Date   : Fri, 29 Apr 1988 16:16:07 GMT
From   : linus!encore!corbin@husc6.harvard.edu (Steve Corbin)
Subject: Help with Xerox 820-1

In article <8792@oberon.USC.EDU> mlinar@eve.usc.edu (Mitch Mlinar) writes:
>In article <7998@pur-ee.UUCP> young@pur-ee.UUCP (Mike Young) writes:
>>
>>       Xerox 820 boards had the WD1771 disk controller. Won't do double
>>density. There are those who have managed to hack the board to do DD, but
>>it involves a daughter board, doubling the CPU clock, hoping the DRAMs
>>can run that fast, etc. etc. Quite a project.
>
>Not quite true.  The Xerox 820 runs at 2.5 MHz which is more than adequate
>for 5.25" double-density.  Although it does not quite hack it for a *generic*

True.  The clock rate is fine for double density.

>8" DD disk driver, the SWP package I mentioned earlier supports 8" DSDD at
>2.5 MHz by clever programming.  The SWP package requires absolutely *no*
>modifications to the Xerox 820-I - although I do recommend a modification to

Whoa!  The 1771 does not perform MFM encoding/decoding so how can you get
double density?  Double sided is trivial but MFM requires a different
controller chip.  The 1793 is very close pin wise to the 1771 and handles
MFM but it requires extra components to set certain timing parameters.  I
believe the add-on card from Emerald Systems uses this chip.

I really would be interested in how a 1771 can be made to handle double
density.

>increase delay for the head settling time (after seek).  This allows a high
>step rate; I run all my 8" drives at 3 mS on the -I.

The step rate is not related to head settling time.  The step rate is the
time difference between step pulses sent out by the controller.  Some
floppys can take step rates down in the microsecond range (buffered seeks).
The drive specifications determine the minimum allowable step rate.

The settling time is the amount of time the head takes to stabilize after
a seek or series of seeks.  You only to need to wait for the heads to
settle after the last seek before performing the read or write operation
so changing this parameter does not cause significant changes in
performance.  Especially considering that rotational delays are around
100ms average (300rpm) and head settling times are around 15ms.

>
>-Mitch

-- 

Stephen Corbin
{ihnp4, allegra, linus} ! encore ! corbin