10 Jan
2010
10 Jan
'10
2:06 a.m.
Hi guys,
I've got a Citizen Z1DE-55A (3.5" 1.44MB PC) floppy drive hooked up to
the disc analyser (it seems my Sony MPF520 has a duff head; it's reading
garbage for one side of the disc). Here's the problem...
- When the drive is selected, irrespective of head position, TRK00 is
always inactive.
- If the drive is given a seek command, it will seek to track 0, assert
TRK00, but the INDEX and READ DATA outputs will go idle (float high).
- The drive remains stuck in this state until power-cycled.
Has anyone seen this kind of thing before? The MPF520 had a head-step
lockout (if track0_sensor == active, then ignore STEP commands) but
didn't crash out if you asked it to step.
For what it's worth, the drive works fine in my PC... I just can't see
why it's locking up like this.
Also, am I right in thinking that pins 12 (DS1) and 16 (MOTEN) are the
two I need to pull low to select the drive, if it's connected before the
twist in the cable? And that DS0 and DS2 would be the select and motor
enable pins (respectively) if the drive was located after the twist?
Citizen are, as always, their usual "helpful" selves: "We don't make
floppy drives, we never made floppy drives. LA-LA-LA, NOT LISTENING!"
Thanks,
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
10 Jan
10 Jan
1:05 p.m.
Hi guys,
I've got a Citizen Z1DE-55A (3.5" 1.44MB PC) floppy drive hooked up to
the disc analyser (it seems my Sony MPF520 has a duff head; it's reading
garbage for one side of the disc). Here's the problem...
It may not be the head, but it shouldn't he hard to check...
Firstly, clean the heads -- not with a cleaning disk (no matter what the
manuals may say), but with a cotton bud and propan-2-ol. I once spent an
afternoon chasing a non-existant fault on the analogue board of a CBM
2850LP that turned out to be drity heads.
Secondly, most dosk heads have 4 connections -- if there are actually 5
wires on the connector, then the last is most likely a screen (do any of
the pins on the drive pCB connector test as grounds? If so, it's almost
certainly a screen). The 4 connections are the 3 wires of a centre-tapped
reand/write winding and one end of the tunnel erase head, the other side
of that goes ot the centre tap.
Just occasionally you come acrooss other arrangements. One is to have the
erase head totally flating wrt the read-write head, so 5 wires. Another
is to include some or all of the switching diodes in the head assembly.
Look for SOT23 pacakges on the flexiprint to the heads. Sony once made a
double-sided head assembly with 6 wires and with the all the switching
diodes on the flexiprint, but they went away from tlat on more modern drives.
You cna test the head windings for continmuity quite easily. If you're
worried you may have magnetised the head by so doing, you can demagnetise
it using a tape head demagnetiser. The head windings are low-resistance
and will essentially show as a short-circuit between all 4 head connections.
If the head checks out, trace form the head connector on the PCB to
switching diodes, often doube diode units in SOT23 packages. Check them,
replace if necessary. The centre tap pin most likely foes to a
transistor, part of the head swtiching circuit, I've had that fail before
now. Again, easy to check and replace.
That Sony model number sounds somewhat familiar to me. Is it a normal-size
(half-height, techniclly 3.5" unit with a logic board iwth 2 large PQFPs
on it (one for the analouge circuitry and one for the digital circuitry)?
And a positiion at the rear right corner of the main chassis that could be
used to mount an eject motor unit, like a Mac drive? If so, grab the
HP9114B (must be B) diagrams from http://www.hpmuseum.net, IIRC that uses
a similar drive.
- When the drive is selected, irrespective of head
position, TRK00 is
always inactive.
- If the drive is given a seek command, it will seek to track 0, assert
Those 2 seem inconsistent. The first says that TRK00 is never asserted,
the latter gices a time when it is. Which is it?
TRK00, but the INDEX and READ DATA outputs will go
idle (float high).
- The drive remains stuck in this state until power-cycled.
