Rebuilding a Roomba Battery for Less

[Kelad]

Hi All, i purchased a 2nd hand roomba 537. Its great.
I have the imax B6AC charger and thought i'd put the battery through its paces as i noticed the run time was only approx 10mins. I used settings recommended by user robocleaner
- charge at 0.5A,
- discharge at 0.4A
- discharge limit set to 12 volts for an older pack
- maximum safe charge time of about 340 minutes per cycle
- maximum pack capacity set to 3000mAh

my results:
chg1 0740 mAh
dchg1 0395 mAh

chg2 0659 mAh
dchg2 0377 mAh

chg3 0634 mAh
dchg3 0357 mAh

ran again

chg1 0544 mAh
dchg1 0352 mAh

chg2 0585 mAh
dchg2 0328 mAh

chg3 0575 mAh
dchg3 0337 mAh

I realise these results indicate a NOK battery, however my question is this - is it worth opening it up and looking for a dud battery and replacing that one or will i find most of them stuffed. I'm happy to give that a crack but is it a waste of time?
Thanks

[vic7767]

If you're going to spend the effort to open up the pack, go ahead and replace all the cells. If you just replace one or two you will be back opening it up over and over.

[mfortuna]

If your charger can cycle individual cells you can get an idea of the overall health of the pack. Chances are even the best cells will not be great and replacing the entire pack may be the only alternative. But if most of the cells are good replacing the bad ones may work for a while.

[Kelad]

If you're going to spend the effort to open up the pack, go ahead and replace all the cells. If you just replace one or two you will be back opening it up over and over.

Thanks vic7767 and mfortuna for feedback. I decided to buy a new Tenergy Replacement Battery 3500mAh from amazon http://www.amazon.com/gp/product/B003MXMWQ2?psc=1&redirect=true&ref_=oh_aui_detailpage_o00_s00

Before i started using it in my roomba 537 i thought i would test it out with my imax B6AC charger.
settings
- charge at 0.5A,
- discharge at 0.4A
- discharge limit set to 10.8 volts
- maximum pack capacity set to 3600mAh

I ran 2 cycles and discharge results were 2633mAh and 2659mAh. Guess you get what you pay for.
I would run some more cycles but the missus wants the machine running. Anyway, i thought these results were bit crap. Thoughts ?

regards
Kelad

[vic7767]

Thoughts ?

I just recently became happier when I converted from NiMH and NiCD to LI-NMC packs. All my Roombas and Neatos now run on 4400 mAH LI-NMC packs and my older Scoobas run on 6600 mAH LI-NMC packs and all is well. No more disappointments.

[mfortuna]

A few more cycles may improve your results but I am surprised the pack's capacity is that low. I am sure the battery will give you decent cleaning results. My old roomba reds clean great with old 3000mah packs.

If I ever spend big $$$ on a new roomba or botvac I may go the LI-NMC route but it doesn't make sense for me to spend money on almost 10 year old roombas that are giving great results with NIMHs. My Neato is only a few years old so there is a chance I could buy a LI-NMC.

[kareenwalton2015]

Hey There. I found your blog using msn.This is an extremely well written article. I will be sure to bookmark it and return to read more of your useful info. Thanks for the post. I’ll definitely return.

[Dipweed]

I too am running Li Ion bats in my 700s and 800s
But
I had one Li Ion pack go bad due to complete discharge or low low discharge. Attempted Fix
Opened Li Ion pack and checked each cell for min voltage. To my supersize all cells were go to go. It was the circuit board that was damaged. To much work to fix.
Question
Hey Kelad
How did you determine/test the capacity on your built amazon sub C cell pack? I have two

Turnigy Sub-C 1.2V 5000mAh High Power Series NiMH Single Cell $2.84/$3.81

Size is good to go and I am testing on a VW but I would like to confirm the cell capacity for 5000mAH. Again how did you go about the Capacity test on the pack you built?
Last
If and when I determine the capacity of these cells are close to 5000mAh then they should out perform any Li Ion or any other pack commercially available via 25% more roomba run time.

