SMPS and EMI

The output side rectifiers also matter to the radiated noise.

It is only a tiny fraction of the power being handled by the DC-DC converter so very inefficient methods of making RF still matter.

When the MOSFETs first turn on, they pass a current spike as they discharge the stray capacitance. This makes a very high di/dt.

If the inductor rings at the turn on, the substrate diode of the MOSFET gets some charges in it. These can make a little step as they go away.

The current in the output rectifiers starts very suddenly making a large di/dt.

If the output is not schottky, the recovery of the diodes makes a little pulse of current flow backwards. This can be a huge di/dt

Purchased DC-DC converters are almost always a source of EMI even when they are name brand. Isolated ones are generally worse than non-isolated but there are a lot of exceptions.

10-20nsec switch times and harmonics thereof, usually. And as MooseFet wrote, diodes help generate some more of those harmonics.

Good ones are expensive. You could ask these guys but insist on being sent the EMC plots:

My experience is that you must filter. Ferrite beads on just about every line that enters or leaves the supply.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

And is the case plastic or metal?

--
Dirk

http://www.transcendence.me.uk/ - Transcendence UK
http://www.blogtalkradio.com/onetribe - Occult Talk Show

I think you might be better off with a converter that has all input hardware physically integrated into it, perhaps with the exception of an on/off switch. This would be marketed with pre-compliance. Something like an SWS75 from Nemic-Lambda is fairly representative of closed frame. GLC75 from SL-Power/Condor is open frame. These are available through standard distribution. Must be a million clones out there.

Bricks are for those who understand more fully the responsibilities of this integration and plan for the thermal and emc budget.

RL

Not unusual. Even the ones that have a page full of certifications only pass in one application, usually full rated output into a purely resistive load. ...hardly an interesting product. ;-)

Sounds possible. Shottkys can switch pretty fast. Inductors can make things go fast, too. Also note the "Existance Theroem". ;-)

Also think about fast transient tests and conducted emissions. They'll bite too. Another thing to watch is long term commitments. It's a PITA to go back through the cert process because the silly vendor dropped the line for this year's, complete with tail fins and a "T" roof.

We had a similar problem - cheap Chinese PSU, just about meets the emissions regs with a resistive load connected directly to the output plug. Add a few more feet of system cabling -> more efficient radiator

-> test fail.

Not surprisingly the PSU vendor weren't interested in fixing it - volumes were similar to yours, a few hundred a year. I suspect the root cause was reverse recovery spikes from the output diodes but I didn't dig too deeply, just found an alternative PSU. A lot of the 'name brand' PSUs just re-package the same PCB so if the dimensions are exactly the same as the one you are using, beware!

We now use Elpac, now part of

They have several units that meet your spec, check them out. The radiated EMI with their unit was unmeasurable.

HTH

y

.com/onetribe- Occult Talk Show

Good question. First attempt to pass EMI was with regular plastic case - failed almost everywhere (the box is size of small microwave oven with bunch of wires). I sprayed it inside with MGChemicals nickel coating (841-340g), sealed joints, grounded at USB connector, sprinkled with holy water, the works.... The instrument passes when lab PS is used, fails when no-name brick is used. The power brick comes with molded ferrite bead (I assume there is a ferrite inside of that hump). I can "silence" the brick by adding ~1lb worth of ferrite beads, I am not sure many people will buy it... :o(

The instrument has power connector on the back for external PS - this is what I am given. There is some pressure to keep the cost down (surprise!).