Has anyone seen this kind of thing before? The MPF520 had a head-step
lockout (if track0_sensor == active, then ignore STEP commands) but
didn't crash out if you asked it to step.
I've seen floppy drives that do odd things if, for example, one of the
input signals is asserted on power-up. But never this. I assume the
control logic is one ASIC :-(, so you can't investigate it.
What signals are you asserting when you try the seek? You don't have
write-gate asserted or anything silly, do you?
Silly suggestion. Connect it to your PC and do some seeks. Use a logic
analyser to see what the signals to the drive are doing. Then see how the
signals from your excersiser differ.
For what it's worth, the drive works fine in my PC... I just can't see
why it's locking up like this.
Also, am I right in thinking that pins 12 (DS1) and 16 (MOTEN) are the
two I need to pull low to select the drive, if it's connected before the
twist in the cable? And that DS0 and DS2 would be the select and motor
enable pins (respectively) if the drive was located after the twist?
Yes, that sounds right.
-tony
3:06 p.m.
Tony Duell wrote:
It turned out to be a logic bug in the FPGA -- the nSTEP output wasn't
being set high/inactive when the seek had completed (though it was being
set high to produce the pulses during the seek). In other words, it
looked like this:
__ __ __
STEP ___| |__| |__| |____________
When I wanted this:
__ __ ____________
STEP ...___| |__| |__|
Added the missing assignment and now it works fine. Interesting that the
Sony drive didn't care about this, yet the Citizen apparently locked up.
In any case it's fixed now and both the Sony MPF520 and Citizen Z1DE-55A
seem happy.
The exact failure mode seems to be:
- If nSTEP is low on startup, the drive will work fine until the next
step operation.
- If a STEP operation leaves nSTEP low (active) then the drive will
leave most of the control outputs in their inactive states.
Firstly, clean the heads -- not with a cleaning disk
(no matter what the
manuals may say), but with a cotton bud and propan-2-ol. I once spent an
afternoon chasing a non-existant fault on the analogue board of a CBM
2850LP that turned out to be drity heads.
I'll strip it again and have another go. But I wouldn't be at all
surprised if the upper head has been wrecked -- the loading mechanism
jammed a while ago. I stripped it, re-lubricated the sliders and
reassembled it, but caught the top head when I was putting the top plate
(with the pin that opens the disc shutter) back in.
The head that's having issues appears to be "side 0" if that means
anything (my copy of the ECMA 3.5in floppy standards is "missing" ATM,
and I can't remember if that's the upper or lower head).
TBH, the timing histogram from the Sony drive's upper head looks like
noise. The timing histogram from the same disc in the Citizen drive (see
below) or a Samsung SFD-321 looks like the normal MFM "three peaks"
histogram.
What I'm seeing is ~81.5ksamples acquired on the "OK" head, and
~67ksamples acquired on the "bad" head, typically with larger gaps in
timing on the latter. It's as if the drive is missing transitions, or
just spitting out noise. Hook up the Citizen drive, and it reads
~82ksamples of timing data for any track.
Just occasionally you come acrooss other arrangements.
One is to have the
erase head totally flating wrt the read-write head, so 5 wires. Another
is to include some or all of the switching diodes in the head assembly.
Look for SOT23 pacakges on the flexiprint to the heads. Sony once made a
double-sided head assembly with 6 wires and with the all the switching
diodes on the flexiprint, but they went away from tlat on more modern drives.
This one has two flexiprint cables with five pins each. No components on
the flexi unless they're hidden inside the plastic body of the head
mountings.
That Sony model number sounds somewhat familiar to me.
Is it a normal-size
(half-height, techniclly 3.5" unit with a logic board iwth 2 large PQFPs
on it (one for the analouge circuitry and one for the digital circuitry)?
It's 1-in high, with an 80-pin QFP labelled "SONY / CXA1803Q / JAPAN /
429A94K", a 32-pin TSSOP labelled "BA6489FS / 435 148" and a 20-pin SOIC
labelled "NEC JAPAN / D16803 / 9423F7" on the logic board.
