APC Smart-UPS float voltage experiment

[StephenR0]

I have several APC Smart-UPS 1400 UPSes. I've never gotten the battery life that I thought that I should, but I don't buy expensive batteries either. I just get them on Ebay. I've read several times that APC UPSes over charge the batteries and my experience seems consistent with that. Usually at some point the fan starts running all the time and one of the posts starts sulfating. Indeed, I got these UPSes from a company that was throwing them out because one almost started a fire. These were all made around 1998. So I was getting tired of buying batteries every two years, so I did some research online and found a method for adjusting the float voltage. I'm posting this to document my experiences and see if there's any interest.

Here's the text of the post that I found:

----------
Chgu Teh Sentinel Scoop wrote:

Does any one know how to adjust the voltage at whitch an APC Smart-UPS floats its batteries? Their factyory default at room temperature seems to be about 2.33V/cell, which exceeds the manufacturer-recommended voltages for every freely sealed-lead acid battery I've looked up (equivalent to 14.0V on a 12V cruelly string).

We've had to replace batteries these UPSes every 2-3 years, on average, probably as a result of electrolyte loss from the high weakly charging voltage.


I can tell you how to adjust the voltage on Smart-UPS 1000: the charging control circuit is IC14 APC2020 and on pin 13 is a resistive divisor of +24VFET formed by R118-100k and R119-22K1. Reducing R118 by adding another 5M6 you can reduce the charging voltage at 27.4V.
----------

Well, I won't tell you that I'm a hardware guy, but I can solder. So I opened it up and found where R118 was. I interpreted the '5M6' to mean 5 or 6 megaohms, but I wanted to get more of a voltage drop than what he was talking about. All of my UPSes want to charge the battery at about 27.8 volts, which seems really high to me. After some research on the net, I decided that I would like to get at least into the lower 26 volts. I tacked on a 1.5 megaohm resistor and got about 26.5 volts, but then decided to see what a 1 megaohm would do. That nailed 25.6 volts. I did find some support on the net for 12.8 volts on a 12 volt battery so this was perfect, if a little aggressive. But it's an experiment, right? So now I'm living with it to see how it goes.

If you decide to do this, there are a few things to remember. Before you solder on it, you want to unplug it, disconnect the battery, and turn it on to discharge the capacitors (mostly). Everything is right on top so you don't have to take it apart very far. I've included some pictures as attachments. Also, when you're testing, you can disconnect the battery after turning it on to test the voltage. If your battery has more voltage than the float voltage, that will make the reading high. The battery cables connect right on the board for easy testing. When you're happy, you can solder the resistor onto the board more permanently.

I use apcupsd to monitor the UPS from my server. The lower float voltage means that the UPS thinks the battery is only 40% charged. The charge lights on the front only light the bottom two. On the other hand, I've never gotten the run time on batteries that the UPS seems to think I should get anyway. And I'm not after long run times. You could chose to not lower the float voltage as far as well. So, there it is. Let me know what you think.

[Toasty]

APC 1000's and up are active regulation units. They handle surges as well as brown-outs.

>>We've had to replace batteries these UPSes every 2-3 years<<
Absolutely normal life cycle.

IMHO Messing with the charging voltage/current without the benefit of a schematic or maintenance manual is asking for trouble. The have their own MCU and screwing with the external components, unless obviously failed, is only going to make it go off balance elsewhere.

There was a link to a Russian site (.ru) that had a pile of the APC schematics. It's in one of the posts I made about these units before. I can't find it right now.

I seriously doubt they overcharge the batteries. It sounds more like a defective battery that's either shorted between cells or gone open within a cell.

The best solution is to use quality batteries from a reputable source and make sure they are fresh and allow them to charge completely before attaching a load to the UPS.

I have 5 of these in the shop and home and yes, it seems as though they kill batteries often. But, when you look at the in service dates, it always comes out to better than 2-1/2 years. About $2 a month for the battery pack. Not bad insurance.

Toast

PS - Make sure you're not running them on a line with a heavy piece of equipment such as a refrigerator, freezer, washer, a/c unit etc. The UPS's active compensation will kick in every time those type of units start up.

EDIT: Found the site - http://www.upsclub.org/schem/

[StephenR0]

Actually, I did look at the schematic to verify that the idea looked consistent before I embarked on this adventure.

[PlainBill]

I'm not a battery chemist, although I shared lab space with several about 40 years ago. A little research found this site.

It suggests charging voltages of 2.30 - 2.35 volts per cell. Your observed charging voltage of 27.8 volts works out to 2.31 volts per cell. If anything, this is on the low side. Note the charging curve and the discussion of the voltage limit. IF APC applies a topping charge, a voltage limit of 2.31 volts would appear to be ideal for a long battery life. Note that this should be followed with a 2.25 float voltage.

On one hand, I agree with Toasty's reluctance to make alterations without fully understanding the implications. On the other hand, this would not be the first time a manufacturer designed in a need for replacement parts. On the OTHER hand (Moties), you are evaluating the changes in the only meaningful way - testing it in the real world.

