3 Nov
2022
3 Nov
'22
4:26 p.m.
GCR stands for "group Coded Record"
On Thu, 3 Nov 2022, Fred Cisin via cctalk wrote:
>>> Also GCR, not MFM. NOT readable with a PC FDC.
>
> On Thu, 3 Nov 2022, Grant Taylor via cctalk wrote:
>> Please expand "GCR".
>
> Sure, . . . (GROSSLY OVER-SIMPLIFIED, such as "pulse" instead of flux
> transition)
> FM is "frequency modulated". Well, it is actually a regular clock pulse,
> with data bit pulse, or no pulse, between each of the clock pulses.
> There is, of course, a limit to how densely packed that can be on a track.
> A signal with all zero bits of the data, and a signal with all one bits of
> the data therefore are two different frequencies.
>
>
> MFM is "Modified Frequency Modulated". Clock pulses really aren't
necessary
> when they fall between two consecutive data pulses. If we leave those out,
> we end up with a much less dense pattern of pulses. (Over-simplified: MFM is
> FM without any clock pulses deemed "unnecessary") We can get away with a
> higher data transmission rate, even TWICE, and still not be much too
> overcrowded on the track. Therefore, twice as much data per track. The
> marketing people called that "DOUBLE DENSITY", and immediately started
> calling FM, "SINGLE DENSITY", although some engineers would argue that FM
was
> "half density" and MFM would be "about single density". If you do
historical
> research, you will find the term "double density" was used in the
literature
> BEFORE the term "single density" was (Just like the phrase "WORLD WAR
TWO"
> was used in newspapers before "WORLD WAR I" was ever applied to the
"great
> war")
>
>
> But, going back to FM, . . . if you look at all of the patterns of pulses,
> you'll see that not ALL of them are dense. In fact, of the 256 possible
> patterns for an 8 bit byte, you can find 32, or even 64, that are low enough
> density that they could be compressed. We can use 5 or 6 bits to represent
> those patterns. But, having only 5 or 6 bits usable to only use the specific
> patterns that were low enough density means that we can't use 8 bit bytes
> directly. but, we COULD recombine, to store 5 8 bit bytes as 8 5 bit
> patterns, or 3 bytes as 4 6 bit patterns. THAT produced low enough
"density"
> of the signal that by upping the data transfer rate, about one and a half
> times as much data scould be stored on a track, admittedly with some
> additional processing overhead. Thus, the Apple2 got about 140K on a disk,
> when the TRS80 got about under 100K (89,600). (Both were originally 35
> track, using Shugart SA400 and SA390 drives)
> "Beneath Apple DOS" has a decent description)
>
>
> The FDC of PC can only directly handle WD/IBM sector and track structure, so
> reading GCR, such as Apple (prior to 1.4M) Victor/Sirius, Commodore, etc.
> calls for different hardware.
> http://www.xenosoft.com/fmts.html has a list of a few of the different
> machines that use formats that CAN be done using the PC FDC. They do still
> have different file systems, with various sector sizes, and directory
> structures.
>
>
> --
> Grumpy Ol' Fred cisin(a)xenosoft.com