Dealing with crappy non-isolated battery chargers on UPS systems

[cte7ds]

I'm working on slapping a couple of big (120Ah) deep cycle lead acid batteries on a Eaton Powerware 5115 1000VA 230V UPS. The integrated battery charger seems to put out 27.50V and about 1 A of charging current. That's not enough for a 120 Ah 24V battery bank. So my idea was to simply put one of those MeanWell NES-350 SMPS directly in parallel with the UPS battery input and set it to 27.6V and set the current limit to around 10A by changing a magic resistor inside

So I did all that and measured that the UPS charging circuitry now was pulling about 0.2A from my MeanWell at 27.6V. Great it works, the UPS starts happily and passes self-test and is able to idle in that state for hours without any smoke.

Now the only major problem I have with this is that when the UPS is hooked to a wall outlet that has safety earth grounding, an AC voltage of approximately 150VAC appears between the battery terminals and earth. I even tried measuring with a low impedance analog multimeter to see if there is any current behind this voltage and was able to pull atleast 2mA without the voltage dropping at all, so that's a problem. I'm guessing the battery charger is not isolated from the mains input :\ Using a power cord without grounding takes care of this problem, but that's not good practice now is it.

As I have my external charger I don't really need the integrated one at all, so completely disconnecting it would be great... if only I could figure out how to achieve that..

A diode between the battery and the UPS takes care of the DC charging voltage from reaching the battery terminals but not the AC component that is the problem.

Any suggestions what I should do? I'll attach a picture of what I have so far. There's the MeanWell NES-350-27 wired in parallel with the battery input of the UPS trough a 50A current shunt board that also monitors the voltage. That all is in parallel with the nominal 24V stock battery pack wired with alligator clips just for testing :p The MeanWell is powered from 230VAC coming from inside the UPS (I just tapped to the IEC mains input connector)

Here's the teardown with pictures from inside the UPS: https://www.eevblog.com/forum/product-reviews-photos-and-discussion/teardown-eaton-powerware-5115-1000va-usb-230v

[xygor]

It looks like the meanwell supply's isolation rating from its output to its frame is 500 VAC.  You should be able to ground both the ups frame and the power supply frame and still do what you want.  Just be careful when probing and don't connect a scope probe without isolating the ups.
It's still not really too safe if the battery is exposed where one might touch it.  Your options there are either enclose the battery or operate the ups from an isolation transformer.  It's also possible there's something wrong with the ups.

[cte7ds]

It looks like the meanwell supply's isolation rating from its output to its frame is 500 VAC.  You should be able to ground both the ups frame and the power supply frame and still do what you want.  Just be careful when probing and don't connect a scope probe without isolating the ups.

Oh yeah, I didn't even realize they had such value in the specs. I'm guessing it's due to one of those small blue capacitors used for EMI reduction.

It's still not really too safe if the battery is exposed where one might touch it.  Your options there are either enclose the battery or operate the ups from an isolation transformer.  It's also possible there's something wrong with the ups.

Guess I'll be using insulated battery terminals and "Do not touch!" signs. And if I ever need to fiddle with the battery pack while the system is running I gotta wear something to insulate me from earth :p

I'm pretty sure this is not a malfunction, see that black piece of plastic going in to the slot in the metal, it is supposed to disconnect both battery terminals from the charger when user opens the battery door to replace batteries. This allows Eaton to save costs and use a non-isolated charger because the terminals aren't user accessible. Damn cheap-asses, they should know people are gonna hack external batteries with poorly insulated terminals to the product :-p

I'll attach a picture of a fun little feature I discovered on another Powerware UPS (pw5110). This one had full 230VAC on the battery terminals depending on which way you connected Neutral and Hot to the mains input (in Europe they are often allowed to be contacted either way) I was even able to light a 40W light bulb between battery terminal and Earth, so plenty of current available

[SeanB]

Pretty common on cheap UPS units. No need for optoisolatotrs except for 2 low speed units for the RS232/USB interface. If the battery is grounded then you need high speed isolators to measure the mains input voltage for dropouts and this raises the cost by a little.

[johnwa]

How much load is going to be applied to the UPS in normal operation? If it is not too much, you might consider not plugging the UPS into the mains at all, and supplying all the power from your external SMPS. There will of course be some loss of efficiency with this approach, and the inverter in the UPS is probably not designed for continuous operation, so it might blow up if you put too much load on it. However, this would permit the battery negative to be grounded, which would be much safer. (I always worry about people upgrading UPS batteries without taking the necessary precautions.)

Expanding on this idea, you might want to think about discarding the UPS altogether, and replacing it with a discrete inverter. I have found that the efficiency of UPS inverters is generally quite poor - some I measured were worse than 50% at low-medium load - so you might gain a worthwhile increase in runtime.