Digital circuits and analog devices

Guy Dawson guy at
Fri Dec 2 16:50:24 CST 2016

There's a Sophie Wilson talking head exhibit at the Science Museum in
London where she says the prototype ARM chip worked well but when they went
to measure it power it used they discovered that the power pin was not
connected. The ARM CPU was running on current leaking from the chips it was
connected to!

On 2 December 2016 at 19:22, Mouse <mouse at> wrote:

> >> I think the original is 'Digital circuits are made from analog
> >> parts'
> >> Another of the laws is 'There is no such thing as ground'
> This reminds me of an incident of my own which is one of my favourite
> examples of why learning about things at every level possible is
> useful.
> I built a PROM reader from discrete logic, to attach to a parallel
> port.  Basically, with 8 output and 3 input pins, you could load an
> address, then read back the ROM's content for that address,
> incremement, repeat.
> But every ROM showed 0xFF as the contents for the last byte, even in
> cases where I knew better.
> Well, the ROM socket was in one corner of the board.  Some of the
> inputs were tied to specific logic levels; I tied those to the Vcc or
> GND pins of the ROM.  The only inputs from the rest of the circuit were
> the address pins of the ROM.
> And, I tied the ROM's Vcc pin to circuit Vcc.  But it turned out I'd
> neglected to tie the ROM's GND point to circuit ground.  As soon as I
> noticed that it all made sense: as long as at least one address line
> was low, it powered the ROM through the input clamping diodes and it
> all worked (apparently one diode drop was not enough to put power
> outside the effective operating range - it may have been slower than
> speed spec, but the circuit was not running anywhere close to that).
> But, as soon as all address lines were high, there was nothing but Vcc
> coming into that corner of the board, so of course there couldn't be
> anything but Vcc coming out.  Grounding that corner's ground point made
> it all work.
> But, if I understood the ROM purely at the logic-gate level, this would
> have been mysterious and cryptic.  If I hadn't known that a low-level
> output could sink a fair bit of current with little-to-no voltage rise,
> if I hadn't known input clamping diodes existed, that would have been
> cryptic, incomprehensible magic.
> /~\ The ASCII                             Mouse
> \ / Ribbon Campaign
>  X  Against HTML                mouse at
> / \ Email!           7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B

4.4 > 5.4

More information about the cctalk mailing list