12 Jan
2011
12 Jan
'11
5:34 p.m.
Hi guys,
Is there a defined standard as to when the READY output on a
Shugart-type disc drive should go active?
I'm (slowly!) working through the rebuild of one of the Amstrad drives
(an EME231 I know to have working mechanical components, potentially
good heads, and a fried control board), and I'm stuck on deciding how to
deal with the READY output.
My initial thought was to hold it inactive until a few INDEX pulses have
passed, and the motor speed was within 5% of 'ideal' speed. The plan was
to use a 32kHz oscillator and a 4040 counter to get a several-Hz signal,
then rig up some logic to check that the disc speed was OK, and after a
few valid index cycles enable the drive.
Then I started wondering... am I over-engineering this? Would waiting
for a couple of full disc rotations be enough to reliably generate a
READY signal?
Irony is that the read-amp will probably be the easiest part of the
whole system... I've got a Motorola appnote which basically says "if
you're using a data rate of X and a rotation speed of Y, these
parameters will work" -- X and Y being the two parameters the Amstrad
drive uses... That just leaves the write amp and erase logic to design
(and maybe a 'write lockout' jumper).
Thanks,
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
12 Jan
12 Jan
7 p.m.
Hi
I believe all 3.5 inch drives without DiskReady will block
both the read data and the index until the drive is considered
to be ready.
Dwight
Date: Thu, 13 Jan 2011 01:34:00 +0000
From: classiccmp at philpem.me.uk
To: cctalk at classiccmp.org
Subject: Disc drive READY output -- any standards?
Hi guys,
Is there a defined standard as to when the READY output on a
Shugart-type disc drive should go active?
I'm (slowly!) working through the rebuild of one of the Amstrad drives
(an EME231 I know to have working mechanical components, potentially
good heads, and a fried control board), and I'm stuck on deciding how to
deal with the READY output.
My initial thought was to hold it inactive until a few INDEX pulses have
passed, and the motor speed was within 5% of 'ideal' speed. The plan was
to use a 32kHz oscillator and a 4040 counter to get a several-Hz signal,
then rig up some logic to check that the disc speed was OK, and after a
few valid index cycles enable the drive.
Then I started wondering... am I over-engineering this? Would waiting
for a couple of full disc rotations be enough to reliably generate a
READY signal?
Irony is that the read-amp will probably be the easiest part of the
whole system... I've got a Motorola appnote which basically says "if
you're using a data rate of X and a rotation speed of Y, these
parameters will work" -- X and Y being the two parameters the Amstrad
drive uses... That just leaves the write amp and erase logic to design
(and maybe a 'write lockout' jumper).
Thanks,
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
7:32 p.m.
On 12 Jan 2011 at 19:00, dwight elvey wrote:
I believe all 3.5 inch drives without DiskReady will
block
both the read data and the index until the drive is considered
to be ready.
I don't know about the 3" drives; they're pretty primitive. And
they're belt-driven, which belt is one of the most frequently-failing
parts on these old drives.
I suspect that I'd check the drive speed (with software) before
formatting/writing tracks when it counts. If the belt is slipping
while you're reading, at least it isn't destructive.
--Chuck
13 Jan
13 Jan
1:33 p.m.
On 12 Jan 2011 at 19:00, dwight elvey wrote:
I don;t think they're any worse than other disk drives. And a lot better
than that darn Shugart thing with the spirally-gooved plastic disk to
move the hean. ARGH!
I believe all 3.5 inch drives without DiskReady
will block
both the read data and the index until the drive is considered
to be ready.
I don't know about the 3" drives; they're pretty primitive. And they're belt-driven, which belt is one of the most
frequently-failing
parts on these old drives.
Not all 3" drives are belt driven, altough I think the Amstrad ones we're
discussing are.
Any particualr reason why the belt i na 3" drive would be more likely to
fail than that in a 5.25" or 8" drive (both of which were commonly
belt-driven)?
-tony
2:44 p.m.
Tony Duell wrote:
Indeed - I seem to remember the one in my Apple /// would occasionally fail
when the head positioner would fall out of the grooved disk.
