically big core but believe me, if you can't get it surplus, it will be BIG in cost.

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
guest

Re: Switcher supplies?

Post by guest »

Hi Richard, Sorry If I digressed a bit in that last post of mine, I guess the point I was trying to make was that I am getting acceptable performance out of 4% silicon steel made for 50-60Hz purposes but in a variable frequency application of around 30Hz to 30KHz with a 1 db drop at both ends. One important difference is that the flux density is a lot lower than for power tranny applications as I have to get acceptable performance down at 30Hz which needs twice the iron as you would have in a 60Hz power transformer, this means that I'm nowhere near saturation at the upper frequencies but my iron loss is greater.
guest

Re: Switcher supplies?

Post by guest »

G'day Richard, I just went and measured some leftover E/I laminations from my audio tranny project, they are around 20 thou but at the upper frequency end I have a low relative flux density as I have to have good performance down at 30Hz. This also gives me some big iron losses but still at an acceptable level.

There is an Australian company: AEM Unicore who have invented a CNC core winder which I understand uses 7 mil GOSS to manufacture C cores with a zig-zag distributed gap that apparently gives core losses of 1.2 W/Kg at 1.7 Tesla or nearly half that of a conventional C core. Most importantly they will manufacture one-off cores to order as it's just a matter of entering the required parameters.

The thing I like most about this kind of C core is that you can wind the bobbin first and then slip the entire core into it and use a large hose clamp type device to lock the core closed.

At this stage I have only just embarked on a few small experimental switchmode designs based around the LM3524 and the SG3525 ic's using FET half bridges to power my PMT's. What I would like to know is whether there's enough advantages in going to a higher frequency to justify the expense and mucking around involved in getting a practical design to work. I would imagine regulation is not too important but a variable output is essential. Assuming you could come up with a good transformer design using ferrite or steel, is there enough advantages in a variable frequency design or should I just stick to the 50Hz of Australia.
Richard Hester
Posts: 1519
Joined: Thu Jul 05, 2001 12:07 am
Real name:

Re: Switcher supplies?

Post by Richard Hester »

I would use a ferrite transformer at 25-30 kHz, high enough frequency to get the full advantage of a switching power supply, but slow enough so as not to be a pain. This is also squarely in the territory of 200nsec fast recovery diodes. Finding ultrafast high voltage diodes is a real pain, doubly so if you're down under... I would definitely ditch the 3524 and 3525 chips and try the TL494. The TL494 is the only old-time switcher chip that I still respect. The dead-time pin gives you an extra handle for controlling the duty cycle. It's prpbably easy to get, even in Oz, as they still use scads of them in cheap bipolar 1/2-bridge AT power supplies. Having said that, I still wouldn't mess with a switcher until cold iron just won't do the job any more. There are enough challenges with the vacuum system , fusor, detection instrumentation, etc., to make this project a full -time job. If a switching HV power supply falls into your hands, use it - otherwise, let the engineers get the grey hairs and RF burns.
guest

Re: Switcher supplies?

Post by guest »

Thanks for the advice re the TL494, I'll keep a eye out for it in future and get hold of a few samples to experiment with, my old texts use the LM3524 in their examples but I think I can make the switch without too much trouble. Neosid Australia actually have a nice big C core for use in briefcase welder systems that would be useful in a switcher.

Well as of today I guess I'm locked into the linear PSU route as I picked up three transformers out of Westinghouse electrostatic precipitator units the size of small shipping containers for about $5 US at the dump. The trannys are about twice the size of 350VA neon sign transformers and put out 1.2KV with 35V input, from that I'd estimate they are good for 8KV or so. They are tar filled with no secondary connection to the casing and appear to be a C core if the box is anything to go by. I will go back tommorrow and get hold of the 10uF paper/oil capacitors to go with them and when they are used with the 2KVA variac I have I should be in business.
guest

Re: Switcher supplies?

Post by guest »

I am working on a 1.2 KVA power supply. I am using two transformers. I am using a Toroid Transform built by Torid Corp. of Maryland (http://www.toroid.com/kits.htm). as an input transform for isolation from AC mains and I am using a HoneyWell PowerLite C-Core (AMC-500) for a output transformer.. I plan to drive the output transformer at a high frequency (AMC-500 is a HF transformer). The Primary coil will be driven with a set of IRF HEXFETS (IRFP260N). HexFreds (HFA30PA60C) will be used as for the switching diodes.

The Toriod transform cost less than $200 and the AMC-500 C-CORE cost about $210.00 from http://www.eastern-components.com. The HexFETs and HexFrads can be obtained from digikey. The IRFP260N HexFETs are about $4 and the HFA30PA60C HexFreds are about $6.. IRFP260N HexFETs are capable handling 300 Watts each. You can run these in parallel to increase power output (two for 600Watts, or four for 1200 Watts).

