How to get 6 watt of 260 VDC from 12V?

[Artlav]

Hello.

I need to get 20mA DC at 260 V from 12-16V input, and fit it into a small box.
Question is, what is the best/most compact way of doing it, without exotic ICs or overly-complex inductors?

I have successfully gotten the voltage and power using a push-pull converter with a transformer, driven by a TL494.
Somewhat like this: http://orbides.1gb.ru/img/inv-sch-hv3.png
However, it does not fit the "small box" requirement, and i can't quite see scaling it down too much without making the transformer unwindably small.
Even the smallest i can make is good for 100W, so way overkill as well.

On the other side, there is a boost converter.
I tried one with MAX1771 ( i.e. http://orbides.1gb.ru/img/max_1771_boost.png ), and it easily fits the size requirements, but it just can't deliver the power needed, dropping out at 2-3W.

So, how can it be done?
Is there a way to make a more powerful boost converter?
Is there some other topology with in-between power level?

[PA0PBZ]

Can you use a transformer from a small laptop or wall wart smps? If you can connect it reversed and drive it with the correct frequency it should work... I think. I mean, who doesn't have a box of these?

[IanB]

You could take a look at the kind of flyback transformer configurations used by camera flash circuits? These circuits pump magnetic energy into a transformer using high current in a low voltage primary winding and then extract high voltage from the discharging magnetic field using a multi-turn secondary secondary winding. This high voltage is used to charge up a reservoir capacitor through a diode.

Flyback configurations are common when a high voltage ratio is required.

[Christopher]

Flyback ! Your "coupled inductor" (Not a transformer) turns ratio defines the output voltage. So it doesn't really matter to the primary-side what the voltage is on the secondary side, allowing you to use any bog-standard controller. I can see the low secondary side current being a problem, but the primary side will have a high-current low voltage so you can use a bog standard fet.

if that sounds too complicated (who wants to wind their own tiny transformer? Planar would be easier possibly!), How about a Boost/SEPIC with the LM3478/3488?

[Artlav]

Can you use a transformer from a small laptop or wall wart smps?

They are about the same size than what i can wind up myself, often larger.
So, no point.

You could take a look at the kind of flyback transformer configurations used by camera flash circuits?

Camera flash circuits are wimpy, nowhere near the current i need.

Flyback do sound interesting, however.
I've only heard of them for really high voltage generation.

What would it look like in my case?
Is it as straightforward as winding a small transformer, and driving it's primary like a boost converter?
Basically, getting away with smaller turn ratio, and thus smaller transformer?

if that sounds too complicated (who wants to wind their own tiny transformer? Planar would be easier possibly!), How about a Boost/SEPIC with the LM3478/3488?

What advantage does it have over LM1771?
Would SEPIC by itself be more powerful than a regular boost?

[NiHaoMike]

What about a current fed push pull converter, as often used for CCFL backlights? (Note that you'll want a different turns ratio to get what you're looking for.)

[Artlav]

Did a quick test on what i think a flyback is - replaced the inductor in the boost converter shown above with a ring wound for the same inductance plus a secondary winding of 5 times the turns.
The diode was connected to the secondary instead of the FET.

The performance is about the same as simply an inductor, but the transistor heats up a lot.

Obviously, i don't know how to cook this topology properly...

[Artlav]

What about a current fed push pull converter, as often used for CCFL backlights?

Could you elaborate?
Google for "current fed push pull converter" does not show anything specific, and for "CCFL inverter" - nothing new.

[NiHaoMike]

Did a quick test on what i think a flyback is - replaced the inductor in the boost converter shown above with a ring wound for the same inductance plus a secondary winding of 5 times the turns.
The diode was connected to the secondary instead of the FET.

The performance is about the same as simply an inductor, but the transistor heats up a lot.

Obviously, i don't know how to cook this topology properly...

The leakage inductance must be very low for a flyback topology to work properly. Also, for your use, an autotransformer boost makes more sense. Even better might be a push pull which makes better use of the core.
http://www.linear.com/docs/4154
For your application, the secondary winding should have fewer turns in order to get more current out of the same size core and rectified to get DC. Then make it regulated based on voltage instead of current.

[T3sl4co1l]

A beefed up version of this would do;
http://seventransistorlabs.com/Images/Discrete_Tube_Supply.png

Tim

[Richard Head]

Flyback, flyback, flyback.

It's simple to design, has no control loop problems and can use a multitude of control ic's.
The dreaded leakage inductance can be significantly reduced by interleaving the primary and secondary windings.
Lp = v^2.t^2/2E. You can use an RM5 at that power level. Run the converter at 50Khz. Set peak flux density to 280mT and use current mode.

Dick

[Psi]

I need to get 20mA DC at 260 V from 12-16V input, and fit it into a small box.
Question is, what is the best/most compact way of doing it, without exotic ICs or overly-complex inductors?

If you have money to spend you could buy four little ~$14 isolated DC-DC modules.
You can get a 12V input version which has DC output taps for 24V@136mA OR 48V@63mA OR 72V@42mA.

4 of those in series (with one using the 48v tap) would get you 264VDC @ 42mA in a small size.