From the label on top of the drive:
9409
E/A30
SONY(R) 5V== [//]
MODEL MPF520-E
10292485 SMM
Made in Malaysia
4-627-742-61 [9U] [CSA] [TUV]
([9U] is the UL Recognised Component mark, [CSA] is the CSA approvals
logo, and TUV is the TUV triangle logo)
On the second label:
[hp] HEWLETT PACKARD
1.44MB Flexible Disk Drive
PART NUMBER: D2035-60031
Made to Hewlett-Packard Standards
5V-0.89A SONY MPF520
("[hp]" is the bracketed HP logo)
And a positiion at the rear right corner of the main
chassis that could be
used to mount an eject motor unit, like a Mac drive?
Centre front is more like it. There's a hole in the main chassis that
leads to a serrated bar that moves when the eject button is pressed.
There are a couple of holes that could (theoretically) be used to attach
an eject motor, but they don't appear to have been tapped.
- When the
drive is selected, irrespective of head position, TRK00 is
always inactive.
- If the drive is given a seek command, it will seek to track 0, assert
Those 2 seem inconsistent. The first says that TRK00 is never asserted,
the latter gices a time when it is. Which is it? I've seen floppy drives that do odd things if, for
example, one of the
input signals is asserted on power-up. But never this. I assume the
control logic is one ASIC :-(, so you can't investigate it.
Probably. The PCB is covered by a metal plate which looks like it'll be
very difficult to remove.
It doesn't even look like the drive can be jumpered for DS0, unless it's
done with SMD zero-ohm links under the metal shield.
What signals are you asserting when you try the seek?
You don't have
write-gate asserted or anything silly, do you?
No, WRGATE and WRDATA are locked high in the FPGA logic (for "data
security" reasons... and because I haven't ported the disc write logic
across yet).
Thanks,
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
4:56 p.m.
Tony Duell wrote:
As an aside, this is a major problem on the full-height Sony 3.5" drives.
They were originally lubricated with a grease that turns hard and sticky
with age. The result is that the eject mechansim doesn't lock in the 'up'
posiiton properly. When you try to eject a disk, the upper head will
catch in the slot of the disk itself and will be ripped off the gimbal
spring. This drive was used in a lot of HP devices, if anyone has ole HP
3.5" stuff and hasn't cleaned off the grease, do so _now_ before you
haveto ifind a new head assembly and align it!
Firstly, clean the heads -- not with a cleaning
disk (no matter what the
manuals may say), but with a cotton bud and propan-2-ol. I once spent an
afternoon chasing a non-existant fault on the analogue board of a CBM
2850LP that turned out to be drity heads.
I'll strip it again and have another go. But I wouldn't be at all
surprised if the upper head has been wrecked -- the loading mechanism
jammed a while ago. I stripped it, re-lubricated the sliders and
reassembled it, but caught the top head when I was putting the top plate
(with the pin that opens the disc shutter) back in.
The head that's having issues appears to be "side 0" if that means
anything (my copy of the ECMA 3.5in floppy standards is "missing" ATM,
and I can't remember if that's the upper or lower head).
The upper head -- the one that points down -- is side 1. Side 0 is the
underside of the disk. The way to rememebr that is that the single-head
drive has the lower head only with a preasure pad on the moveable part of
the arm. And yes, single-head 3.5" drives do exist (as for that matter to
40 cylinder ones -- 67.5 tpi)/
TBH, the timing histogram from the Sony drive's upper head looks like
Do you mean 'upper head' or 'side 0 head' here?
noise. The timing histogram from the same disc in the
Citizen drive (see
An open-circuit head will produce noise, as will anything that causes it
not to be connected to the read amplifier.
This one has two flexiprint cables with five pins
each. No components on
the flexi unless they're hidden inside the plastic body of the head
mountings.
That;s normal for later drives. The switching components are on the PCB.
That Sony model number sounds somewhat familiar
to me. Is it a normal-size
(half-height, techniclly 3.5" unit with a logic board iwth 2 large PQFPs
on it (one for the analouge circuitry and one for the digital circuitry)?
