Interest in forum designed HV supply?

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.
kbonin
Posts: 123
Joined: Sat Jan 04, 2003 9:54 pm
Real name:
Contact:

Interest in forum designed HV supply?

Post by kbonin »

Like many people on this board, I’ve been lurking on eBay for some time, and too few fusor friendly supplies seem to be available.

I’ve got a question – given the specialties of various members of this board, how feasible would it be for some of the board members to collaborate on the design of a reasonable supply for amateur physics applications?

Given a baseline specification of: 120v single phase driven, continuous duty 50kV @ 50mA, voltage and current limited, arc and short-circuit protection, precision external control and feedback.

How hard would it be do design this thing? Given a full-wave Cockcroft-Walton multiplier driven through a step-up transformer driven by a IGBT stage in front of some sort of power supply controller IC hiding behind the appropriate levels of opto-isolation?

Its beyond my own current analog (RF and HV) design skills, but it seems that given the identification of some reasonably priced key components (like the diodes and caps in the multiplier stack and the appropriate IGBT’s) it should be possible to back into a first-order design that a few people could build and tinker with.

Anyone interested in discussing it? It seems like there would be interest in the amateur community for applications beyond the fusor…
ChrisSmolinski
Posts: 147
Joined: Thu Jun 13, 2002 2:46 pm
Real name:
Contact:

Re: Interest in forum designed HV supply?

Post by ChrisSmolinski »

I'd say it is do-able. At a previous job I designed the controller for an x-ray power supply (uP controlled, it handled two sources actually) that regulated both the HV and beam current. Current regulation was 0.02% and voltage was 0.005%, I doubt we'd need that. It drove a voltage multplier that was either half wave or full wave. Up to 250 kV and up to 20 mA, though not both at the same time ;-) Standard 120VAC input.

If we came up with a group design, we could do a group buy on the PCBs and major components.

My design was PC/104 based, I'd suggest maybe one of the Rabbit uC boards today. Cheap, and you can get ethernet, which would let you control via a PC rather than knobs and dials if you wanted to.
guest

Re: Interest in forum designed HV supply?

Post by guest »

I'm definitely interested. I think cheap HV components can be difficult to find. Were you thinking more along the lines of an X-ray transformer as your step-up or winding it yourself?

50kV at 50mA is 2500W. Are those Voc and Isc values, or do you really mean running at 2500W? If so, I think you would have to move to 220V to handle the power. I think 1000W is more reasonable for a 110/120V supply.

How would you provide feedback to your controller? A resistor divider for voltage feedback? What about current feedback? If you can be sure current only flows in the HV cables, then you could measure the current on the return path either with a resistor or a hall effect sensor. If you wanted to use the power supply in an application where the air is ionized and travels in many paths to ground (e.g., ionocraft), then you would have to put your current sensing element on the HV end. But how to do that? Hall effect sensor?

It seems like the factors to consider are:
-obtaining or creating a HV transformer
High freq. or low freq? Air core, iron core, ferrite core, or piezoelectric transformer? The frequency of the input signal will determine the core type for a ferromagnetic transformer (or vice versa if you already have a transformer).

-obtaining HV diodes & HV capacitors
They need to be able to handle large enough currents.
-obtaining HV resistors
-obtaining a power controller IC
-Creating the low voltage power supply end.
-Closing the loop
-Others...
User avatar
Richard Hull
Moderator
Posts: 14992
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: Interest in forum designed HV supply?

Post by Richard Hull »

Dave is correct. If you are gonna' control this thing and not use massive capacitors needed for a 50ma rating at 60hz, then you are going to have to go to a ferrite core unit driven at high frequency. Costs here can be very prohibitive. Do we have the interest and the bucks behind the interest here?

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
kbonin
Posts: 123
Joined: Sat Jan 04, 2003 9:54 pm
Real name:
Contact:

Re: Interest in forum designed HV supply?

