Square waves from UPS into transformer increase heating?

[daestrom]

Yup... The faster the step freqency the more steps per full wave
at 60 or 50 hertz ... So the more steps the less filtering required
to clean up the wave form... So you can have any freq ya want..

So has to be much higher at 60hz you would get a square wave
at 120hz you get 1 step each 45deg of rotation of the wave form
and so on.....

So, as a guess, the step frequency is a nice even harmonic of 60. So the
frequency source for the step circuits can be evenly divided and also act as
the time base for the 60 hz. One frequency to regulate/control would make
for simpler design, right?

daestrom

 

[jk]

Actually ANY standard cheap simple switchmode power supply will be
happiest when seeing a square wave input. A PF corrected one won't be,
but that is a different problem.

This is patently untrue. I'm sure you can find me devices that really do
object to so-called 'modified' sine waves, and I can imagine the UL has
enough spare time on their hands that they write up standards that are
useful for certain applications, but I can also guarantee you that there's
tons of really cheap inverters out there making pretty nasty looking
pseudo-sine waves that's used to power lots of dirt cheap equipment, and the
vast majority of the time nothing 'bad' happens.

Although you probably could convince me that statistically there might be
some impact on the life time of the equipment... but still, for many
applications, a true sinewave output inverter isn't worth the (significant)
additional cost.

---Joel Kolstad

jk

 

[DarkMatter]

Actually ANY standard cheap simple switchmode power supply will be
happiest when seeing a square wave input. A PF corrected one won't be,
but that is a different problem.

If the first thing it hits is a full wave bridge, then even DC can
be applied to its input terminals! (for the non PF units, of course)

 

[Nam Paik]

Actually ANY standard cheap simple switchmode power supply will be
happiest when seeing a square wave input. A PF corrected one won't be,
but that is a different problem.

Actually, I ran into some "cheap" 100 watt switchmode power supplies
which interpreted modified squarewave 115VAC as 230VAC. There are
some inexpensive 115/230V voltage sense & "auto-select" the voltage
doubler (automatic version of the 115/230V manual slide switch we used
to see in 80's PC's) which uses the dV/dT (voltage slope). 115VAC
modified squarewave has as high (or higher) dV/dT as 230VAC sinewave -
therefore it did not engage the voltage doubler. Result was SMPS did
not work at all (since the rectifier output was only about 170VDC
whereas it should have been 340VDC if the voltage doubler would have
been turned ON as it should have). Problem with 170VDC rectified DC
in a "cheap" switchmode power supply is that some "real cheap" designs
may not even have Low Voltage Shutdown protection (to save money, of
course) and then to produce same watts output at half voltage, current
has to double. With double current through its switching power
transistors, diodes & transformers overheating can damage (smoke) them
in a few minutes. Let's caculated the heat generated by transistor at
double currrent. Since Rds(ON) remains same, P = R times I squared -
this means 4 times heat generation !!! Looking at the wimpy heat
sinks which barely keep temperature within SOA even when current is
lower than rated maximum, do you think they have a chance when
transistor is overheating by factor of 4 ???