PDA

View Full Version : North Star Single Density Controller and Teac Drives



deramp5113
November 11th, 2015, 05:32 PM
I keep a number of half-height Teac 5.25" drives around, both 96 tpi and 48 tpi, as a reliable reference point when working on vintage computers and older drives. While full height drives always seem to need some work, these Teac drives are remarkably consistent and reliable.

However...

I have one situation in which the Teac drives aren't reliable. Writes fail most of the time when used with a North Star single density controller. They work fine with a single density MITS controller, with the DD North Star controller in the single density mode, but writes are unreliable with the single density North Star controller. Anyone else experienced this?

Mike

Chuck(G)
November 11th, 2015, 05:57 PM
Hard-sector floppies?

deramp5113
November 11th, 2015, 06:07 PM
Yes, the standard 10 sector floppies used with North Star controllers. Pretty much every machine I have uses a hard sector controller!

Mike

Chuck(G)
November 11th, 2015, 09:19 PM
Have you scoped the index/sector pulse WRT pulse width? Do you have a logic analyzer to capture one of the written sectors? A card such as a Catweasel should also give you the same information. That may give you a hint. Later HH Teacs have a bit more "smarts" in the drive that may be getting in the way.

deramp5113
November 12th, 2015, 06:32 AM
Have you scoped the index/sector pulse WRT pulse width?

The Teac drives work fine when reading disks, and the same drives work fine with numerous other types of hard sector controllers, so I don't believe there is an issue with re-shaping of the index/sector pulses. Also, the problem is identical with or without my Virtual Sector Generator generator in line, and I know the index/sector pulses from VSG are valid for the controller.

The problem is specifically related to the write process, and it's not the same for all Teac drives. Some drives fail all the time, and others fail just some of the time. I can change the rate of failure by switching to different power supplies for the drive. This all tells me something about the write process is on the "hairy edge" of working or failing since such minor variations affect the failure rate.

Presently, I suspect the problem is the write pulse width used by the North Star single density controller. The clock and data pulse width is exactly 500ns as set up by a 2 mhz clock and digital counters. Every other controller I've looked at - including the North Star DD controller - has pulse widths of 700ns or longer. I've dug into the schematic and figured out a simple board mod (just lifting one pin out of the socket) that will extend the pulse width to 1000ns. I will be able to try this later today.

Mike

Chuck(G)
November 12th, 2015, 08:39 AM
It wouldn't surprise me--I've used Teac drives to read Northstar floppies without issue. Fortunately, the FD55 series is well-documented--you should be able to find an OEM manual for one that calls out the specs. But then, at this late stage, I'm mostly concerned with reading stuff, not writing it.

deramp5113
November 12th, 2015, 10:59 AM
Turns out the Teac 55Gxxx and 55Bxx drives don't like the short, 500ns clock/data pulses on the write data line that are generated by the single density North Star controller. I extended the pulses to 1000ns and all drives now write reliably.

The mod is simple. Lift pin 3 of IC 2C out of the socket. This makes the write pulse assert for two cycles (500ns each) instead of just one cycle. This mod assumes that pin 3 (an input to a 3-input AND) will reliably look like a "1" while floating. Probably a safe enough bet for hobby work. To be safer, after lifting pin 3 out of it's socket, tack it to pin 4.

Mike