Post by kbonin »

On input / output ratings:

I just rescanned for numbers, Richard Hull said in this thread viewtopic.php?f=11&t=4260#p27098 "50kv at 30 ma" - that’s 1500W, or 12.5A @120VAC @ 100% efficiency.

I think that single phase is a good target, even if it means scaling down, so that leads to these questions:

What’s a reasonable efficiency to expect for such a device?
What’s a reasonable input load to maintain a single-phase device? 1500W?
For an average fusor's losses, is there a minimum current? 20mA?
Where would that move the Voc/Isc targets?
kbonin
Posts: 123
Joined: Sat Jan 04, 2003 9:54 pm
Real name:
Contact:

Re: Interest in forum designed HV supply?

Post by kbonin »

I'm curious if there are ways to come up with a lower cost unit, or to have enough interest that a bulk order of a few critical components could overcome the availability limitations most of us have – kwtubes in Lithuania doesn’t seem to get his hands on enough large caps for more than one or two decent multiplier stacks a year, for example…

Even if not, it seems like an interesting exercise – I for one have wanted to know more about the inner workings of HV supplies for at least 20 years! :) I got to help design and build an insanely high current arbitrary waveform generator once; this looks like far more fun.
ChrisSmolinski
Posts: 147
Joined: Thu Jun 13, 2002 2:46 pm
Real name:
Contact:

Re: Interest in forum designed HV supply?

Post by ChrisSmolinski »

By going to a moderate freq (say 20 kHz) we can still use iron core magnetics, and reduce capactor values to inexpensive (and safer!) values. I think some scrounging around may turn up a lot of relatively inexpensive caps and diodes for the multiplier.

Of course, we all may have different values of 'inexpensive'
kbonin
Posts: 123
Joined: Sat Jan 04, 2003 9:54 pm
Real name:
Contact:

Re: Interest in forum designed HV supply?

Post by kbonin »

That sounds like a very good thing - iron core is within the reach of a moderate budget home lab, ferrite core means were back to scrounging for a rare critical item, or even worse, paying for custom. Iron core transformer seems like an addition to the design constraint list, and 20 kHz is easy to deal with.

I figure the multiplier stack parts are the most expensive element, and also the most difficult to locate. Any ideas?

I also agree on the computer interface, thats a given for all my lab work nowadays. I've been using AVR parts in my designs to date, its a nice single chip do everything microcontroller, and its going to add Ethernet soon. I've heard of the Rabbit boards, I'll check them out!
Richard Hester
Posts: 1519
Joined: Thu Jul 05, 2001 12:07 am
Real name:

Re: Interest in forum designed HV supply?

Post by Richard Hester »

Iron cores for 20khz operation are really not within an amateur's reach unless he is very lucky, and they are still not very useful at that. 2 mil (0.002") cut cores or laminations will work at 20kHz for inductor applications if the flux swing is fairly low. The flux swing will be too large for an economical transformer design (i.e., it will burn up). The 2 mil material is not readily available anyway. The cut cores are made to order, and any overstock is quickly snapped up by institutions with deeper pocketbooks than ours. 2 mil laminations may be available at a minimum order of about 100lb.The commonly available silicon steel material is 12 mil or 16 mil, suitable for 60Hz operation. At 20kHz it will melt down very quickly.
For a 20kHz transformer, ferrite or amprphous metal is the best choice.
Having said that, it is hard to beat a 60Hz transformer and variac for adjustability and ruggedness. I have over 20 years of direct experience with switching power supplies (it's my job), and this is the path I'm choosing. There are more than enough challenges with the fusor without the added aggravation of messing around with a cranky home-built high voltage switcher. Look in the junkyards for X-ray supplies, or very politely and circumspectly ask your local x-ray repair establishment (consult the phone book) if they have a junked head or transformer they would be willing to sell. Come across as sober and responsible, and you may have yourself a deal. Come across as a screwball or hotdogger, and they will (rightly so) hang up on you. There is an Ebay seller offering 150kV diode stacks at a very reasonable price. Add a big variac and a set of ballast resistors, and you're in business (mostly). A 60Hz transformer-based supply is really the best way to go for a beginner - anyone who thinks they can put together a 2.5kW, 50kV switching power supply from a PC board and wishes is really asking for a lot of trouble and expense, with no reward.
ningauble
Posts: 104
Joined: Sat Jun 16, 2001 8:17 am
Real name:

Re: Interest in forum designed HV supply?

