I’ve foolishly volunteered to take over electrical maintenance duties at the local squash club and inherited a problematic backup power system. The club has a newish Synapse branded Axpert MAX-7.2K and almost every time loadshedding hits the computers reboot or shut down. It’s this one:
I’ve set setting 3 on the inverter to “UPS” (for a claimed 15ms transfer time), and I’ve also replaced the power supply of the main computer with a Corsair RM650x, which according to reviews has an excellent hold-up time, in excess of 24ms at full load (which the modest computer will never pull):
The battery bank is 4 x Pylontech UP5000 batteries connected with 50mm^2 cables to the inverter. There doesn’t seem to be a correlation between the inverter load and whether the computers reboot when loadshedding hits or not. There are 4 courts, each drawing 1kW with the lights on for a max load of 4kW. Sometimes the computers reboot with only 1 court on, and sometimes they don’t.
The club is quite far from the road and it appears that the Eskom supply cable is undersized for the distance. With all four courts on and the inverter recharing the batteries (total load of 7kW), I’ve measured the incoming supply at 195V.
My next step is to put a double-conversion UPS between the inverter and the computers, but before I spend more of the club’s money is there anything else I can try?
I’m guessing the inverter does not meet the transfer time claim. This sounds like a job for those cheap UPSs that you get for ~R1k, no need for double conversion.
The hang time can vary a lot depending on the power draw. With a lower voltage you will also see more current being drawn which will also reduce the time.
If you add another UPS before the PC that could also negatively impact the PC as they need to sync the power wave.
At our squash club we had to remove the UPS bacause with the inverter and UPS the PC would also loose power each time loadshedding stop/start.
If you are drawing 1000W per court then I would guess you are still using normal flurecent tubes. Swop those out for LED tubes will bring it down to about 550W per court. We did that 2 years ago and it has made a huge difference. Just note you also want to use less LED tubes as they are more bright. This would be my first suggestion.
The club’s got one of these in storage that I could try out:
It claims a worst case transfer time of 10ms, which is 5ms better than the inverter.
I suspected as much. The RCT UPS mentioned above that is standing around is a modified sine wave. So what’s going to happen when loadshedding hits is that after about 10ms the clean Eskom sine wave from the inverter bypass is going to change to a modified sine wave from the small UPS, and then presumably a couple of seconds later to the sine wave(ish) output from the inverter. Hopefully that doesn’t mess the computers up too much.
I’ll give it a shot and see what happens.
They’re actually already LED, but clearly not very efficient stuff. It was done before I took over, so I take no blame.
It depends on how PFC is done, but generally there is no syncing, it just uses the incoming voltage to modulate the booster current. And they are designed to work with these UPSs (I recall there were issues with modified sine when PFC just came out, but I don’t think it’s a problem anymore?)
This is what I was thinking. If the PSU has a PFC controller they are very fussy, especially with cheap UPSs. In my use case, the solution was a cheaper, more basic power supply.
Have these bulbs cool LED bulbs in the house.
Switch them on, color 1.
Flick the switch twice quickly, color 2 … you get the idea.
What I have noticed distinctly this last week forcing the balancing, is if the batts are low, the changeover is very noticeable, and the bulbs change color.
If the batts are 100%, no change of color most of the time, times it does, suspect a “bad” Eskom switch-off or whatnot.
Also have an Online 2.4kVA UPS, bugger, just sitting there, it draws up to 100w to be “ready”.
For me, not negotiable, when a computer switches off running off an inverter, that is bad, really really bad.
Idea:
Have asked our kids to ask around for “broken” UPSs, got 4 for free … replaced the small batts, they work again.
So I was digging around in the rats nest of cables and found a tiny UPS wired between the inverter and the computer. Being newly educated that UPS and power supply incompatibilities is a thing, I removed the little UPS and in the week since there hasn’t been any rebooting of the computer. Yay!
After careful reconstruction of the timeline, it appears that what happened here is that the hold-up capacitors in the ancient computer power supply was at some point no longer able to carry the computer through the 15ms+ switchover time of the inverter, so the small UPS (with a claimed 6ms switchover) was purchased. This must have worked for a couple of years until the capacitors in the power supply had degraded to the point where even 6ms was too long. This is where I entered the picture and bought the new power supply, only to find that it didn’t help - until I found and got rid of the UPS.
