Yes, I have to run it at 3.3V because the other guys' logic stuff is
operating there. About 60% or so would be good, to avoid being too close
to the threshold. On my boards I'll use 74HC logic since we don't have
to break any Mbit/sec records. Personally, I am a fan of efficiency and
thus AC termination but I don't call the shots on this one. I found that
most digital guys are scared of AC termination for some reason. Probably
because of data-dependent jitter but if the C is small enough I never
had any issues.
Do you guys still run your VME stuff at 5V?
Heretic! VME is 5 volts! We use linear or switching regs on each
module to get the 3.3, 2.5, and 1.2 volts (so far) that most of the
logic needs. It's sort of nice to start at +5, because we can make the
lower voltages very clean.
One of our customers, name withheld, uses VME crates, name withheld,
whose +5 spikes up to +9 if you tease the power switch just right. The
old Xilinx 4000-series 5-volt FPGAs don't like that.
BTW, we only use part of the bus because it's a serial comms scheme plus
a dozen address lines.
What's the serial data rate? Keep in mind that most of the VME lines
are actually under-terminated (typical loaded line Z < 100 ohms,
equivalent termination 200 ohms at each end.)
I think so but not sure since it is very densely packed and I wasn't
able to bend a resistor array far enough. But on Monday I can ohm it out
somehow.
The standard VME thevenin termination is 200 ohms, one of same at each
end of the bus.
Guess I'll have to look for a L272 version or equivalent that can safely
operate down to 3V.
How about yanking the opamp and shorting the termination rail to 3.3?
That would give you something like 200 ohms to +3.3 on each end of
each bus line, perfectly usable.
We've done dozens of various VME designs, using most every conceivable
sync/async interface over lots of logic families, sold to scores of
different customers, and VME always works. Pretty good for a
25-year-old bus.
http://en.wikipedia.org/wiki/VMEbus
John