Caution with Eaton 5SC UPS models - live line voltages on the battery!

[cuisinart]

Hi group, this is a new one for me, and I'm posting to see what youall think about it. Safe? Concerning? A design flaw?

I was recently working on battery troubleshooting a 5SC series UPS - 5SC750. While taking some signals measurements I was shocked to find that the battery was AT LIVE LINE LEVELS while the system was in normal operation.
Typically one assumes that the battery in UPSs that don’t have live voltage warning labels are at most at their own potential with respect to ground - 24 - 48V and of course can source a lot of current to be cautious of. However in the 5SC750G system, a 230 V UPS, the battery terminal voltage drops to over -320V with respect to ground, and can sink AMPS of current! This seems to be a serious risk.

Just a word of caution to those out there who may be working on them, be careful! There is no sticker warning around the battery that it is reaching negative line voltages, battery cases aren’t typically designed/rated to withstand hundreds of Volts, and the surge potential isn’t just from the battery voltage but from line power.

Because the battery is housed in a grounded case, this means the walls of the battery are withstanding cyclic -340V of potential in the 240V model of UPS. 
This also means that if a user is doing tests and connects the battery while it is sitting outside the unit, they could be electrocuted if they touch the battery terminals and ground (the case).  To me, this is just nuts, but what do you think? If there's interest I could post scope plots, etc.

[bitwelder]

Thanks! I'm owner of a 5SC750i and while I'm generally not doing any maintenance (mostly: batteries replacement) with mains still plugged, at least I'll put a warning sticker on the unit.
Did you check if there is even galvanic continuity between the input terminals and the battery?

[jrmymllr]

Hi group, this is a new one for me, and I'm posting to see what youall think about it. Safe? Concerning? A design flaw?

I was recently working on battery troubleshooting a 5SC series UPS - 5SC750. While taking some signals measurements I was shocked to find that the battery was AT LIVE LINE LEVELS while the system was in normal operation.
Typically one assumes that the battery in UPSs that don’t have live voltage warning labels are at most at their own potential with respect to ground - 24 - 48V and of course can source a lot of current to be cautious of. However in the 5SC750G system, a 230 V UPS, the battery terminal voltage drops to over -320V with respect to ground, and can sink AMPS of current! This seems to be a serious risk.

Just a word of caution to those out there who may be working on them, be careful! There is no sticker warning around the battery that it is reaching negative line voltages, battery cases aren’t typically designed/rated to withstand hundreds of Volts, and the surge potential isn’t just from the battery voltage but from line power.

Because the battery is housed in a grounded case, this means the walls of the battery are withstanding cyclic -340V of potential in the 240V model of UPS. 
This also means that if a user is doing tests and connects the battery while it is sitting outside the unit, they could be electrocuted if they touch the battery terminals and ground (the case).  To me, this is just nuts, but what do you think? If there's interest I could post scope plots, etc.

Unrelated question since you have a 750. Do you get the runtime claimed, or haven't you pushed it that far? If so, is it more, or less?

[amyk]

While taking some signals measurements I was shocked

Literally or figuratively?

I'm actually not surprised; UPS are not isolation transformers. As for the battery case being subjected to that voltage, look at how thick typical mains cord insulation is. This isn't kilovolts, 340V is nothing to worry about through a few mm of plastic.

[cuisinart]

I haven't pushed it in terms of runtime.  Looking to sell this one because it's the 230V variant and needs 230V input to operate normally. Otherwise it's a nice design and highly efficient.

Regarding the shock, this is figurative, fortunately I was paying close attention to probing first before touching things and only using ground scope clips on areas that are at true ground.

There is actually a lot of isolation in this system - use of isolated transformers, use of isolating relays, and opto-isolators.  Where the design isolation 'breaks down' is at the gate drive of the low-side output phase-control FETs.  The source pin of the FETs is the reference for gate drive and rather than use opto-isolators in the gate drive control, they chose to connect a single long trace from isolated logic and control low Voltage power plane to the un-isolated HV drive low Voltage reference power plane.  This HV drive plane drops negative with respect to earth as a result of the phase control FETs' internal diodes conducting when normal power is applied to the unit. So that single trace between planes is what eventually connects to the heatsinks' pads on the 'isolated' logic and control side and makes them hot along with the batteries themselves.

[SeanB]

Saves money in manufacture, they only isolate the interface to control, which is easy to do with a few optocouplers, and a USB to serial chip with custom VID:PID to mark it as UPS specific, not a bog standard HID serial port so the computer side does not use it as serial port. Yes in most of the cheap ones that do not offer battery expansion the batteries will be connected to mains in some form, either to save wire in winding the transformer, or to simplify control circuitry of the power interface and measuring system. Very common to have mains neutral connected to one side, as that makes voltage sensing very easy, just a simple diode and resistor divider, and the MCU ADC running to sample the voltage on mains peak, with the same pin also used to trigger a zero crossing interrupt on the MCU, so it can measure mains peak voltage for regulation of the output, and to sense voltage loss fast and switch over, plus on mains return allows slewing the output switching so that going off battery does not involve a large current spike in the transformer.