Hugo Holden
Veteran Member
I thought I would post my experiences here with this card. One I have just restored is the MDS -AD2, double density card.
Generally most of these cards were fitted with TI brand sockets, these grab the pin from side to side, rather than across the flat.
A range of corrosion issues happen due to dissimilar metals of the socket pin and the IC claw. This means that every IC needs to have both sides of every pin cleaned and the sockets need cleaning, checking and lubricating.
I start by removing all the IC's, cleaning the sockets with CO contact cleaner. Then I spray some Inox mx-3 into a small container. I make a tool, using a pin removed from a defunct IC , soldered to a wire handle a couple of inches long. I dip that into the mx-3 and introduce that into the socket claws one at a time, to feel the tension of the claw. (Sometimes the claws have been damaged by rough insertion of IC's previously). This has two uses; it checks the socket claw is normal and it introduces a small amount of the mx-3 lubricant. This takes a while as there are over about 650 socket claws to check.
Then after that I start on the IC pins which need to each be cleaned on the two side surfaces which contact the sock claws. The pins are different on different brand IC's. The TI IC's are silver plated steel. Some other IC's , such as Signetics types have thick tin coated leads, or a pre-soldered surface, others, like Motorola have a satin tin plating. In many cases a chemical reaction has occurred and a line of dark grey or black oxide is seen where the claw has been touching the side of the IC pin. It must be removed. I have found it better to simply use a folded strip of 1200 grade paper to clean this off. But both sides of the pins need doing and with the over 650 pins to do, that is over 1200 cleaning procedures. I do it with bright light and high magnification. Also for the TI IC's, I clean the black silver oxide off the flat sides of the pins as well, to improve the appearance. It takes about 3 hours to "process" all the IC pins so they are oxide free on the surfaces which contact the socket claws.
In addition where many of the IC's were initially fitted, they bent the pins at the junction of the thin and wider part of the pin, not ideal as the pin enters the socket at an angle and sometimes even scroll up in the socket housing. So I straighten this bend and bend the pins instead at the junction of the IC body a little, by rotating the IC with the pins against a flat surface, so that the pin spacing is correct for the socket and the pins enter the socket perpendicularly.
Once the pins are cleaned up on all the ICs, I wash them with CO contact cleaner and apply mx-3 to the pins, the IC's can then be re-fitted, being confident that every pin, on every IC has a good connection to known good socket claws. And the chance of it cropping up with intermittent problems are substantially reduced. But its a lot of work, though I think probably less work than trying to find an intermittent fault due to bad IC pin-socket connections.
The 5V regulator (closest to the edge connector) on this card runs extremely hot. Even with the input voltage exactly on 8V and no higher, it is far too hot to touch. The pcb gets discolored by heating (see attached photo).
On two of these cards now I have added some additional heat sinking with a good amount of thermal paste on both surfaces of the added heatsink. This helps, but on my computer I run it from a variac s that the 8V supply never gets too high, which helps, but it needs to be 7.5V at least for the 5V regulators to function properly.
Luckily, after I gave this N* card the above treatment, it worked first pop. Even if it had worked, without these efforts, most likely it would have been unreliable. Probably, the least reliable components remaining would be the tantalum capacitors.
When the N* disk controller card came out it was said that they used "A lot of clever and unorthodox hardware tricks" (editors of 73 magazine). They used memory mapped I/O rather than 8080 I/O ports. This makes the disk controller look like a 1k block of variable speed ROM to the system software and disk commands decoded from data not sent out from the accumulator, but instead from the contents of the address bus. Status and disk data are given to the CPU on the memory data in lines, as if the data was retrieved from ROM, and when the data is not fast enough, wait states are introduced. So the card doesn't tie up any of the computer's I/O ports. The more I learn about these cards, the more I like them. It seems that at the time is was a very economical way to enter the world of floppy disk storage with S-100 computers.
Generally most of these cards were fitted with TI brand sockets, these grab the pin from side to side, rather than across the flat.
A range of corrosion issues happen due to dissimilar metals of the socket pin and the IC claw. This means that every IC needs to have both sides of every pin cleaned and the sockets need cleaning, checking and lubricating.
I start by removing all the IC's, cleaning the sockets with CO contact cleaner. Then I spray some Inox mx-3 into a small container. I make a tool, using a pin removed from a defunct IC , soldered to a wire handle a couple of inches long. I dip that into the mx-3 and introduce that into the socket claws one at a time, to feel the tension of the claw. (Sometimes the claws have been damaged by rough insertion of IC's previously). This has two uses; it checks the socket claw is normal and it introduces a small amount of the mx-3 lubricant. This takes a while as there are over about 650 socket claws to check.
Then after that I start on the IC pins which need to each be cleaned on the two side surfaces which contact the sock claws. The pins are different on different brand IC's. The TI IC's are silver plated steel. Some other IC's , such as Signetics types have thick tin coated leads, or a pre-soldered surface, others, like Motorola have a satin tin plating. In many cases a chemical reaction has occurred and a line of dark grey or black oxide is seen where the claw has been touching the side of the IC pin. It must be removed. I have found it better to simply use a folded strip of 1200 grade paper to clean this off. But both sides of the pins need doing and with the over 650 pins to do, that is over 1200 cleaning procedures. I do it with bright light and high magnification. Also for the TI IC's, I clean the black silver oxide off the flat sides of the pins as well, to improve the appearance. It takes about 3 hours to "process" all the IC pins so they are oxide free on the surfaces which contact the socket claws.
In addition where many of the IC's were initially fitted, they bent the pins at the junction of the thin and wider part of the pin, not ideal as the pin enters the socket at an angle and sometimes even scroll up in the socket housing. So I straighten this bend and bend the pins instead at the junction of the IC body a little, by rotating the IC with the pins against a flat surface, so that the pin spacing is correct for the socket and the pins enter the socket perpendicularly.
Once the pins are cleaned up on all the ICs, I wash them with CO contact cleaner and apply mx-3 to the pins, the IC's can then be re-fitted, being confident that every pin, on every IC has a good connection to known good socket claws. And the chance of it cropping up with intermittent problems are substantially reduced. But its a lot of work, though I think probably less work than trying to find an intermittent fault due to bad IC pin-socket connections.
The 5V regulator (closest to the edge connector) on this card runs extremely hot. Even with the input voltage exactly on 8V and no higher, it is far too hot to touch. The pcb gets discolored by heating (see attached photo).
On two of these cards now I have added some additional heat sinking with a good amount of thermal paste on both surfaces of the added heatsink. This helps, but on my computer I run it from a variac s that the 8V supply never gets too high, which helps, but it needs to be 7.5V at least for the 5V regulators to function properly.
Luckily, after I gave this N* card the above treatment, it worked first pop. Even if it had worked, without these efforts, most likely it would have been unreliable. Probably, the least reliable components remaining would be the tantalum capacitors.
When the N* disk controller card came out it was said that they used "A lot of clever and unorthodox hardware tricks" (editors of 73 magazine). They used memory mapped I/O rather than 8080 I/O ports. This makes the disk controller look like a 1k block of variable speed ROM to the system software and disk commands decoded from data not sent out from the accumulator, but instead from the contents of the address bus. Status and disk data are given to the CPU on the memory data in lines, as if the data was retrieved from ROM, and when the data is not fast enough, wait states are introduced. So the card doesn't tie up any of the computer's I/O ports. The more I learn about these cards, the more I like them. It seems that at the time is was a very economical way to enter the world of floppy disk storage with S-100 computers.
Attachments
Last edited: