27 Apr
2017
27 Apr
'17
10:42 a.m.
Hi, All,
I'm cleaning/refurbing a TRS-80 Model 4 I picked up at VCFe that was
in dirty-but-mostly-working shape. I've completely dismantled,
cleaned and reassembled the keyboard, I fixed the bad cable to the
floppy controller, I cleaned and lubed the TM-100-1 floppy drive, and
had it all nice and working, then I bumped the power strip it was
plugged into and the momentary surge took out the electronics on the
floppy drive. I swapped it out with another TM-100-1 unit (borrowed
from a Model 1) and it's booting again, but when
using this toasted
drive as :1, I get either an ERROR 3 or ERROR 4 from TRSDOS 1.3
(my
primary testing disk for the moment). The drive still seeks and spins
but it won't read disks that it used to read before the power hit.
I also have an original NEWDOS/80 disk and a copied MULTIDOS disk. I
have not yet fixed up a PC with a 40-track 5.25" drive for making
fresh disks, but it's on the list of solution paths.
I have the TM-100 service manual PDF (which includes schematics), so
it shouldn't be difficult to work through the functional subsystems of
the drive electronics. My question is are there any specific issues
with the parts on the TM-100 PCB to look for? There are a handful of
reasonably common ICs, and dozens of discrete components. Of course I
can trace through each section looking for where the results are
unexpected, but for such a common thing as a TM-100, perhaps there are
known pain points and perhaps someone here has repaired a few and
could highlight what parts might be "fragile".
Additionally, for a testing framework to poke signals through the
drive for debugging during the repair, what's a good platform? A PC
running MS-DOS? The TRS-80 Model 4 itself? Besides doing
directories, are there any good bits of software anyone can recommend
for exercising floppy drives on a sub-system-by-subsystem basis?
(move the heads, do a read, do a write...)
I expect like the last repair (shorted tantalum filter cap), this
repair is going to be a small number of components. Parts of the
drive are known to work - the motor turns on and off when it should,
and it does seek back to track zero when manually moved off of track
zero prior to doing a DIR :1 or when booting it as :0. At first
glance, something appears to be toasted in the read electronics.
It's not impossible to find another TM-100-1 or replace it with a
TM-100-2 (more common, owing to its appearance in the IBM 5150 PC),
but I'd like to just repair this one and get back to TRS-80 hacking.
Thanks for any tips or pointers.
-ethan
27 Apr
27 Apr
10:58 a.m.
On Thu, Apr 27, 2017 at 6:42 PM, Ethan Dicks via cctalk
<cctalk at classiccmp.org> wrote:
Hi, All,
I'm cleaning/refurbing a TRS-80 Model 4 I picked up at VCFe that was
in dirty-but-mostly-working shape. I've completely dismantled,
cleaned and reassembled the keyboard, I fixed the bad cable to the
floppy controller, I cleaned and lubed the TM-100-1 floppy drive, and
had it all nice and working, then I bumped the power strip it was
plugged into and the momentary surge took out the electronics on the
floppy drive. I swapped it out with another TM-100-1 unit (borrowed
from a Model 1) and it's booting again, but when using this toasted
drive as :1, I get either an ERROR 3 or ERROR 4 from TRSDOS 1.3 (my
primary testing disk for the moment). The drive still seeks and spins
but it won't read disks that it used to read before the power hit.
I also have an original NEWDOS/80 disk and a copied MULTIDOS disk. I
have not yet fixed up a PC with a 40-track 5.25" drive for making
fresh disks, but it's on the list of solution paths.
I have the TM-100 service manual PDF (which includes schematics), so
it shouldn't be difficult to work through the functional subsystems of
the drive electronics. My question is are there any specific issues
with the parts on the TM-100 PCB to look for? There are a handful of
reasonably common ICs, and dozens of discrete components. Of course I
can trace through each section looking for where the results are
unexpected, but for such a common thing as a TM-100, perhaps there are
known pain points and perhaps someone here has repaired a few and
could highlight what parts might be "fragile".