Post by ningauble »

I agree with Richard Hester (How are those PMT boards
going?) if you look at :
viewtopic.php?f=11&t=4268#p27106
You'll see a lot of this conversation has been already covered in
detail (where I argued the toss a lot with Richard Hester but
followed his advice anyway) . If you just have to make
something I'd be inclined to custom make a HV tranny, rectifier
and inductor/cap filter. I ended up being very patient and
scoring most of my stuff commercially made except for the
rectifiers which are resin encapsulated strings of 60 x 1N4007
diodes.

Mark Harriss
guest

Re: Interest in forum designed HV supply?

Post by guest »

Has anyone tried making their own transformer and putting together their own core from scratch (iron-core, ferrite)?
ChrisSmolinski
Posts: 147
Joined: Thu Jun 13, 2002 2:46 pm
Real name:
Contact:

Re: Interest in forum designed HV supply?

Post by ChrisSmolinski »

I'll have to see what I can dig up on the iron core magnetics I used for my design, this was over 5 years ago. There are some ferrite options as well, I don't want to completely dismiss them. We most likely will be stuck with winding our own transformers. I still think the pain of this is well worth it, compared to the many difficulties we will face going with 60 Hz operation. 60 Hz sounds very easy at first, you can use off the shelf magnetics. But your caps get to be huge. Lots of stored energy. Ripple and regulation are terrible.

I've mostly played with the PIC uC, I did some Atmel as well (mostly the Mega series, some of the smaller stuff), and most recently the Rabbit. Digikey has some nice cheap boards & dev stuff in their catalog.

I think computer control is the only way to go today. Ideally, when/if I build a fusor, it will be 100% computer controlled. With sensors monitoring *everything*, logging all runs. Then the data analysis comes later. Automated control, stepping through various voltage/pressure levels would be a great way to find out what happens where, and discover any magic operating points. (Though my understanding is that most fusor runs are too short lived to do much of this, due to heating of the grids if I am correct)

All the stuff we need is expensive if bought new, surplus, both used and NOS, is the way to go. I think the odds of finding already build multipliers that are suitable are close to zero. Just find the caps and diodes, building the multiplier, especially for low voltages like 50 kV, is easy. Ask me about 150kV multipliers sometime ;-) I'd suggest mounting them, along with the transformer preceding them, in an oil filled enclosure.
Richard Hester
Posts: 1519
Joined: Thu Jul 05, 2001 12:07 am
Real name:

Re: Interest in forum designed HV supply?

Post by Richard Hester »

The Atmel has to have something to control before you can get your HV. Designing a robust HV switcher is no joke. Actually constructing one is again quite another thing. If you don't get it right, the thing destroys itself with the first arc. And don't tell me there'll never be an arc...
User avatar
Richard Hull
Moderator
Posts: 14992
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: Interest in forum designed HV supply?

Post by Richard Hull »

MULTIPLE RESPONSES

Kerry: Fusors will do decent fusion at as little as 3-5ma! More ma at any given voltage, more fusion.

Richard Hester: Right on! I really hate the limitations of a linear supply ala 1950, but as you say, it is really tough to kill a magnum iron core 60hz tranny backed by a beefy variac. No need to really filter, per se, to flat lined DC.

A switcher is pretty much for a well defined, fully understood load situation. (which plasma sputterings are not)

First pass on a power supply should be for a demo system anyway and linear is the only way to fly. Besides, you only need about 7kv or so at a few mils. Later can come the monster supply with the big iron on your hip. Neon sign transformers are the unit of choice for the first pass demo fusor. Just as much as they are totally inappropriate for real fusion systems. Build simply, robustly, and only to the level actually needed for the task at hand. You will learn in well defined baby steps.

