On Wed, Sep 28, 2005 at 05:19:18PM -0400, Timothy Miller wrote:
> You may recall that I'd asked for help with power supplies, and some
> of you had responded with questions. Those questions didn't get
> answered as a result of me and Howard failing to communicate well.
> Anyhow, we're back on track, and tomorrow, Howard will be writing
> something up on what he's been working on with this.
>
> The chip we've been planning to use is the "MAX8578". Howard wired it
> up on copper clad and experienced total failure. Then he got a board
> blank for something that had had a similar pws before and wired it up
> on there. Things got a little better, but still don't work.
>
> It's 5V in, 3.3V out. What we see is it starting up with some narrow
> pulses (to charge the supply) that slowly get wider. When it reaches
> about 1.5V on the output, the chip registers an over-current error and
> shuts off. Some time later, it resets and tries again, with the same
> result.
I just took a quick look at the data sheet. Boy, that's an unusual
switcher circuit. I sure didn't grasp it all at first glance.
Well, one place to look is at interactions with the current limit
circuit. The R and C might be off, or there could be too much stray
inductance in series with the high-side FET, causing spikes to be added to
the current limit detector, or lead inductance on a bypass cap, for
instance. This beast probably can't tolerate leaded caps at all. Also,
where the data sheet calls for ceramic parts, it probably need them to be
SMT ceramic parts to keep the series inductance down and the self-resonant
frequency up. If it's something like voltage spikes on the current limit
circuit, you'll need a fast scope to see it. One difficulty observing fast
transients is that the inductance of the scope probe ground lead tends to
add transients to the scope input that aren't really there, due to ground
currents flowing up the probe cable's outer conductor through the scope
chassis to the power cable. One way to deal with that is to mount a BNC
connector on the ground plane with all four screws, then run a lead or a
resistive divider from the test point to the BNC inner pin. This works
particularly well if the divider presents 50 ohms to the BNC. Then use a 50
ohm cable with BNC connectors at both ends to bring the signal to the scope.
> If someone can investigate this chip, perhaps find a SPICE model,
> etc., that would be great. If that isn't enough information, we'll
> put up a JPG of our schematic. If we can't get this working, we'll
> have to investigate a more expensive chip.
Betcha it's layout effects, which wouldn't show up in a SPICE model.
Besides, one reason to choose this chip is its fast load transient response.
A conventional buck regulator, even a current-mode version, is better at
suppressing line transients, which isn't really the issue here.
>
> Thanks.
>
> _______________________________________________
> Open-graphics mailing list
> [email protected]
> http://lists.duskglow.com/mailman/listinfo/open-graphics
> List service provided by Duskglow Consulting, LLC (www.duskglow.com)
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
