-=> Mike Ross said to George White
 -=> about "BAG OF CHIPS" on 02-07-04  11:52.....

 MR> I thought comparators already had a little bit of feedback internally,
 MR> not sure though? Otherwise both why create a special purpose device
 MR> called a comparator ic and why wouldn't one simply use an opamp
 MR> instead?

1. Comparators don't normally have any internal feedback. If they
   had some positive feedback, the positive and negative trigger
   levels would be different and they would be called Schmitt
   Trigger Buffers.

2. Comparators chips don't usually have any internal compensation
   to make them stable when they are working in the linear part
   of their transfer characteristic (non-saturated output). This
   makes them a lot faster than most op-amps, but liable to some
   oscillation as the input goes through the transition range.

The input transition range is the range of input voltage for
which the output is non-saturated. It is roughly equal to the
output voltage swing divided by the gain. My data sheets do not
give a figure for the gain of the 399, merely stating that the
amp is "high gain".

If the gain in the linear region is as low as 1000 and the output
swing is from +5 to -5v = 10v, the input transistion range is
only 10mV. If the gain was higher -- which it almost certainly is
in the case of the 339 -- this range is correspondingly smaller.

In this case we are driving it from unipolar logic with fast rise
and fall times. Typical output rise and fall times for CMOS logic
on a 5v rail is 200nS.

So the time spent in a transistion range that is only 10mV wide
near the middle of the logic level swing would be less than 1nS,
and positive feedback would not be necessary to stabilize the
comparator during such a short time.

   ,-./\
  /     \ From Greg Mayman, in beautiful Adelaide, South Australia
  \_,-*_/    "Queen City of The South"    34:55 S  138:36 E
       v

... DAMN IT, I GOTTA GET OUTTA HERE!
___ Blue Wave/386 v2.30

--- FLAME v2.0/b