On 12 Jan 2011
at 19:00, dwight elvey wrote:
I don;t think they're any worse than other disk drives. And a lot better
than that darn Shugart thing with the spirally-gooved plastic disk to
move the hean. ARGH! I believe all 3.5 inch drives without DiskReady
will block
both the read data and the index until the drive is considered
to be ready.
I don't know about the 3" drives; they're pretty
primitive. And Not all 3" drives are belt driven, altough I
think the Amstrad ones we're
discussing are.
Any particualr reason why the belt i na 3" drive would be more likely to
fail than that in a 5.25" or 8" drive (both of which were commonly
belt-driven)?
I think the belts aren't quite as tall in the 3" drives as they are in 5.25"
and 8" ones, though - I wonder if they're more prone to stretching or
breaking? (whether they're typically under more stress at start-up or not too,
I'm not sure)
cheers
Jules
14 Jan
14 Jan
4:10 a.m.
On Thu, 13 Jan 2011, Jules Richardson wrote:
Tony Duell wrote:
Indeed - I seem to remember the one in my Apple /// would occasionally fail
when the head positioner would fall out of the grooved disk.
That's why the cover is never fastened to my Northstar Horizon. Head arm
has a steel ball under tension that keeps walking out of the groove on the
cam.
--
On 12 Jan
2011 at 19:00, dwight elvey wrote:
I don;t think they're any worse than other disk drives. And a lot better
than that darn Shugart thing with the spirally-gooved plastic disk to move
the hean. ARGH! I believe all 3.5 inch drives without DiskReady
will block
both the read data and the index until the drive is considered
to be ready.
I don't know about the 3" drives; they're pretty
primitive. And
13 Jan
13 Jan
3:45 p.m.
On 13 Jan 2011 at 21:33, Tony Duell wrote:
Not all 3" drives are belt driven, altough I
think the Amstrad ones
we're discussing are.
Any particualr reason why the belt i na 3" drive would be more likely
to fail than that in a 5.25" or 8" drive (both of which were commonly
belt-driven)?
I don't know enough about the composition of the belt. But I've got
plenty of 8" drives with their original belts running just fine, but
I haven't seen a single 3" drive from a Joyce that didn't have a
failed belt somewhere in its history.
--Chuck
5:57 p.m.
On 13/01/11 21:33, Tony Duell wrote:
I don;t think they're any worse than other disk
drives. And a lot better
than that darn Shugart thing with the spirally-gooved plastic disk to
move the hean. ARGH!
Sounds a bit like the helicoid in a camera lens...
Not all 3" drives are belt driven, altough I
think the Amstrad ones we're
discussing are.
Indeed they are. Cheap cassette motor tied to an even cheaper thin
plastic "flywheel" by means of a shoddy little drive belt. On the plus
side it's a fairly standard cassette drive belt so fairly easy to replace...
The worst part about them is that damn flywheel. Most 3.5in drives I've
seen use a honking great slab of heavy metal as a combined flywheel /
rotor (the ring magnets for the brushless motor are inside). There's
enough weight on one of those to smooth out any reasonable level of
speed variation...
Any particualr reason why the belt i na 3" drive
would be more likely to
fail than that in a 5.25" or 8" drive (both of which were commonly
belt-driven)?
In the case of the Amstrad drives? Because they're cheaply-made piles of
junk? :P
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
10:59 p.m.
> I don;t think they're any worse than other
disk drives. And a lot better
> than that darn Shugart thing with the spirally-gooved plastic disk to
> move the hean. ARGH!
On Fri, 14 Jan 2011, Philip Pemberton wrote:
Sounds a bit like the helicoid in a camera lens...
Flat disc with a spiral, NOT a helix. Used in the Shugart SA400. Just
like a primitive music storage system that was called a "record".
11:28 p.m.
On 2011 Jan 13, at 10:59 PM, Fred Cisin wrote:
Spiral in the disc drive, but I think Philip is drawing an analogy to
the mechanism in digital cameras that moves the lens in and out and
does the zoom. These are concentric cylinders with pins in one cylinder
riding in helical grooves in an adjacent cylinder. I think the point is
in both cases the objective is to transform rotational motion (easy to
generate with a small motor) into linear motion (arguably more
difficult to generate directly).
