On 03/18/2011 09:46 PM, Bitrex wrote:


Thanks for taking the time to create this model. I was doing some
experimenting with it and comparing it to the standard bipolar 555
model, and noticed that the CMOS model seems to have significantly
greater shoot-through current than the bipolar model. I haven't used a
CMOS 555 or measured those currents "in real life", so I'm curious if
that is a real difference, or perhaps the bipolar 555 models don't model
it effectively?

Can't say about this specific family, but we had a HORRIBLE experience
last year with a recent 5 V single-gate part, in the 74LVC1G family.
We had a pair of analog comparators and a gate and FF from this family
in each channel of a circuit, packing 32 channels per 6-layer board. We
tried all SORTS of schemes to isolate the triggering of one channel from
others. Eventually, I thought to try measuring the shoot-through pulses
of the parts. I measured them as a complete channel, so it was one gate
PLUS one FF. I was able to estimate the shoot-through current at 2 A
for about 2 ns! Yikes, 2 AMPS!!! No WONDER it was crapping up the
analog section.

So, I started scanning data sheets for low equivalent switching
capacitance, and found some 3.3 V parts that showed much lower equiv
capacitance. That was the 74AUP1G family. I had to use 3 parts to get
the gating configuration I needed, and re-design the whole board to run
off 3.3 V instead of 5 V, but in the end it was a much cleaner board, as
well as fixing the inter-channel interference problem. These chips are
so quiet I can't really measure the shoot-through pulse. With the
others, I observed a substantial Voltage dip in the local decoupling caps.

I kind of think the 74LVC1G parts had some really crummy design, and
made no effort whatsoever to control the turn-on and turn-off times of
the output stage transistors, just had one driver stage feeding both P-
and N-channel devices. My guess is that the 74AUP1G parts have separate
drivers for the P- and N-channel output transistors and have some R-C
tweaking via sizing the drivers so that they can tune the turn-on and
turn-off precisely to get rid of shoot-through.

I was astounded at 2 A shoot-throughs, and can imagine this could bite
even a purely digital design, even if only in the EMI compliance area.
Man, a 2 A pulse at some clock frequency could really turn into a major
transmitter!

Jon