A new topology for building a low-cost DIY HV power supply ... 30kV/80W for $25.

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.
User avatar
Chris Bradley
Posts: 2931
Joined: Fri May 02, 2008 11:05 am
Real name:

A new topology for building a low-cost DIY HV power supply ... 30kV/80W for $25.

Post by Chris Bradley » Fri Aug 24, 2012 6:51 pm

This is the first article describing practical outcomes of my epicyclotron project, and possibly the most useful one for future fusor applications. This is the power supply I developed for the project (thread reference; viewtopic.php?f=8&t=2029#p9250 epicyclotron project #5). I've been using variations of this circuit for 3 years, and have applied for a UK patent for it.

With this topology, it is possible to use many feeble HV AC power sources, and add their collective power and voltages together into a substantial power supply.

I have a small degree of reservation in putting forward this construction because, in the normal scheme of things, the voltages and powers that amateurs have access to are usually limited by availability and cost so the progression to high volts and power is slowed and there is time to acquire the native skill and cautions for HV work (and the consequent X-ray spectrum they may generate between electrodes in a vacuum). However, as you will see, this topology means that one can buy readily available parts, brand new, and move straight on to having one's own high power, high voltage circuit for just a few $10's. All the risks of any multi 10's kV x multi-mA supply go along with this, so please excuse me to just briefly flag up the usual cautions.

The history/origin of this circuit:- Part of the construction of the circuits in my epicyclotron involve applying a pulse on top of an HVDC voltage. It dawned on me that because one can capacitively isolate *any* time varying signal with a capacitor and a diode in a charge-pump arrangement, and 'stick' it on top of an HV level, then a natural consequence of this is that you can simply keep adding HV stages on top of other HV stages providing you give each supply its own capacitive 'Greinacher' type isolation and that they share a common reference potential.

So I arrived at the circuit which is, generically, shown below. I have also attached a copy of the patent application which shows, and describes in detail and practice, other configurations of this circuit.

...
Attachments
hv_psu_pat.pdf
(138.15 KiB) Downloaded 290 times
psu_concept.jpg
psu_concept.jpg (28.66 KiB) Viewed 5730 times

User avatar
Chris Bradley
Posts: 2931
Joined: Fri May 02, 2008 11:05 am
Real name:

Re: A new topology for building a low-cost DIY HV power supply ... 30kV/80W for $25.

Post by Chris Bradley » Fri Aug 24, 2012 6:57 pm

Prototyping the circuit:

Initially, I wasn't entirely sure this circuit would live up to the job, so I ran a 4 stage device built from 4 x 4W 12V (car) neon light supplies for several months before being satisfied it was a robust design. I bought inverters from this site;

http://shop.wiltec.info/product_info.ph ... -40mm.html

Unfortunately, the price has gone up since I bought 100 at E1.25 each. However, they are able to put out a lot more power than I was expecting, so I didn't need as many as I thought and six of these is fine for a 30kV supply capable of delivering 80W at those voltages. (6 of these inverters will put out higher powers for a short time but will overheat, so for higher currents you'd need to parallel up more such supplies.)

With a nominal rating of 4W each I figured I'd be needing dozens. In fact, performance testing showed they were happy enough operating continuously up to 20V and 1A (at the same time). At 20V they are putting out ~2kV RMS unloaded. (They can be run to 'military' power to 25V and 1.25A, but anything higher they overheat and burn out.) They operate at around 65% efficiency, across a wide range of output voltages/currents. So I rate one unit can reliably deliver 13W, whilst dissipating 7W of heat, though they may not perform that well in an unventilated enclosure so I avoid running them that hard continuously.

The sum total of the electronics parts here that I paid, which are still commonly available on the internet, was 1xinverter ($1.50), 1x100pF 30kV cap ($0.45), 1x10nF 6kV cap ($0.30), 2x8kV 10mA diode ($0.25) = $2.50 per stage. 6 stages = $15. Box, terminal block, nuts and bolts, plus heat shrink were an extra $10. Total for this 30kV 80W device, built from all-new, readily available parts = $25.