Mathematically, I think a helix and spiral can be shown to be
topologically equivalent, at least as long there is a hole in the
center of the disc with the spiral (imagine expanding the hole, folding
the surface of the disc out to become the cylinder wall: one side of
the disc becomes the outside of the cylinder, the other side becomes
the inside, and the spiral turns into a helix).
(Fixed a digital camera for a friend awhile ago which had been dropped
with the lens extended. Had to be disassmebled extensively, to find one
of the guide pins had been knocked out and was jammed between the
cylinder walls.)
> I don;t
think they're any worse than other disk drives. And a lot
> better
> than that darn Shugart thing with the spirally-gooved plastic disk to
> move the hean. ARGH!
On Fri, 14 Jan 2011, Philip Pemberton wrote:
Sounds a bit like the helicoid in a camera
lens...
Flat disc with a spiral, NOT a helix. Used in the Shugart SA400. Just
like a primitive music storage system that was called a "record".
14 Jan
14 Jan
11:52 a.m.
On 13/01/11 21:33, Tony Duell wrote:
Not really. I am talking about a spiral, not a helix. The Shugart
mechanism I am thinking of (it was used in the Apple Disk II for example)
has a flat plastic disk with a coarse spiral groove in it. The disk is
mounted on the stepper mtoro spindle, and is parallel to the floppy disk
itself. The head carriage has a follower (looks like a ball bearing on a
leaf sprign) that runs in the groove. As the plastic disk rotates, the
follower runs in the goove and moves the head.
And you can imaging, it was not a prcise mechanism when new. And after a
bit of wear it's next-to-hopeless..
Some floppy drives, certain;y 8" ones, had a helical leadscrew on the
stepper motro shaft that moved the head cariage via an anti-backleash
nut. This mechanism seems to have no major problems.
I don;t think they're any worse than other
disk drives. And a lot better
than that darn Shugart thing with the spirally-gooved plastic disk to
move the hean. ARGH!
Sounds a bit like the helicoid in a camera lens... Not all 3" drives are belt driven, altough I
think the Amstrad ones we're
discussing are.
Indeed they are. Cheap cassette motor tied to an even cheaper thin
plastic "flywheel" by means of a shoddy little drive belt. On the plus side it's a fairly standard cassette drive belt so
fairly easy to replace...
The worst part about them is that damn flywheel. Most 3.5in drives I've
seen use a honking great slab of heavy metal as a combined flywheel /
rotor (the ring magnets for the brushless motor are inside). There's
Sure. Driect-drive 5.25" drives have that too.
The belt-driven 5.25" and 8" drives that I've worked on have quite
massive metall spindle pulleys that act as flywheels.
enough weight on one of those to smooth out any
reasonable level of
speed variation...
I think if I was rebuilding this drive as you are, I'd be tempted to make
a metal spindle pulley, which should have some flywheel effect.
-tony
12 Jan
12 Jan
8 p.m.
On 1/12/2011 10:00 PM, dwight elvey wrote:
Hi
I believe all 3.5 inch drives without DiskReady will block
both the read data and the index until the drive is considered
to be ready.
Dwight
My experience echos this at least on standard PC 3.5" floppy drives.
When read data starts pulsing, the data is immediately correct and
valid. While the drives spin up, read data isn't active at all. The
spec sheets I've read all point to 500ms worst case for spin up. I
assume the drive waits 'til the RPMs are within tolerance, and then
starts pulsing read data. There aren't any invalid pulses, like time
between edges too large due to too slow rotational speed.
For my purposes (a read-only amiga controller), the initial data coming
off the track isn't valuable anyways because you are ALMOST always in
the middle of the track at the beginning of the read. There are some
corner cases that need to be caught and dealt with appropriately.
I've not bothered to use diskready or index in my application, although
they are connected to the FPGA -- but I use them to look at RPM
consistency, etc. I simply read slightly more than a full track to
guarantee I have all sectors in a sequential buffer rather than screw
with wrapping the end of the buffer w/ the beginning. I pay a slight
per track performance penalty for this but this is negated because I can
start reading prior to receiving an index signal.(where some other
solutions would just wait)
Keith
7:29 p.m.
On 13 Jan 2011 at 1:34, Philip Pemberton wrote:
Then I started wondering... am I over-engineering
this? Would waiting
for a couple of full disc rotations be enough to reliably generate a
READY signal?