[Kelad]

Some info for people from Australia. I had in the past purchase batteries from Amazon, but the current US dollar and postage costs make it pricey. I thought i would give ebay a go where delivery is free and from Sydney. It cost $AU 40.86.

http://www.ebay.com.au/itm/Battery-For-iRobot-Roomba-500-3-3Ah-Ni-MH-HeavyDuty-510-530-537-550-560-580-630-/131533009597?hash=item1e9ffa72bd:g:PnMAAOSwNSxVd5j5

Anyway, brand is "PowerPro". I used my imax B6AC charger and put it through its paces. 3 Discharge results, 2615, 2756, 2615 mAh respectively. Listing says 3300mAh, so again came up short. Anway, my 530 ran for >1hr easy and i had to shut it off early.

[a1robotrepair]

Study of OEM and aftermarket IRobot batteries, (R3 and above excluding the IRobot Li-ion).
OEM batteries use a very good cell, typically tests to > 3 .3Ah new @ 50mA/14.4V. The new Blue E-Life > 3.4Ah.
I can't find a reputable battery manufacturer that makes a SubC Ni-MH above a 3.3Ah or Ni-Cad above 2.5Ah, So I doubt there is such an animal.

Testing of the OEM thermistor revealed that the thermistor is +2 to -0 deg. across the charging temperature range. These thermistors aren't the $1.29 +-5 dog. +-5%. They appear to be +2 to -0 +-0.1%, $15 to $20 range if bought individually.

Testing of the aftermarket thermistors revealed many were the $1.29 thermistors. I now know why the aftermarket batteries don't last long and vent acidic fumes causing the green high resistance shorts in the charging detection circuit, CE3. I will still use aftermarket batteries but from now on will replace the T with an OEM. Don't throw those OEM thermistors out, reuse them! If you spent a lot of $ on a Robotic Vac and don't want to mess with replacing the T, buy the OEM batteries and keep your vacuum PCA performing well.

Aftermarket Li-Ion: Reference read first: http://electronicdesign.com/boards/keep-eye-temperature-trends-during-li-ion-battery-charge-and-discharge-cycles
http://www.robotreviews.com/chat/viewtopic.php?f=1&t=2757
Battery Balance/Protection Board = BatPTB

All of the Li-Ion packs I've tested and tore apart have had no over/under temperature detection circuit, one did have an 80C fuse link. They utilize a resistance indicating to the robot <35C. The resistance changes when charging is complete indicating >50C fooling the robot into high temp no charging mode. Running a home built pack using the above method caused the battery to charge well above the 40C recommended max charging temperature of the cells I used. I now have redesigned my "SMART" circuit to take advantage of a thermistor whose circuit is in the specified charging temperature range and also does not automatically shut off charging when BatPTB stops charge. It will emulate the dT/dt at a specified charged voltage before the BatPTB shuts down charging, Trickle Mode for the final charge to the batteries.

Conclusion: Unless you're prepared to test your aftermarket battery for adherence to quality standards, you should only use an OEM battery. DONT BUY NI-CADS!
Specifics and testing procedures are IP and will not be disclosed. I also like the IRobot Corps. and I'm sure they don't want their IP revealed.
Not all brands tested.

[mfortuna]

There are a number of manufacturers that make 5000 mah Subc NIMH packs for RC vehicles. Not sure how good they are. Just 10 years ago I was running 3700 and 3800mAH NIMHs in mod racing class. Those were matched cells and met their specs (> 5 minutes at 30A discharge). It is hard to believe there has been no improvement in 10 years.

[a1robotrepair]

https://na.industrial.panasonic.com/sites/default/pidsa/files/nimh_batteries_series_datasheets_merged_11.24.14.pdf

Just one of the reputable manufactures. You should look up Uniross also.

A 5A Ni-MH SubC?
A 5A Li-Ion 36650 I would believe, I have them in a Scooba battery pack. Their max continuous output is 5A for 56min. Even Lithium batteries have to give up high drain for high capacity.

