Received: from life.ai.mit.edu by karazm.math.UH.EDU with SMTP id AA07093
  (5.65c/IDA-1.4.4 for <glove-list@karazm.math.uh.edu>); Mon, 14 Oct 1991 22:12:01 -0500
Received: from cauda-equina (CAUDA-EQUINA.MIT.EDU) by life.ai.mit.edu (4.1/AI-4.10) id AA19676; Mon, 14 Oct 91 23:07:38 EDT
From: giszter@ai.mit.edu (Simon Giszter)
Received: by cauda-equina (4.1/AI-4.10) id AA02969; Mon, 14 Oct 91 23:07:26 EDT
Date: Mon, 14 Oct 91 23:07:26 EDT
Message-Id: <9110150307.AA02969@cauda-equina>
To: glove-list@karazm.math.uh.edu
Subject: PC Hi Res
Cc: giszter@ai.mit.edu

I played with the Hi Res mode on a 25MHz 386 tonight. I found much of the 
jitter in values disappeared when I added in a small delay in the 
readbit macro.

namely:

#define readbit(x) for(i=0;i<5;i++); rd=inportb(0x379); ...

I still couldn't push the thing beyond the 4/21 for constants N & D 
which Greg Alt found. It was actually most stable a little below that
speed (i.e. the dummy values were all stable). I didn't have obvious 
echo problems unless I shielded the glove with my hand. If your glove
is skittish it may help to add this delay.

ciao,
Simon