[time-nuts] PicoPSU power supplies

[John Ackermann N8UR]

Sorry for the cross-post, but I know this is of interest to some folks 
on the HPSDR (high performance software defined radio) list, and I 
suspect it may be helpful for some time-nuts as well.

The important message first: the little "PicoPSU" switching power 
supplies that plug directly into an ATX socket say they are for 12V 
input.  It turns out that they mean it -- the supply will shut down if 
the input voltage is more than 13 - 13.5 volts.  So using these in a 
typical battery-powered environment is potentially a problem unless you 
somehow drop a volt or so.

Here's the sad story...

I recently built several low power Atom-based computers that use the 
PicoPSU supplies.  The theory was that I'd run these machines, along 
with all the other 12V gear in the basement, off a couple of big 
batteries on a float charger.  The battery voltage varies from about 
13.5 to 14.2 depending on the charger state.

Built the machine on the bench using an Astron supply, and everything 
was cool.  Put on the battery bus in the lab, and things got very funky. 
  Sometimes the power LED would come on, but nothing more would happen. 
  Take the machine to the bench, everything was fine.  Take it the lab, 
didn't work.

I checked all the power connections and found no problems, but the 
battery supply simply would not run the computer.

Finally, I found a manual for the PicoPSU on the web, and discovered 
that when they say it's a 12V supply, they mean it -- there is an 
overvoltage shutdown that's supposed to trip between 13.0 and 13.5 
volts.  In practice, it seems that at least some of them are good for a 
little bit more than that, because the 13.8V from my bench supply worked 
OK.  But when the battery voltage hit about 14.1, that was too much and 
the supply shut down.

So, I now have the computers running from an Astron supply that I 
adjusted for good measure down to 12.6 volts, and all is well.  For the 
long term, I'm considering putting a diode or two in series with each 
computer's power lead to drop the voltage that way.

John