Date   : Fri, 21 Oct 2005 16:12:35 +0100
From   : Richard Gellman <splodge@...>
Subject: Re: Dual Floppies On A Master

[Replying to two emails]

Jules Richardson wrote:

> Richard Gellman wrote:
>
>> Something to keep in mind, the floppy connector, unlike say, an IDE 
>> or SCSI connector, is not a bus - i.e. it doesn't have "data" lines, 
>> there's no strobe line, REQ/ACK, etc. Its a *very* low level interface, 
>
>
> as you qualified in another post, the data's essentially a serial bus, 
> with various other control lines paralleled. It's just a simple bus 
> arrangement with very little in the way of protocol (in a similar way, 
> ST506/412 interfaced hard disks on the earlier PCs and other systems 
> were much the same - it's only much later that hard disks became 
> intelligent as chip costs came down)

To Pete Turnbull: This is the point I was making in another post. The 
floppy interface is more of a "controllable" type than a "commandable" 
type (as clarified by my pointing out the procedure being shifted from 
drive (commandable) to PC (controllable)). I appreciate that neither is 
more conventional than the other, but my point was that in the modern 
world of connections, its unusual to find a device attached by ribbon 
cable that uses a "controllable" bus. As I said before, IDE and SCSI 
devices are addressable (the commandable type) - the overall point being 
that one should not expect a floppy control system to operate on 
similiar principles to that of IDE/SCSI devices - e.g. the Master/Slave 
selection (IDE) or ID bits (SCSI).

As a matter of opinion, I would disagree with your statement regarding 
PC system buses (PCI/ISA/Memory). Whilst I grant there is lot more going 
on than the average CPU bus, the basic principle behind data flow is 
that of an addressable data bus with flow control lines (e.g. data 
strobes). When I said "exactly like a chip on the system board", I 
primarily referred to device control chips rather than the system buses 
themselves, e.g. the keyboard controller. the PCI/ISA bus is (at the 
simplest explanation) an extension of the CPU bus (its more complex than 
that, but its the easiest way to visualise in terms of data flow).

>
>> (Oh, and the lil' red light on the front? Tied to the motor enable 
>> circuit).
>
>
> Usually! Some drives do very strange things...

I conceed to Pete Turnbull that this not necessarily the case. His post 
regarding Motor Enable being sent to both drives at the same time, with 
the select being used to determine the drive in question does ring true 
with my setup. I would note however that in my twin drive configuration, 
even though the light for one lights up (and works fine), always both 
drives spin at the same time.

>
>> In fact it all make me wonder why the circuit board in floppy drives 
>> is so complex. I'm sure most of it is handled by the c    omputer - 
>> surely just some current stepping would do the job?
>
>
> Hmm, I guess there's head stepper and spindle motor control circuitry, 
> head amplification, processing of signals from the write protect 
> switch, and density select switch, head load circuitry (where 
> applicable), control of signals based on write gate line, track zero 
> detection (to clean up signal from sensor), index hole detection (to 
> clean up signal from sensor), reset circuitry (so the drive doesn't 
> wipe data on power on/off; often a problem on very old machines!) etc.
>
> Not complex overall, but I suppose a handful of components for each 
> job adds up! I suppose there's enough logic (15 - 20 gates) that they 
> could justify a custom chip on later drives to do all that work.

Ah, but therein lies my point. The drives I've seen have several chips 
and ridiculous numbers of control wires - in fact, the spindle motor has 
its own board! I acknowledge that there's a lot more going at the low 
level than is covered by the interface (such as, as you say, signal 
cleanup) but surely 4 4-sided chips, 20 wires, several resistor packs, 
and two circuit boards is a little overboard? I say this of course not 
knowing a thing about what those boards are doing, and accepting that 
there very well be some very complex operation going on that just hasn't 
sprung to mind yet....

-- Richard