Dave: I am unaware of anyone here actually rolling their own transformer system. It is a great up hill battle where HV is concerned.

Gerneral:

Multipliers are just great, but lose a lot in the 60hz region as the caps reguired for even slipshod functionality are expensive and hold too much energy. So..... multipliers are best used in switcher systems to boost voltage. At 60 hz, the transformer is where the voltage needs to be produced.

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
ChrisSmolinski
Posts: 147
Joined: Thu Jun 13, 2002 2:46 pm
Real name:
Contact:

Re: Interest in forum designed HV supply?

Post by ChrisSmolinski »

Richard Hester wrote:
> The Atmel has to have something to control before you can get your HV. Designing a robust HV switcher is no joke. Actually constructing one is again quite another thing. If you don't get it right, the thing destroys itself with the first arc. And don't tell me there'll never be an arc...

You're certainly correct, there is always an arc. Arcs were a major cause of failure in our x-ray HV supplies, and the sources themselves, both in damage to multipliers and the tube itself (we blew a hole in the glass more than once). Current limiting is your friend.

The president of the company I worked for when I designed the x-ray power supply insisted on personally testing out how it handled arcs.

We rigged up a 160 kV source (minus x-ray tube for radiation safety and installed high voltage power resistors in place to simulate the load). All tanked in oil, with the top lid removed for access. He got a 6 ft lucite rod, and attached a ground wire to one end, and proceded to draw arcs fast and furious. A whole crowd gathered on the test floor. You could see the oil swirling around in the tank like it was a deep fryer, due to the electrostatic fields.

There was a great cheer when he produced a rather good arc, and the HV shut down. Then sighs after it came back up a few seconds later.

The test results were a tie, both the HV supply and company president survived.
kbonin
Posts: 123
Joined: Sat Jan 04, 2003 9:54 pm
Real name:
Contact:

Re: Interest in forum designed HV supply?

Post by kbonin »

Here’s a perfect example of why I like this board so much – I both just publicly exposed my ignorance of something seemingly simple, and learned something! :)

In all the iron core transformers I’ve played with, I always assumed (in the alternative meaning of the word, I now realize) that the use of multiple plates in the core construction was solely for manufacturing simplicity, it being cheaper to stack thin stamped sheet than to use thicker plates. I never realized there were reasons for the plate thickness.

So is the thickness and alloy used herein sufficiently unusual that it wouldn’t be feasible for an amateur group to attempt it? You mention 2 mil as appropriate for 20kHz…

Questions
Is 20 kHz a good target for iron-core HV in our power range?
Is 2 mil a good target for cut cores for this transformer?
You mention possible 100# minimum orders, any pointers to sources? Prices?
User avatar
Richard Hull
Moderator
Posts: 14992
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: Interest in forum designed HV supply?

Post by Richard Hull »

I'll; respond briefly on only one point. The laminated steel and related thickness is strictly there for magnetic loss reduction and not convenience of construction! (more new knowledge) Manufacutrers would do hand springs for joy if they could just cast solid blocks of steel or iron for cores.

But No!

They gotta' watch out for magnetic losses in the form of hysteresis losses and saturation losses. This means strap or sheet steel laminations of special composition needed to be carefully stacked, insulated in some cases, and interlaced inorder to improve efficiencies and cut losses.

Steel doesn't magnetize instantly and likewise doesn't shed its magnetism right away either. The faster the cycling (alternations) the thinner must be the sheeting and the more complex the core until you get to a frequency where only a powdered iron matrix is adequate to the task (ferrite). Nothing is more electrically frustrating than having to reverse the flux in a chunk of iron you just magnetized a few hundred microseconds before. You now have to fight the very field you just created to demagnetize it PLUS force the steel's mag field back in the other direction. This is massive loss and only 1500 piece sheet steel laminated cores can help solve this problem....EVEN AT 60hz!

