Will The Grand Master Of Disk Controllers step foreward?

From: Richard Erlacher <edick_at_idcomm.com>
Date: Fri Apr 16 10:17:08 1999

Oddly enough, though it hasn't been my stock-in-trade for about 15 years, I
still have the manual for the original WD-1000 controller. I may, in fact,
even have one of the boards around somewhere. I certainly have a couple of
the ones used in the TVI TS-806-20's I have sitting in the driveway (under
4" of snow, at the moment).

Dick

-----Original Message-----
From: Richard Erlacher <edick_at_idcomm.com>
To: Discussion re-collecting of classic computers
<classiccmp_at_u.washington.edu>
Date: Friday, April 16, 1999 9:07 AM
Subject: Re: Will The Grand Master Of Disk Controllers step foreward?


>
>-----Original Message-----
>From: Eric Smith <eric_at_brouhaha.com>
>To: Discussion re-collecting of classic computers
><classiccmp_at_u.washington.edu>
>Date: Friday, April 16, 1999 1:45 AM
>Subject: Re: Will The Grand Master Of Disk Controllers step foreward?
>
>
>>> Specifically the WD-1000-5 disk controller. These are the ones that
>>[...]
>>> If you are intimately familiar with this legendary interface, I would
>>> like to hear from you. I need to figure out how to modify it for 8"
>>> harddrives.
>>
>>Regrettably I no longer have the manual or schematics for these, so a lot
>>of this is from memory.
>>
>>The WD1000-5 was the WD1000 repackaged on an 8" * 5.75" board, to match
the
>>form factor of 5.25" drives.
>
>
>There's more to the difference than that. For one thing, the WD1000 series
>used the WD1100 chipset and an 8X300 microcontroller to run the whole
thing,
>while the WD1000-05 and -08, as well as the later models, used the WD1010
>chip along with other combinations of the 40-pin support chipset of which
>WD1014, which was, in at least one incarnation, an 8041.
>
>>The original WD1000 and WD1001 had both 34 and 50 pin drive control
>>connectors. I'm guessing that the WD1000-5 left the 50 pin connector out.
>>However, you only need to scramble the pins appropriately, as the actual
>>signals are the same. All odd pins are ground on both connectors; the
>others
>>should map thusly:
>>
>> 34-pin 50-pin signal
>>
>> 2 2 *RWC reduced write current
>> 4 4 *HS2 head select 2
>> 6 40 *WG write gate
>> 8 8 *SC seek complete
>> 10 42 *TK0 track 0
>> 12 44 *WF write fault
>> 14 14 *HS0 head select 0
>> 16 NC
>> 18 18 *HS1 head select 1
>> 20 20 *IDX index
>> 22 22 *RDY ready
>> 24 36 *STEP
>> 26 26 *DS1 drive select 1
>> 28 28 *DS2 drive select 2
>> 30 30 *DS3 drive select 3
>> 32 32 *DS4 drive select 4
>> 34 34 *DIR step direction (in when asserted)
>
>
>This should look pretty much like 8" floppy disks. An early controller I
>built used an FDC chip to drive these control signals, as the 8"
Winchesters
>had the same maximal step rate back then as the 8" double-headed FDD's.
>That's overkill, and the FDC expects to see things from the data stream
>which close the loop, and it won't see them. Open-loop, e.g. simple
>head-positioning command operation is possible, at least to see if the
>drive's mechanical functions are working. An enterprising approach would
be
>to operate the drive with a pair of small single-chippers one fairly slow
>one to handle the head positioning, and the other a fairly quick one to
>modulate the data, e.g. with ERLL code as was used in the PERSTOR
>controllers.
>
>>The radial data connectors are the same for both drive sizes.
>
>
>Yes except that some drives extracted clock locally and sent it on the data
>cable as well. For that reason, it would be advisable to stick with the
>4.34 MHz data rate. Keep in mind, also, that while the wide cable is
driven
>with open collectors, the data cable is intended to be driven with
>differential drivers/receivers of the MC3486/87 or 26LS31/32 type.
>
>>The bigger problem is that 8-inch drives used a data rate of 4.34 Mbps
>rather
>>than 5 Mbps. I seem to recall that the WD1000 had a jumper setting for
>this.
>>If they removed the 50-pin drive control connector, they probably also
>removed
>>the jumper and supporting circuitry.
>
>
>Western Digital was somewhat confused about how they should number their
>controller models back in those days, and the scheme got muddled, but as I
>recall, and I have some controllers to prove it, the data rate was fixed on
>the board at the factory, in some cases, particularly the larger WD1000
>boards with the WD1100 chips + 8X300 on board, had a discrete VCO as
opposed
>to the 74S124 or the LS624 they later used. These VCO's had to be tuned
>quite carefully and a procedure was included in the instruction manual.
>There were jumpers for accomplishing this tuning operation on nearly every
>type of board in this entire family, but one needed both a crystal and a
>retuned VCO for the PLL, not to mention changing the passive components in
>the integrator (LPF) of the clock extraction circuitry. The process of
>setting the VCO center frequency was not terribly difficult, but one had to
>know which jumpers to remove at which stage of the operation because there
>were inputs to the circuit which had to be active and other which had to be
>passive at different stages of the operation.
>
>>> Also, does anyone have docs for the Quantum Q-2040 8"
>>> Winchester? I dunno what kind of power to feed it (24v sounds correct,
>>> but I seem to recall it used 110vac also!), and so on.
>>
>>No data here, but almost certainly not 110 VAC. Probably 24V AC and 5V
DC.
>
>
>The power connections are precisely what is used with an 8" FDD, including
a
>110 VAC supply for the spindle motor. Don't forget that some 8" drives had
>to be fed a negative 5V supply while others could swallow either -12 Vdc
>or -5, depending on a jumper because they had on-board regulation. You
have
>to look for the regulator and ensure it has the jumper which bypassed it in
>the correct position.
>
>>You *might* be able to get a Q2040 to run at 5 Mbps, but I've never
>personally seen it done.
>
>We, meaning my people and I, tried this several times and never got it to
>work reliably with Shugart drives. They did some signal processing on the
>data, and included clock on the cable, so you might be able to skip the VCO
>retuning if you can find the "right" place to inject this conditioned clock
>in the data/clock recovery circuit. I'd be surprised to find the Quantum
>drives did things much differently, as they had to be compatible with the
>Shugarts, and some controllers, e.g. Intel's, relied on the drives or other
>external circuitry to extract clock.
>
Received on Fri Apr 16 1999 - 10:17:08 BST