3 Jun
2011
3 Jun
'11
1:51 p.m.
Hi guys,
Someone has just donated an old ST506 hard drive to the DiscFerret
project... This time, it's an Amstrad drive, with part number 40095/A
"DRMD20A12A".
Thing is, I can't seem to find any information on it -- and at the very
least I need to find out what the CHS settings are in order to create a
DriveSpec file.
Here are a couple of photos:
Top of drive:
http://twitgoo.com/2bnvcy / http://i52.twitgoo.com/n7yp.jpg
PCB:
http://twitgoo.com/2bnvdl / http://i53.twitgoo.com/2cgzc0k.jpg
Major chips are:
Hitachi HD63B01Y0P
SanKen STA435A
Hitachi HA13426
CF77143FT
SSI 540-3CH S8714 32119.1.1F
Hybrids: SC619 and SC628
Text at top corner of PCB near power connector: PY349C
Head stepper motor is labelled:
TYPE 103-4902-0320
10.5VDC 0.26A 0.9 DEG/STEP
LOT NO. 05709I
6035095-1
Form factor is 3.5-inch, half-height (about the same height as a 5.25 HH
MFM drive)
From what I can gather, it's a 20MB MFM drive, with four heads --
there's a sticker on the top which reads:
DEFECT:CYLINDER
HEAD0
HEAD1 327 386 426 504
HEAD2
HEAD3 216
(The numbers are handwritten -- badly)
There's a second sticker along the edge:
W A R N I N G
WARRANTY VOID
IF THIS SHIELD
BROKEN
Does anyone have any idea what this thing might be?
Thanks,
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
3 Jun
3 Jun
2:04 p.m.
On 03/06/2011 21:51, Philip Pemberton wrote:
Hi guys,
Someone has just donated an old ST506 hard drive to the DiscFerret
project... This time, it's an Amstrad drive, with part number 40095/A
"DRMD20A12A".
Thing is, I can't seem to find any information on it -- and at the very
least I need to find out what the CHS settings are in order to create a
DriveSpec file.
Here are a couple of photos:
Top of drive:
http://twitgoo.com/2bnvcy / http://i52.twitgoo.com/n7yp.jpg
PCB:
http://twitgoo.com/2bnvdl / http://i53.twitgoo.com/2cgzc0k.jpg
I have one of those, it was the first working hard drive I had on my
PC-200. I believe it's a standard 615x4x17 type 2, 20MB drive, or that's
how I've always formatted it, I think it will also format up as a 32MB
if formatted RLL.
Cheers.
Phill.
--
Phill Harvey-Smith, Programmer, Hardware hacker, and general eccentric !
"You can twist perceptions, but reality won't budge" -- Rush.
4 Jun
4 Jun
7:36 a.m.
On 03/06/11 22:04, Phill Harvey-Smith wrote:
On 03/06/2011 21:51, Philip Pemberton wrote:
> Someone has just donated an old ST506 hard drive to the DiscFerret
> project... This time, it's an Amstrad drive, with part number 40095/A
> "DRMD20A12A".
[...]
I have one of those, it was the first working hard
drive I had on my
PC-200. I believe it's a standard 615x4x17 type 2, 20MB drive, or that's
how I've always formatted it, I think it will also format up as a 32MB
if formatted RLL.
Whatever it is, it doesn't have that many cylinders. The head stepper is
a 0.9deg/step thing, and the circuitry appears to single-step it. After
~380 steps, the flag on the motor shaft longer moves.
It's starting to look like this thing is a Type 01 or 23. But a 3.5in
half-height ST506 drive with a 10MB MFM capacity?! That seems too
strange to be true...
For bonus points, read_winchester is producing 130K of acquisition data
per track, and a rotation speed measurement of about 3550RPM. So the
data density is (in theory at least) roughly twice what the CDC MFM
drive produces.
Hmm.
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
8:37 a.m.
On 03/06/2011 21:51, Philip Pemberton wrote:
Someone has just donated an old ST506 hard drive to
the DiscFerret
project... This time, it's an Amstrad drive, with part number
40095/A "DRMD20A12A".
Take a closer look--the drive should be "DRND20A"--it's an Alps 20MB
unit--615/4/17 are the nominal ratings. 60msec. average access.
10MB isn't unusual for a 3.5" MFM drive; I believe that Rodime even
offered a 5MB model (RO-351).
--Chuck
9:20 a.m.
On 04/06/11 16:37, Chuck Guzis wrote:
Take a closer look--the drive should be
"DRND20A"--it's an Alps 20MB
unit--615/4/17 are the nominal ratings. 60msec. average access.
Nope. The label definitely says DRMD20A (with an "M").
Although the similarity with the Alps part number is making me very
suspicious... unfortunately photos of Alps MFM drives seem very scarce.
Curiously enough, there's also nothing on TULARC about it...
One thing I've found out: this thing does NOT support Buffered Seek.
Turn that on and it misses steps like crazy... So I need to figure out
what value to use for the track-to-track seek delay... Ugh.
10MB isn't unusual for a 3.5" MFM drive; I
believe that Rodime even
offered a 5MB model (RO-351).
... And the RO-352, a 4-head 10MB drive.
