(Yet another) transformer question

[Kat Manton]

Hi,

I'm currently throwing a few ideas around for generating a high voltage from a low voltage DC supply.

Not wishing to buy anything, I'm thinking about ways to use bits I've already got. Something I've got several of (after buying a job lot a while back), is output transformers. Mostly small ones, probably originally used with a single ECL82, EL84 etc.

So, the core's gapped and I'd guess they're rated around 3W.

As a rough guesstimate, would I be relatively safe driving one in reverse to produce around 250V, and could pull around 10mA (2.5W) without melting anything?

I was also thinking of driving the transformer with something approximating a square wave at about 1kHz (easy to generate), which (if my thinking is vaguely along the right lines) would mean the reservoir capacitor would be smaller than it would for 50Hz.

(Maybe I'd be better off abusing a mains transformer... I'm just wondering how well an audio o/p transformer would work... or not...)

Cheers, Kat

[M0FYA Andy]

Kat,

This question is closely related to my recent question of using a transformer at 1500Hz. The issue seemed to boil down to the turns-per-volt not being correct for the frequency of operation, even if the turns ratio is correct. The question then becomes 'what is the consequence of this?' I have ordered a book on transformer design from Amazon (Practical Transformer Design Handbook by Eric Lowdon) after seeing a copy, so hopefully will understand the physics a bit more. I also plan some experiments to see what happens, but I have too many projects and not enough time so it won't happen quickly!

Andy

[jimmc101]

The Comark 1905 insulation meter uses an output transformer in the way you propose (followed by a quadrupler to get 1kV).
Low current, but it should scale. I think mine is fitted with a Radiospares 'Economy' output transformer.

Circuit is here https://www.vintage-radio.net/forum/...5&d=1249937033

Jim

[julie_m]

You could easily modify my CFL inverter circuit here:
http://blog.earthshod.co.uk/?p=37
This is originally made for driving a single-ended transformer (as in your case, one out of the junk drawer; otherwise I'd have gone for a centre-tapped one). You may be able to get away with less-chunky transistors in the bridge as you are only getting about 3 VA out of it.

[kalee20]

I don't think you'll have much trouble driving your O/P transformers backwards. But watch the voltage (see below).

The presence of the gap will mean that the standing off-load magnetising current is higher than it need be, that's all.

Pulling 10mA should be possible with most small O/P transformers - unless designed for battery valves. But as you reckon 3W then this can be ruled out.

As to frequency - a higher frequency would be well worth investigating. It's possible that the transformer may even saturate at 50Hz, with 250V on the primary - if it was designed for a small loudspeaker there'd be no point in extending the bass response much below 80Hz or so, the turns may have been chosen according to this, and the space saved used to have beefier wire. So, if you can choose your frequency, go for a few hundred Hz. At 1kHz square wave, expect a fair bit of ringing on the edges!

[Ed_Dinning]

Hi Kat, I think the basic physics has already been answered, but beware of the ringing with square waves, if big enough it could damage the insulation.

How about a different sort of transformer, the ignition coil from a car or bike. This on relies on the L di/dt to give about 400v on the pri and then steps it up to about 25KV (old types) or 40KV+ for the modern coil packs. Some of these actually give + & - outputs at the same time (the - output not being used as it fires the plug during the exhaust stroke.

Lower inpuy volts or sine wave drive will give lower op volts.

Ed

[amp_mangler]

absolutely no reason why this cant be done, the only thing you will have to watch is your rectifier selection, some garden variety types get real cranky with square waves, try some from a old PC power supply

Joe

[Kat Manton]

Hi,

Now the slight diversion into the world of Philips signal generators seems to be drawing to a close; it is time for some more pondering...

Thanks for the comments so far; "square wave" was just a thought as it's easy to generate and "1kHz" was a figure pulled out of the aether based largely on the thought that it's an audio transformer, so it should work at frequencies higher than 50Hz.

As I'm now thinking of controlling the output voltage by altering the drive voltage, I may end up with sine wave (or a close approximation thereof) drive. PWM is all well and good but that's getting complicated; varying the drive amplitude and bunging the drive transistors on heatsinks is simpler. Power levels are low and it's unlikely to need water cooling

I'm wondering how well a few measurements relate to... well; anything, really. I've dug out a random output transformer. The primary (which would be used as the secondary) measures about 10H on my bridge and DC resistance is around 700Ω. Turns ratio is about 37.4:1, so the impedance ratio is about 1400:1

I've shoved that into LTSpice (as two coupled inductors) and I'm playing with a few circuit ideas. But I'm not sure if I know enough about the transformer for anything which works in a simulation to work in the real world. However, I have managed to produce over 500V from a doubler, regulation is terrible, but it will also deliver 5mA; which is acceptable. That's without really trying; most components were guesstimates rather than calculated, so it's promising.

(It may turn into an excessively complicated capacitor reformer, by the way. I could build a nice simple one but this is more interesting, I might learn something and it's an excuse to play with LTSpice a bit more. A nice thing about computer simulation is virtual parts cost nothing and they're harder to blow up...)

Cheers, Kat

Whats the input voltage? If it is around 12V and you don't need isolation use one of the readily available boost chips with an external switch. 20:1 boost isn't too much for one inductor. LTspice (wonderful isn't it) has all (Linears) smps chips ready to model with. Should fit in 1/2 a matchbox.

[kalee20]

I think that your simulation should model the real world fairly well.

A couple of things... winding resistances should be modelled (measure at DC, that'll be good enough) and include in your model. Leakage inductance should also be included - best way here is short-circuit the secondary (which you are using as your primary) with a short piece of thick wire, soldered on (contact resistance can really throw measurements!) and measure the primary (your secondary) inductance. Will probably be some tens of millihenries - include this in your model. If you do this, you can consider your transformer as two coupled inductors with 100% coupling, with external resistances, and an external inductor in series with your output winding.

Then, with your virtual scope, you can probe around and see where you are losing regulation.

[Kat Manton]

Hi,

I should explain; I like to set myself little challenges to keep myself amused. The rules for this one include not making any inductors/transformers (not that I'm averse), not buying anything if at all possible and not using any integrated circuits (been there, done that, too easy...)

Hence the question's about using output transformers backwards, rather than a general question about generating 500V or so at a few mA. I could do that with an SMPS chip, two mains transformers back-to-back, an HT transformer...

Cheers, Kat

"I should explain...", much more fun, how about a home made vibrator using a relay with a weight and spring on the armature? A lot of calculator solar panels in series and a 1kw lamp? Way too silly(!) Back to transformers, I have used a weeny 1.2VA mains one for interstage coupling, worked OK for audio.