Neil Cherry wrote:
> I've lots of silly mistakes but I learn a long time ago that you can
> never trust communications. I learned that in my college course
> (where the experiments failed more than they succeeded - Wow I learned
> a lot in those labs). This was an on the job moment, I only screwed up
> big time there! ;-)

One time in college I was building a device to interface with the I/O bus o=
f
a minicomputer (sortof like a PIC, but slower, larger, more expensive, and
required its own room with dedicated power transformer and air conditioner)=
..
One of the signalling lines was open collector to be pulled down by the
device as part of some handshake.  The minicomputer manual specified a
surprisingly high max sink current, something like 50mA.  All I had at the
time as I was wire wrapping this late one night in the dorm was a open
collector NAND gate that could sink a lot (7438?), but not quite as much as
the worst case spec.  I'd always gotten away with "close" before, and
figured there was margin built in on both ends.

By the time I got the rest of the device built and got time on the
minicomputer (you had to sign up for a 2 hour slot days in advance) to test
it, I had forgotten about the late night shortcut.  I got the driver
working, wired it into the OS, and things actually worked fine.  Then days
later people were reporting flakiness.  As I'm sure you guess by now, I
eventually traced it back to the NAND gate not always pulling the signal
line low enough.  I had to add a whole extra chip on the board just to get
enough current sink.  Everything worked reliably after that.

There wasn't the web at the time, but I still wouldn't have written up a
page telling people how I ran into this gotcha trying to make what I did
somehow sound reasonable and it was just that Murphy got me.  What I did wa=
s
stupid, and I got caught.  In fact, I don't think I ever told anyone until
now what exactly I did to fix the flaky behavior.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.