Below I have attached an example of one of the 6-stage 30kV supplies I have made with these little 12V inverters. (This is one of the two PSU's shown in the photo in viewtopic.php?f=8&t=2029#p9250 .) In the demonstration shown here, the DVM is reading 1/200th of the output voltage (it is a 1M meter in series with 200M), so it is showing 20kV, which is across a gang of 100M resistors totalling 5MOhm at 4mA (80W output, about its max continuous power rating).

Clearly, the circuits I'm presenting here can be used with any transformer type. Even mixed types. Preferably, you'd want to use a high frequency type to keep cap values low. The following is the important element of implementing this type of circuit with common lighting inverters: These high frequency inverter type transformers usually have a shunt capacitor in at least one of the output lines. It is NOT possible simply to attach the linking capacitors (caps C1, C2 & C3 in the diagram) on to these outputs, because the built-in caps will take on too high a voltage when they isolate the stage they are feeding. In the case of these little inverters I've used, when I got them I was somewhat frustrated that they are fully potted! Ach! Initially that seemed to make circuit modifications difficult. Fortunately, only one output of the output transformer within it had a capacitive shunt, and the other wasn't earthed. So what I have done is drill through the case at the right point, then screw in a bolt which shorts the output [that used to have the capacitive shunt] to the case of the unit. The original shunt capacitor is now eliminated from the circuit. Then I put the high voltage 'link' capacitor onto the other line, which is therefore then directly connected to the output transformer within.

So the characteristics of the driving inverter you will need for this circuit is that one line is a shunt capacitor that you can remove from the circuit, and then you need to be able to set up an earth on one of the output transformer's lines, and add your own high voltage capacitor to the other.
Attachments
P1070928s.jpg
P1070928s.jpg (97.67 KiB) Viewed 5730 times

User avatar
Chris Bradley
Posts: 2931
Joined: Fri May 02, 2008 11:05 am
Real name:

Re: A new topology for building a low-cost DIY HV power supply ... 30kV/80W for $25.

Post by Chris Bradley » Fri Aug 24, 2012 7:00 pm

Circuit Variations:

If you have more powerful inverter units, then you can use this capacitive linking from each supply into a number of the stages, rather than just one per supply (see Figure 8 of the 'HV_psu_pat' attachment to explain). If you end up with just one supply feeding all the stages through serial caps (to increase their isolation voltage), then you end up back with a regular CW multiplier (as shown in Figure 7(ii))!! You can see how this circuit comes 'full circle' back to the conventional design.

A very useful outcome I have realised, even for a regular multiplier, is that capacitors in a CW multiplier do two distinct jobs. They don't do the same work. The caps in direct series with the output load need to be high values to smooth the load output. However, the other caps, the 'link' capacitors, can be significantly smaller than those. This is because the caps connected to the AC supply get pulled up and down in each cycle anyway, and it doesn't matter if their voltage varies more widely that the 'storage' caps in series with the load as they are only serving to follow the input voltage. It's surprising how small these 'link' caps can be – I use two serial 100pF 30kV caps to isolate my 60kV supply, and they feed 'storage' caps of 10nF, that's 200 times smaller capacitance value for the 'link' caps than the 'storage' caps yet without detriment to the output ripple!

The other particular thing to note (which you might have noticed in the previous picture already) is that with this capacitively isolated PSU design, it is also possible to capacitively isolate the ground itself! This is a bit bizarre at first sight, because it seems that the output stack is not coupled to anything, so how can it generate a potential!? I think this is a **Really Big Deal** for this invention – it provides a means to construct a supply with a fully DC isolated output. It is not 'floating', but it does mean you can freely connect ground to either end (or a middle) output. It also means you can stack these supplies in series to add up their output voltages with other such supplies – providing the link capacitors can give enough isolation. I think this works because capacitively coupling multiple nodes of the stack to the ground causes each stage to 'push' against each other. Collectively, they can then drive power through this completely capacitive decoupling. I figure that particular feature, and together with the use of multiple small supplies on one stack, is useful and patentable.

Another great thing about this topology is that power control can be achieved by simply turning each supply on or off. This is not a continuous control, but provides a degree of discrete power/voltage levels that may be sufficient control for some applications.

