Power Supply for X-Ray Tube

Hello Everyone,

Project I currently working is about a power supply for powering X-Ray tube. I would like to get more information about this topic and some help from you guys, if you are so kind. I apologize for my bad English.

So, my goal is to build a power supply with max. output voltage 100 - 120 kV with power of 80 W. Voltage should be adjustable (ideally from 0V but it can be from 20 kV or so), but it does not have to be regulated (it would be great anyway).

I have transformers FLYPVM400 from this site http://www.amazing1.com/transformers.html, so i designed Half-bridge driver for it.
**broken link removed**
**broken link removed**
It's using IC SG3525 with adjustable frequency and duty cycle. I also prepared feedback compensation parts, but they are not mounted yet, so values of R14, R15 and C11 are not critical.
Remote pin is for switching the the supply On/Off by external voltage.Whole IC part of the driver will be powered from SMPS Mean Well S-60-12 **broken link removed**

Power part of the supply is connected to Mains Voltage (230 V in our country) by this filter from ATX power supply.
**broken link removed**

I made Gate-Drive Transformer for switching MOSFETs (IRF730). It's wound by 3 twisted enameled copper wire (0,4 mm diameter) on white toroid core (26mm diameter, I dont know exact type of material, but its working so... It's probably from feroxxcube). 36 Turns working just fine (I tested many different core shapes, materials, winding techniques and number of Turns, but this one has best results, but not ideal).

One of Gate-Source voltage waveforms is on this picture.
**broken link removed**

For power transformer, I wound 90 turns with 0,7 mm diameter enameled copper wire. FLYPVM400 has secondary 4000 turns. I measured this voltage waveforms with HV scope probes. Yellow is primary, blue is secondary.
**broken link removed**
Here, first issue appears. Why is there sinusoidal waveform on secondary? Is it influence of parasitic capacitance between secondary turns or leakage inductance of transformer? There is also significant voltage drop on primary waveform. Rectified 230V is about 310Vdc and measured primary voltage waveform aplitude is about 280V. Why is that so?

Amplitude of sinusoidal waveform is dependent on switching frequency. (more Hz - less Voltage) so im using this for adjusting output voltage. Also rectifired output Vdc voltage is dependent on freqeuncy and independent on duty cycle. Realy weird for me =/

FLYPVM400 has rating of 20 kV on output. For getting higher voltage i'm using villard cascade multiplier with 10 stages. Capacitors are 1nF / 30 kV (http://hvstuff.com/30kv-1000pf-high-voltage-ceramic-disc-capacitor) and diodes are 20kV / 10 mA (http://hvstuff.com/10ma-20kv-high-voltage-diode-hv-rectifier-tesla-ham) Whole Multiplier is in epoxy.
**broken link removed**

Whole high voltage part (transformer, multiplier) including x-ray tube is in lead box isolated also by epoxide and filled with insulating oil. I conected whole thing like this:
**broken link removed**

Is it posible to measure current through x-ray tube like that? Is there better option?
I measured voltage with multimeter HV probe. System worked good, I also made few pictures with dental x-ray tube.
**broken link removed**
But then, multiplier crashed at 60 kV and 1mA (two multipliers crashed like that) under the load. After crash, theres only small voltage on the output of multiplier (6 kV - 9 kV).

How can I measure ripple on High DC voltage? When im using resistive divider from HV resistors and scope probe i get distorted results (realy huge ripple).

If anyone has any idea or remarks for any part of my project, please, let me know. Thank you all for reading. Posting some feedback would be great.

Those look like old CRT flyback transformers but at super high prices!

I'm not in the least suprised it flashed over, you need far more spacing between components in the multiplier and I would not recommend epoxy as a filler, it probably can't withstand such high voltage and in any case many commercial epoxy products contain or absorb moisture. The capacitors look OK but what diodes are you using? Normal silicon types will not work properly, you need 'fast' types to be able to switch at 15KHz.

Yes, you can measure current like that but I would strongly recommend you connect some sort of protection across the meter and also bypass it with a capacitor (100nF will do, it doesn't need to be high voltage rated). Incidentally, you should also have a capacitor from the output of the multiplier to ground for best results.

WARNING - sorry but I have to add this, high voltages can be lethal and so can X-rays, both are invisible until it's too late! Be VERY careful.

