Dave said:
Sure. Clouds block the longwave IR from escaping to space.
Clear nights are always colder.
Concerning the Sept 2 EDN, page 55 longwinded article on digital
common noise, why in the heck didn't anyone just fix the totem pole
driver circuitry to achieve virtually zero spike current during
switching?
At Fairchild, back in the 1980s era, IBM was complaining of timing
errors created by scads of totem pole drivers switching all at the same
time (wide parallel bus, differential output - making for double the
drivers).
The spike current was in many amp region at room temperature, and with
a very stiff +5V supply, could exceed 100 amps at 85C.
Good thing the problem was moved from the digital group to the analog
group.
The pseudo long-tailed pair that was used to convert the single-ended
input to a differential signal was the first poorly designed section.
The threshold was not temperature compensated correctly, a low value
resistor was used for the common emitters, and the small gain difference
between the two transistors was nor compensated.
We made it a "perfect" long-tailed pair by using a current source for
the emitters (just like in an op-amp) and fixed the reference design as
well; and used the correct collector resistor imbalance to achieve equal
and opposite signals (in the linear region).
Having equal and opposite signals fixed the timing error in the
output.
And the drivers for the output totem pole needed some improvement,
along with some added circuitry to rapidly remove stored charge in the
base of the pull-down during turnoff.
As i vaguely remember it, that spike current was reduced to about an
amp at 125C, and almost zero at 25C.
In the original driver, spikes created during 0-1 transitions happened
at a different time than spikes created during the 1-0 transitions;
making for a *very* noisy supply.
Naturally, with 32 sets banging away, the supply had spikes that went
negative a substantial amount, and far too positive (7V can destroy a
lot of TTL devices).
Not to mention the masking / skewing of other data signals...
Too bad there is no way for me to get this feedback to the EDN author
(i found it impossible to get anywhere on the EDN site - even by
stooping to IE).