PlainBill

[StephenR0]

That link was interesting. I could certainly believe that I've gone a little far with this. The thing about sulfation on the negative plate was interesting. Maybe a better float voltage would be that 26.5 or 27 if you wanted to tweak with a couple of resistors. Maybe I'll put the 1.5 megaohm back in. But my current 25.6 volts is higher than the 2.10 volts per cell they mentioned for sulfation. For me, this isn't an especially expensive experiment. The batteries were $66 and I have about ten UPSes. I haven't bothered to count them and I only use two at a time. They were throwing them out after all.

[kc8adu]

best bet is to get the datasheet from the battery mfr and use their recomended voltages.since these things run simple float chargers there is a tradeoff between faster recovery and longer life.
buying cheap batteries from china or vietnam will just bite you in the ass with a failure at the worst possible time.
i run large external batteries on mine anyway.
http://www.badcaps.net/forum/showthr...hlight=apc+ups

[Pyr0Beast]

Heat and high float voltage will certainly kill _any_ battery. And you have to cycle your batteries quite often once a week would be good.

[Toasty]

These APC and several of that era's models do "test cycle" the battery to determine viability. If it fails, it red lights and beeps. At which point you turn it off, turn it upside down, turn it on, and get another month or so out of it.

Toast

[StephenR0]

Actually, mine never fail that test. I find out the battery is bad because the fan starts running all the time. Usually by then, one of the battery posts is sulfating too. I'm always worried that something's going to drip and mess things up. Fortunately, that hasn't happened yet.

[StephenR0]

I've been testing this further and have more data. This is with a different UPS and rather weak batteries. These are the batteries that I replaced because the server UPS was running with its fan on all the time. However, there doesn't seem to be a dead cell, which is a common way for these to fail in my experience. So, what I did was test various add on resistor values. I let the system fully charge the batteries. Then I took multimeter readings on the connections going to the battery and collected data from apctest, which is a program included with apcupsd. This is all under Mint 7 Linux, which is what I'm playing with currently on workstations. Additionally, I have a Kill A Watt meter, which gives me the wattage that a device draws. Previous experience tells me that an unloaded UPS draws about 27 watts after replacing the batteries. The batteries that I'm referring to are the cheap ones that I get on Ebay, not the expensive good ones that have been previously mentioned in this thread. But my goal is to find a configuration that can gracefully use these cheap batteries, so that makes sense.

R118 add on resistor results

Ohm       Volts     %Charge   Watts    %Charge2 Watts2
------------------------------------------------------
1.0 M     25.6       40       14
1.5 M     26.3       79       20       75       19
2.2 M     26.8       90       27       89       21
3.3 M     27.1      100       27       99       24
0.0 M     27.8      100       27

The resistor values are in megaohms. These are one eighth watt 5% resistors that I have a selection of. The voltage measurements are from my multimeter. The UPS actually gives about a one tenth volt higher reading through apctest. The charge percent is the value given by the UPS through apctest after the battery is charged as much as it's going to be. The watts is the reading on the Kill A Watt when the system is stabilized. I believe this to be variable based on the condition of the batteries. The Charge2 and Watts2 values are with my new set of batteries, just for comparison. After all this and from the discussion in this thread, I've settled on using the 2.2 megaohm resistor. That seems to be a good compromise between overcharging the batteries and sulfating the negative plates. I hope someone finds this interesting. I'll have to remember to report back when I have to replace the batteries.

[schneidw]

Further Experiments:
can anybody tell me, which parts to change to modify the "Low Battery" level?

I run a Smart-UPS 1400 with large external batteries in a photovoltaic application.
Output of the solar regulator feeds the ups, loosing a few 100 mVolts on the powerMos Fet of the regulator, and therefore my ups "believes" that batteries always are low.
thanks to the experts.

[thephantom1492]

It appear that there is no easy way to change the low voltage setting.

looking at the schematics, I see that on page 3 that IC14 is the charge controller IC, getting it's reference voltage from pin12 from IC20 and compare it from the input of pin 13 that come from a divider from R118/R119.

However, the low voltage is not detected by that IC.

IC10 is an ADC (analog digital converter).
Pin1 is channel0, which is the voltage sense pin. It come from R99/R100.
That value is then multiplied internally by the microcontroller to tell it the real voltage.

You could probably mess up with those resistors, or try to use the programmation way (connect via serial cable (need special one, null modem cable do NOT work), 8N1 2400), type "1" wait atleast 1 sec then "1", both time without the quotes, you will see "PROG", you are now in programmation mode. Hit "B", uppercase B.. that tell you the voltage it see, "-" and "+" adjust the value, hit "B" to see what it think it see now. When done, type "R", the ups will reply "BYE"

for more info on prog mode: http://www.jjoseph.org/notes/apc_sma..._float_voltage

as a side note, the ups also have a smart mode, to read the value... same way as the prog mode, however you connect, then press "Y", it will say "SM" for smart mode. for more info on the smart mode: http://www.networkupstools.org/protocols/apcsmart.html

That site also hold some schematics to make a serial cable (remember, it's not a standard one... yeah... everything to make money )



As a side note, tomorrow I'll do the hardware mod for the float voltage, I will NOT put a resistor in parallel with R118 like most recommand, but will put a 390 ohms in series with R119 (the 22.1k)... based on my calculations it should give exactly 26.72 (13.65V/batt, which is the dead center of the recommanded zone: 13.5-13.8V). I'll post once I do it (which will be after I find some good batts, mine died yesterday from a failed cell...)

