Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!sdd.hp.com!spool.mu.edu!uwm.edu!linac!att!pacbell.com!ames!haven!umd5!cogsci!wjb
From: wjb@cogsci
Newsgroups: comp.sys.ibm.pc.hardware
Subject: Re: Can I Low Level Format my MFM drive with 17 < SECTORS < 26 ?
Message-ID: <20.Mar.91.203931.33@cogsci.cog.jhu.edu>
Date: 21 Mar 91 01:39:31 GMT
References: <5832@trantor.harris-atd.com> <1991Mar16.022315.28142@digi.lonestar.org> <5855@trantor.harris-atd.com>
Organization: JHU Cognitive Science Center, Baltimore, MD
Lines: 36

[Lots of people talk about why you can/can't format more then 17 sectors with
an MFM controller, 26 with RLL etc.]

My answer is: I don't think so.

	One thing which might help, is to remember that the controller clocks
the data signal to the drive at a fixed rate (at least during formatting)
based on a crystal on the board.  This means that depending on the clock
rate of the crystal and the RPM of the drive, there are just so many bits
(magnetic flux changes) that can be written by that controller on a single
track.  Some of those bits are used by headers, trailers, and error
correction codes.  You might be able to increase the capacity of your system
by increasing the clock rate.  This would require that the rest of the
electronics on the controller and the drive could handle the faster state
changes.  I've been told that some drives have lowpass filters which would
actually filter out any signals which changed at a faster rate then that for
which they are designed.  If it did work, you could then try format the
drive with more sectors.

	Another interesting note is that each flux? time does not actually
store a data bit.  Bits are encoded and actually take up more then one flux
time.  I think the big difference between MFM and RLL is that the encoding
scheme is more efficient (i.e. has a smaller expansion ratio).  I think this
expansion is used by the hardware to help recover the clocking/phase of the
data signal when the data is being read back from the drive.  This allows for
small variations in the RPM of the drive not to affect the ability of the
controller to reread the data.  Perstor(?) sells controllers which store
even more data then MFM or RLL controllers on the same drives.  Does anybody
know what methods they use to increase capacity?  Is it just better encoding
or do they actually increase the rate of magnetic flux change?

				Bill Bogstad

P.S.  All of the above is by an intelligent laymen.  I think the general
theory is correct, but some of it is probably wrong.  Hopefully a real EE or
CE will correct me.  I became a programmer a long time ago. :-)