David W. Hess (dwhess@banishedsouls.org) on 9/27/07 wrote:
---------------------------
>Jonathan Kang (johnbk@gmail.com) on 9/26/07 wrote:
>---------------------------
>>Transistors in non-saturation switch a lot faster than going back and forth between
>>active and inactive regions. This is especially true for current-steering logic.
>>The problem is you open yourself up to be much more sensitive to noise and latch-up problems.
>
>I completely agree. At the time I ran across this, ECL was not that common (if
>it ever could be considered such) and the comparator in question was not a latching
>comparator making it unusual given its speed.
>
>Why would latch-up be an additional problem? Isn't that just an aspect of circuit
>design and process? I have only had to deal with it in old op-amp and comparator
>designs as well as CMOS processes that suffer from their parasitic SCR structure.

Sorry, I meant for CMOS. I've hardly worked in any other process so my thinking automatically goes to what problems (bi)CMOS would have.

>Testing fast devices is actually more difficult from one perspective because parts
>like tunnel diodes which can be used in fast pulse generators have gotten much more
>difficult to find. The replacement in these applications is naturally the emitter coupled multivibrator.

I've only worked in high-speed digital so it may be different for RF or other types of signals, but we generally test a high-speed driver/receiver with an LFSR signal going in rather than a pulse. This, of course, means we'd have to build both a transmitter and receiver but a lot of the technology that goes into one can go into the other.