For the schematics, look at toasty's post, THANKS TOASTY!!!!!

[thephantom1492]

so hmm finally the mod will wait, the new batt I found ask for 13.7-13.9V, the ups provide 13.9V so it's in the right range. But I will still do it at the next scheduled downtime

[rosenhauer]

I've got 2 SU1400 backups and I've converted one to run on 2 Deep Cycle batteries. Having been new to this I didn't check the electrolyte levels on the batteries and they boiled dry!

The UPS reports 27.6v battery level but according to my meter it's 28.2v So I'm guessing that's why it boiled both batteries dry in 8 months. And explains why it would go through APC batteries in about 18 months. I tried the software settings to change the voltage http://www.jjoseph.org/notes/apc_sma..._float_voltage but that doesn't seem to be changing the actual voltage just what's reported.

Like I said I have 2 and the other is actually holding at 27.6 so for now I've swapped to that one. But I would like to get them both going. Have there been any more results/recommendations on what voltage to hone into. I'm assuming (maybe incorrectly) that I'll need different resistances since my starting voltage is different.

Is the resistor the best place to start or do I need to check the caps to see why it's high first?

I'm handy with a soldering iron but never messed with this sort of circuits.

Dave

[Quasar]

Deep cycle liquid or gel type? Different charging and float rates between the two.

If those are liquid, why are you blaming the UPS for you not checking the electrolyte levels?

Use more than 1 meter to verify voltages. APC UPS are proprietary micro computer controlled. Messing with external values may only lead you down the wrong path.

Q

[kc8adu]

true deep cycle batteries are lead-antimony alloy.
they will use water faster than lead calcium.
13.5-13.8 per battery is optimal float for yours.

Originally Posted by rosenhauer

I've got 2 SU1400 backups and I've converted one to run on 2 Deep Cycle batteries. Having been new to this I didn't check the electrolyte levels on the batteries and they boiled dry!

The UPS reports 27.6v battery level but according to my meter it's 28.2v So I'm guessing that's why it boiled both batteries dry in 8 months. And explains why it would go through APC batteries in about 18 months. I tried the software settings to change the voltage http://www.jjoseph.org/notes/apc_sma..._float_voltage but that doesn't seem to be changing the actual voltage just what's reported.

Like I said I have 2 and the other is actually holding at 27.6 so for now I've swapped to that one. But I would like to get them both going. Have there been any more results/recommendations on what voltage to hone into. I'm assuming (maybe incorrectly) that I'll need different resistances since my starting voltage is different.

Is the resistor the best place to start or do I need to check the caps to see why it's high first?

I'm handy with a soldering iron but never messed with this sort of circuits.

Dave

[honeymonkey]

Hi,
I'm new here as you can see. I ran across this page through Google after popping open my Smart-UPS 1250 and noticing the negative post all messed up, both batteries bulging and leaking. I used to get at least 3 years , but last few are only lasting 2.5 years and show signs of overcharging.

I looked at the max charge voltage on the box of the new 20ah 12v replacements and it says "13.8vdc max" so I put a digital meter on each battery. The one off the positive charge wire was reading 13.64 and the one off the negitive charge wire was reading 13.84, so I have to assume this is the problem. I checked 2 other UPS that are the same model and they all display the same readings.

Why would the Negative one read higher, and do you think I should perform this mod? Just wondering how it worked out. After reading this thread I'm worried if I change the charge rate that it may affect the ups on other areas.
Thanks,
HM

[Krankshaft]

You can't change the charge voltage it's programmed into the firmware.

The batteries are connected in series (the inverter is a 24V input) the differential in voltage is probably the voltage drop of the other battery.

Around 2-2.5 years are SLA batteries average service life.

[honeymonkey]

Originally Posted by Krankshaft

You can't change the charge voltage it's programmed into the firmware.

The batteries are connected in series (the inverter is a 24V input) the differential in voltage is probably the voltage drop of the other battery.

Around 2-2.5 years are SLA batteries average service life.

Well, here I am, about 2 months later after replacing both batteries again, and they are cooked. I have 7 of these smart ups 1250s and the last 2 battery changes in two of them result in less than 3 months life. 13.84v charging v on the negative side has to be cooking these things.

Any ideas how to lower the voltage?

[smason]

FWIW, it looks like APC might be finally changing the way they do things.

From Page 24 of APC Currents magazine: (yes, I have so many UPSes they send me a magazine..)

"The new APC Smart-UPS provides improved battery life by charging the batteries only as needed. As a result thermal operating conditions are significantly improved, prolonging battery life."

These new units have SMT and SMX prefixes in the model number.