The output is labeled as negative but that doesn't matter, just connect it up with -264V as the ground and 0V as positive. They're isolated modules so there's no issue.

The modules are 21.8mm x 7.6mm x 11.2mm

http://www.digikey.com/product-search/en/power-supplies-board-mount/dc-dc-converters/4325599?k=1272sC

Note: I edited this post because i discovered the output was much higher when other outputs are not used.
It can do 42mA @ 72V so they would provide your 20mA no problem!

[Artlav]

Flyback, flyback, flyback.

Ok, did the math.

24mm diameter ring with 140 of permeability, 30 turns primary, 80 turns secondary.
Should carry the current, and need only 200V on the FET.

In practice, however, it's way less than yesterday's cobbled-together attempt.
I'm getting barely any output, the ring gets hot (!), the transistor gets hot, the dummy load stays cold.

What am i missing?

You can get a 12V input version which has DC output taps for 24V@136mA OR 48V@63mA OR 72V@42mA.

Nice... And wasteful.
Just feels wrong, like duct taping a wheel to the car instead of using bolts.

[Artlav]

Pop quiz - Is ferrite conductive? It sure is not.
And permalloy that my ring is made of? Turns out it is!
That was the whole problem - the ring itself is quite conductive, and have enough sharp edges to scrap off insulation on the windings, shorting the inductor all over the place.

A wrapping of tape and a rewind later, all works like a charm.
Nothing gets hot anywhere, and a quick check shows about 180V across the FET.
Looks like i'm in love.

Thanks all.

I ran your input paremeters through an online calculator. I targeted complete energy transfer to keep the turns to a minimum. The power is low and primary peak current only hits a little over 2A; d<0.45. You should be able to use a 55-80V fet. Use an RCD snubber to 0.75-0.8 of Vds of your fet.

Well, between winding 200 more turns on a small ring and using a 200V FET i choose the 200V FET.
At this power level it's excess of on resistance does not matter enough.

[Richard Head]

24mm diameter ring with 140 of permeability, 30 turns primary, 80 turns secondary.

Why such a big core? Perhaps RM5 is a little small but RM8 will easily handle it.
I used to manufacture a DC-DC conveter with a 600VDC output at 10W and that used an RM8 at 50khz, with room to spare in the bobbin.
Also, when the number of secondary turns gets high you will have a parallel resonance problem in the secondary due to inter-winding capacitance.

Dick

[akis]

Hello.

I need to get 20mA DC at 260 V from 12-16V input, and fit it into a small box.
Question is, what is the best/most compact way of doing it, without exotic ICs or overly-complex inductors?

I have successfully gotten the voltage and power using a push-pull converter with a transformer, driven by a TL494.
Somewhat like this: http://orbides.1gb.ru/img/inv-sch-hv3.png
However, it does not fit the "small box" requirement, and i can't quite see scaling it down too much without making the transformer unwindably small.
Even the smallest i can make is good for 100W, so way overkill as well.

On the other side, there is a boost converter.
I tried one with MAX1771 ( i.e. http://orbides.1gb.ru/img/max_1771_boost.png ), and it easily fits the size requirements, but it just can't deliver the power needed, dropping out at 2-3W.

So, how can it be done?
Is there a way to make a more powerful boost converter?
Is there some other topology with in-between power level?

How small is small ? What efficiency do you require? For example with 12V input you could use a simple oscillator at 5V peak (say a wein bridge op-amp) -> buffer (say a push pull complimentary pair) -> transformer at 1:40 (say an RM12 hand wound) -> rectifier ----- that would give you 250V-280V at the output of a 10K resistor which is what you need and you'd spend about 3W on the two buffer transistors (say MJE15028/MJE15029) and a little bit on the transformer and rectifiers. Assuming you go up to 200KHz the transformer could be a RM10 or similar. I have already got one of these designs working at 200KHz, but I use a 24V supply and the transformer is an RM12 with a 1:6 ratio - however with very little changes it could do what you want.

Here is a diagram of a simulation I just did

[Richard Head]

AcHmed

If inter-winding capacitance on the secondary side becomes excessive the sinusoidal reactive secondary current is reflected through to the primary side and your current ramp is superimposed on a sinusoid. Very irritating and it totally messes up your control. The only solutions I found was to allow as high a flyback voltage as tolerable across the power switch to reduce the turns ratio required and space the turns out. Also, like you suggest, break the secondary into several layers and insulate well between layer (to reduce capacitance between layers).

I wonder if one could use a seperate winding to pick up a sample of the sinusoid and mix it in anti-phase with the secondary to neutralise the irritating phenomenon?

Dick

[T3sl4co1l]

For a flyback converter of this size, I would say a well-coupled winding with up to 300 turns will have little trouble making correct waveforms (tolerable overshoot, ringing).  To achieve that coupling, probably a P-S-P winding layup would be needed.  R+C filtering and snubbing can be used to reduce the ringing.

I'd probably run it a bit over 200kHz because I'm cool like that, but you can be slow and do it at 50-100kHz.

I would say resonant (and quasi-) transformers and methods only become real important in the kV range.

Tim

[Richard Head]

T3sl4co1l

I agree completely.

Dick