On 11/27/22 12:54, Jon Elson via cctalk wrote:
Anybody need some 1/2" 9-track tape drives?
I have a Pertec T9640 tape drive (vacuum column, 75 IPS 800/1600 Pertec
unformatted interface). DEC sold this drive as the TU45. I also have
an MDB MLSI-TM11 Q-bus tape controller that will run this drive on
either LSI-11 or uVAX II systems. DEC does not supply a driver for this
setup on the uVAX, but I have source of the driver I used on VMS 4.7.
I also have 2 CDC Keystone 92185 drives with formatted Pertec
interface. These drives handle 1600/6250 BPI, and can start/stop at 25
IPS or stream at 25 or 75 IPS. The tape path is all air-bearing.
I am located in the St. Louis, MO area.
If anyone cares, I've been working on a Pertec tape controller design.
The initial version worked remarkably well with only a few bodge wires.
I'm assembling the respin of the design and do not anticipate any issues.
It's basically a 4x6 inch PCB with a bunch of through-hole TTL hooked to
a piggyback MCU. I've selected the STM32F407 Chinese boards as they're
inexpensive and easily available from Aliexpress, unlike a lot of the
other MCU boards which seem to have been hit by the chip shortage.
My current controller is getting a bit cranky and I don't want to have
to rely on it.
In essence, any MCU with 3 or 4 8 bit 5v tolerant GPIO ports, with one
of them being bidirectional can probably work. The MCU board that I'm
using has a battery-backed RTC as well as integrated SDIO. Interface to
the host is pretty much dumb terminal CLI-type with data transfer to the
host using good old YMODEM, using either (depending on firmware
assembly) async serial or USB 2.0 FS. Transfer uses a microSD card and
has been verified to work with cards of up to 64GB capacity. The file
system is exFAT, so this should work with Linux, Windows, etc.
I've got a bit more assembly to do and then I'll be tweaking the
firmware (the STM32F407 runs at 168 MHz and has 512K of program memory,
as well as 192K of SRAM onboard).
To date, I've tested it on the fastest drive that I have, a Fujitsu
M2444AC streamer. The 1MB/sec transfer rate doesn't appear to be a
problem even at 6250 GCR.
Rather than using open-collector outputs to drive the formatter
interface, I'm using tristate bus buffers with drive capability of 60ma.
Until the interface is initialized, the outputs float, so, in theory,
multiple controllers can share the bus. Power is supplied by a 5V 2A
"wall wart" supply.
The reverse direction is standard TTL level, terminated with 220/300
networks.
That's about all I can think of tonight. If anyone is interested, I'll
put you on my email status list. Eventually, I'd like to host the whole
thing on Github.
A special "thank you" is due Mattis Lind, who has done a super job of
PCB layout (something that gives me a raging headache). I could not
have done what you did in the time that you did it!
All the best,
Chuck