Floppy drives of that vintage consist of a number of almost independant
systems (there's a common enable/drive select but that's about it).
It looks like a problem in the read chain, but is it? If the spindle was
turning at the wrong speed I think it would fail to read.
So I would check each subsection of the drive, just in case it's not
the obvious one.
In the case of the read amplifier, you could start with a differential-
input 'scope on the pins used for head alignment (reading a normal
disk). If there's a good signal there the head select diodes (AFAIK
these are present even on the single-sided drive) and first amplifier
IC are OK. If not, then debug that part.
I don't know of any obvious common failures, and anyway
murphy's law will ensure that your drive has something odd failed :-)
Stock fault lists are useful if you have many indentical units to repair,
you might get 90% of them off the bench quickly but fairly useless for
one-off jobs.
Additionally, for a testing framework to poke signals through the
drive for debugging during the repair, what's a good platform? A PC
running MS-DOS? The TRS-80 Model 4 itself? Besides doing
directories, are there any good bits of software anyone can recommend
for exercising floppy drives on a sub-system-by-subsystem basis?
(move the heads, do a read, do a write...)
There are/were drive exercisers but no idea where you'd find one now.
If you were nearer to me, you could use mine....
More seriously, the signals involved are fairly slow speed. Before
I had a drive exerciser, I linked the important drive inputs (select,
motor on, step, direction, etc) to a PC parallel port and wrote a
short program (I think in Turbo Pascal, but it was a long time
ago) to turn the motor on, step the heads around, etc. I think
I monitored some of the drive outputs using the printer port input
lines too, but a logic probe is all you really need to check index,
track 0, etc.
You need a 'scope to debug the read amplifier of course (that
applies even if you have a drive exerciser).
Do check the spindle speed. Often there is a strobe disk on the
spindle pulley in these drives (if it's missing, I am sure you can
get an image of it somewhere, print it, cut it out and stick it on the
pulley). But also check ('scope) that you are getting 5 index pulses
per second, just in case the darn thing is running at half speed or
something.
-tony
6:36 p.m.
On Thu, Apr 27, 2017 at 1:58 PM, Tony Duell <ard.p850ug1 at gmail.com> wrote:
On Thu, Apr 27, 2017 at 6:42 PM, Ethan Dicks via
cctalk
> (diagnosing and repairing a) TM-100-1 floppy drive...
Floppy drives of that vintage consist of a number of
almost independant
systems (there's a common enable/drive select but that's about it).
It looks like a problem in the read chain, but is it? If the spindle was
turning at the wrong speed I think it would fail to read.
Yes. I did consider the spindle speed, but I'm reasonably certain it's good.
So I would check each subsection of the drive, just in
case it's not
the obvious one.
Well, yes. I'm really just asking about known weaknesses as a starting point.
In the case of the read amplifier, you could start
with a differential-
input 'scope on the pins used for head alignment (reading a normal
disk). If there's a good signal there the head select diodes (AFAIK
these are present even on the single-sided drive) and first amplifier
IC are OK. If not, then debug that part.
Right.
I don't know of any obvious common failures, and
anyway
murphy's law will ensure that your drive has something odd failed :-)
Perhaps. The last failure with this unit had the most obvious cause.
I literally tested one component and it was exactly that part that had
failed.
> Additionally, for a testing framework...
There are/were drive exercisers but no idea where
you'd find one now.
If you were nearer to me, you could use mine....
Right. I don't have a drive exerciser (and I live in the States) so I
didn't figure that was an easily achievable path. I'm figuring on
having to use a live machine to make the drive dance and I don't fancy
writing my own code on top of it all. Rather than go to that extent,
I'm more likely to toss the drive on the "to be fixed someday" pile
and get to it (much) later.
You need a 'scope to debug the read amplifier of
course (that
applies even if you have a drive exerciser).