A couple of one-shots work fine--I've also used an 8-pin
microcontroller--extremely simple and very stable. You could even
use a 6-pin uC.
Have a look at the schematics of any 8" drive for examples of one-
shot ready circuits.
--Chuck
7:35 p.m.
Anent the issue of microcontrollers vs. analogue solutions, there's
an article that doesn't say much in ED today:
http://bit.ly/gDcDx1
--Chuck
13 Jan
13 Jan
1:35 p.m.
Anent the issue of microcontrollers vs. analogue solutions, there's
an article that doesn't say much in ED today:
http://bit.ly/gDcDx1
As you said, pretty much content-free. I think it can be sumarised by
'The best design depends on the applciation, the designer's skills and
the contents of the junk box' :-) Something which is hard to disagree with.
-tony
1:29 p.m.
Hi guys,
Is there a defined standard as to when the READY output on a
Shugart-type disc drive should go active?
Not that I a maware of. Are there any standards for anything on that
interface :-)
I'm (slowly!) working through the rebuild of one of the Amstrad drives
(an EME231 I know to have working mechanical components, potentially
good heads, and a fried control board), and I'm stuck on deciding how to
deal with the READY output.
My initial thought was to hold it inactive until a few INDEX pulses have
passed, and the motor speed was within 5% of 'ideal' speed. The plan was
to use a 32kHz oscillator and a 4040 counter to get a several-Hz signal,
then rig up some logic to check that the disc speed was OK, and after a
few valid index cycles enable the drive.
Then I started wondering... am I over-engineering this? Would waiting
for a couple of full disc rotations be enough to reliably generate a
READY signal?
I think that's massively over-engineered :-). Most drives just waited for
a couple of index pulses after motor-on which was enough time for the
thing to get up to speed.
Irony is that the read-amp will probably be the easiest part of the
whole system... I've got a Motorola appnote which basically says "if
you're using a data rate of X and a rotation speed of Y, these
parameters will work" -- X and Y being the two parameters the Amstrad
That is nto too suprising. Amstrad may have done a lot of odd things, but
these drives did turn at 300 rpm and use the standard data rate.
drive uses... That just leaves the write amp and erase
logic to design
(and maybe a 'write lockout' jumper).
I asusme you've looked at some old drive schematics. The write circuit is
normally pretty simple, a divide-by-2 D-type clocked from the WD signal
and reset by WG (to get a consistent starring phase when writing), a
couple of traisnstors to drive the heads and a current sink. And a bit of
enable logic. Often the transistors were part of a tranistor array chip.
Many older 5.25" drives (and 8" ones) didn't use a microcontroller at
all. it was all SSI/MSI chips. So it's fairly easy to work out what's
going on.
-tony
14 Jan
14 Jan
6:11 a.m.
On 13/01/11 21:29, Tony Duell wrote:
Not really... other than "read data is a 1us pulse every time there's a
transition" and similar stuff like that.
It's the minutiae that cause the trouble.
Hmm, fair enough. Doing it that way gets rid of a 555 timer, or a binary
counter on the CPLD and a 32kHz crystal oscillator.
Is there a
defined standard as to when the READY output on a
Shugart-type disc drive should go active?
Not that I a maware of. Are there any standards for anything on that
interface :-) Then I started
wondering... am I over-engineering this? Would waiting
for a couple of full disc rotations be enough to reliably generate a
READY signal?
I think that's massively over-engineered :-). Most drives just waited for
a couple of index pulses after motor-on which was enough time for the
thing to get up to speed. That is nto too suprising. Amstrad may have done a lot
of odd things, but
these drives did turn at 300 rpm and use the standard data rate.
Never a bad thing :)
I asusme you've looked at some old drive
schematics. The write circuit is
normally pretty simple, a divide-by-2 D-type clocked from the WD signal
and reset by WG (to get a consistent starring phase when writing),
That's the easy part :)
a couple of traisnstors to drive the heads and a
current sink. And a
bit of
enable logic. Often the transistors were part of a
tranistor array chip.
Hm, I don't think I have any discrete transistor array chips (unless the
ULN2003 and 2803 High Current Darlington Arrays count).. and the only
transistors I have any great number of are BC857s (BC547s in SOT23
surface mount packages).