Incidentally, if you have doubts that such 'weedy' looking inverters can do this collectively, I can confirm that the voltage reading showing is accurate. Firstly, I can test that DVM arrangement with my calibrated 30kV high meg tester and it reads accurately (similarly, the resistor values have been similarly checked). Secondly, this thing reeks of ozone when it is operating, and the noise of coronal discharge is usual as I have not sought to particularly reduce the localised breakdown of the thin internal wires. I dare say I am losing a mA or so to coronal discharges themselves! Thirdly, it generates copious detectable x-rays across electrodes in vacuo, as expected for the measured voltage.

On that last point, please may I just note that if you use this topology and gang up a total of a few 100W worth of lower-power cold cathode inverters, which are pretty easy to get a hold of these days, then you WILL end up with a few 100W of HV power. I speak from experience in saying that you get a bit blasé about handling these lower powered units on their own, but the HV and X-ray hazards when summed like this are as real as any other high power HV supply.

If you have an interest in trying to build one of these, then the attachments will provide you with a fuller description of what's involved. Remember that your HV safety is your own concern, it's real easy to get into trouble with lots of HV power around, both direct and indirect contact and from X-rays, and this topology makes 'lots of HV power' readily accessible on a budget.

There are some weaknesses to the design, the main one being that if the stage connected to the potential furthest from ground is shorted to ground, then the capacitor jumps through the full voltage which causes the full electrical stress across that inverter's output transformer. However, this is generally OK unless you are deliberately using it to make big fat sparks directly to ground. Just avoid 'sparking' as best you can, and the occasional arcing should be tolerated. I've stress-tested the PSU shown at its 'full' nominal value (36kV), and generally the 'top' one, or two, inverters blow up and 'protect' the rest of the stack. Broken inverters are easily swapped out. Sometimes you may not even notice there are faulty inverters because the stack will still keep operating even if one, or a few, of the stages is no longer producing any power. If you have ballast in series then the circuit generally survives and protects the inverters. I've described this issue along with some further mitigations and circuit variations in the patent application documents, as attached above.

User avatar
Chris Bradley
Posts: 2931
Joined: Fri May 02, 2008 11:05 am
Real name:

Re: A new topology for building a low-cost DIY HV power supply ... 30kV/80W for $25.

Post by Chris Bradley » Fri Aug 24, 2012 8:53 pm

Just a bit more eye candy for you here, so the first pic is showing how seriously risky such a build can get!!! HV wires everywhere! This dodgy looking layout sat under my main work-bench for several months before I got around to building the enclosed modules to keep it all safe and tidy. These 4 inverters gave me a working envelope of 24kV/10mA, up to 60W or so, and performed well whilst I 'stress tested' them during regular experimental operation.

Second pic is showing the testing of two modules, six inverters each, one with 30kV of capacitive isolation in series with a second with 60kV isolation. I was testing total output to beyond 60kV here, but there were certain difficulties, such as all the LCD displays with a meter turning black so I got no readings from them, not having suitable test loads, and the copious production of ozone which became a bit unbearable quite quickly!

Third pic is a typical installation, all neat and tidy once again, during an experiment. (This is a close up of item #5 in the photo in the link at the top of this thread).
Attachments
P2220974s_psu_in_exp.jpg
P2220974s_psu_in_exp.jpg (95.27 KiB) Viewed 5730 times
hv_module_testing_1.jpg
hv_module_testing_1.jpg (99.22 KiB) Viewed 5730 times
P8150618s_prebuild_psu.jpg
P8150618s_prebuild_psu.jpg (112.44 KiB) Viewed 5730 times

User avatar
Donald McKinley
Posts: 186
Joined: Wed Nov 02, 2005 3:53 am
Real name: Donald McKinley

Re: A new topology for building a low-cost DIY HV power supply ... 30kV/80W for $25.

Post by Donald McKinley » Sun Aug 26, 2012 12:02 am

Wow Chris,

Thanks for posting. Beautiful work. I can't wait for the rest of your project details. It's very inspiring. It looks like this design could be made ultra compact.

Don M

nicolas leboucher
Posts: 14
Joined: Mon Aug 20, 2012 5:04 pm
Real name: leboucher nicolas
Location: france
Contact:

Re: A new topology for building a low-cost DIY HV power supply ... 30kV/80W for $25.

