PDA

View Full Version : Trying to fix a Tandon TM-100-2A



amadain
June 11th, 2010, 08:32 AM
Hi,
I'm trying to revive an XT with dual Tandon TM-100-2A floppy drives. The XT boots and I can run DOS(3.3). One Tandon works fine, if a little noisily. When the other is connected to the PC PSU the PSU refuses to start. I assume it is the protection against a short kicking in. The dud drive works fine if the main logic board of the working drive is fitted to it.

Resistance measurements, with a cheap DVM, between pins 1 & 2 - the 12V supply, I believe - of the Molex power connector of the faulty main logic board appear to show virtually zero resistance. I have a copy of the Sams TM-100 Computerfacts booklet. However my electronics knowledge and reading of schematics are not great. I suspect a shorted capacitor in the early part of the power supply circuits of the main logic board.
Am I way off base in this assumption? I would be greatful for any hints of where to look to find the problem.
T.I.A.
Simon

MikeS
June 11th, 2010, 08:54 AM
Hi,
I'm trying to revive an XT with dual Tandon TM-100-2A floppy drives. The XT boots and I can run DOS(3.3). One Tandon works fine, if a little noisily. When the other is connected to the PC PSU the PSU refuses to start. I assume it is the protection against a short kicking in. The dud drive works fine if the main logic board of the working drive is fitted to it.

Resistance measurements, with a cheap DVM, between pins 1 & 2 - the 12V supply, I believe - of the Molex power connector of the faulty main logic board appear to show virtually zero resistance. I have a copy of the Sams TM-100 Computerfacts booklet. However my electronics knowledge and reading of schematics are not great. I suspect a shorted capacitor in the early part of the power supply circuits of the main logic board.
Am I way off base in this assumption? I would be greatful for any hints of where to look to find the problem.
T.I.A.
SimonNo, sounds like you're on the right track; bad tantalums do seem to be the most common cause of shorts these days. There are various ways to find the culprit, but the easiest and safest is probably just to follow the 12V distribution and remove all the caps along the way one by one since there aren't that many; there's also a switching transistor somewhere IIRC. See the similar thread for pointers to more documentation.

You could of course just connect an unprotected source of +12V and wait for the explosion... ;-)

amadain
June 11th, 2010, 09:12 AM
Hi MikeS
Tempted as I am by your second suggested approach to the problem, I think I will go with the first.
Simon

MikeS
June 11th, 2010, 09:15 AM
Hi MikeS
Tempted as I am by your second suggested approach to the problem, I think I will go with the first.
SimonWell, although it's usually unintentional it's actually quite effective, as Tez and others can attest ;-)

tezza
June 11th, 2010, 11:52 AM
Hi MikeS
Tempted as I am by your second suggested approach to the problem, I think I will go with the first.
Simon

Yes, sometimes you can get the explosion AND not immediately find the short as in my little adventure here: :)
http://www.classic-computers.org.nz/blog/2009-03-25-other-apple-disk-drive-repairs.htm

The good thing is you have a duplicate working board for comparison so even without a schematic you can trace the fault from the power pins as Mike suggests (and as I did above).

Tez

paul
June 11th, 2010, 12:40 PM
Start with the circuit on page 17. J2(1) is the incoming 12 V, so check (remove) C35, C36, and C43 first, then C1A on the small servo board (or just unplug the PCB.)

I've never tried it but it must be possible to locate a shorted cap by warmth with a carefully-applied current-limited power supply set to a low voltage.

Chuck(G)
June 11th, 2010, 01:55 PM
Given that large tantalum caps age ungracefully, I'd probably unsolder all of the ones on the +12, starting with the largest values, checking for a short after each one. And then I'd replace the ones I've removed with new caps--they're not that expensive.

modem7
June 11th, 2010, 03:16 PM
I've never tried it but it must be possible to locate a shorted cap by warmth with a carefully-applied current-limited power supply set to a low voltage.
That reminded me of a similar technique I learned of donkeys years ago, but have never used. I've seen others use it for 'troublesome' shorts. A regulated power supply with user-adjustable current limiting. Set power supply (disconnected) to the nominal voltage of the shorted line (or lower). Set current limiting control to minimum (minimum current). Connect shorted line. Slowly adjust the current-limiting control so as to slowly increase the amount of current that the power supply provides. Eventually the shorted component blows open.
There is a risk though that the current gets large enough to blow open a track, and on a multi-layer board, that's a big risk to take.

Another technique, which I've tried and not had success with, is to use a low-ohms meter. Because the tracks have some resistance, the theory is that the closer you get to the short, the lower the resistance you will measure. Requires a meter that is designed to measure very low values of resistance.

Chuck(G)
June 11th, 2010, 05:19 PM
PCB short circuit sniffers are a lucrative business. Low-voltage op-amp circuits like this one (http://www.edn.com/article/473626-Quickly_find_pc_board_shorts_with_low_cost_tracer_ technique.php) are sometimes used. Another way is to inject an audio frequency between the shorted traces and then "sniff" with a probe attached to a high-gain amplifier; the advantage is that you don't have to make good electrical contact with your probe.

If you've got a short between power and ground on a multilayer board where the power and ground planes are the innermost layers, most of these approaches don't work very well because of the low resistance of the planes.

amadain
June 12th, 2010, 02:06 AM
Hi all,
Many thanks for all the input. I have ordered some replacement caps. I will wait until I have those to hand before attempting to track down the dud. I will let you know how I get on.

