isolated DC/DC converter

The transformer is actually a DRQ127-331, which is 330 uH with about 4 uH of leakage inductance. I'm using it as a 1:1 transformer, with a bridge rectifier-capacitor-load on the secondary. It is NOT a lousy transformer!

Neither end of the inductor has a current spike because... it's an inductor!

OK, next step is to slow down the gate drives.

John

Why? I always wondered why people buy fast drivers and then place tens of ohms of gate resistors in the outputs. That decreases efficiency and increases heat.

OK, thanks for saying which one. The full transformer model and bridge rectifier should be included, along with the expected load.

Miller might help. But I think there is a much better way using resonant power transfer. The spikes and fast edges are gone, and the current drain drops to the tens of mA when idling instead of hundreds of mA. All very good for noise. I'm sure Joerg would have some excellent suggestions.

Mike

Tens of ohms? I'm using 270. [1]

That decreases efficiency and

And it radically reduces switching noise. I'm getting 75% efficiency, and a lot of the loss is in the transformer and the rectifiers.

John

Anything worth doing is worth overdoing.

- Meat Loaf

Efficiency doesn't count much if it makes too much noise. How's the current spikes and fast edges? There's not much change in LTspice.

I have given up on switching supplies in low noise applications. Resonant is definitely the way to go.

Mike

The switching waveform is nice

ftp://jjlarkin.lmi.net/DCBB_3.JPG

I'm using big gate resistors to soften up the switching, which compromises the efficiency some.

Custom magnetics and, probably, poor load regulation. And you still have peak currents somewhere.

John

Yikes! But I geuss it's small FETs that are driven.

I'd try that secondary inductor. 3uH leakage inductance isn't enough to take its job.

Oh, hey, here's an idea: If you kill more efficiency you can achieve zero noise by the time it's at 0% :-)

I was rather surprised he is still alive, considering his lifestyle when he was young.

I was thinking I could get my isolation with a heater on the 24V side, and then a stack of thermocouples to make +-10.

John

That would be a thermopile at that point, a big one at that :)

Jamie

You showed that before.

Fast edges. Current spikes. You are going to have trouble with noise.

I tried Miller. You won't like it.

Nope. I can use exactly the same transformer as you are using. TL431 regulates the drive level and controls the output voltage. Very simple circuit. Should fit in about the same space as your circuit.

No spikes or peak currents anywhere in the oscillator. The only current spikes are at bridge diode turnon. As Joerg mentioned, an inductor in series will kill them.

Incidentally, you have not talked about load regulation until now. That will be difficult with switching, so you have to put in a LDO after the bridge. That will increase the losses and kill the efficiency.

Have fun. Wait till you need to go to 14 bits:)

Mike

I wasn't planning on it. I have done this sort of thing before.

Post it.

I'm putting a heap of regulators after the isolation stage. Nine to be exact.

John

I don't need to model it. I built it, and it works.

John

On a sunny day (Thu, 10 Nov 2011 17:21:24 -0800) it happened John Larkin wrote in :

Yes, commonly done with plutonium as heater. Voyager 1 and Voyager 2 use it.

You will need radiation measurement equipment..

John Larkin a écrit :

Hmmm, right now I have a board with a 400kHz 20W buck switcher (get 24V down to 2-20V at 1A) along with a 1500 switchable gain preamp stage, followed by a 16bits 1.5MSPS ADC.

Just good PCB routing gives below 50nV input referred SMPS noise (no shielding).

Oh, and the SMPS has better than 98% efficiency and zero gate resistances.

Good PCB practice rules...

Same reason we use BJTs with hfe >=3D 200 in amplifiers with gains of 2-- for control.

Yep. Life's a trade-off.

-- Cheers, James Arthur

MHz

the

it is

nt

LEDs and PV might qualify you for green grants.(*)

(*) Or fan + windmill.

-- Cheers, James Arthur

One of my guys proposed doing a multi-channel isolated DAC by running an insulated shaft across the board, with a motor on one end and 16 little generators...

Something like that could be practical if, say, you needed a few watts of power a couple hundred KV off ground. Motor, long fiberglass shaft, generator. Stepper motors make nice little generators.

Or riffing on your idea, compressed air through a plastic tube, and a turbine.

Some people do sell laser-fiber-PV setups that will move a watt or so. Envision $$$.

John

is

Sounds like an ESD problem waiting to bite.

Going out or in (think Geithner)? You will, of course, have to give some of it back to the DNC.

How about a combo--LEDs on a tubular PV? That way you could blow air thru the tube and run a windmill too.

Brand-new factories for building tubular PVs are available, cheap. Federal loans too. Call it a "hybrid" and you're good for millions.

I used a laser last night studying cooling flow in an enclosure. It was ... illuminating. The flow was counter-intuitive--straight through cooling fins and other obstacles, with zero flow through wide- open areas. Cool.

-- Cheers, James Arthur