"David W. Schultz" <[email protected]> writes:

> Use a relay instead to provide v_pyro.

Mechanical relay lack of reliability is part of the reason we designed
an all-solid-state system.  I guess we all have different impressions
about what's "good" based on what's burned us the most in the past?

I think the real issue you're identifying here, that we agree with, is
that it'd sure be nice to be able to detect that there's a FET that's
failed shorted, or is for some other reason "stuck on".  After talking
about it yesterday, Keith reminded me that at one point he suggested I
add one more resistor per channel so that we can detect that, but that
suggestion didn't make it into the current board revision.  If that ide
works out as expected, I'll make the change for production TeleFire boxes.

> The alarm outputs are then not controlled by a GPIO output but by the
> presence of 12V on the output of this relay. You get a warning when
> the system arms or when the relay fails.

The advantage of having the alarms controlled the way they are now is
that we can use them to start encouraging "strays" to get clear even
before there's full arming.  Our current approach has the beeper
"chirping" when the local box is armed, going to a full-on "scream"
along with the flashing strobe when the LCO fully arms that bank.

I get the point about a welded relay causing the alarm to be on as a way
of knowing something is wrong with the relay, but if our idea about how
to detect FET failures works out, we can easily have that trigger an
alarm when any output is "broken".  Which seems better in that it would
also prevent the situation where a rocket is attached to a pyro channel
and launches immediately when the box is armed by the LCO?

> The DARS launch controllers have used this idea since 2004. No relays
> have died yet.

Thanks for the pointer to the DARS design.  It's always interesting
reading about the experiences others have had!

> The gate drive (3.3V) to the output FETs is a little weak. I like the
> MIC5018 driver in situations like this to make sure that they are fully
> enhanced. Not so much for extra output current but to reduce power
> dissipation.

I've used and like various Micrel parts, but for this design, with the
FETs we're using, at 3.3V we're already well over the knee in the gate
charge curve... so driving the gate harder seems unlikely to make much
difference here.

Thanks for your interest in the system, and all the inputs.

Bdale

Attachment: signature.asc
Description: PGP signature

_______________________________________________
altusmetrum mailing list
[email protected]
http://lists.gag.com/mailman/listinfo/altusmetrum

Reply via email to