Testing bus transceivers
jnc at mercury.lcs.mit.edu
Thu Nov 19 08:03:02 CST 2015
> From: Guy Sotomayor
> A number of data books have circuits that were used to measure the
> parameters .. so you might want to look at some of them
Thanks for the tip! Alas, I don't know of any data books with circuits like
that - any pointers? (And it's only the leakage current I need to check; the
other 3 key DC parameters I have already managed to check.)
> From: Mark J. Blair
> Do you have any sort of precision, regulated, adjustable lab bench
> power supply that you could use to vary input voltage?
Well, I have some Lambdas, but I'm not sure they are 'precision' (any more -
the pots are kind of dirty/flaky, so they jump when you turn them).
> If not, using a potentiometer as you mentioned would also work.
Yeah, that's what I did - running a 500 ohm pot from ground to +5V provided a
nice input, and I was able to verify the 'maximum 0 input voltage' and the
'mininum 1 input voltage' quite easily.
I also realized that in a system with 'normal' single QBUS box with
termination, one has a 330 ohm pull-up in the CPU's termination, and 180 ohm
pull-up in the BDV11 terminator, which adds up to 42 mA into the transceiver
when pulling to low voltage; if one adds another 180 ohm pull-up, that adds
another 28 mA, getting me to the required 70 mA for the 'maximum 0 output
Since the QBUS specs allow another termination, if you have a second box,
that's probably why they spec it at 70 mA, actually.
> Also, how many of these did you buy?
Umm, like 400? :-)
(Dave Bridgham and I are winding up to produce a QBUS card which uses SD
memory cards along with an FPGA and micro-controller to emulate a range of
DEC disk systems - RK11, RP11, etc - and we wanted to assure a goodly supply
of our selected transceiver chip before we spin out a PC board.)
> If it's something like a dozen, manual testing will be practical. If
> it's something like a hundred, then automating the testing might be a
> good idea, and a fun project all by itself if that sort of thing
> interests you.
Well, not really, to be honest (I'm not an analog person), plus to which I
have a zillion other projects (e.g. a bunch of UNIBUS machines to refurbish,
plus other non-computer things) which are backlogged, the last thing I need
is another one! ;-)
Also, you might be able to automate the _testing_, but one still has to plug
the chips in and out, which is a certain amount of work, so it's not like
automated testing would allow me to trivially check really large numbers of
chips. Which is why I've adopted the 'test randomly selected units for
meeting specs' approach (remember, I'm just trying to make sure these aren't
counterfeits), and for that, manual testing works fine.
More information about the cctech