homemade switchmode 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.
Post Reply
UG!
Posts: 116
Joined: Thu May 19, 2005 4:21 pm
Real name:
Contact:

homemade switchmode HV supply

Post by UG! »

i have gathered most of the bits to make one of these, but some areas of my understanding remain cloudy:

first, how are people varying the output voltage? PWM control seems to be popular, but it seems to me that this will only vary the average power delivered to the multiplier stack, rather than the actual output voltage. does this not result in a very high ripple output voltage from the multiplier?

what degree of voltage controll should i aim for? the supply (multiplier output) will probably be about 30Kv max, 20-30, 10-30, 0.1-30? there must be a point where PWM control stops working well?

is a current control mode worth implementing?

what transformer output voltage should i aim for?

and finally, where does one obtain polyester sheeting to insulate the transformer lairs? i can;t seem to find unmetalized and un-sticky gunked stuff anywhere (at least in UK)

any other comments?

thanks
Oliver
:)
User avatar
Mike Beauford
Posts: 419
Joined: Wed Apr 25, 2007 2:24 pm
Real name:
Location: Morton Grove, IL

Re: homemade switchmode HV supply

Post by Mike Beauford »

Hi Oliver - I've been working on building my SMPS as my first project in regards to building my fusor. I've been using this web site to figure out all the relavent parameters, you just plug in your numbers of what you like and poof, there it is. ( http://schmidt-walter.fbe.fh-darmstadt. ... mps_e.html ). It even gives you the transformer and inductor type to use, wire size and windings per primary/secondary. Myself, I'm aiming for a 10Kv output at 500mA, and from there I will use a 10 stage CW stack. I'm also going to use current sense feedback because it gives you complete control regarding the output voltage. My plan is to build possibly 3 of these and connect them together so I'm building extra control circuitry to slave them together to get roughly 300Kv. :) I picked 10kv because going any higher on the output voltage, the parts required for the CW stack start to get very expensive above 20Kv per stage(I'm using 20Kv because I'm doubling the values of the diodes/caps used for safety and wear-and-tear on the parts). Radio spares has all the parts that are suggested from the web site calculator. Regarding transformer insulation, try McMaster Carr (http://www.mcmaster.com). They have all types of tape you can use to insulate between windings.

Hope this helps!
Mike Beauford
Jon Rosenstiel
Posts: 1494
Joined: Thu Jun 28, 2001 1:30 am
Real name: Jon Rosenstiel
Location: Southern California

Re: homemade switchmode HV supply

Post by Jon Rosenstiel »

Oliver-
I control the series-loaded-resonant topology switcher I built with a variac on the input. Not too elegant, but it works for me.

Jon Rosenstiel
AnGuy
Posts: 208
Joined: Tue Jul 20, 2004 7:49 pm
Real name:

Re: homemade switchmode HV supply

Post by AnGuy »

>and finally, where does one obtain polyester sheeting to insulate the transformer lairs? i can;t seem to find unmetalized and un-sticky gunked stuff anywhere (at least in UK)

Try US Plastics. You can buy clear non-metalized Mylar from them. Although you may want to consider appling a varnish or polyethene glue to encase the layers too. You can by the expense electrical varnishes (aka Dow corning 994 varnish) but polyethene works fine can can be purchased at hardware stores. This prevents the enamal insulation wearing off from vibration. The biggest problem you may run into is arcing from the sides of the windings. If you put to many windings per layer it increases the voltage differental more than narrow coils. You may have to use segmented coils in series. Its also a good idea to vacuum impregnate to fill any small voids between the windings.


Mylar
http://www.usplastic.com/catalog/produc ... 5Fid=20800
UG!
Posts: 116
Joined: Thu May 19, 2005 4:21 pm
Real name:
Contact:

Re: homemade switchmode HV supply

Post by UG! »

thanks for advice :)

i am planning on vacuum impregnating the secondary (which is why i want plane mylar sheet, rather sticky tape, which i discovered tends to seal in the inner most windings, and prevent potting compound getting to them.)

i did consider useing a VARAC, to control voltage, it would certanly work, but doesn't leave any option for computer controll, which is a bit of a shame. i may use it none the less though!

i havn;t decided exactly what design to use yet, but was thinking a center-tapped primary, with one half wound reverse to the other, so a single poalarity supply can be used and the transformer driven from 2 semiconductors rather than 4 as in a H-bridge design. the MOSFETs or IGBTs will be on the ground side of the primary, to eleviate the need for transformer coupling the gate drive, which i can never get to work properly, and will have the added advantage of easy and safe probeing with a 'scope when it doesn't work!
i also like the idea of capacitively coupling the primary, to prevent total melt-down of one of the semiconductors shorts.