Yes. Of course I have a scope. At least one.
Do check the spindle speed. Often there is a strobe
disk on the
spindle pulley in these drives
There is. It works.
But also check ('scope) that you are getting 5
index pulses
per second, just in case the darn thing is running at half speed or
something.
Worth verifying, but there's a visual difference between 300 RPM and
150 RPM (plus they sound different).
-ethan
12:27 p.m.
On 04/27/2017 10:42 AM, Ethan Dicks via cctalk wrote:
It's not impossible to find another TM-100-1 or
replace it with a
TM-100-2 (more common, owing to its appearance in the IBM 5150 PC),
but I'd like to just repair this one and get back to TRS-80 hacking.
The issue with Jugi's artistry is that there isn't a single version of
the PCB for the TM-100. I have drives with 3- 4- and 6-wire steppers on
them, all taking different PCBs.
But a TM-100 is a very simple-minded device. You can easily check its
basic mechanical operation off the system by grounding various signal
lines (all are active-low). So you can ground the motor line to make
sure that the motor spins; you can ground the drive select to see if the
LED illuminates. With the drive select operating, you cna use the STEP
and DIRECTION lines to move the head carriage; you can check the
operation of the track zero sensor.
Checking the read-write channel is a bit more challenging and a 'scope
can be very useful.
For lack of anything else, if you've got a PC with a floppy controller,
you can use Dave Dunfield's ImageDisk utility to perform simple checks.
He includes an RPM tester, though if the drive still has the strobe disk
attached to the spindle flywheel, a simple line-operated neon lamp can
do the same.
--Chuck
1:43 p.m.
On Thu, 27 Apr 2017, Ethan Dicks via cctalk wrote:
It's not impossible to find another TM-100-1 or
replace it with a
TM-100-2 (more common, owing to its appearance in the IBM 5150 PC),
but I'd like to just repair this one and get back to TRS-80 hacking.
Actually, the IBM 5150 PC originally (August 1981) mostly came with
TM100-1. It wasn't until about nine months later, when IBM released
PC-DOS 1.10 (May, 1982, which supported double sided drives) that they
added TM100-2 availability.
Why not use the TM100-2? It will work for any of the single sided use,
and some versions of the operating systems supported double sided drives.
(On the Tandy version of LDOS, double sided support was undocumented,
because Tandy didn't sell a double sided drive, until the model 4D?)
Remember that the PC uses all drives jumpered as if they were :1/DS1/B:
Radio Shack used an "unused" pin of the interface for drive :3/DS3/D:,
Unfortunately, that was the pin that was "reserved" for side select!
That poses a problem for model 1 (with four drives on one cable), but not
for III, 4, nor 4P
If you have a model 1 drive that has the modification for drive select
using the side select signal, don't use it if you use any double sided
drives in the machine, or clip that wire on the drive.
You can use "half-height" drives in a model 3 or 4.
7:34 p.m.
On Thu, Apr 27, 2017 at 4:43 PM, Fred Cisin via cctalk
<cctalk at classiccmp.org> wrote:
On Thu, 27 Apr 2017, Ethan Dicks via cctalk wrote:
Ah. I had forgotten that detail. I only ever worked machines with
double-sided drives.
It's not impossible to find another TM-100-1 or replace it with a
TM-100-2
Actually, the IBM 5150 PC originally (August 1981) mostly came with TM100-1.
It wasn't until about nine months later, when IBM released PC-DOS 1.10 (May,
1982, which supported double sided drives) that they added TM100-2
availability. Why not use the TM100-2? It will work for any of the
single sided use, and
some versions of the operating systems supported double sided drives.
(On the Tandy version of LDOS, double sided support was undocumented,
because Tandy didn't sell a double sided drive, until the model 4D?)
I know it will work.
Remember that the PC uses all drives jumpered as if
they were :1/DS1/B:
Right. I well remember how to jumper drive selects.
