jeff@abnji.UUCP (jeff) (02/26/85)
[I don't believe in bug-killers, but I'll bow to peer pressure]
Let me begin my dissertation on Uninterruptible Power Supplies
with a quote from Tom Lehrer:
"Life is like a sewer. What you get out of it depends on what
you put into it"
The same is true of UPSs. Do you want +5 volts DC only,
+/-5, +/-12, or 120VAC?
I have some switching power supplies that take ~16 VAC or ~20
VDC. (To operate on 120 VAC, they use a step-down transformer and a
full wave rectifier). I figure that I can put a battery between the
rectifier and the power supply. When AC is on, the battery is charging.
If the AC drops or cuts out, power is drawn from the battery.
Rather than having the battery constantly charging (which
I believe is bad for some batteries whereas some may not have any
adverse effects) there could be a charging circuit that would
charge the battery only when needed, cut out the battery when
charged but the AC power is okay and connect the battery when
the AC drops (just like the emergency lights). The only trick
is to connect the battery fast enough for the power supply not to
drop its output.
If you want 120VAC(**), then how about this:
use the aforementioned "emergency light" to provide
uninterrupted 12 VDC to a power converter (that converts 12VDC to 120VAC -
popular with campers to run appliances from a car battery) and run
everything from the converter. Under normal circumstances, it is
wasteful to convert 120VAC to 12VDC to 120VAC again, but the noise
and spikes shouldn't get through (it'll act as a filter) and the
moment the AC goes away, the battery will continue powering the
already operating converter (no time needed to "warm up",
less likelihood of transients).
Of course, you could just BUY ONE (there are many manufacturers,
some advertise in Byte) but that's no fun :-) (but it will come with
a guarantee/warranty that it will work first time, but that's what
you're paying for!)
(**) at approximately 60 Hz, although most power supplies are not
very frequency sensitive. The output may not be a pure sine-wave,
thus the RMS voltage and total power may be greater than a sine wave.
Nice power supplies should forgive this.
Some food for thought from
Jeff 'soft and furry' Skot
at ATTIS in Somerset New Joisey
{ ihnp4 | mcnc | cbosgb } abnji ! jeffdyck@alberta.UUCP (Terry Dyck) (02/27/85)
On the low voltage input power supplies, why not put a diode in series with the batteries. They won't charge from the AC supply, but will always be on line if the power 'browns out' or quits. Because they are right after the rectifiers they will only feed power into the circuit when the voltage of the batteries exceeds the voltage of the rectifier output. There is a problem using a cheap 12 volt inverter. They put out a square wave. The duty cycle of the ac is about 50% and this will sometimes cause overheating and very poor efficiency in devices with a transformer input. There is also a problem of the very high noise associated with square waves. From experience I know this can cause very strange disk errors. I would think you would be better off using the battery approach mentioned. It would be more efficient, generate no noise, and with a diode isolator, no problem using any type of charger you require. Terry Dyck U. of Alberta ihnp4!alberta!dyck
john@hp-pcd.UUCP (john) (03/14/85)
<<< < Why not put a diode in series with the batteries? > The problem with this is that it cannot tell whether the power failure was due to an actual failure or simply turning the machine off. In one case you want to supply power and in the other you don't. You need to wire it into the power switch so that it can be disabled and should put in a trickle charger to keep it topped off. Another problem with diodes is their forward voltage drop. If you drop .6 Volts across the diode from a 6 Volt battery then you use 10% of your battery energy in the diode. This means you have to use larger batteries. John Eaton !hplabs!hp-pcd!john