Oliver
:)
DaveC
Posts: 2346
Joined: Sat Jun 30, 2001 1:13 am
Real name:

Re: homemade switchmode HV supply

Post by DaveC »

Oliver -

You use the PWM to control the DC level that goes to your switch mode HV driver. The PWM usually can work from about 10 % of full output, on up to full output. A second switcher produces square wave AC to the IGBT, MOSFETs or whatever you're using. Often people use a set of gate drivers between the switcher/driver and the IGBT gates for better control.

As Jon Rosenstiel has mentioned you can get the variable DC any way that's convenient - variac or the PWM approach.

For the transformer, Mylar is a good dielectric, but you might have some problems with it, if you intend to work the transformer very hard. I typically use Kapton tape - 3 - 5 mil thick for the interwinding insulation. It adds a bit to the overall winding size, and the windings may slip on it, if you aren't using enough winding tension. But that's all handle-able.

What ever you plan to do with the transformer when it will be working, oil bath, fluorinert or ??? a good vacuum impregnation with some insulating varnish first, will help everything stay together. What I have done, in the past, is to vacuum impregnate in a vacuum oven with the secondary in a beaker of the varnish. Pull a vacuum that can get below 100 microns, and let the part sit till the pressure bottoms out and all bubbling ceases. I usually will bring the pressure up to atmosphere and then pump down again, a couple times to help the varnish penetrate all the windings.

A good insulating varnish will be solventless, so it should easily get to 50 microns (eventually) without excessive evaporation. But anything near 100 microns will be fine. When the winding is fully impregnated, you pull it ouf of the varnish and let it drain a bit, before baking it at 100 - 150 C for about a few hrs.

Note: This is where you have to be careful about the insulation and bobbin construction. Never exceed these material's temperature ratings, but stay 10 to 20 C below, during the bakeout.

All that lower temperature will do is increase the time needed to get a full cure of the varnish...At 60 to 80 C cure time could be up to 24 hrs or even more. Most HV insulating varnishes will not harden (cure) at room temperature.. so a bakeout is necessary.

Hope this helpful.


Dave Cooper.
UG!
Posts: 116
Joined: Thu May 19, 2005 4:21 pm
Real name:
Contact:

Re: homemade switchmode HV supply

Post by UG! »

sorry for delay in replying, i have been very busy with my dissertation.

very helpfull, thanks Dave :) i suspected as much about the PWM voltage controll, unfortunatly it doubles the complexity of the whole project, i'll probably end up useing a variac.

i was planning to use something like this for vacuum impregnation:
http://www.rapidonline.com/productinfo. ... leno=61766

Oliver
AnGuy
Posts: 208
Joined: Tue Jul 20, 2004 7:49 pm
Real name:

Re: homemade switchmode HV supply

Post by AnGuy »

>but was thinking a center-tapped primary, with one half wound reverse to the other, so a single poalarity supply can be used and the transformer driven from 2 semiconductors rather than 4 as in a H-bridge design. the MOSFETs or IGBTs will be on the ground side of the primary, to eleviate the need for transformer coupling the gate drive, which i can never get to work properly, and will have the added advantage of easy and safe probeing with a 'scope when it doesn't work!
i also like the idea of capacitively coupling the primary, to prevent total melt-down of one of the semiconductors shorts.

I would recommend going with an H-bridge over the Push-Pull (Center-tap with two transistors). With Push-Pull ,your transistors need to operated in double the working voltage. (ie if your input voltage is 170 VDC, your transistors will need to be rated at a miniumum of 340V). You will also have to deal with significantly higher leakage voltage (voltage spikes) that require larger snubbing or even higher voltage rated transistors. Finally you have to deal with flux walking. If the pair of primary windings are not exactly equal, or coupling to your secondary differs, the magnetic flux will walk. This can cause a sudden current surge if your transformer core become saturated cause by flux walking.

To drive the high side transistors in a H-bridge you can use a pair of half-bridge drivers, like IR2103 or HIP4082 (full bridge), which makes your job of driving an H-bridge very easy. I would recommend looking at using a PWM SMPS driver with a current sense to prevent you from blowing an expensive set of transistors caused if your transformer core become saturated. You can lock the PWM SMPS controller if you wish to use a variac, but virtually all SMPS controllers with current sense include PWM circuits. The problem with fuses and breakers is that they are simple too slow to react to sudden current rushes if your transformer saturates. Before the breaker/fuse trips, you'll very likely blow a transistor or two. I would also include a breaker (preferably a magnetic breaker instead of a thermal breaker because they trip much faster). Breakers are excellent for safety. Another issue with breakers is be sure to use DC rated breakers. if the DC voltage is high enough, and you use an AC breaker there is a chance that it will sustain current through a steady arc. You can opt to use an AC breaker, but besure to attach it on the AC side of your input power not on the DC side. For the current sense it may be easier to work with a shunt instead of a current transformer, if you put the current transformer on the DC side, the core of the current transformer will quickly become saturated. You could put it in between your output transformer and the one side of the H-bridge, but if you run into a driver switching problem or issues with voltage spiking from current leakage, your could run into complications.

