[kwlug-disc] UPS sized battery purchasing locally?

[Andrew Kohlsmith (mailing lists account)]

On Jun 1, 2014, at 7:02 PM, John Johnson <jvj at golden.net> wrote:
> The output of some UPSes is, as you noted, not a sine wave. For these types of UPSes the output s is generated by a rectangular pules in opposite polarities.
> The amplitude of the pulses is usually fixed. The width of the pulses are modified in a Pulse Width Modulated manner such that the resulting power curve approaches the nominal 110 Vac. 

Generally speaking, *no* modern UPS outputs a voltage waveform coming even remotely close to a sine wave. You used to have the old 6-step type systems but I don’t think anything like that has been manufactured for many, many years. Nowadays UPSes use high-power MOSFETs to alternately connect either the positive or negative battery terminal to the AC plug marked “battery backup” using a high-frequency pulse width modulation scheme that may or may not be modified to reduce specific harmonics typically problematic with this type of application.

Better UPSes will have load reactors or perhaps DC chokes which help to slow down the edges of the PWM waveform and “absorb” the high frequency components so the load doesn’t have to. The *current* output of almost every UPS will resemble a “fuzzy” sine wave, but the voltage waveform will be very ugly indeed.

AC motors are great filters but the older ones tend to burn out over time when driven from VFDs due to the high frequency switching components breaking down the insulation in the first few turns of the stator coils. I haven’t done much analysis on the effects of fast edge rates on PC power supply front ends but I expect that there are similar issues, which is why the good UPSes employ the heavy and expensive load reactors on their outputs.

Source: I spent 13 years in industrial power electronics designing soft starters, variable frequency drives, switched capacitor harmonic filters and other fun three phase devices in the dozen-to-couple-thousand kW range. Embedded system design is more fun when a software bug could launch the device off the wall in a fireball. :-)

> The nominal 100 Vac sometimes called 117 Vac or 120 Vac has a wide tolerance, ranging from about 90 Vac to something around 126 Vac.
> I would expect that UPSes are calibrated to supply their output at the low end of this range.

Some of them (older APC units for sure, not sure about new ones or other mfrs) allow you to bump the output up or down 5-10% to help accommodate equipment which can’t handle exactly 120VAC. This is likely due to bad assumptions made about voltage droop on long source feeds. When I worked in this area we would sometimes have to adjust things for this reason.

> Pure sine wave AC is important for some synchronous motors.

Harmonics can cause a lot of issues. I do not have direct experience with traction systems but I’ve heard that subway/railway controllers employ slightly modified algorithms which reduce or eliminate negative-torque-producing harmonics which can make the riders feel slightly ill.

-A.