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
12:57 p.m.
On 04/06/11 17:20, Philip Pemberton wrote:
One thing I've found out: this thing does NOT
support Buffered Seek.
Turn that on and it misses steps like crazy... So I need to figure out
what value to use for the track-to-track seek delay... Ugh.
Another thing I just found out: the head actuator stepper motor isn't
stepping properly. Every time I issue a STEP pulse, it seeks out *two*
tracks instead of just one.
So I end up with a 615-track image, which only contains 308 tracks of
valid track data.
Does anyone have any ideas why the drive might be doing this, and if
it's likely to be fixable?
Thanks,
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
2:39 p.m.
On 04/06/11 20:57, Philip Pemberton wrote:
So I end up with a 615-track image, which only
contains 308 tracks of
valid track data.
OK, I'm an idiot. This was caused by an off-by-one bug in libdiscferret...
Basically, the STEP command accepts a number between 0 and 127 for the
number of steps to move. 0 is one step, and so on up to 127 for 128 steps.
Turns out libdiscferret was sending a value of '1' when it wanted to
move the head one step in or out...
Then there was a lovely bug which caused the first seek from track 0 to
another track to effectively be doubled-up. So if you did a seek from t0
to t4, then from t4 to t10, you'd actually end up at t14...
Bugfixed version uploaded with my apologies and a large dose of "D'oh, I
can't believe I did something that stupid!"
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
3 p.m.
On 04/06/2011 22:39, Philip Pemberton wrote:
On 04/06/11 20:57, Philip Pemberton wrote:
[Snip!]
Doh! We've all been there and done that :(
So it turns out that drive is 615x4x[17,26] then ?
Cheers.
Phill.
--
Phill Harvey-Smith, Programmer, Hardware hacker, and general eccentric !
"You can twist perceptions, but reality won't budge" -- Rush.
So I end up with a 615-track image, which only
contains 308 tracks of
valid track data.
OK, I'm an idiot. This was caused by an off-by-one bug in libdiscferret...
4:19 p.m.
On 04/06/11 23:00, Phill Harvey-Smith wrote:
Doh! We've all been there and done that :(
I also managed to lean over a powered up CDC Wren-II/HH 5.25in
half-height drive while I was probing the DiscFerret's STEP line with my
scope.
For those of you who have never seen a Wren-II/HH, they have a pair of
small-ish HDA interface PCBs on the back of the 'drive enclosure'. There
are a dozen or so metal pins joining these two boards.
I wear a metal watch.
No points for guessing what happened.
Thankfully the drive survived, with minimal damage -- a little burning
around the Molex socket, and the DiscFerret "power take-off" cable was a
little singed, but that's it.
Guess I shorted a power rail to the drive casing...
So it turns out that drive is 615x4x[17,26] then ?
Yeah, it's a 615-cylinder, 4-head drive. PC-AT drive type 2.
If you format it MFM, you get 17 spt (which is how this drive has been
formatted). If you format it RLL, you get 26 at the cost of reliability.
I don't think I'd bother wiring this drive up to an RLL controller --
the scatter plot for heads 0 and 1 on every track after 75 or so show
extreme amounts of 'fuzz' towards the latter half of the track.
I'll be surprised if the data on this thing decodes at all. It was
probably due for a low-level reformat YEARS ago. Speed accuracy is poor
at best -- assuming it's a 3600RPM drive, it's running at least 60RPM
slow. Compared this to the CDC, which is barely 2RPM off speed at any
given time...
Yuck.
Oh well, if nothing else it'll probably be handy for stress testing the
data-separator and decoder engine.
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
4:33 p.m.
On 5 Jun 2011 at 0:19, Philip Pemberton wrote:
I'll be surprised if the data on this thing
decodes at all. It was
probably due for a low-level reformat YEARS ago. Speed accuracy is
poor at best -- assuming it's a 3600RPM drive, it's running at least
60RPM slow. Compared this to the CDC, which is barely 2RPM off speed
at any given time...
60 RPM isn't bad at all--less than 2%. Your data separator should
have a capture range of at least 5 times that. What matters more is
ISV.
--Chuck
5:26 p.m.
On 05/06/11 00:33, Chuck Guzis wrote:
60 RPM isn't bad at all--less than 2%. Your data
separator should
have a capture range of at least 5 times that. What matters more is
ISV.
If only it was that easy!
The timing error in the bad sections is over 20% +/- on the 'good'
tracks, and over 60% on cyl 0 head 0.
Ho hum.
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
5 Jun
5 Jun
11:01 a.m.
Doh! We've
all been there and done that :(
I also managed to lean over a powered up CDC Wren-II/HH 5.25in
half-height drive while I was probing the DiscFerret's STEP line with my
scope.
For those of you who have never seen a Wren-II/HH, they have a pair of
small-ish HDA interface PCBs on the back of the 'drive enclosure'. There
are a dozen or so metal pins joining these two boards.
I wear a metal watch.
No points for guessing what happened.
4949
days inactive
4951
days old
11 comments
4 participants
participants (4)
-
afra＠aurigae.demon.co.uk -
ard＠p850ug1.demon.co.uk -
cclist＠sydex.com -
classiccmp＠philpem.me.uk