RickNel
Veteran Member
Following the previous post on my Catheter Project, I got advice in the Vintage Hardware area of this Forum on how to get an AT controller (mounted in a PC) to recognise floppy drives that are hardware-configured as drive A (DS0 in Tandon docs).
The problem is that the PC-AT FDD controller standard anticipates that all drives are hardware-configured as B, then routes one of them to the controller's port for A by having the famous "twist" in the FDD 34-pin cable. The twist transposes pins 12:14 and 10:16. A flat cable without twist, such as used in most pre-AT, will only recognise the drive that is configured as B.
Mike Stein suggested a special cable to reconcile the differences in signal usage between the AT and Shugart-standard controllers. The chart below shows the patching that resolves this. (It seems you need to double-click on the image to see it at proper size - I don't know why)
View attachment 6313
Note that the PC-AT controller identifies Dsk A at pin 14, where the Shugart pinout has DS0 (Dsk A) at pin 10 and uses pin 14 for DS2 (notionally disk C).
Using a cable patched this way, a PC-AT controller will recognise floppy drives hardware configured as DS0 (A) and DS1 (B), which is what I wanted for the purposes of the Catheter setup, because it maintains the hardware configuration compatible with their (normal) usage with a S-100 controller. The DPDT switch that swaps the disk identities (A-B) is no longer necessary for normal Catheter use, but does no harm and can still be used as a hardware faultfinding cross-check.
CAUTIONARY TALE on HARDWARE MODS.
I fumbled to this good situation via several hardware blunders that caused confusion to me and to my advisers on the forum!
1. In building the "Mike's Twist" cable, I accidentally inverted the 34-pin IDC connector at one end, thus delivering every signal except 12 to the wrong pin. The result was a "coffee-grinder" effect that had the heads vibrating backwards and forwards until I cut the power. After that, neither drive would respond in any situation.
Signal failure was tracked back as far as a 7407 open-collector buffer at 3E on the TM-100 controller board. There was nothing between this and the edge-connector except the 8-way DIL switch at 1E that controls disk selection.
I replaced the 7407s after a long wait to get them. Still nothing. I checked the DIL switch at 1E again. The switch was fine, but I found the socket was connecting intermittently with the switch at pin 2 (DS0). Desoldering the socket to replace it, I found the problem.
The sockets supplied on the TM-100 controller boards are low-quality enclosed plastic boxes with single-sided contact blades inside. I made the original switch-exporter for the Catheter using header pins (see previous post) in socket pins 2,3 and 15. These header pins are thicker than the pins of the DIL switch that previously used that socket. Insertion of those thicker pins was enough to bend the contact blades back about .5mm. The extra shaking (and perhaps some heat) associated with the "coffee grinder" episode described above seems to have weakened that contact further. When I put the DIL switch back in the socket for diagnosis, pin 2 made no contact at all.
So I replaced the sockets at 1E on both drives with more robust, double-sided contact sockets.
I also remade the switch-exporting headers by soldering the 3 wires into pins 2,3 and 7 of 8-pin machined sockets. This gives a much stronger contact in the sockets at 1E, with more mechanical stability.
My fumbling was not yet over, however.
The 7407 buffer at 3E uses a pull-up voltage supplied by 5vdc through a 150 resistor at R14. On the schematics, this looks as if it is adjacent to 3E, but in fact 3E get this voltage through a trace on the component side of the TM-100 controller board that runs from pin 1 of 1E (the switch block), under 1E, 2E and 3E.
According to Murphy's Law (what can go wrong, will go wrong), in desoldering and removing the old sockets at 1E, I had somehow broken that trace under 1E on one of my pair of TM-100s. As a result, 3E was not getting the pull-up voltage. I only discovered this after soldering in the new socket. Rather than do another 32 desolder/solder operations to lift the new socket and fix the trace, I ran a wire direct from R14 across the solder-side of the controller board to 3E.
Fumbles corrected and lessons learned.
PS. Anyone who reads this far might be interested in a comment on the SAMS Computerfacts TM-100 documentation, available online. It's a great visual format and trouble-shooting guide, and helped a lot with signal tracing, BUT you need to double-check its accuracy against Tandon's own schematics. In the area I was using for this problem, the 7407 buffer at 3E was incorrectly drawn as an inverter, which it is not. This confused me for a while until I cross-checked the 7407 datasheet.
