TG43 signal and the Nec 765 (Jules Richardson)

Allison ajp166 at
Tue Aug 2 06:02:27 CDT 2005

>Subject: Re: TG43 signal and the Nec 765 (Jules Richardson)
>   From: "Herbert C. Williams" <herby1620 at>
>   Date: Mon, 01 Aug 2005 01:14:58 -0700 (PDT)
>     To: cctech at
>The TG43 signal has NOTHING to do with pre-compensation.  The original 
>IBM specification for single density (128 byte per sector, 26 sector
>track, 77 tracks) floppies has an option to change the write current on

IBM spec aside that, had little to do with what was done.  

>the drive head for the more densely packed sectors on the interior of 
>the disk.  because of the constant angular velocity (360 rpm to be 
>exact) the interior tracks were more dense than the outer ones.  To 
>prevent the bits from blasting into each other they (IBM) reduced the 
>write current on the inside (greater than 43, thus TG43) of the disk.  
>This was purely a drive option.  Some later drives had more smarts and 
>could count track position on their own, and didn't need the signal.

Precomp wasn't _required_ for Single density,however double density 
it was required.  Also some vendows found that modifying the 
precompensation amount (due to bit crowding) based on TG43.

Now I'd also add that TG43s effect on write current also had a similar
effect on the bit shift that precomp would and the combined effect could
be very profound.

>Precompensation was NOT used on single density 8 inch drives.  When 

Look at what you have already stated.  Reduced write current was used
to compensate for bit crowding on inner tracks..  if that not write
compensation?  Precomp is used for the the same reason only its based
on the bit pattern.

>things changed to double density drives (MFM encoding) precompensation 
>was necessary when too many transitions were next to each other.  The 
>floppy formatter chip usually did this internally, and since it had the
>track register, it could do it "automagically".  Some did it 
>differently, but it basically shifted the bits a bit to one side or the
>other to "compensate" for bit crowding.

Actually the first generation (1791 and 765A) double density FDC chips 
did NOT do the precomp internally but instead supplied the signals needed
to do it externally using fast TTL.


More information about the cctech mailing list