It's 1-in high, with an 80-pin QFP labelled "SONY / CXA1803Q / JAPAN /
429A94K", a 32-pin TSSOP labelled "BA6489FS / 435 148" and a 20-pin SOIC
labelled "NEC JAPAN / D16803 / 9423F7" on the logic board. Those 2 seem
inconsistent. The first says that TRK00 is never asserted,
the latter gices a time when it is. Which is it?
It turned out to be a logic bug in the FPGA -- the nSTEP output wasn't
being set high/inactive when the seek had completed (though it was being
set high to produce the pulses during the seek). In other words, it
looked like this:
__ __ __
STEP ___| |__| |__| |____________
When I wanted this:
__ __ ____________
STEP ...___| |__| |__|
Added the missing assignment and now it works fine. Interesting that the
Sony drive didn't care about this, yet the Citizen apparently locked up.
5:20 p.m.
Tony Duell wrote:
Side 1 head. Upper head. Side 0/lower reads OK, side 1/upper reads as
garbage.
As an aside, this is a major problem on the
full-height Sony 3.5" drives.
They were originally lubricated with a grease that turns hard and sticky
with age. The result is that the eject mechansim doesn't lock in the 'up'
posiiton properly. When you try to eject a disk, the upper head will
catch in the slot of the disk itself and will be ripped off the gimbal
spring. This drive was used in a lot of HP devices, if anyone has ole HP
3.5" stuff and hasn't cleaned off the grease, do so _now_ before you
haveto ifind a new head assembly and align it!
Sounds similar to what happened with this one -- I had to extract the
disc by opening the drive up and manually forcing the eject mechanism to
work. Unfortunately I pushed the slider too far and knocked the top tray
off the mounting. When I put it back in, I suspect I hit the top head
and either broke it or knocked it out of alignment.
TBH, the
timing histogram from the Sony drive's upper head looks like
Do you mean 'upper head' or 'side 0 head' here? An open-circuit head will produce noise, as will
anything that causes it
not to be connected to the read amplifier.
What's odd is that there seems to be some data there -- one or two
sector headers occasionally decode OK, so maybe the head is just out of
alignment.
That, alas, is nothing like the drive I am thinking
of. Mind you, the
drive I was thinking of is a 720K one (actually the version I know is a
80- cylinder, double hard _600 rpm_ unit, but it came in all sorts of
versions.
That sounds like the drive in my HP 1651B logic analyser...
The drive in the 16500B appears to be a fairly standard laptop-style
(i.e. thin) 300RPM 1.4MB drive, though.
I would still be incined to trace back from the head
connectors. If they
go straight to one of the chips, you may well have big problems, but if
there are any discrete components involved, it's worth checking them.
To be honest I'm not overly concerned about fixing the MPF520. It's a
fairly nice drive in that it has physical jumpers for DS0/DS1 and
density selection (AUTO or PIN-2), but I have Panasonic and Citizen
drives that (at the very least) have the DS0/DS1 switch.
The drive I'm using now (the Z1DE-55A) doesn't have the jumpers, but it
does have a rather fetching black front panel and a green ACTIVE LED...
which I'm very tempted to replace with a blue LED... :)
Without knowing a lot more about the internals of the
drive I can't begin
to guess what is going on :-)
:-)
My best guess is that the LSI contains a flip-flop that's cleared when
STEP transitions H->L, and cleared when it transitions L->H. The FF
output is then used as an enable signal for the RD_DATA and INDEX drivers.
As I've said before: "It works now, or at least seems to."
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
6:51 p.m.
Tony Duell wrote:
Ah no. In the full height one, the eject mechanism will normally still
move if you press the button, but the latching pawl dowsn't drop into
place. So the disk holder -- and upper head -- go down again when you
release the button. if the disk is partially ejected when this happens,
it's almost certain the top head will be ripped off the gimbal.
There are some photss in my filckr account (tony_duell) of this drive in
bits.
As an aside, this is a major problem on the
full-height Sony 3.5" drives.