The problem is that the PC-AT FDD controller standard anticipates that all drives are hardware-configured as B, then routes one of them to the controller's port for A by having the famous "twist" in the FDD 34-pin cable. The twist transposes pins 12:14 and 10:16. A flat cable without twist, such as used in most pre-AT, will only recognise the drive that is configured as B.
Mike Stein suggested a special cable to reconcile the differences in signal usage between the AT and Shugart-standard controllers. The chart below shows the patching that resolves this. (It seems you need to double-click on the image to see it at proper size - I don't know why)
View attachment 6313
Note that the PC-AT controller identifies Dsk A at pin 14, where the Shugart pinout has DS0 (Dsk A) at pin 10 and uses pin 14 for DS2 (notionally disk C).
Using a cable patched this way, a PC-AT controller will recognise floppy drives hardware configured as DS0 (A) and DS1 (B), which is what I wanted for the purposes of the Catheter setup, because it maintains the hardware configuration compatible with their (normal) usage with a S-100 controller. The DPDT switch that swaps the disk identities (A-B) is no longer necessary for normal Catheter use, but does no harm and can still be used as a hardware faultfinding cross-check.
CAUTIONARY TALE on HARDWARE MODS.
I fumbled to this good situation via several hardware blunders that caused confusion to me and to my advisers on the forum!
1. In building the "Mike's Twist" cable, I accidentally inverted the 34-pin IDC connector at one end, thus delivering every signal except 12 to the wrong pin. The result was a "coffee-grinder" effect that had the heads vibrating backwards and forwards until I cut the power. After that, neither drive would respond in any situation.
Signal failure was tracked back as far as a 7407 open-collector buffer at 3E on the TM-100 controller board. There was nothing between this and the edge-connector except the 8-way DIL switch at 1E that controls disk selection.
I replaced the 7407s after a long wait to get them. Still nothing. I checked the DIL switch at 1E again. The switch was fine, but I found the socket was connecting intermittently with the switch at pin 2 (DS0). Desoldering the socket to replace it, I found the problem.
The sockets supplied on the TM-100 controller boards are low-quality enclosed plastic boxes with single-sided contact blades inside. I made the original switch-exporter for the Catheter using header pins (see previous post) in socket pins 2,3 and 15. These header pins are thicker than the pins of the DIL switch that previously used that socket. Insertion of those thicker pins was enough to bend the contact blades back about .5mm. The extra shaking (and perhaps some heat) associated with the "coffee grinder" episode described above seems to have weakened that contact further. When I put the DIL switch back in the socket for diagnosis, pin 2 made no contact at all.
So I replaced the sockets at 1E on both drives with more robust, double-sided contact sockets.
I also remade the switch-exporting headers by soldering the 3 wires into pins 2,3 and 7 of 8-pin machined sockets. This gives a much stronger contact in the sockets at 1E, with more mechanical stability.
My fumbling was not yet over, however.
The 7407 buffer at 3E uses a pull-up voltage supplied by 5vdc through a 150 resistor at R14. On the schematics, this looks as if it is adjacent to 3E, but in fact 3E get this voltage through a trace on the component side of the TM-100 controller board that runs from pin 1 of 1E (the switch block), under 1E, 2E and 3E.
According to Murphy's Law (what can go wrong, will go wrong), in desoldering and removing the old sockets at 1E, I had somehow broken that trace under 1E on one of my pair of TM-100s. As a result, 3E was not getting the pull-up voltage. I only discovered this after soldering in the new socket. Rather than do another 32 desolder/solder operations to lift the new socket and fix the trace, I ran a wire direct from R14 across the solder-side of the controller board to 3E.
Fumbles corrected and lessons learned.
PS. Anyone who reads this far might be interested in a comment on the SAMS Computerfacts TM-100 documentation, available online. It's a great visual format and trouble-shooting guide, and helped a lot with signal tracing, BUT you need to double-check its accuracy against Tandon's own schematics. In the area I was using for this problem, the 7407 buffer at 3E was incorrectly drawn as an inverter, which it is not. This confused me for a while until I cross-checked the 7407 datasheet.