Post by nicolas leboucher » Tue Aug 28, 2012 6:33 pm

hello chris,

I ve bought some of the same inverters for my diy scanning electron microscope, in order to build the hv power supply for accelerating column. Can you please give more precise explanation for shorting the output capacitor with the screww that can be seen on your pictures ? thanks

User avatar
Chris Bradley
Posts: 2931
Joined: Fri May 02, 2008 11:05 am
Real name:

Re: A new topology for building a low-cost DIY HV power supply ... 30kV/80W for $25.

Post by Chris Bradley » Tue Aug 28, 2012 7:01 pm

Hello Nicolas,

Sure. I have some nice x-ray images of the internal parts of these inverters to show how to do this, but, sorry, can't get to them right now. Should be able to get them later in the week and I'll explain in more detail.

There is a certain hit-and-miss with getting the screw in the right place, but, usefully, there is a label on top which is very useful to guide where to put the hole! I've lost about one in 20 to getting the hole in the wrong place ... :}

One thing to note is that I got two different types from this supplier, and the potting in the second batch was a bit variable. I guess a cost-reduction design change. ~ a third had no potting above the board, ~ a third were mostly potted and the rest fully potted. A bit variable. Quality versus price, I'm afraid! But this can be dealt with easily enough because if the potting is bad you can remove the aluminium cover and solder direct to the transformer. There's usually a way where there is a will!...

When you get your inverters, please post a photo of the black potted bottom, and of the end with the white wires, of one of the inverters in your batch then I'll know which type you've been sent so I can advise you appropriately.

(PS - thanks for your intro post and the efforts you've posted on your website. Hope your projects go well.)

nicolas leboucher
Posts: 14
Joined: Mon Aug 20, 2012 5:04 pm
Real name: leboucher nicolas
Location: france
Contact:

Re: A new topology for building a low-cost DIY HV power supply ... 30kV/80W for $25.

Post by nicolas leboucher » Wed Aug 29, 2012 4:20 am

chris,

thanks for your reply and i ll send pictures of the inverters as soon as i ll get them.

I m on the way to build th HV power supply for the accelerating column of my homemade scanning electron microscope, starting from your design. I m also working on the new design for my vacuum system around an Edwards diffstak 63 diffusion pump and a couple of inficon gauges and pneumatic valves, the whole being controlled by a labjack and labview software.

Concerning my website i will had within the next days some pages about homemade cyclotrons and electrostatic accelerators, going back on my very first intend which was to give an overview in french of the more significant amateur projects in physics.
I m also inserting english translation within all the pages, realizing that the amateur particles physics enthousiasts community is mainly from US and UK.

I d like to point out the great job you ve done on your epicyclotron. Building a cyclotron was my very first intend, but i give up last year due to the lack of used electromagnets on the french market. You create an alternative way and i m really thinking of building one after my vacuum system completion.
One question : does your patent prevent anybody from using the same design without your permission ? I ve translated all the posts on your project for my own, and i also wonder if i can publish a page about your project on my site.

Thanks a lot for all the shared inforamtion !!!

User avatar
Chris Bradley
Posts: 2931
Joined: Fri May 02, 2008 11:05 am
Real name:

Re: A new topology for building a low-cost DIY HV power supply ... 30kV/80W for $25.

Post by Chris Bradley » Wed Aug 29, 2012 5:36 pm

My patent has no effect on anyone outside UK and US. I'll update that thread with a quick comment about it, just in case a reader were to want to try it.

But please let me restate to ensure I do not mislead anyone in any way; my epicyclotron is an experimental design and I have yet to figure out its exact behaviour, even for myself yet!

nicolas leboucher
Posts: 14
Joined: Mon Aug 20, 2012 5:04 pm
Real name: leboucher nicolas
Location: france
Contact:

Re: A new topology for building a low-cost DIY HV power supply ... 30kV/80W for $25.

Post by nicolas leboucher » Thu Aug 30, 2012 5:07 am

hello chris,

if think there is really no confusion about that, your project just offers an opportunity to try a new concept, without any garantee of success. the main interest in amateur physics is to build uncommon things while learning, and the most exciting thing is probably to open new research paths with limited means.

I have two practical questions:

what is the value of the resistors placed in serie with the linking capacitors? You mention a value of ten times the impedance of the inverters, but what is that value?

the inverters are powered in series with a DC power supply, what are the characteristics of the one that you are using?

Post Reply