I am in the process of winding the HV secondary coil and constructing the heat sinks for the HexFETs.

Hopefully this information should provide you a starting information for your power supply.

Good Luck
guest

Re: Switcher supplies?

Post by guest »

Hi Guy, That sounds an impressive power supply, I'd be very interested in how it comes along, I'm a fan IGBT's as I'm of the opinion they are rugged enough for the job and not too hard to drive. Unfortunately (or fortunately) I've chickened out and gone for the low tech but reliable approach as I scored three suitable transformers for about $15 US total!! : they are 9.2 KVac 400 VA single phase transformers made by Westinghouse for their electrostatic precipitators, according to the control box they came out of they are rated at 13KV DC @ 30 mA continuous. They appear to be a lot more conservatively rated than the neon transformers as they are twice as big as similar VA rated neons so I'd be more inclined to believe the spec. These things are designed to run continous duty at full power so I'm hopeful of dragging a little bit more current out of them, they don't appear to be current limited like a neon tranny as the circuit had massive high value resistors in series with them for current limiting. In theory I should be able to get up to about 38KVdc and say 30-40mA current out of them.
guest

Re: Switcher supplies?

Post by guest »

I've often thought about doing just what you are doing.
What is the design output voltage? Amperage?
This doesn't sound too portable.... How heavy is it going to end up? Ive even thought about starting from 12v and using a bank of batterries in parrallel to get needed amperage. It sounds like a workable approach.. I'm getting tired of trying to spec off Ebay..
Might be cheaper to design exactly what you need rather than hunt forever on the net.

Larry Leins
Physics Teacher
guest

Re: Switcher supplies?

Post by guest »

Well, I am shooting for at least 40KV for the output. It should be able to pump out at least 30-50 ma at that voltage. It really depends on how efficient the HexFETs and my custom design output transformer end up. The HexFETs have a low Power-On impediance. but I am not exactly sure how much heat loss I will be generated by the HexFETs. I don't have any expensive modeling tools to find out, so I have to do it the old fashioned way, by trial and measurement. I am also concerned with the output HV diodes. They probably will not be able to operate at high frequencies. So I probably will need to drive the output transformer at a less than optiminal frequency. I am still searching for a decent set of HV diodes. There aren't too many manufacturers of HV diodes in the 60+ KV range that aren't for industrial use. Many of the manufacturers offer HV diodes in the 10-1000A range. I dare not think of how much those babies would cost! I am looking to avoid tube based recifiers because 1. I haven't done any work with tubes, and 2. I am also aware that running recifier tubes at HV produces lots of X-rays.

I am going to include variable voltage output by using a combination of changing the driver frequency and duty cycle, so I can run at different voltages for testing. I am also including two circuit breakers (one on the primary winding of the input transform, and another on the primary winding of the output transformer) to limited excessive current draw. Beware of the different types of circuit breakers. Avoid "thermal" breakers, since they will permit overcurrent operation to run for several minutes. You want to use "magnetic" breakers that trip within seconds.

FYI: You might want to consider using HexFETS instead of IGBTs. I investigated the using IGBTs for my design and choose HexFETS over IGBTs because the HexFETS have built in load balancing. The Rds or ON-resistance increases with the temperate of the device. so if you have multiple HexFets coupled together they will automatically distribute the load. HexFETS are very easy to drive and don't have latency problems of the IGBTs. If you look on International Rectifiers web site http://www.irf.com , they have several white papers on HexFETs, including how to drive them in parallel.

As far as the weight, I would expect the whole unit to be between 30-35 pounds. I plan to mount the power supply in a standard 19 inch equipment case. The weight isn't the problem, its cooling! I suspect that this power supply will have a dual use as a rack mountable space heater!
guest

Re: Switcher supplies?

Post by guest »

HEXAFETS get pretty toasty in high load conditions.
I built an inverter out of them and ended up using an old hairdrier with a bad element to blow high speed air on the semis... they can easily burn themselves into oblivion.
Pretty neat on the weight I must admit.
The x ray problem on tubes is fairly easy just put it in a steel box. Russian tubes are very cheap but they also have a high voltage diodes in the 500 - 1500 volts at 200 amps.
Line them up into a diode stack. I'm using diode stacks from x-ray machines good for 100 kv at 40 ma.

Larry Leins
Physics Teacher
guest

Re: Switcher supplies?

Post by guest »

I am going to build special heat sinks for the HexFETs. I've purchased some High CFM 12V fans that I will attach to the heat sinks. The Heat sinks will be constructed out of copper since it's really good at heat transfer.

If I can't find any acceptable semiconductor HV diodes, I will look into using tubes. Do know of any good documents/URLs that cover tube recifiers?
Post Reply

Return to “High Voltage - Fusor Input Power (& FAQs)”