The problem I'm trying to figure out is: how much current do I need to
put through the head, and how much current did the Amstrad board put
through it?
I'd rather like to avoid overloading the head -- permanently magnetising
or frying it would not be a good start.
Thanks,
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
12:13 p.m.
On 13/01/11 21:29, Tony Duell wrote:
Not really... other than "read data is a 1us pulse every time there's a
transition" and similar stuff like that.
Is that 1us guaranteed? IIRC it comes from an RC one-shot in the drive,
and there's considerable tolerance there.
Is there
a defined standard as to when the READY output on a
Shugart-type disc drive should go active?
Not that I a maware of. Are there any standards for anything on that
interface :-)
It's the minutiae that cause the trouble.
Indeed yes. As I said in another message, at one time you could get OEM
and service manuals for floppy drives, at least one of which would
docuemnt that drive's interpretation fo the interface. Not any more :-(
That is nto
too suprising. Amstrad may have done a lot of odd things, but
these drives did turn at 300 rpm and use the standard data rate.
Never a bad thing :) I asusme you've looked at some old drive
schematics. The write circuit is
normally pretty simple, a divide-by-2 D-type clocked from the WD signal
and reset by WG (to get a consistent starring phase when writing),
That's the easy part :)
a couple of traisnstors to drive the heads and a
current sink. And a
bit of
enable logic. Often the transistors were part of
a tranistor array chip.
Hm, I don't think I have any discrete transistor array chips (unless the ULN2003 and 2803 High Current Darlington Arrays
count).. and the only
transistors I have any great number of are BC857s (BC547s in SOT23
surface mount packages).
You mean you don't keep 2N3904s and 2N3908s in stock? I am seriously
suprised, I use those by the dozen...
The problem I'm trying to figure out is: how much current do I need to
put through the head, and how much current did the Amstrad board put
through it?
I thought we estimated this some time ago from the resistors on the
Amstrad board and a reasonable assumption that the write current was
passed through one of the resistors.
Alternatively, look at th shcemaitcs for some normal 5.25" and 3.5"
drives that use discrete components in the write amplifier (I'll have a
go if you can't find any..) Work out the write curtrent from those. It
won't be exactly right, but I;ll be it'll be close...
I'd rather like to avoid overloading the head --
permanently magnetising
or frying it would not be a good start.
Frying it would be a darn bad idea, I susepct the only sourve of spare
heads comes witrh logic boards attached :-)
However, you can demagnetise disk drive heads with a tape hard
demagnetiser ('defluxer' in some books). Basically a coil with a soft
iron cor that's connected to the mains through a push switch. Put the tip
of th core on the head, press the switch, slowly move it away to arm's
length and then (only then) release the button. I have no idea if you can
still buy such things, but they must turn up on E-bay, or I guess you
could make one.
-tony
4:37 p.m.
On 14/01/11 20:13, Tony Duell wrote:
No, I've seen datasheets spec anything from 250ns to 1us or so as a
"typical" value. The DiscFerret will pick up anything down to 1ns, if a
pulse that narrow gets through the I/O buffers. Guaranteed maximum data
rate through the LVX14s is ~5MHz, though if you used push-pull TTL (i.e.
ignore the pull-up resistor pack) you could probably go to 20MHz or
beyond. As for propagation delay, the IDC cable will probably cause more
trouble than the length of the tracks on the PCB...
"I believe in standards. Everyone should have one!"
-- George Morrow, "Quotations from Chairman Morrow"
They are. And I put a Farnell order in the other day.... without
ordering any.
Oh, I have other heads. Two more in fact (if you count the other drive
which apparently has a working head and logic board but seriously poor
speed regulation), plus the single-sided drive. The plan is to design a
double-sided logic board, but try it on the SS drive, then move it to a
DS drive to check the head-switching stuff.
Not really...
other than "read data is a 1us pulse every time there's a
transition" and similar stuff like that.
Is that 1us guaranteed? IIRC it comes from an RC one-shot in the drive,
and there's considerable tolerance there. Indeed yes. As I said in another message, at one time
you could get OEM
and service manuals for floppy drives, at least one of which would
docuemnt that drive's interpretation fo the interface. Not any more :-(
The Samsung SFD321B datasheet is 'out there'. Somewhere on
www.techtravels.org/amiga/amigablog/ as I recall.