Brian.

"How can I measure ripple on High DC voltage? When im using resistive divider from HV resistors and scope probe i get distorted results (realy huge ripple)."

On a 10X scope probe you have seen a compensation capacitor to adjust the frequency response of the probe, I assume that you are familiar with that.

Well, on a 1000X probe the compensation is far more complex. Many years ago, I used a Tektronix 1000X probe and it had five interacting adjustments, so the compensation circuit must be quite complex.

"But then, multiplier crashed at 60 kV and 1mA (two multipliers crashed like that) under the load. After crash, theres only small voltage on the output of multiplier (6 kV - 9 kV)"

If you build anything and then encapsulate it in epoxy, any humidity present will be trapped. The assembly have to be baked at 100C for 30 minutes to remove any humidity, and immediately apply the potting compound.

Have you measured the voltage waveform directly across your primary, rather than just from the half bridge output to ground? The primary may be resonating with C7, which would explain the sine wave. Have you characterized the winding self inductances of the transformer winding?

For getting a measurement of the HVDC ripple, you can just capacitively couple to the output with a very high voltage capacitor (good luck finding a 120KV cap), and use that with a standard scope probe. Otherwise you'll need an attenuator, either capacitive or resistive. Even if you try to DIY your own attenuator you're looking at a decent amount of work and cash to do it, unless you get lucky on ebay.

Why don't you believe the results you get with your existing attenuator? Your tube alone probably is a very small capacitance, so high ripple shouldn't be surprising. And current can find a lot of ways to escape at 100KV. Did you turn off the lights and search for corona everywhere?

I suspect the high ripple is actually 50/60Hz and not at the frequency (or harmonic) of the switching frequency. If my guess is right, the very high impedance of the probe attenuator is effectively making it open circuit and susceptable to AC mains pick-up or capacitive coupling of the transformer signal. The only sure way is to do as suggested in post #4 but HV capacitors are very difficult to find and very expensive. The highest I have here are rated at 8KV although some older TVs used ones across the tube final anode and were rated at 30KV, you might find a source of those to wire in series.

Brian.

There is -or was- a h.v. cap manufacturer. Made out of glass made by a company called plastic capacitors.
Last time I used them was in early 1980s

EDIT: just googled it, the company still exists

betwixt said:

The capacitors look OK but what diodes are you using? Normal silicon types will not work properly, you need 'fast' types to be able to switch at 15KHz.

Click to expand...

I'm using these diodes https://hvstuff.com/10ma-20kv-high-voltage-diode-hv-rectifier-tesla-ham
Those should have Maximum Reverse Recovery Time 150 ns, so I think it should be Ok, but I'll have better diodes(higher voltage and current rating) and caps soon.

mtwieg said:

Have you measured the voltage waveform directly across your primary, rather than just from the half bridge output to ground? The primary may be resonating with C7, which would explain the sine wave. Have you characterized the winding self inductances of the transformer winding?

Click to expand...

Yes, I measured it directly across primary winding. I also tried to remove C7 (which is blocking capacitor to avoid flux imbalance. Or atleast I hope it is ) Result was flatten top of the square waveform, but same secondary waveform.

mtwieg said:

Why don't you believe the results you get with your existing attenuator? Your tube alone probably is a very small capacitance, so high ripple shouldn't be surprising. And current can find a lot of ways to escape at 100KV. Did you turn off the lights and search for corona everywhere?

Click to expand...

I don't belive it because i measured it also on unloaded multiplier and with different values of divider resistors with really different and weird results.

I din't search for corona, I'll do it with next multiplier which i will also dry thoroughly.

betwixt said:

I suspect the high ripple is actually 50/60Hz and not at the frequency (or harmonic) of the switching frequency. If my guess is right, the very high impedance of the probe attenuator is effectively making it open circuit and susceptable to AC mains pick-up or capacitive coupling of the transformer signal. The only sure way is to do as suggested in post #4 but HV capacitors are very difficult to find and very expensive. The highest I have here are rated at 8KV although some older TVs used ones across the tube final anode and were rated at 30KV, you might find a source of those to wire in series.

Click to expand...

Ripple is actually at exactly same frequency like the switch frequency (and is sinewave). I got few 60 kV capacitors, so I will try them.

Thank you all.