5 May
2021
5 May
'21
9:06 a.m.
On 05/05/2021 12:07 AM, Chuck Guzis via cctalk wrote:
Small S/360 systems were operated from standard AC
distribution. I
don't know where the breakover was for IBM S/360; probably not for the
model 30, 40, or 75. The 195, I suspect did use an MG set.
"Real" 360's did not use MG sets. By real, I mean the ones
built with SLT technology, the discrete transistors and
diodes on 1/2" square ceramic hybrid modules. The logic was
essentially DTL. As best as I can figure, the 360/20, /30,
and /40 used some 60 Hz transformer power supplies. (Same
for the /22 and /25, later replacements for the model /30.)
I'm not so sure about the model /44, a non-microcoded and
stripped-down machine intended for process control. (No
decimal or character instructions.)
The /44 was a pretty big machine, about the same size as a
360/50.
The /50 and /65 definitely used the converter-inverter
system to supply the logic power supplies with 120 V 2500 Hz
regulated sine wave power. I would guess the model /75 used
the same scheme.
The 360/85 was a prototype of the 370/165, and was built
with IBM's own take on ECL, called MST.
It was water-cooled, and ran from 415 Hz power. The /85 was
slightly repackaged to become the
370/165, and used larger static RAMs in the cache and
control store. But, it was still quite similar.
The 360/91 and /95 were the same CPU, but the /95 had a
non-destructive readout memory system, which offered higher
speed due to not needing a writeback of just-read data.
These were built with an ECL variant, but using a packaging
technology similar to SLT. There isn't a whole lot of
detail on the hardware of these machines. They were
supercomputers, not microcoded, and with multiple functional
units, out of order execution, etc. The basic design was
later redone with MST for the model 195. These all ran off
415 Hz power, I think. most installations had massive UPSes
for them.
Jon