EHT Transformer

[IconicPCB]

I need to get together a 30KV 10mA power supply.
Never looked at EHT transformers, last time transformer design attempted was way back when my father was a young man.

So any experts /practitioners/ etc out there please comment on salient issues when designing such a transformer... for example inductance of the primary winding as I plan to drive it from a full bridge.
What is typical efficiency of such a beast being driven at say 25KHz or thereabouts.
Any hints will be greatly appreciated.

Ah yes the core will be akin to an old fashioned TV EHT transformer.

[Zero999]

That's 300W, which is quite a lot of power.

It's not something I would consider making. I would look into buying one. Perhaps you could find new, old stock for a CRT projector?

[Wolfram]

DC or AC? 30 kV AC is exceedingly difficult at high frequency, as these kind of AC voltages are very hard to insulate reliably. Oil immersion is by far the simplest.

If DC, I would strongly consider a voltage multiplier powered by some 4 - 6 kV RMS, which is much easier to design a transformer for. CRT LOPTs usually used a diode-split transformer operated in flyback mode, with a sectioned winding, each section with a diode in series. This reduces the AC voltage stress between each section and the core. I'd prefer the multiplier option as the number of turns (and correspondingly the resonant frequency of the transformer) can be greatly reduced.

[IconicPCB]

I am looking to be educated in the arts of EHT transformer design/ specification. I don't wish to wind or modify an existing transformer.
Basically i wish to patch up my ignorance before clashing with a designer/manufacturer over what is needed.

Therefore an idea of primary inductance range of values that is realistic and within expected range of values is probably a good starting point.
An indication of the expected volume of the transformer may be in order as well as a specification on primary voltage range.

One more point it is indeed 30kV DC.
Forgive my ignorance.

[IconicPCB]

No takers?

[Harry_22]

Hi!

Just a few photos of how the pros do it.
Here 70/80kV each shoulder, can be less, current up to 1mA.
This is for X-RAY.
Everything is in oil, including the keyboard now

[Postal2]

... Ah yes the core will be akin to an old fashioned TV EHT transformer.

A big FBT transformer will fit without modifications, but you need 3..4 of them for 300 watts. The rectifier will limit the power. Just connect them in parallel.

[TimFox]

X-ray tubes are often run from split  +/-  HV supplies to reduce insulation problems.
I like the ceramic-bobbin HV secondaries for the voltage multipliers fanning out from the transformer cores in the center:  nice layout.
What is the history of the burnt marks on the second unit's PCB?

[coppercone2]

I think the primary should be similar to a normal 300W transformer because they are all the same on the primary, unless the secondary won't fit in a given geometry.

The secondary is probably going to be wound with way too many layers, some people that do ELF radio figure out the inductance of alot of turns, its ugly


Otherwise you find the maximum turn overlap you are comfortable with and figure out how long its gonna be for 30kV. I think if you just keep stacking layers up to make it whatever convenient size it starts getting weird impedance properties

I think reasonable would be 15-20 layers of windings. Then you figure out how big it might be.

Lets say you settle for 15 layers of windings, and a wire gauge, and a similar primary to a regular transformer. all you neeed to do is figure out how big it gets from there. I think you might be able to get somewhat close to a estimate.

A kind of math optimization problem that if you solve will at least give you theoretical first order extrapolated specifications. Then you can use a transformer calculator to get extrapolated theoretical impedance. 

Your turns ratio is 136. Lets say 300 turns on the primary as a guess. Then you need 40,000 turns on the secondary.

If its 20 turns deep, then you need 2000 turns down. A clueless guess is 36AWG. That is 0.005 inches diameter. Meaning, 2000*0.005 = 10 inches . That seems kinda big. I guess they do it deeper and it gets worse. At 40 turns you get a reasonable 5 inch coil with a added diameter of 0.2 inches all around the secondary.


I don't know if any of this holds for high frequency transformer. maybe its nicer and more compact. But I think that should at least theoretically give you a 60Hz 30kV transformer


I think the # of turns on the primary for a high frequency transformer goes down. At 25Khz what could it be, like 50 turns? Then you need a measly 7000 turn secondary. But I don't think you can stack the layers so much at higher frequency? I will totally guess and say... 10 layers. Then its a 3.5 inch long secondary .

I wonder how  I did

[Postal2]

... Your turns ratio is 136. Lets say 300 turns on the primary as a guess. Then you need 40,000 turns on the secondary.  ...

I read with rapture, and now I'm waiting for the chapter about the diode bridge and capacitor.
https://www.cic-ltd.com.tw/en/product/CIC-PT_EPF-30SA_B.html

[IconicPCB]

Guys and other pronouns,thank You for the comments.

50Hz is fine and safe, however i think something in the tens of kilohertz will be a bit more compact.
I managed to cobble together a statement of what is needed and am presently awaiting a shock to the system.
I think i am likely looking down the throat of US $40+ transformer.
I hope the offer comes in soon, I need to have the costing promptly.


My expectation is primary winding  to be less than 17mH, and everything else will come out of that starting point.

Once again thank you all for your advice, i shall keep you informed as to how this progresses.

[Harry_22]

What is the history of the burnt marks on the second unit's PCB?

This is dirt in the oil. I don't know where it got sucked from, everything was hermetically sealed. Unfortunately (or fortunately) I don't do the disassembly of these units.

[Zero999]

Your turns ratio is 136. Lets say 300 turns on the primary as a guess. Then you need 40,000 turns on the secondary.

Where did you get that figure from?

The original poster hasn't said anything about the primary voltage, or the topology.

If it's just straight from the mains, then that's reasonable, although the voltage will be a little over 30kV, when the primary is 230V.

He's now talking about 10kHz, so presumably it's run off an oscillator, which would make flyback a possibility, in which case the turns ratio can be much smaller, thanks to the inductive kick-back.

[coppercone2]

yeah I did say I have no idea about switching transformers , that is a mysterious unusable coil I collect.

I imagined a sine wave drive but I guess thats not realistic. Just a fast inverter to make it simple.