25 Nov
2014
25 Nov
'14
6:49 p.m.
On 11/25/2014 03:02 PM, William Donzelli wrote:
IBM 360's used 2500 Hz single-phase power produced
internally by a
converter-inverter unit. it made the individual power
supply modules QUITE
small. The problem is they had totally minimal energy
storage in the DC
link of the converter-inverter unit, so the machines were
extremely sensitive to
the slightest power bobble on the incoming 208 V 3-phase 60
Hz power. The
SLIGHTEST flicker seen on a fluorescent light meant the 360
crashed.
The 370's went to 415 Hz provided by an internally-mounted
motor-generator
set, this made the machines able to ride through significant
power dropouts.
At least, the KL10B power supply DID work reliably. Our
university didn't
care about the power waste, but obviously it was a big
problem for
Compuserve.
Jon
400Hz 3-phase
power also helps. That gives a ripple frequency of 2400Hz in
a fullwave "star" configuration, with only 13% ripple, no? The same holds
for early model Cybers (even before they were called "Cyber") which used a
different variety of logic.
Well, yes, of course. Which leads to the question why
DEC did not go
that route. I guess the KL10 power supply people were drunk (again).