Never a bad
thing :)
Try sayign that to an Apple ][ fanatic and watch the fireworks! Hm, I
don't think I have any discrete transistor array chips (unless the
I think they're still available. You mean you don't keep 2N3904s and 2N3908s in
stock? I am seriously
suprised, I use those by the dozen...
My standard transistors are the BC547, BC557 and the SMD variants
thereof (BC847 and BC857). I also use 2N7000 and 2N7002 N-MOSFETs like
they're going out of fashion, and the Infineon BSS84 low-power
logic-level P-MOSFET is a particular favourite :)
The BD139 and BD140 are my standard medium-power NPN/PNPs, and I've got
a few Mospec MJE2955 and MJE3055 10A/75W TO220 power transistors too.
Note that all of these are European parts... the BCs were (AIUI)
originated by Philips/Mullard, the BSSes by Siemens/Infineon.
In terms of 2N-series parts, I have a couple of 2N2222s and that's about
it. Oh, and some 2N3819 JFETs. Apparently there are a couple of 2N3904s
and 2N3906es in my "unsorted parts" box, though whether they've been
used for something in the meantime I have no idea.
I thought we estimated this some time ago from the
resistors on the
Amstrad board and a reasonable assumption that the write current was
passed through one of the resistors.
Indeed we did. Brain no workie...
Alternatively, look at th shcemaitcs for some normal
5.25" and 3.5"
drives that use discrete components in the write amplifier (I'll have a
go if you can't find any..) Work out the write curtrent from those. It
won't be exactly right, but I;ll be it'll be close...
The Shugart SA800 or the Siemens drive Al Kossow mentioned might be a
good starting point...
I'd rather
like to avoid overloading the head -- permanently magnetising
or frying it would not be a good start.
Frying it would be a darn bad idea, I susepct the only sourve of spare
heads comes witrh logic boards attached :-) However, you can demagnetise disk drive heads with a
tape hard
demagnetiser ('defluxer' in some books). Basically a coil with a soft
iron cor that's connected to the mains through a push switch. Put the tip
of th core on the head, press the switch, slowly move it away to arm's
length and then (only then) release the button. I have no idea if you can
still buy such things, but they must turn up on E-bay, or I guess you
could make one.
A degausser in other words?
I don't think I've ever seen one small enough to go between two disc
heads... or even into a tape deck for that matter.
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
15 Jan
15 Jan
10:10 a.m.
Oh, I have other heads. Two more in fact (if you count
the other drive
which apparently has a working head and logic board but seriously poor
speed regulation), plus the single-sided drive. The plan is to design a
Woudln;'t it be easier to fix the speed cotnrol of that drive? Perhaps
it's a motor problem (if it really is a cheap cassette-type motor, I could
believe one of the windings was open). I can't beleive the speed control
circuitry is that complicated.
double-sided logic board, but try it on the SS drive,
then move it to a
DS drive to check the head-switching stuff.
They were veryt common at one time. The 'business end' was a soft wiron
rod about 7 or 8mm in daimeter with a shaped tip. That sould get between
the heads of most drives I think.
As I said, I have noi dea if they ar still made, but I can't beleive
they are rare on Ebay...
-tony
However, you can demagnetise disk drive heads
with a tape hard
demagnetiser ('defluxer' in some books). Basically a coil with a soft
iron cor that's connected to the mains through a push switch. Put the tip
of th core on the head, press the switch, slowly move it away to arm's
length and then (only then) release the button. I have no idea if you can
still buy such things, but they must turn up on E-bay, or I guess you
could make one.
A degausser in other words?
I don't think I've ever seen one small enough to go between two disc
heads... or even into a tape deck for that matter.
5090
days inactive
5092
days old
19 comments
9 participants
participants (9)
-
ard＠p850ug1.demon.co.uk -
cclist＠sydex.com -
cisin＠xenosoft.com -
classiccmp＠philpem.me.uk -
dkelvey＠hotmail.com -
hilpert＠cs.ubc.ca -
jules.richardson99＠gmail.com -
keithvz＠verizon.net -
snhirsch＠gmail.com