Radio Shack used an "unused" pin of the
interface for drive :3/DS3/D:,
Unfortunately, that was the pin that was "reserved" for side select!
That poses a problem for model 1 (with four drives on one cable), but not
for III, 4, nor 4P
Yes. I was reading about that the other day. I don't have any
4-drive units, just multiple dual-drive units.
If you have a model 1 drive that has the modification
for drive select using
the side select signal, don't use it if you use any double sided drives in
the machine, or clip that wire on the drive.
Good to keep in mind. I borrowed a half-height "Bachelor" FD-104 from
a Model 1 for drive/controller testing. I cannot find docs for it,
and barely a mention, even in TheRef. There is one jumper block by
the 34-pin edge connector - marked D.S. / 0 / (1) / 2 / 3 / MX. (the
'1' is damaged/missing because of a PCB via). It seems the first
jumper is likely to be single/double-sided (there are 2 heads), then 4
drive selects, then the MX one-drive-only-select jumper.
You can use "half-height" drives in a model
3 or 4.
Yes but at the moment, I only have one half-height 40-track drive,
that Bachelor FD-104. My only loose drives are 80-track HD units.
-ethan
8:17 p.m.
On Thu, 27 Apr 2017, Ethan Dicks via cctalk wrote:
Additionally, for a testing framework to poke signals
through the
drive for debugging during the repair, what's a good platform? A PC
running MS-DOS? The TRS-80 Model 4 itself? Besides doing
directories, are there any good bits of software anyone can recommend
for exercising floppy drives on a sub-system-by-subsystem basis?
(move the heads, do a read, do a write...)
You should check out a couple of disk utility programs for the Model 4:
Super Utility 4/4P
http://tim-mann.org/trs80/su4a.zip (CMD version)
https://fjkraan.home.xs4all.nl/comp/trs80-4p/dmkeilImages/util/97-0005.ace (bootable
version)
http://tim-mann.org/trs80/doc/su4man.pdf
Hyperzap 3.2C
https://fjkraan.home.xs4all.nl/97-0002.ace
If you hook up a working drive to boot from, one of those should help
to exercise/exorcise the bad drive. :-)
Mike Loewen mloewen at cpumagic.scol.pa.us
Old Technology http://q7.neurotica.com/Oldtech/
10:49 p.m.
On Thu, 27 Apr 2017, Ethan Dicks via cctalk wrote:
> Additionally, for a testing framework to poke signals through the
> drive for debugging during the repair, what's a good platform? A
> PC running MS-DOS? The TRS-80 Model 4 itself? Besides doing
> directories, are there any good bits of software anyone can
> recommend for exercising floppy drives on a sub-system-by-subsystem
> basis? (move the heads, do a read, do a write...)
I'd still recommend Dave Dunfield's ImageDisk--it's got several good
functions--test RPM, Clean heads, check alignment, etc.
Runs on MS-DOS and, AFAIK, even on the 5150.
I'm actually a bit surprised that nobody's done a floppy exerciser for
the lowly Arduino. Easy enough--if you use the timer "capture" feature
on the AVR, you can even do simple disk read testing.
--Chuck
10:56 p.m.
On Thu, Apr 27, 2017 at 11:49 PM, Chuck Guzis via cctalk
<cctalk at classiccmp.org> wrote:
KryoFlux has something like that. https://www.kryoflux.com/ has the
details, but it is commercial.
Warner
On Thu, 27 Apr
2017, Ethan Dicks via cctalk wrote:
> Additionally, for a testing framework to poke signals through the
> drive for debugging during the repair, what's a good platform? A
> PC running MS-DOS? The TRS-80 Model 4 itself? Besides doing
> directories, are there any good bits of software anyone can
> recommend for exercising floppy drives on a sub-system-by-subsystem
> basis? (move the heads, do a read, do a write...)
I'd still recommend Dave Dunfield's ImageDisk--it's got several good
functions--test RPM, Clean heads, check alignment, etc.
