> > I have one of their units; I haven't used it much since it only comes with
> > 1Kx24 of storage and additional storage (added via FIFOs) is rather pricey.
>
> They're quoting $39 for an upgrade to 8Kx24 and $89 to 32Kx24. Doesn't
> sound too pricey to me. Maybe this is new pricing.
>
> Other than the buffer limitation, does the unit live up to its claims?

Those prices don't sound too bad; when I'd looked at the parts cost for
upgrading the prices were rather more than that.

Otherwise, I really haven't used the unit all that much and they may have
improved the design since I got it, but here are some restrictions which
you might want to know about:

[1] The "pattern trigger" has to look for a specific value on the first 8
    inputs; you can't set "don't care" bits within those 8 bits.  If you
    don't want to use all 8 bits for pattern matching, you have to make
    sure the bits not used for that purpose are in a known state when the
    unit triggers.

[2] There is no "back-triggering"; logging will start 1-2us after the
    trigger word is detected.  In the application in which I'd intended
    to use this device, I was interested in what had transpired just
    before the system crashed, but there was no way to have the unit
    record "UNTIL" the trigger condition and then stop recording.

[3] There was not, at least when I got the unit, any convenient means of
    watching the input pins real-time [e.g. while manipulating the CPU,
    watch to see what happens].

[4] The unit, at least when I got it, does not use any form of data comp-
    ression when sending the recorded data to the PC.  At the 9600-baud
    data rate used by the device, a 32Kx24 stocked board would take over
    a minute to retrieve a capture [at least I would expect it too--I
    haven't expanded mine].

[5] The pods for the unit are all hard-wired to the circuit board and are
    the largish-size clips; while such clips are nice for fastening one or
    two wires to a chip they can be unmanageable when dealing with too
    much more than that.  A plug for a ribbon cable/header connector would
    have made things much more convenient.

[6] The inputs are connected directly to the FIFO chips; having burned my
    finger on a FIFO chip which went into major-power-sink mode in another
    project for no clearly visible reason (probably input glitch) this
    concerns me greatly, especially at the $20+ cost per chip.

[7] At least on my unit there does not appear to be any external clock input
    option.  For looking in detail at, e.g., bus activity it would have been
    very useful to have been able to look at the bus state on consecutive
    cycles.  Otherwise, it's necessary to set the sample rate to twice the
    bus rate and sample the bus clock along with the bus data.

Maybe I'm being overly negative about this device; I do know that it fell
substantially short of my expectations and I personally regard the money
I spent on it as $200 down the drain.  If anyone else can manage to make
good use of it, I'm happy for them but unfortunately it is not suited to
my needs.

Were I designing a logic analyzer-style device, I would have used three
74HC574's on the inputs and three 22V10's behind them; the 22v10's would
indicate whether the input data had changed and if so, trigger a store
to memory.  Along with the latched data I'd store the output of an 8-bit
synchronous counter to time-stamp it.  If the system were running at 50MHz,
a single x8 memory array would allow 12.5 million transitions per second to
be logged [but with 20ns positional accuracy]; using two x8 memory chips
and sending bits 0-15 to one and 16-23 and timestamp to the other would
allow 25 million transitions per second to be logged [again, 20ns positional
accuracy].  If the inputs to the device were changing at less than 200Khz,
the system would log "nothing happened" records; a single 32Kx8 chip would
fill up in about 50ms; two chips would last about 100ms.  If this is an
acceptable capture window, then there is no need to vary the 50MHz "sample"
clock.

Personally, this is what I'd been hoping to get from ProBoard circuits, as
I'd sketched out this plan (and had been contemplating building it) when I
saw the ProBoard circuits ad.  Perhaps I'll build byself such a device
someday.