Browsed by
Month: November 2020



I had one of these old herculus MDA video cards that had artifacts on the screen.. random characters in a few places and some random attributes. That usually means bad video memory. The ram chips were soldered on in this case, so I unsolder all of them. MT (Micron) 4264s.

I was talking to Clay on the phone and he laughed and said ‘All of the micron 4264s go bad!’. Pretty easy to replace, but I was curious which ones were bad and how they were failing.

I grabbed an Arduino and built up a quick DRAM tester. Pretty easy to do given the interface of the DRAMS, and these particular ones have 4ms refresh intervals.

Sure enough, 4 of the 8 work perfectly, and 4 of the 8 have bad bits. You can see in the scope shot a couple of bad bits during a read bank filled with ‘1’s. Most of the bad ones had 400-800 bad bits (out of 64k), and a few had 3-4k.

It was also pretty easy to measure the actual access time, given that these are 120ns chips. It measured about 60ns from CAS fall, which is right on given the spec of 80ns max.



Speaking of vintage hardware.. and vintage floppy disks. I got a device from UK called kyroflux that is a USB attached floppy controller interface. It is interesting in that it digitized the flux signals directly from the floppy interface and sends it to the PC over USB. This makes it possible to not only to read pretty much any format or protection method, but allows for more advanced software to try and recover sector information that is damaged.

I was able to get it up and running with a Fujitsu 360K floppy drive and read and write raw images from both DOS and TRSDOS floppies.

There is another tool called HxC that can build the flux patterns from an image file (that you can edit, add files etc), and that pattern can be written back to the disk.

A really useful tool if you have a stack of old floppies laying around you want to read from, especially if they have errors. Support for HD drives (1.2M ones), and of course all of the 3 /12 and 8″ formats.

It is interesting to look at the raw flux streams and see how tenuous floppy magnetic medium is. I have to admit I am surprised it worked as well as it did for so many years!



Hard drive formatted and installed in 1988, still works great. Lotus 123 and Word Perfect , and turbo pascal included. And the miniscribe isn’t making funny noises Clay Cowgill, Christopher Neil Bradley!



I decided I needed something interesting to decorate my office with. The 5150 is the perfect choice for the discerning engineer.

This is a mid 1984 production IBM PC, the PC that started it all. It originally came with either one or two full height floppy drives, and those look to have been replaced with the half height ones to make room for a hard disk.

The motherboard looks to be in good shape, and it is the 64-256K variety ( 1984 production ). The add on AST SixPackPlus card it typical of the era, giving you extra serial and parallel ports as well as additional memory up to 640K.

The async card (serial) appears to be original IBM 5150 vintage. The display adapter is a Paradise Systems MultiDisplay card which was a nifty adapter back in 1984 as it had both MDA and CGA outputs ( although you could only use one at a time).

I took a close look at the motherboard and it doesn’t appear that the tantalum caps have been changed.. so I’ll do that as the old ones tend to explode.

The power supply is not the original 5150 as that original power supply was only 63 watts, so not enough for a hard drive. This looks to be a Power Tronic 150W power supply used in the XT as well as the PC when upgrading to a hard disk.

The hard drive is a Microscribe 3425, 20MB ST506 interface MFM drive (similar to the classic ST225). 625 KB/sec transfer rate.

I think Clay might have a few of these buried in his lab somewhere.. and perhaps even some memory for it!