[mfortuna]

A couple examples:

http://www.maxamps.com/proddetail.php?p ... hgodLCAAKw

http://www.amazon.com/Tenergy-5000mAh-B ... B002PEBM8S

[a1robotrepair]

My thinking is to build a Li smart circuit that would mimic the dT/dt of a Ni-Mh battery and fool the robot into CI/CV mode.
test w/o thermistor...im working on it now.
1. Switch robot into trickle mode @>16.6V.
2. Disable mode after trickle complete, battery depletion, and off charger.
3. Circuit in battery or in robot availability. requires elimination of thermistor or alternate wiring to battery.
tested on 880 because they play a sad song before the protection circuit switches.
also wanted to see if it was somehow in li charging mode with the data I obtained from another test.

start-charge: 2013-11-20-1456-L
do-charging-checking-fets @ minutes 160
bat: min 160 sec 49 mV 14027 mA -8 tenths-deg-C 252 mAH -1 state 1
Charging FET test passed.
do-charging-wait-initial @ minutes 160
do-charging-on @ minutes 160
880 ran for 145minutes with 2p4s Li pack and rested for 2 hours.
They quit counting time when asleep?
bat: min 160 sec 50 mV 14444 mA 1133 tenths-deg-C 248 mAH -1 state 5
bbox vars saved!
Switched Discharge circuit increasing T over t.
bat: min 323 sec 18 mV 16615 mA 1180 tenths-deg-C 245 mAH 2696 state 5
bat: min 323 sec 19 mV 16615 mA 1188 tenths-deg-C 246 mAH 2696 state 5
bat: min 323 sec 20 mV 16587 mA 1188 tenths-deg-C 246 mAH 2696 state 5
bat: min 323 sec 21 mV 16587 mA 1188 tenths-deg-C 246 mAH 2696 state 5
bat: min 323 sec 22 mV 16587 mA 1188 tenths-deg-C 246 mAH 2696 state 5
bat: min 323 sec 23 mV 16615 mA 1188 tenths-deg-C 246 mAH 2696 state 5
bat: min 323 sec 24 mV 16615 mA 1188 tenths-deg-C 246 mAH 2696 state 5
bat: min 323 sec 25 mV 16615 mA 1188 tenths-deg-C 247 mAH 2696 state 5
bat: min 323 sec 26 mV 16615 mA 1188 tenths-deg-C 247 mAH 2696 state 5
bat: min 323 sec 27 mV 16615 mA 1180 tenths-deg-C 247 mAH 2696 state 5
bat: min 323 sec 28 mV 16615 mA 1180 tenths-deg-C 248 mAH 2696 state 5
bat: min 323 sec 29 mV 16615 mA 1188 tenths-deg-C 250 mAH 2696 state 5
bat: min 323 sec 30 mV 16615 mA 1180 tenths-deg-C 254 mAH 2696 state 5
bat: min 323 sec 31 mV 16615 mA 1180 tenths-deg-C 268 mAH 2696 state 5
bat: min 323 sec 32 mV 16615 mA 1180 tenths-deg-C 284 mAH 2696 state 5
bat: min 323 sec 33 mV 16615 mA 1880 tenths-deg-C 315 mAH 2696 state 5
bat: min 323 sec 34 mV 16615 mA 1880 tenths-deg-C 384 mAH 2696 state 5
bat: min 323 sec 35 mV 16615 mA 1880 tenths-deg-C 384 mAH 2696 state 5
do-charging-trickle @ minutes 323
This thing will run my hour and a half, every day and may never trip the protection circuit.
Passive balance happens here.
bat: min 323 sec 41 mV 16587 mA 55 tenths-deg-C 305 mAH 2696 state 7
saving bbox vars
bat: min 328 sec 0 mV 16532 mA 55 tenths-deg-C 245 mAH 2696 state 7

i'd like to see data from an aftermarket Lion.EDIT
This works about 1 out of 3 times, I need a longer time decay.

