Vintage IBM 3420 Reel-to-Reel Tape Drive.

Hi,

Would there be any chance of getting one of these to work, even if in an
extremely limited way, (possibly through contorted means) with a PC or
PC-like machine? I've read something about a related IBM 3803 control
unit. Would that be required or is it just for multiple drive clusters?

Yours,
Gary Hayward.

Gazza writes:

Would there be any chance of getting one of these to work, even if in an
extremely limited way, (possibly through contorted means) with a PC or
PC-like machine? I've read something about a related IBM 3803 control
unit. Would that be required or is it just for multiple drive clusters?

While there were a few special-case exceptions, the IBM mainframe
peripherals generally were designed with a model-specific interface
used to attach them to a control unit, which had the responsibility
for presenting the standard S/360 (or S/370) interface to the channels
in the CPU. None of these interfaces are directly compatible with
anything you're likely to find for a PC, if for no other reason than
the extremely tight timing tolerances in the signal sequences.

I don't recall the data flow in the 3420 logic, but essentially the
tape drives attached to mainframes had relatively little intelligence,
responding to instructions from the controller to perform mechanical
actions, and performing character-by-character transfer of data between
the controller and the tape head with little buffering.

Something else to consider is that the 3420 (at least all the 3420
systems I recall) used three-phase power, which isn't usually found
in a residential environment, and usually isn't available in offices
either, even if it is used in the office building. (I do recall
events where the idiots in the electrical shop reversed phases while
working on the building power. Hint: tape drives don't work too
well when they are equipped with pressure columns.)

In other words, it might be possible to attach a 3420 to a PC, but
it will be one hulluva big project.

Joe Morris

Joe Morris wrote:

Gazza writes:

Would there be any chance of getting one of these to work, even if in an
extremely limited way, (possibly through contorted means) with a PC or
PC-like machine? I've read something about a related IBM 3803 control
unit. Would that be required or is it just for multiple drive clusters?

While there were a few special-case exceptions, the IBM mainframe
peripherals generally were designed with a model-specific interface
used to attach them to a control unit, which had the responsibility
for presenting the standard S/360 (or S/370) interface to the channels
in the CPU. None of these interfaces are directly compatible with
anything you're likely to find for a PC, if for no other reason than
the extremely tight timing tolerances in the signal sequences.

I don't recall the data flow in the 3420 logic, but essentially the
tape drives attached to mainframes had relatively little intelligence,
responding to instructions from the controller to perform mechanical
actions, and performing character-by-character transfer of data between
the controller and the tape head with little buffering.

Something else to consider is that the 3420 (at least all the 3420
systems I recall) used three-phase power, which isn't usually found
in a residential environment, and usually isn't available in offices
either, even if it is used in the office building. (I do recall
events where the idiots in the electrical shop reversed phases while
working on the building power. Hint: tape drives don't work too
well when they are equipped with pressure columns.)

In other words, it might be possible to attach a 3420 to a PC, but
it will be one hulluva big project.

Joe Morris

Hi,

Thanks for the info.

BTW, here's the web page I saw the 3420 for sale on (quite good info.):

http://cgi.ebay.co.uk/ws/eBayISAPI.d...e=STRK:MEWA:IT

and the 3803:

http://cgi.ebay.co.uk/ws/eBayISAPI.d...4&ssPage Name

Yours,
Gary Hayward.