IDE <-> MFM

[Tony Duell]

> 
> what I had been thinking was shouldn't it be somewhat *easy* to make 
> an IDE drive into an MFM. Just strip off what you don't need! How do you 
> make an elephant...take a piece of stone and remove everything that 
> don't look like an elephant! And it's done!

Well, it would be _if_ a modern IDE drive still consisted of a WD1003 + 
a simple nead drive. The problem is that it doesn't. All the drive has to 
do is look to saftware as if it was a WD1003 + a simple hard drive. The 
software interface assumes things like the same number of sectors/track, 
etc. 

But the hardware might well not do that at all. Provided the drive can 
accept a cylinder/head/sector number and return a buffer full of bytes, 
the software will be happy. The fact that sectors that (to softwre) 
appear to be on th same culinder in reality aren't doesn't matter. But it 
sure does matter if you want to bypass the built-in controller.

There is then the issue that the signals you want to access may only 
exist inside (cusatom) ICs. Not the easist thing to connect to.

And the fact that you won't get real technical information on an iDE 
drive. That's the only part of my PC that I don't have scheamtics for. I 
tried just aout every manufacturer about 10 years ago, none would supply 
them, I douvt that's changed. And the ICs on the sort of drive I was 
looking at seemd to consist of a (mask-programmed) microcontroller, an 
SRAM (presumably a data buffer at least in part), a little glue, some 
recognisable analogue bits, at least one large digital ASIC and at least 
one analogue ASIC. Not conducive to being modified.

-tony