They were originally lubricated with a grease that turns hard and sticky
with age. The result is that the eject mechansim doesn't lock in the 'up'
posiiton properly. When you try to eject a disk, the upper head will
catch in the slot of the disk itself and will be ripped off the gimbal
spring. This drive was used in a lot of HP devices, if anyone has ole HP
3.5" stuff and hasn't cleaned off the grease, do so _now_ before you
haveto ifind a new head assembly and align it!
Sounds similar to what happened with this one -- I had to extract the
disc by opening the drive up and manually forcing the eject mechanism to work. Unfortunately I pushed the slider too far and
knocked the top tray
off the mounting. When I put it back in, I suspect I hit the top head
and either broke it or knocked it out of alignment.
Does the gimbal spring (the spring leaf that uspports the upper head)
look bent or distorted? It is quite delicate, although often bending it
back so it looks flat will get the drie working again.
TBH, the
timing histogram from the Sony drive's upper head looks like
Do you mean 'upper head' or 'side 0 head' here? An open-circuit head will produce noise, as will
anything that causes it
not to be connected to the read amplifier.
What's odd is that there seems to be some data there -- one or two
sector headers occasionally decode OK, so maybe the head is just out of
alignment. That, alas, is nothing like the drive I am
thinking of. Mind you, the
drive I was thinking of is a 720K one (actually the version I know is a
80- cylinder, double hard _600 rpm_ unit, but it came in all sorts of
versions.
That sounds like the drive in my HP 1651B logic analyser... 
7:02 p.m.
---snip---
My best guess is that the LSI contains a flip-flop that's cleared when
STEP transitions H->L, and cleared when it transitions L->H. The FF
output is then used as an enable signal for the RD_DATA and INDEX drivers.
As I've said before: "It works now, or at least seems to."
Hi
This is normal behavour for the newer drives that don't have a DskRdy
signal.
The way they indicate not ready is to turn off both the data and the
index signals. I think this works from seeing the disk spin but I don't
recall exactly how this works.
I've cross wired the index to the dskrdy for some of the older machines
and that seems to work but you do need to check to see if it uses
active or passive loading.
Dwight
_________________________________________________________________
Hotmail: Trusted email with powerful SPAM protection.
http://clk.atdmt.com/GBL/go/196390707/direct/01/
10:58 p.m.
The upper head -- the one that points down -- is side
1. Side 0 is the
underside of the disk. The way to rememebr that is that the single-head
drive has the lower head only with a preasure pad on the moveable part of
the arm. And yes, single-head 3.5" drives do exist (as for that matter to
40 cylinder ones -- 67.5 tpi)/
They are very common on european MSX machines. And most of them has
drive belts! :oO
11:08 p.m.
the arm. And yes, single-head 3.5" drives do
exist (as for that matter to
40 cylinder ones -- 67.5 tpi)/
The Shugart SA300 single head 3.5" drive was ued in the Gavilan, with a
custom bezel. The SA350 DS 3.5" drive could use that bezel and there was
eventually DS support (GAVILAN MS-DOS 2.11 K?)
Which SS 3.5" drives were out before that?
The only 67.5 TPI drive that I have is an Epson. Who else made them?
11:35 p.m.
On 10 Jan 2010 at 15:08, Fred Cisin wrote:
The only 67.5 TPI drive that I have is an Epson. Who
else made them?
Sony had at least one half-height single-sided 600 RPM 40 cylinder
drive. I used to have a few (0AD or something like that). 26 pin
connector on the rear. I think they were used on the Preis CP/M box.
--Chuck
6:54 p.m.
the arm. And yes, single-head 3.5" drives do
exist (as for that matter to
40 cylinder ones -- 67.5 tpi)/
The Shugart SA300 single head 3.5" drive was ued in the Gavilan, with a
custom bezel. The SA350 DS 3.5" drive could use that bezel and there was
eventually DS support (GAVILAN MS-DOS 2.11 K?)
Which SS 3.5" drives were out before that?
The only 67.5 TPI drive that I have is an Epson. Who else made them?
