R: R: Floppy disk: one drive per face

[Fred Cisin]

On Fri, 6 Nov 2020, Enrico email.it via cctalk wrote:
> First of all I would like to thank you all for every suggestion that comes
> from so much experience. I only know my Ferguson bigboard 1 pretty well
> The z80 based machine was called GENERAL PROCESSOR Model T year 1981.
> You can see it here: https://www.vintagesbc.it/?page_id=877 and the cards it
> was equipped with here: https://www.vintagesbc.it/?page_id=879 even if it is
> old information because I, together with a competent hardware technician and
> another enthusiast who has already written an emulator HERE:
> https://nippur72.github.io/gpmodelt-emu/ (you have to press CR to start then
> BD and enter ...)
> and the emulator video here:
> https://www.youtube.com/watch?v=cRmmUYkaDxs&feature=youtu.be we are
> proceeding to derive electrical schematics, save firmware, software, and try
> to reconstruct and document everything.
> You have to think that they used a WD1791 which has an inverted data bus and
> in the firmware they use CPL instruction to invert it again so on the disk
> the CP / M  (CCP + BDOS + bios) is written all INVERTed.
> Based on the firmware on eprom configuration it could manage:
> - 2 drives 8 "(128x26x77) or 5" (128x18 (maybe 16) x40) all in FM and seen
> as A, b, c, d
> - 1 8 "DSDD single B unit and 1 HD 10MB Winchester drive (divided into 2
> halves A and C)
> At the beginning the engineers made ibm 3740 in side A. Only after they
> passed to make a drive double face so we think they didn't change things
> also because it seems to be possible to boot from both side A and side B and
> we expect the side B of drive A it would in IBM 3740 format.
> We noticed that the sectors always start from ZERO and not from 1 so it is
> interesting what you say: “you really should try unplugging the heads of the
> drive and swapping them. That will put "side A / 0" on the second / B side
> of the disk.! "

You did an excellent job of answering the questions!

The NAME probably doesn't matter, but it's always good to include, in case 
anybody here might possibly have previously dealt with it.

With a WD179x, it is a little surprising that they didn't use MFM ("double 
density").

On the other hand, 8" SSSD is "THE STANDARD", so must be supported, and/or 
if earlier models had used the WD1771 FDC, this could be a legacy format. 
The extreme standardization of CP/M 8" SSSD is another reason to justify 
the replication onto second side, instead of changing the format.  I once 
pleaded with Gary Kildall to officially declare a 5.25" secondary disk 
format standard.  His reply was, "THE STANDARD is 8 inch Single Sided 
Sigle Density."; it was understandable that he would not tolerate creation 
of a "secondary standard".   Although, we ended up with about 2500 
different microcomputer disk formats, due to the lack of standards for 8" 
DSSD, 8" SSDD, 8" DSDD, 5.25" SSSD, 5.25" DSSD, 5.25" SSDD, 5.25" DSDD.
(MOST of which could have been prevented with an official proclamation 
from DRI.)


Swapping the head cables might be a simple way around it, IFF the second 
side is formatted exactly the same as the first.  One obvious possible 
difference is the head number field in the sector headers.  If EXACTLY the 
same, then both would have 0 in that field.  BUT, it could have 1 on the 
second side (which will make it easier to read with an NEC chip).  And, 
the WD179x controllers didn't really care much about that field, so it 
could be ANYTHING, including wildly wrong numbers.  In most cases, the WD 
chip wouldn't even notice if that field had the wrong value, but the NEC 
chip needs to know what value to expect and look for.

If the imaging that included the second side was done successfully with 
IMD, then that implies a 1 in the head number field on the second side.
Not insurmountable, but it means that just swapping the heads will still 
require some special software.
If the imaging that included the second side was done successfully with 
IMD, then consider the possibility of either doing the file transfers from 
those images, or writing some minor software to split those images into 
two separate single sided images.

The inverted data isn't a serious problem, although it is inconvenient 
when looking at the data to see whether it is successful.  But, it is easy 
to work around.

Another possible complication, if you are trying to do the transfer on a 
PC is that the WD179x controllers could start the beginning of a track 
sooner after the index pulse than the PC NEC765 controllers can handle. 
That is usually not too hard to work around, but is something to watch 
for, particularly if you get failures to read the first sector of each 
track.

Not all NEC floppy controllers can do FM and/or 128 byte sectors.  Dave 
Dunfield has created some test routines for checking that.


--
Grumpy Ol' Fred     		cisin at xenosoft.com