Resonant power supply for voltage multiplier / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Resonant power supply for voltage multiplier

Move Thread

LAN_403

Yandersen

Fri Mar 28 2014, 04:03AM

Registered Member #6944 Joined: Fri Sept 28 2012, 04:54PM
Location: Canada
Posts: 340

Recently come to an idea of making a relatively simple sine-wave generator to feed high-frequency AC to a voltage multiplier. Here is a simplified schematic:

It is a basic LC circuit (L1-C3) with energy pumped into it every half of the period: one of the inductor's ends is connected either to the Gnd or to the Vboot. Neglecting the losses, the peak voltage will rise by Vboot volts every other oscillation cycle. So when current flows to the C3, the high side MOSFET opens connecting inductor to the Vboot - during that half of the oscillation period energy is sucked from the supply. When current ceases and reverses, the low side MOSFET opens to lock the energy inside the LC circuit until current reverses again. The direction of the current flow can be determined by the voltage of the C3 - if it rises, the current flows into it and high side MOSFET must be open; if voltage of the C3 goes down, the low side MOSFET is open. To shift the voltage control point close to the ground level for driver circuitry, the C2 is used (DZ1 clamps the voltage swing to the -0.6V...+2.5V range for safe and easy reading). C1 and resistor divider are used for peak voltage reading: driver must prevent further pumping as soon as desired voltage is achieved, otherwise uncontrolled rise of voltage may blow up the caps.

By now the circuit based on that schematic was successfully tested with 12V for Vboot and -500V...500V voltage swing of AC out was achieved. Howether, the MOSFET driver MCP14E3 was used directly to swing the inductor, so the converter was limited to a very low power. Still, running at 13KHz (due to a low frequency ferrite I had) it was able to support a voltage multiplier consisting of 20 stages of 100nF 1KV ceramic caps producing 13KV of output voltage. Right now I am ordering parts to build a full version of this circuit with addition of IRS2184 to drive two power MOSFETs. The higher frequency ferrite core will be used for 100KHz operation, 1KV of AC output voltage peak instead of 500V and, maybe, higher than 12V for Vboot if required.

Most of the parts will be ordered from digikey, howhether, choosing the right capacitor C3 is a PITA: there is only one cap I found that is able to work on 700VAC (990V peak then, right?) at 100KHz critically close to the AC voltage deration limit:

As I calculated, the capacitance I actually need is 6.8nF, so C3 will be composed of 3 of such caps. I even gone on EPCOS website and scrolled through ALL of their cap products - there is no better choice (for lower capacitances only). May anyone suggest the capacitors in 1nF...6.8nF range, able to continuously operate on 100KHz (or even more?..) sine signal with 1000V peak voltage (700Vrms)? Mica caps are perfect but unaffordable, so it will either be film cap or ceramic one. Howether, as I compared performance of a ceramic disc cap and bag-shaped wound film cap of the same capacitances, the wound film cap gave me better performance due to lower losses, I guess. Any suggestions about cap for C3 or tips to how and where to search for the HV HF caps?

johnf

Fri Mar 28 2014, 05:21AM

Registered Member #230 Joined: Tue Feb 21 2006, 08:01PM
Location: Gracefield lower Hutt
Posts: 284

Yandersen.
These caps are what all HF transmitters use ie high voltage and very high RF current capable. I have Toshiba, Murata , Philips examples here from recycling old am transmitters 4 -10kW tube type.
Some of these have ratings such as 15kV DC and 5kA RF. Ham fests etc are a good place to hunt.
Also I have got a couple of vacuum variable caps 30- 1200pF and these are rated at insane RF current and would make good trimmers for a fixed value C3.

Dr. Dark Current

Fri Mar 28 2014, 10:15AM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Russian caps type K15U (look on ebay). Those are RF power ceramic caps with extremely low dissipation factor. I have seen exactly the capacitance you are looking for - 6.8 nF rated at some 3 kV. This cap will have no problem with hundreds of kHz.

BTW, your circuit is a series resonant converter. You will need a current -> frequency control (detuning based on output current - PLL is the best solution for this) to avoid exploding your converter under no load and to allow power transfer with load.