The only one I've seen is the one used in the Epson portable floppy drive
(PF10 or something like that) for their CP/M, etcc, laptops. That's a
strange unit, the drive doesn't have its own logic board, therer's a
single board with the cotnroller, CPU, firmware, read/write amplifier, etc
on iot. Alas the CPU chip in mine has failed. Fortunately it's ROMless,
so there's some hope of getting a replacement, it's one variant of the
6202 in a surface-mount package. Again I can dig out the number if anyone
thinks they know a source...
-tony
11 Jan
11 Jan
6:57 p.m.
On 1/10/10 8:56 AM, Tony Duell wrote:
I have the service manual for the single-head model and an unofficial
scheamtic of the double-head one. Since they're indentical mechancially
apart from the head carriage (and no service manual is going to cover
stripping that down), I have all I need :-)
Electroncially, they're quite different (and there are 2 versions of the
double head PCB that I've seen -- FC9 and FC16 IIRC). The single-head PCB
is quite unconventional in some areas, there's a matching transformer
betwee nthe head and the read amplifer, for example. The double-head one
is more normal.
Thanks for asking, thouygh.
-tony
As an aside, this is a major problem on the
full-height Sony 3.5" drives.
Do you have the service manual? I was just about to upload it since it is part
of the Apricot service manual.
10 Jan
10 Jan
9:57 p.m.
On Sun, 10 Jan 2010, Philip Pemberton wrote:
I'll strip it again and have another go. But I
wouldn't be at all
surprised if the upper head has been wrecked -- the loading mechanism
jammed a while ago. I stripped it, re-lubricated the sliders and
reassembled it, but caught the top head when I was putting the top plate
(with the pin that opens the disc shutter) back in.
The head that's having issues appears to be "side 0" if that means
anything (my copy of the ECMA 3.5in floppy standards is "missing" ATM,
and I can't remember if that's the upper or lower head).
Unless you install it on edge or upside down, the lower head is #0, and
the upper head is #1 (or #1/#2, A/B, front/back, . . . )
11 Jan
11 Jan
6:47 p.m.
Unless you install it on edge or upside down, the
lower head is #0, and
the upper head is #1 (or #1/#2, A/B, front/back, . . . )
For a 3.5" drive it's safe to say that the side where the motor spindle
engages with the disk is the head 0 side. That's normally true of other
sizes of floppy disks too, but with those it's technically possible for
the motor spindle to be on either side of the media, with the 3.5" it
isn't.
-tony
10 Jan
10 Jan
11:29 p.m.
On 1/10/10 5:05 AM, Tony Duell wrote:
Firstly, clean the heads -- not with a cleaning disk
(no matter what the
manuals may say), but with a cotton bud and propan-2-ol.
Be EXTREMELY careful doing this. At least on 800k Mac floppies it it very
easy to bend the flexure on the upper head so that it no longer has enough
downward pressure to make contact with the disk.
11 Jan
11 Jan
6:59 p.m.
On 1/10/10 5:05 AM, Tony Duell wrote:
Odd. All 3.5" double-head drive service manuals -- in fact just about all
double-head flopply drive service manuals -- that I've read warn against
cleaning the heads 'by hand' with a cotton bud and insist you use a
cleaning disk.
I've always used a cotton bud though, and had no problems at alll.
Perhaps I am jsut very careful, but I have never distorted that gimbal
spring during cleaning.
-tony
Firstly, clean the heads -- not with a cleaning
disk (no matter what the
manuals may say), but with a cotton bud and propan-2-ol.
Be EXTREMELY careful doing this. At least on 800k Mac floppies it it very
easy to bend the flexure on the upper head so that it no longer has enough
downward pressure to make contact with the disk.
5474
days inactive
5475
days old
17 comments
7 participants
participants (7)
-
aek＠bitsavers.org -
ard＠p850ug1.demon.co.uk -
cclist＠sydex.com -
cisin＠xenosoft.com -
classiccmp＠philpem.me.uk -
dkelvey＠hotmail.com -
pu1bzz.listas＠gmail.com