I mentioned insulated above. THAT's RIGHT. Insulated! The core is metal. Metal is conductive and the core has to close around the coil. As such, the core would make a giant 200 million ampere shorted turn just sucking on the power grid. So, each lamination or little strip needs to be insulated from every other little strip, in many cases, to avoid this problem. (conduction and or induction losses)

Power transformers are not pretty to design. They are an artform, though most transformer companies design by cookie cutter prior work design extrapolation.

Note: Maxwell's equations DO NOT allow for intuitive transformer design! Professor Parry Moon of MIT found this out as a young engineer in the 1930's and reworked maxwell back to a modified form of Webers original equations to satisfy all physical scenerios involving ELECTROMAGNETIC INDUCTION.

The method of teaching flux cutting fails miserably in transformers. Not one line of flux cuts or moves through any copper in a transformer, yet induction takes place in spite of this. Maxwell knew this and dreamed up the "potential vector" to handle this problem. Trouble is you gotta' know just when to apply certain of Maxwell's work and where it is going to pass muster. Parry Moon and his mathematician wife devised modifications needed to handle all INDUCTION scenarios and discovered, to their surprise, that they had just created a modernized version of Weber's work!!! There is a superb article with the Moon's work fully shown in a back issue of Infinite Energy Magazine for those who care to wade through the mathematics. It features several scenarios where waxwell fails to compute accurate induction results and their modifcations of Weber gives good answers across the board.

Unfortunately for the Moons and Weber, one needs to know far too much precise info regarding charge positioning to apply the work in simple day-to-day practical work and so maxwell's effective short cuts remain to this day provided you are smart enough to choose the right path through his work.

It is interesting to note that induction as done by weber and the Moon's needs no B field (H) in any of the equations!!! Why? Because due to first principles, the B field is a secondary effect and is fully dealt with by knowing the exact position and action of charge and charge alone.

Magnetism and light are not primary entities but instead secondary junk. Use them as primaries in calcs and you are bound to limit the value and validity of the results at the expense of ease of use and application.

Sorry for the protracted theoretical machinations, but even going back to the 1860's we took some wrong turns to simplifiy things which are biting us in the ass to this day.

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
Richard Hester
Posts: 1519
Joined: Thu Jul 05, 2001 12:07 am
Real name:

Re: Interest in forum designed HV supply?

Post by Richard Hester »

I specifically said in my previous post that 2mil laminations would probably not be suitable for a 20kHz transformer because of excessive eddy current and hysteresis losses. Maybe one could make the thing work by using a big core and/or lots of primary turns to push the flux density down, but it's not worth it. The only place I've seen silicon steel work in a 20kHz application was for the output inductors of 1000-1500W 5V switching power supplies, where the flux was mostly DC with only 10-20% current ripple.This usage quickly went by the wayside with the advent of higher switching frequencies. Ferrite or amorphous metal is the only way to go for a 20kHz transformer. For a HV transformer, the secondary must be specially designed to reduce capacitance, or you will spend gobs and gobs of power just to charge the secondary stray capacitance. As an example, assume a stray capacitance of 100pF, which is a conservative guess for the stray capacitance of a couple of thousand turns of non-optimized secondary. Assume also a secondary voltage of 10 kilovolts. At 20kHz, the secondary stray capacitance will sop up at least 125ma just to charge the stray capacitance. This translates to an enormous current on the primary side of the transformer. Smoking MOSFETs, anyone? No one should try to design and build a HV switching transformer without a clear knowledge of the tradeoffs and pitfalls.
User avatar
Richard Hull
Moderator
Posts: 14992
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: Interest in forum designed HV supply?

Post by Richard Hull »

I was responding to Kerry's surmise on why he thought they used laminations in transformers. I also thought I would explain why they do it at all. I wouldn't push a laminated iron core unit above 10khz. The audio guys did it in the 60's but that was with difficulties and losses accepted. If one needs higher freqs just bite the bullet and go to the specialized powder cores, then up your freqs and save on the caps needed to make DC.

