SMT1000 with RBC7 Batteries / Poor Runtime (RESOLVED)

[bimmerdriver]

I have an SMT1000 with an APC9630 NMC. I use this UPS to allow my computers to shut down cleanly in the event of a power failure. The normal load is approximately 425-460 W / 510-540 VA at 96% efficiency. When I first purchased this UPS, at a 400 W load, the runtime was approximately 15-20 minutes, plenty of time for an orderly shutdown. According to the chart on the APC website, the runtime at this load should be ~15 minutes, so it was working within specification.

Recently, the runtime started to decrease. It was reporting 7 minutes of runtime. Then it started reporting that one or both of the batteries were failed. Then it started reporting 3 minutes of runtime. Then it started reporting that the UPS did not have sufficient capacity for the load and that it would shut down immediately in the event of a power interruption. When it ran the self-test, the output power switched off. Finally, it was reporting 0 minutes of runtime. Obviously there was a problem, presumably with the batteries. Since the batteries are a few years old, I decided to replace them.

I understand APC used to offer an SMT1000XL, which had the RBC7 batteries, like the SMT1500. I saw a youtube video of a guy who installed these batteries in a regular SMT1000. The UPS worked fine and had extended runtime. Since the RBC7 batteries have quite a bit more capacity than the RBC6 batteries (17 Ah, compared to 11 Ah), I bought a set of XNB SN12020 batteries and installed them. These batteries are rated at 20 Ah, so they should have even better runtime than the APC RBC7 batteries. (Just to be clear, the SN12020 batteries are not a direct fit. An adapter is required.)

After I installed the batteries, I let the UPS charge the batteries overnight, then I did a runtime calibration using a 450 W load. At this load, the runtime of the SMT1500 should be 27-28 minutes. However, the calibration only ran for a few minutes and afterwards the runtime was only around 3-4 minutes, worse than it used to be with RBC6 batteries at that load. Since this runtime is nowhere what it should be, I removed the batteries and charged them externally at 14.4V, then left them to float for 24 hours at 13.6V. Then I reinstalled them and another runtime calibration using a 70 W load. The calibration ran for 25 minutes and discharged the batteries to less than 20%. This discharge rate corresponds approximately 400 W load, so something is not right.

The supplier claims the batteries are fine, but I can’t think of any reason other than the batteries for the poor runtime. Other than the poor runtime, the UPS isn’t giving any strong indications that it has a problem. The efficiency at 440 W load is 96%. This is on the low side, according to the graph on the APC website. At full charge it’s reporting that the battery voltage is 27 V, corresponding to 13.5 V per battery, which is reasonable.

I’m wondering if anyone has an idea what could be causing the poor runtime, other than the batteries being defective. If anyone has tried installing RBC7 batteries in an SMT1000, what kind of result did you get?

[Brad_C]

On 1/3/2020 8:20 AM, Daryl said:

I’m wondering if anyone has an idea what could be causing the poor runtime, other than the batteries being defective.

That's an easy one. The calibration routines in the UPS don't have much more leeway than needed to allow for tolerance in the as-supplied batteries. You've significantly increased the capacity of the battery bank but haven't told the UPS it has more battery, so the calibration constants will be maxed out.

On the older SmartUPS you could tell them it had external batteries even if it wasn't an XL model. On the newer SmartUPS you could tell it about external batteries by breaking into the comms link between the UPS processor and the front panel/interface processor, but that was a bit of a process. Don't have any idea if it's possible on the new-new UPS.

The other thing is a runtime calibration requires a load somewhat > 30% of the rated VA value. 70W load on an 1000VA UPS doesn't come close, so that was never going to work. A value somewhere around 50% seems to work nicely.

[bimmerdriver]

Thank you for your reply. I understand that the UPS would reasonably assume the batteries are RBC6. So if that was the case, the runtime would be the same as before, which they are not. The runtime is nowhere near the factory specification. I also would have thought that when the calibration is done, the UPS would see that the batteries are performing better and adjust accordingly. I can see the second assumption might not be valid, but I don't understand why the larger batteries aren't giving at least the same runtime as the RBC6 batteries.

[Brad_C]

On 2/3/2020 11:19 PM, Daryl said:

I don't understand why the larger batteries aren't giving at least the same runtime as the RBC6 batteries.

Because the calibration routine is fussy. On an older unit the trick would be to manually reset the battery constant but you can't do that on a newer unit. One thing worth trying is put a ~50% constant load on it (like a heater or something resistive), pull the mains and let it run until it shuts down completely. I'd use a 500W flood light for a 1KVA unit.

[bimmerdriver · Accepted solution]

On 3/2/2020 9:33 PM, Brad said:

Because the calibration routine is fussy. On an older unit the trick would be to manually reset the battery constant but you can't do that on a newer unit. One thing worth trying is put a ~50% constant load on it (like a heater or something resistive), pull the mains and let it run until it shuts down completely. I'd use a 500W flood light for a 1KVA unit.

I took the batteries back to the supplier and had them tested. One failed, so it was replaced.

Last night, I finally got around to making another attempt at calibrating these batteries. I used lights, adding up to 350 W, which is the minimum that can be used for calibration (50%). The PF was exactly 1, according to a Kill-A-Watt. Starting from fully charged batteries, the calibration ran for 44 minutes and ended up at 15% remaining charge. According to the runtime graph for the SMT1500, the runtime is supposed to be 40.7 minutes at 350 W, so I'm happy with 44 minutes. (Not sure what % remaining charge is used for the runtime graph.)

After externally charging the batteries and reinstalling them, with 434 W / 500 VA load, the runtime is 18 minutes. That's good enough for me.

Based on these results, the UPS calibration does seem to recognize the larger battery, so for anyone with basic skills, this is a pretty easy modification. However, the usual disclaimers apply... YMMV.

[Anonymous user]

Hello! 

How had your SMT1000 held up with upgraded batteries after a year of use? 

I just acquired a SMT1000 and am looking to upgrade my batteries to what is inside a SMT1500. (7.5ah vs 9ah)

Any advice? Thanks!!

[bimmerdriver]

The UPS has worked fine since I replaced the batteries. I've only had a few short power outages, but it's still showing the same runtime as before. If you have an SMT1000 and you need to replace the batteries, I can't see any reason why not to put in the highest capacity batteries that will fit. Just be sure to carefully calibrate it with a resistive load.