Yandersen

Fri Mar 28 2014, 01:15PM

Registered Member #6944 Joined: Fri Sept 28 2012, 04:54PM
Location: Canada
Posts: 340

johnf wrote ...

Yandersen.
These caps are what all HF transmitters use ie high voltage and very high RF current capable. I have Toshiba, Murata , Philips examples here from recycling old am transmitters 4 -10kW tube type.
Some of these have ratings such as 15kV DC and 5kA RF. Ham fests etc are a good place to hunt.
Also I have got a couple of vacuum variable caps 30- 1200pF and these are rated at insane RF current and would make good trimmers for a fixed value C3.

Any ideas what are their series could be called like, so I could search them out? Actually, I do not need voltages higher than 1KV peak, because of a voltage multiplier limitation: the diodes in it must be double-V rated, and the affordable fast diodes have the highest Vr=2KV - those are RGP02-20E. Size of the cap does matter too - no need to make it bigger than the whole pcb, you know. :)

Dr. Dark Current wrote ...

Russian caps type K15U (look on ebay). Those are RF power ceramic caps with extremely low dissipation factor. I have seen exactly the capacitance you are looking for - 6.8 nF rated at some 3 kV. This cap will have no problem with hundreds of kHz.

Oh, well, no doubt Russian parts can do everything and more than the crippled ones we can get here, but my unfortunate preferences are to order parts available from reliable distributors around there.
Besides, I heard ceramic caps may not be the best solution due to a capacitance fluctuations. The increase in capacitance will lead to the increase in energy for the same voltage - it will increase the supply current and may result in ferrite saturation. From the other side, if capacitance will drop with the voltage applied, the power output will decrease, which is not desired either. So one of the additional requirement from C3 is a parameters' stability. Here is a good explanation about behavior of the ceramic caps, good to know:

After reading this I've got convinced not to use ceramics as C3.

Dr. Dark Current wrote ...
BTW, your circuit is a series resonant converter. You will need a current -> frequency control (detuning based on output current - PLL is the best solution for this) to avoid exploding your converter under no load and to allow power transfer with load.

Sorry, didn't get it. As far as I understand the operation of my design you called a series resonant converter, it can safely operate without a load if voltage peak value is controlled - as soon as the desired voltage achieved, the low side MOSFET locks in ON position and the energy inside LC resonator slowly decays up to some level, where pumping starts up again. Every new cycle the voltage rises by no higher than Vboot, so if it is 12V and output is 1KV, the output voltage fluctuations from one cycle to another will be no more then 12V, which is just 1%. I tested it's operation already on 12-500V, don't worry. Works fine. :)

Steve Conner

Fri Mar 28 2014, 03:05PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Yes, it already has a control loop, just not a frequency shift type. It is more like the cycle-by-cycle current limit I use on my DRSSTCs.

A converter like this will work, but it can never be quite as efficient as a step-up transformer.

Yandersen

Fri Mar 28 2014, 09:38PM

Registered Member #6944 Joined: Fri Sept 28 2012, 04:54PM
Location: Canada
Posts: 340

Of course. The only advantage of this design is that it generates sine-wave signal of predetermined amplitude - the ideal solution for voltage multiplier as a load. Fluorescent light seem to like this too, as I already tested. :)

Oh, finally I found out how are those caps are called and made - those are wound capacitors, with dielectric film and metal foil electrodes. Some are also have double-side metalized dielectric film and metal foil electrodes. Here are some examples from KEMET:

The best option here is 2.7nF, 2000DC/700VAC caps from PHE448 series - the deration of their AC voltage starts right at 100KHz operation frequency. Well, almost the same as 2.2nF caps from EPCOS B32652 series.

I also checked Panasonic's "bags" - they give a "permissible current vs frequency" charts instead of voltage derating ones. But none of the Panasonic caps are able to handle the currents my calculations predicted at 100KHz frequency. Should scroll through Vishay also. Never realized that caps made with thick metal foil as electrodes are so rare...
Oh, well, Vishay's caps seem to be the worst solution for high frequency applications.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob