Any experience with Fairchild FSFA2100?

Folks,

Has anyone used the FSFA2100 primary switcher chip? We are seeing some failures upon a "dirty" input power start-up. That shouldn't happen because per datasheet it has cycle-by-cycle current limit.

-- Regards, Joerg

Joerg, Not good engineering practice to depend on "current limit" at startup with all the chaos in the system. Implement a "power on reset" at startup for maybe 100 mS to allow all nodes to settle. Good candidates for POR input are LVcc (1.0V). Or hire a consultant. Cheers, Harry

Yes -- go hire a consultant, Joerg! Whatever were you thinking designing that part in that way!

Maybe I should hire a consultant to help me with my next digital servo loop -- what do you think?

You certainly need analog help ;-) ...Jim Thompson

I _am_ a consultant :-)

Ain't my design. It's one of those cases where I was approached late in the game, unfortuntely too late for a radical design change. Personally I never use fully integrated switcher ICs with on-chip power devices for anything more than a few watts.

:-)

However, in the not too distant future, for meaning

No, time :-)

Then we could do a more discrete design and be done with it.

I _am_ a consultant :-)

Ain't my design. It's one of those cases where I was approached late in the game, unfortuntely too late for a radical design change. Personally I never use fully integrated switcher ICs with on-chip power devices for anything more than a few watts.

Joerg, That FSFA2100 has been updated to a FSFR-XS2100. Looking at the upgrade may solve your problems. Or, HAC. Cheers, Harry

:-)

Regards, Joerg

That one has a much higher Rdson and is functionally very different. Feedback is via the timing resistor input.

If we have to redesign I'd certainly not use this kind of chip anymore, then we'd go with external FETs.

What is HAC?

Harry, you newsreader doesn't quote right. Not a problem but could confuse some people.

HAC = Hire A Consultant.

(He knows you're a consultant, Harry's just having a bit of fun. :-)

Cheers, James Arthur

Maybe a diode clamp on CS....or check LVCC behavior. Not much hysterisis there.

RL

That always checked out ok. Right now it looks like we might have had some bad chips, we can't make them fail anymore.

I certainly hope it's not one of these fine examples of Chinese counterfeit engineering:

Definitely not. My client only buys at reputable distributors.