On Mar 11, 2010, at 9:28 AM, Dave McGuire wrote:
On Mar 11, 2010, at 11:19 AM, Guy Sotomayor wrote:
Passives are not an issue. The assembly house
supplies them as part of the assembly fee.
Ahh, an assembly house...another expense. I build my own boards...far cheaper. But
then I've not dealt with big enough quantities for that to become a problem; the
largest run I've done was fifty units.
For me it's not a matter of expense, it's getting it done. I'm not good
enough any more to solder these parts with the fine pin pitch and forget about trying to
do BGAs outside of an assembly house (or with sophisticated BGA rework stations with
re-ballers).
Even if they didn't, I'd pay them for it
- I don't want to deal with parts that I have a hard time seeing much less handling.
FPGAs, level shifters (or more appropriately 5v tolerant LVTTL transceivers), LVDO
regulators, etc. cost $$s no matter who you buy from (and I order from a lot of different
suppliers). I also design using readily available (ie in production) parts (the only
exception are the unibus transceivers and I go on the secondary market for those and
typically can only buy in lots of what's available...ie if somebody has 350 parts for
$7/ea, that's what I get...yes I have a lot of $'s tied up in various inventory).
Yes, I have that problem too. :-( They sure do come in handy, though.
Speaking of those 5V-tolerant transceivers, what's your favorite part these days?
For "drivers":
SN74LVC244 (2x4), SN74LVC541 (1x8) and SN74LVC162244 (4x4) or for the truely insane
SN74LVCH322244 (8x4) but the last is BGA only.
For fully registered transceivers:
SN74LVCH16543 (2x8).
I sometimes use the registered transceivers on outputs from the FPGA to keep the
simultaneous output signal switching in spec (ie stagger the switching of the output
signals by a clock cycle and then once all of the outputs are stable, clock it into the
registered transceiver).
TTFN - Guy