Again thanks to all - I'm a little overwhelmed by your generosity.

Simon

amadain
June 14th, 2010, 04:12 AM
Hi all,
Got the capacitors this morning and replaced C36 on the logic board, a 4.7uF tantalum. The original had zero resistance when removed from the board. Luckily spotted that they had a polarity before I fitted the new one! Now the power seems o'k. The drive spins and the activity led lights. However the heads do not appear to seek. Further investigation obviously needed.
Simon

MikeS
June 14th, 2010, 04:02 PM
Hi all,
Got the capacitors this morning and replaced C36 on the logic board, a 4.7uF tantalum. The original had zero resistance when removed from the board. Luckily spotted that they had a polarity before I fitted the new one! Now the power seems o'k. The drive spins and the activity led lights. However the heads do not appear to seek. Further investigation obviously needed.
SimonIt works properly when you put the other board on it, and the jumper block and terminating resistor are installed identically to the working drive? You want to isolate first whether it's the motor, the board (and the jumper & terminator) or the cable.

If it is on the board it shouldn't be too hard to trace since the schematics are available; the control signals are all open-collector so you can ground them without any problems (they should normally all be at +5V). Grounding pin 12 should select the drive and grounding pin 20 should step the heads (pin 18 determines the direction); if you happen to have a dual-mode floppy cable with both edge-connectors and header plugs that makes it convenient to access the signals with little jumper wires. Check that the +12V does indeed make it to the motor driver.

I've never yet seen a stepper motor go bad, so hopefully it'll be a simple repair. Good luck!

Chuck(G)
June 14th, 2010, 05:07 PM
I've never yet seen a stepper motor go bad, so hopefully it'll be a simple repair. Good luck!

I've only seen it once, on a Calcomp 103 8" drive--the steppers on those ran hot.

However, on the TM100 drives, ICs 4E and 4D (SN75462, 8 pin DIP) are common casualties. Fortunately, they're still easy to get.

amadain
June 15th, 2010, 04:37 AM
Thanks MikeS and Chuck(G),
Yes everything is fine if I just swap in the working main board from the good drive, using the same terminating resistor and jumper block setup. With the good board the drive passes. without error, the Adv. Diagnostics (v 2.25) floppy test, including multiple iterations of the test. So I'm assuming the problem on the dud main logic board, or is there something I'm missing? When I boot the motherboard the drive led lights, and spins the floppy, as normal but there is no sound of the heads doing a seek. If I push the heads away from their rest position, reassemble, and reboot. they do not move. I assume they should move to find track 0.
I'm hoping that, like you said MikeS it will be a simple repair. I'm in deepest rural France at the moment and total facilities consist of a soldering iron, or two, some desoldering braid, and a small DMM. However I have some other bits & pieces on order,
Simon

per
June 15th, 2010, 08:57 AM
You may need a logic probe or an osclioscope if it's the logic board that is failing. The Sam's computerfacts for this drive* should contain expected osclioscope output and schematics.

(* Easy to find on Google)

Dwight Elvey
June 15th, 2010, 09:19 AM
PCB short circuit sniffers are a lucrative business. Low-voltage op-amp circuits like this one (http://www.edn.com/article/473626-Quickly_find_pc_board_shorts_with_low_cost_tracer_ technique.php) are sometimes used. Another way is to inject an audio frequency between the shorted traces and then "sniff" with a probe attached to a high-gain amplifier; the advantage is that you don't have to make good electrical contact with your probe.

If you've got a short between power and ground on a multilayer board where the power and ground planes are the innermost layers, most of these approaches don't work very well because of the low resistance of the planes.

Hi
My method of using the plane as a voltage divider works with traces as well as planes ( just increase the current
a little ). It does work best with a 5 digit meter but I've had success with more current and a 3 or 4 digit.
With a plane, one connects the supply at two opposite corners and then marks the 0V line. Then move the supply to
the 90 degrees corners and form a new line. Where both lines cross if where the short is.
Dwight

MikeS
June 15th, 2010, 10:49 AM
You may need a logic probe or an osclioscope if it's the logic board that is failing. The Sam's computerfacts for this drive* should contain expected osclioscope output and schematics.

(* Easy to find on Google)Nice to have of course, but a 'scope won't help much more than just a multimeter or even just a (current-limited) LED when finding a relatively simple static logic problem like this. Just ground pin 12 to permanently select the drive and then follow the signal through a couple of gates and flipflops to the motor drive lines while you repeatedly ground pin 20.

As I mentioned, there are links and pointers to most of the available TM-100 documentation in this parallel thread:
http://www.vintage-computer.com/vcforum/showthread.php?20881-Tandon-TM-100-1-TM-100-2-Service-manuals-and-tech-info

Chuck(G)
June 15th, 2010, 12:11 PM
Hi
My method of using the plane as a voltage divider works with traces as well as planes ( just increase the current
a little ). It does work best with a 5 digit meter but I've had success with more current and a 3 or 4 digit.
With a plane, one connects the supply at two opposite corners and then marks the 0V line. Then move the supply to
the 90 degrees corners and form a new line. Where both lines cross if where the short is.
Dwight

Thanks, Dwight! I never would have thought that there'd be enough resistance in power planes
with. a component short to produce enough of a difference in emf to show this.

This could come in very handy.