Analog Devices isoPower devices (ADuM524x)

We used transformer coupled supplies, at 60 KHz with deliberately slowed edges, to avoid problems like that:

Probably cheaper, too.

We do use their logic isolators, which seem fine, but not the self-powered ones.

John

MHz Presumably ;)

All the switching currents are within the chip (other than maybe transient input) so there may not be much you can do externally to reduce EMI. Have you tried shielding it?

Robert

Switch to alternative ?

Maybe check out the various isolators at NVE

Not opto nor RF.

D from BC British Columbia Canada.

Apparently I wasn't thinking clearly. A conductive shield across the isolation gap might not be very good for the isolation.

I take it you've implemented there suggestion about providing 100Ohms or so of impedance at 300MHz and adding caps to slow down the transition edges?

Robert

NVE

Similar technology to the AD part. I think the problem he is seeing is coming from the isolated power supply portion of the chip, I don't recall any of the NVE parts providing that. It would be rather nice not to have to add a transformer or an Isolated DC-DC but not if you can't get past the EMI requirements of course.

Robert

Some interested party, ADI or TI, published a paper showing that the NVE parts are buggy and can get confused about output states.

I think Steve's problem is with the power isolation stuff, not the logic isolator.

John

Can you show us a schematic how you tried filtering the noise problems?

Did your filters have any effect on the noise problem?

Maybe you should forgot about the integrated power converter and build external one (or use off-shelve from power-one, Synqor, Tyco, etc.). We used ADUM1200 for isolated DC/DC converters as separation of synchronisation pulses for rectification; however we also notice that sometime, during power-up, outputs doesn't reflect input state. ADuM use inductore as isolation barier. Later we replaced with capacitor isolators (ISO from TI). Performance is better and there is no more "strange" behavior during strat-up. Also MTBF is much better for ISO than for ADuM.

Janusz

I was thinkn: Some isolator + clean power = possible solution

Isn't an onboard 300Mhz converter like trying to have a quiet meeting in one room with fire crackers going off in the adjacent room.. :P

D from BC British Columbia Canada.

That's the way to do it.

I mostly use a modulation scheme, essentially a clock and then a DC restore (clamping) on the system side. That way there are no single-sourced parts in the machine.