Greg Neff wrote in message . ..
On Tue, 06 Apr 2004 10:40:11 -0700, Jim Stewart
wrote:

(snip)

Another issue you may want to look at is the size
of any output capacitor that you put on the unit.
I know that I didn't read the data sheet carefully
with some Power Conversions units and burned up a
couple samples before I caught the problem. Too
large of an output capacitor will lunch the driver
transistors on startup. This might have happened
to your tray of parts.





We have been using ceramic capacitors (a few uF) on the output.

I would also recommend reading the application notes listed on the
DCP02 product page:

http://focus.ti.com/docs/prod/folder...dcp021205.html

These things can generate a lot of RFI, so you have to watch this. The
transformer has some healthy ringing at around 55MHz. Using ferrite
bead and ceramic cap filters on the outputs is strongly recommended.
If you have more than one DCP02 in a box then make sure that they are
synchronized (best to use polyphase) to avoid beat frequencies, and to
avoid simultaneous current peaks on the input supply.


================================

Greg Neff
VP Engineering
*Microsym* Computers Inc.


Had very similar experience with EMI and these parts. During startup,
current spikes reach 2.5A, regardless of output loading (or the lack
thereof). If your power supply comes up slowly, the converters will
produce these spikes for the duration. Very high fields are found
near the center of the modules, and parasitic capacitance from input
to output is something like 30pF, IIRC. So, there's a need to bypass
the common mode current. For us, the most effective approach was to
put a patch of copper immediately underneath the module, unconnected
to any pins. This provides a return path for the CM currents without
injecting any into either ground plane. We also ended up using a high
capacitance MLCC (10uF Y5V) to bypass the primary side. Lytics and Ta
caps had too much ESR.
Paul Mathews