The best design approach would be to test your circuit using low input voltages (24/36/48 VDC) and a low voltage secondary. This will allow you to tweak your circuit (snubbing, frequency, PWM controll before stepping up into the higher and more difficult operating voltages. Be sure to test driving with a dummy load on your secondary as this will allow you to tweak your snubbers. If you drive with out a load there won't be any leakage current to generate spikes. Current leakage is a big problem with HV SMPS because the primary-secondary magnetic coupling is very poor do to the isolation between the two,and you'll have to deal with huge voltage spikes. I did read an article years ago about a company that designed high isolation Brick SMPS that copper plated the core to attenuate current leakage, although I've never tried it myself.

IR Half-Bridge MOSFET/IGBT transistor drivers.
https://ec.irf.com/v6/en/US/adirect/ir? ... 4294841704

Watchout for high thermal dissipation with HV MOSFETS and IGBT transistors. Pwr loss = I^2*Ron (Ron= transistor resistance) + Transistor switching time. The higher voltage rated MOSFETs have higher Ron. Its also a good idea to use TVS (availanche diodes) across the gate and source to prevent excessive gate voltages that could destroy the transistor. Also use a low ohms resistor between the Vs Pin and the center tap of the H-Bridge transistor pair and another low ohms resistor between the transistor gate the the output of the MOSFET gate driver to prevent self-oscillation.
AnGuy
Posts: 208
Joined: Tue Jul 20, 2004 7:49 pm
Real name:

Re: homemade switchmode HV supply

Post by AnGuy »

>i was planning to use something like this for vacuum impregnation:

I am not sure this will pull a hard enough vacuum to pull out the air from your windings. I used a glass jar attached to an automotive break line vacuum pump. Repeativity pumped down and backfilled until I saw no more air bubbles coming out. I don't think the plastic will create enough vacuum to pull out all of the air.

Another issue is if the additives to the polyeurthane reduce the dielectric breakdown voltage. If it doesn't then it would be ideal. If you can't find out, you might be better off using regular polyeurthane. I have a suspection that the potting fluid was never intended for HV transformer use (pure speculation on my part!)
User avatar
Doug Coulter
Posts: 1312
Joined: Sun May 27, 2007 3:18 pm
Real name: Doug Coulter
Location: Floyd, VA, USA
Contact:

Re: homemade switchmode HV supply

Post by Doug Coulter »

Anonymous guy is obviously a good, current engineer. I can take no exception to his remarks, other than they don't take into account what can be scrounged on the cheap. Here's an example -- the IGBTS and 4700uf/400v capacitor bank (from another huge broken switcher) were nearly free, the 50a 1kv bridges didn't cost much either. This design is push-pull.

See the pic of the one I'm working on, that will be a HV supply, and with different output coupling, also an induction heater for treating metals in a vacuum.

The IGBTs I scrounged cheap are 1200v/300a (continuous with a big enough heatsink, not shown here) so no worries there -- the objection to needing a high voltage device is utterly gone. I will be using IXDD414yi 14 amp parts in TO-220 as gate drivers (you can see one of them in the picture, not wired in yet). The rest, consisting mainly of an LM3524 and lv power supply, will be in a separate box.

The idea here is to simply brute force rectify the power line for starters, so all this stuff will be "hot" and in a cage for my safety. If it starts getting thermally hot, I will add copper flashing plates to the backside for heatsinking and add a fan. The LM3524 takes care of shoot through issues, and could even give me overcurrent shutdown, and will give me variable frequency and duty cycle with nothing more than a couple of (isolated knob!) pots. If I need more, I'll hook it to 220v, or if less, some stepdown transformer. The logic chips will have their own power supply of -5,+15v which is what the gate of the IGBT wants to see for quick switching.

So, this is just a big fat building block with several potential uses, depending on the output transformer.

Anonymous guy is also right about needing something very fast for protection (and maybe it should be latching and require some manual reset) -- fuses aren't fast enough for anything semiconductor, and this can be the hardest part of a job to get right. Here, I'm lucky enough to be on solar power, and my inverter auto shuts down even in mid-cycle when it thinks it is shorted at about 75a ac or so. You don't even get a spark if you short out a "suicide cord" it goes down so fast. This stuff is all so big in comparison and thermally massy I think I'll get away with "using the wires as fuses", but I do have some spare IGBTs anyway. Given that 1200*300=360,000watts...I think the IGBTs might be pretty safe running on a ~7kw inverter at a tiny fraction of their ratings.
Attachments
switcher.jpg
Why guess when you can know? Measure!
User avatar
Mike Beauford
Posts: 419
Joined: Wed Apr 25, 2007 2:24 pm
Real name:
Location: Morton Grove, IL

Re: homemade switchmode HV supply

Post by Mike Beauford »

Very nice work Doug.
Mike Beauford
Post Reply

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