When components have a cross over point and need to change form, it is ludicrous to operate at the edge or in no man's land with either form. Either stay low with laminations or go to the moon with powder. It would be like finding the ragged edge of a power switching mosfet and specifically designing it to work for a bit in its micro-narrow linear region using hand selected units. I have actually seen this done!! It worked as long as the good ship lollipop steamed the way you wanted it too. As near as me and my fellow engineers marveling at this piece of work could figure, it was to save about 7-10 parts in a circuit! Oh those bean counters working to a price point.

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
Richard Hester
Posts: 1519
Joined: Thu Jul 05, 2001 12:07 am
Real name:

Re: Interest in forum designed HV supply?

Post by Richard Hester »

Powder materials are not suitable for 20kHz transformers because of the low permeability and relatively high losses, but they make OK inductors. They work at RF freqiencies because the suitable permeabilities are very low, as is the operating flux density.
User avatar
Richard Hull
Moderator
Posts: 14992
Joined: Fri Jun 15, 2001 9:44 am
Real name: Richard Hull

Re: Interest in forum designed HV supply?

Post by Richard Hull »

OOPs. I have always refered to all non solid metal cores as powder cores whether iron powder as in tuning slugs or small Rf inductors or ferrite powder cores found in switchers. Big difference. I have been called on my oversimplification before by my peers here. Sorry. When I say powder cores, for power, I mean properly selected ferrite (iron oxide). Even ferrites must be selected for frequency by composition based on permeability. Losses in circuits at frequency are often acceptable engineering-wise to achieve an otherwise near impossible goal.

Bottom line........Stay with solid, lower frequency, real steel cores for fusor work.

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
kbonin
Posts: 123
Joined: Sat Jan 04, 2003 9:54 pm
Real name:
Contact:

Re: Interest in forum designed HV supply?

Post by kbonin »

OK, lemme ask this question - transformer wise, whats used in the 50kV 500-1500W Glassman units that are only 3.5" high and weigh < 20#?
Richard Hester
Posts: 1519
Joined: Thu Jul 05, 2001 12:07 am
Real name:

Re: Interest in forum designed HV supply?

Post by Richard Hester »

What's used in the Glassman supplies is a custom designed and built ferrite transformer. The secondary is likely wound on a sectioned bobbin (probably a proprietary design tooled by Glassman) to reduce stray capacitance. There may be some careful interleaving of primary and secondary sections to reduce leakage inductance. The whole secondary may be a potted assembly that is slipped on to the transformer core (I've seen it done this way in some designs). The transformer would certainly be designed by an old hand who has made his mistakes far in the past and has HV, high frequency design down pat. Most of these (HV power) companies wind their own transformers in-house, as most commercial winding houses don't have the expertise for this kind of work and can barely even handle high frequency, low voltage, low power designs. They (the winding houses) can wind from your specific design, but Lord help you if you give them free rein with a transformer design - you will most likely get back a Frankenstein monster. I say this not sitting on a high horse, but from my own and my customer's experiences over the years.
guest

Re: Interest in forum designed HV supply?

Post by guest »

HF Cores in the >1000W are relatively easy to locate. These cores are at the core of an electric welder (TIG, MIG, etc).

The difficulty of building a HV transformer is constructing the secondary coil. I found it extremely time consuming because you must seal each level of winding to prevent leakage and internal arcing. The rest is easy.
plasmann
Posts: 72
Joined: Wed Jul 18, 2001 1:26 pm
Real name:

Re: Interest in forum designed HV supply?

Post by plasmann »

What I've seen used in their supplies are multiple, small ferrite core transformers, looks like around 500W per coil, connected to a couple output buss wires. The cores are about 1/4" dia and the secondary winding is about 2" dia and 1/2" wide. The secondary is floating to allow full wave multiplication. No HV potting on the coils, they are just enameled magnet wire, in the form of a flyback transformers without the potting, probably 20ga. wire, and of course the secondaries only go to a couple kV. The winding has a chevron shape when viewed perpendicular to the coil axis(>>>>) probably to lessen the inter winding capacitance. These are driven in a voltage controlled mode to drive the output full wave multiplier.
Post Reply

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