Runs on MS-DOS and, AFAIK, even on the 5150.
I'm actually a bit surprised that nobody's done a floppy exerciser for
the lowly Arduino. Easy enough--if you use the timer "capture" feature
on the AVR, you can even do simple disk read testing.
28 Apr
28 Apr
10:01 a.m.
On Thu, Apr 27, 2017 at 11:17 PM, Mike Loewen via cctalk
<cctalk at classiccmp.org> wrote:
On Thu, 27 Apr 2017, Ethan Dicks via cctalk wrote:
Thanks for the tip!
Additionally, for a testing framework to poke
signals through the
drive for debugging during the repair...
You should check out a couple of disk utility programs for the Model 4:
Super Utility 4/4P
Hyperzap 3.2C If you hook up a working drive to boot from, one of
those should help to
exercise/exorcise the bad drive. :-)
Right. I have one perfectly working mech and one that's got read problems.
I did locate the other boxes of TRS-80 floppies I was searching for (I
put them away, on a bottom shelf. Doh!) In there was a set of
(original) Dysan test floppies and an original Floppy Doctor disc
(instructions at
https://archive.org/download/Floppy_Disk_Diagnostic_1980_Dave_Stambaugh/Flo…)
I learned a couple of things last night - On the working drive, the
speed (which looked OK on the strobe disc) was clocking in from
199.8-201.2 ms per index. The other drive is around 198.9-199.8, a
bit slow but not much. The not-working drive passes the "A" test
(write protect, status bits and index strobe), but fails the "B" test
claiming there is no data received from the drive. I did swap the
boards between the two mechs. The problems stay with the suspect
mech, and the good mech works with either boardset. Since the index
detector and the motors are good in the bad drive, I'm left with the
hypothesis that the head has failed. Moving a scope upstairs will
likely be highly informative.
So that's where I'm at - the drive was working before the power was
jiggled and now it's not and it doesn't appear to be the electronics.
Thanks for all the responses.
-ethan
10:18 a.m.
On 04/28/2017 10:01 AM, Ethan Dicks via cctalk wrote:
So that's where I'm at - the drive was working
before the power was
jiggled and now it's not and it doesn't appear to be the
electronics.
Make sure that the head is making good physical contact with the medium.
Sometimes the head suspension can get damaged and lose contact.
Use an ohmmeter to check for continuity in the head windings.
--Chuck
12:35 p.m.
On Fri, Apr 28, 2017 at 1:18 PM, Chuck Guzis via cctalk
<cctalk at classiccmp.org> wrote:
On 04/28/2017 10:01 AM, Ethan Dicks via cctalk wrote:
I will check for that, but it was working minutes before I bumped the
power strip and it stopped working immediately after (it was the boot
drive - the machine failed to reboot after the momentary power loss
and the diskette was not damaged - the same diskette boots in the
other TM-100 when it's strapped as :0/DS0.
So that's where I'm at - the drive was
working before the power was
jiggled and now it's not and it doesn't appear to be the
electronics.
Make sure that the head is making good physical contact with the medium.
Sometimes the head suspension can get damaged and lose contact. Use an ohmmeter to check for continuity in the head
windings.
Yes. That was my plan. I have a dead 1541 because the read head is
open, so I've seen it before.
I accidentally ran a drive test with the head unplugged - same results
as when plugged in, so I'm really leaning towards a toasted head,
sadly.
Checking current auction prices on TM-100-2 drives... wow... they sure
got pricey. They are a long way from "how many do you want? Take
two, they're small" days. I think I have a bare drive on the
bookshelf out at my farm. Have to go look for that now.
Thanks,
-ethan
2795
days inactive
2796
days old
11 comments
6 participants
participants (6)
-
ard.p850ug1＠gmail.com -
cclist＠sydex.com -
cisin＠xenosoft.com -
ethan.dicks＠gmail.com -
imp＠bsdimp.com -
mloewen＠cpumagic.scol.pa.us