Capacitive load with a High voltage 5kV, High Frequency 10-500kHz power supply

[ysaabe]

Hello everyone,

I'm a Biomedical Student, the team in which I'm working is currently developing biosensors for disease detection. We have an application in which we have a biosensor that requires an input voltage of 5000V AC pure sinusoidal signal with a frequency between 20 kHz - 500 kHz. The biosensor can be modeled or could be reduced as a lossy capacitive load of some picoFarads (around 5-20 pF).

Till now we have been connecting our biosensor with CCFL inverters, these can output 2000 V AC with a fixed frequency value between 20 kHz to 70 kHz and current limited to 5 mA, this has give us some success, but the limited frequency and not enough potential difference is problematic (https://www.jkllamps.com/pdfs/BXA-12579_MOD5.pdf). We require a bigger voltage and an adjustable frequency power supply.

I have been looking through internet different solutions, but my knowledge in power electronics doesn't help me , what I'm looking is an already made commercial power supplies that fit these needs, a good guide to start building one or someone that could build one for us. In few words we need :

1. Sinusoidal output signal with an adjustable voltage control from 0 to 5kVpp
2. An adjustable frequency control from 10 kHz to 500 kHz
3. Capable of working with Capacitive loads of 5 to 20 pF with Voltage drop.

Any advice would be really appreciated.
Thank you very much

 

[D.A.(Tony)Stewart]

10pF presents 30kOhm reactive impedance at 500kHz

5kV on 30kOhm , yields 167 mA reactive current and you must consider biosensor ESR for conduction losses as well as parallel leakage.

reactive power of P = V^2/Zc= 0.83 VAR= 5kV*0.167A

Thermal resistance and electrical resistance will restrict your biosensor to short wavelet bursts.

Perhaps best way to generate this is with a high voltage step-up high frequency transformer from a low impedance sinewave source. Since coupling factors tradeoff with eddy current losses, you may need to use low inductance Litz wire or multi stranded magnet wire.

When choosing a suitable toroidal core with distributed gap, use several turns of Mylar tape for high voltage insulation.

But you may be better to choose from an excellent supplier such as Murata or TDK

Is this a gas analyzer on a chip?

 

[FvM]

reactive power of P = V^2/Zc= 0.83 VAR= 5kV*0.167A

Reactive power is in fact 0.835 KVar. The calculation shows, that the said quantities (5 kV, 500 kHz, 20 pF) surely won't be observed simultaneously, not in a biosensor. Realistically, you assume that the sensor impedance will be lossy, so a considerable percentage of the reactive power will be dissipated as losses, may be 10 or 100 W or even more? It would be reasonable to calculate a set of intended sensor operation points in this regard.

Apart from the power handling problem, it will be extremely difficult to implement the AC source with the imagined large bandwidth. At first sight, the specification isn't but a kind of maximum demand, not asking for technical feasibilty.

Still demanding, but essentially simpler would be a small bandwidth resonant circuit that compensates the sensor capacitance with a respective transformer inductance. It has to be switched to different frequencies.

Or as the ultimate brute force method, a linear HV amplifier that drives the sensor without a transformer.

 

[chuckey]

More questions?-
The capacitor how lossey is it - phase angle?
Why do you need a variable frequency?
Do you need a high voltage just to get a measurable amount of current?
To generate those voltages with a linear amplifier is not for the faint hearted or for engineers with dependents.
You can use a high Q inductor to step up the voltage from a signal generator, by a factor of 500 or so, depending on your coil. So a few watts of RF will generate the voltages you say you need. The coil will need retuning with the generator, or you could power it with a very short pulse allowing the circuit to ring, so if the loaded Q is still 500, each ring will be 1/500 down from the previous, so if you fed it with a 50 KHZ PRF, it would produce a stream of pulses at its resonant frequency, only decaying by 10% before it is pumped up again. So the whole sweep frequency would be done by the tuned circuit. I have cut out a previous article about high Q coil construction. :-

Years ago I worked with Low frequency airfield beacon transmitters that were extremely limited in the size aerial they were allowed (they operated on airfields). So the dummy aerial we used was a 10 ohm resistor with a 100PF capacitor in series. The magic was that we needed large basket weaved coils to resonate this capacitor at 210 KHZ. About 4" diam and 10" long wound with 120 core litz wire with a Q of over 470. The actual transmitter would make 80 watts into a 50 ohm load, but only 10 watts into the 10 ohm resistor. Oh yes, and by the way you could draw arcs of greater then a inch from the coil!!. The 70 watts was lost in the coil.
The coils were adjusted by having a smaller rotatable coil inside of it, so its inductance and the mutual inductance added or subtracted from the main coil.
So a couple of slip rings and a motor rotated small coil, Voila, a swept frequency high voltage generator(Fmin/max = .8?)
Frank

 

[D.A.(Tony)Stewart]

Reactive power is in fact 0.835 KVar.

Thanks FVM for my oversight....

Something seems wrong with your sensitivity requirements if you need 5kV. With charge effects, partial discharge and breakdown voltages in a tiny sensor, the requirements seem to conflict.

Can you illuminate why you need such high levels? This far exceeds the safe levels of X watts/ cm^3 for humans.

Given you possibly cant answer these questions for some non-technical reason. THese will most certainly cause ionization of any plasma or at least sub-ionizing molecular permanent changes, which may be damaging.

It may be much more effective to pulse the target with a laser pulse than raise the EM energy .