[a1robotrepair]

I found a data sheet on a 5A Tenergy C Ni-MH.
I was wrong, apparently there are 5Ah Sub Cs.

http://www.amazon.com/Tenergy-5000mAh-B ... B002PEBM8S

https://www.all-battery.com/datasheet/c%20size%205000mah.pdf
Check out discharge rate and falloff at .5C//.2C. Not good for high rate discharge. Running these above .5C will shorten their lives. These should work good as a Roomba battery, .25C charge and .4C discharge and deliver over 4.5Ah @ .4C when new.

[a1robotrepair]

Neat battery thermistor test with a 1 off 8K / 25C high beta very accurate thermistor circuit. I put together 4 IRobot thermistors in series, installed then with good contact between different cells from the bottom of the pack. I soldered lead wires from the series array in // with another thermistor.
15C tells robot 19.6C
25................29.9
35............,.,.40.2
45................50.6
This was a trial for my 1 off 880 li-Ion home build.

That would be cool if I were building a Ni-Mh pack.
Better would be parallel resistor making my Tcircuit look like 40 when its 35.
1T // 27K
45=48.6
35=40
25=32
15=24.8
Now my // indicating over temp circuit resistance can be narrowed.
Adding series resistance and smaller // resistance can narrow it farther.
15=

[a1robotrepair]

Duo cell from an aftermarket RR member sent me. Bad cell, bad pack.

http://liionbms.com/pdf/bostonpower/swing4400.pdf
https://datasheets.maximintegrated.com/ ... AX5419.pdf
http://aosmd.com/pdfs/datasheet/ao4407.pdf
https://www.fairchildsemi.com/datasheets/TI/TIP42.pdf
https://1b774386-a-62cb3a1a-s-sites.googlegroups.com/site/isrendering/fiscourse.comp.pdf?attachauth=ANoY7cqKv5yHw-Ay8BS2_1DK--bdCdzeiwT2BIYa5_IK-ndwlpBjOcihDcu_3Bj81e6-WoZwEy4yxgLrDuUYGif1rhrKS2cmU29-2bNttUY9B3SmCWfI7pNeFY8xUwVG-5nj3z_MkvX2InQ8LOeQzGyei07yIQJnn4eXVh-1LSnEn6Cvc72FXWe82-UQTXdGZajYK5kMgf1ZpTmJh3un4BQzVg5kPZqN-w%3D%3D&attredirects=1
http://www.milliman.com/uploadedFiles/insight/2013/using_leakage_as_CV.pdf
http://datasheet.sii-ic.com/en/battery_protection/S8254A_E.pdf
Reassembled pack with 2 // IMR18650E batteries of differing chemistry to originals. Their curves are nearly identical but their internal impedance is 2X.
Manually balanced cells of pack and pack checked.
Put on charger and supervisory ic switched bot to trickle after 2 minutes by slowly decreasing resistance to thermistor circuit. Ran bot for 80 minutes and docked. Charged correctly to 16.3V after rest.
Trickle charged to 16.6 V after 33 min. rest. sat overnight on charger 16V by term software.
Opened pack found B1=4.131V, B2=4.137V, B4.168V, B4=4.169V.
It appears balance is good side to side but not end to end.
The supervisory could recognize the different Chemistry and not allow balance of B1/2 to B3/4 or the board has suffered damage from the vented battery, or the board caused the battery to over discharge / overcharge.
Proper cell on the way, we'll see how it works.
Testing reveals 2X impedance is the problem for now.

[a1robotrepair]

Here are some pics of pack showing balance/protection board.
One duo cell, B2 is different, 2 2.2//cells IMR.
After sitting, B1=4.087V, B2=4.093V, B3=4.123V & B4=4.121V.
After going thru the circuit it appears the best balance will be 35mV with already unbalanced cells.
The Bal/Pt PCA is working correctly. Original problem with pack was cell infant mortality. After buying a Boston Power SONATA 4400, which I was told was identical to the SWING 4400, I looked up their data sheets and they are not. The SWING is a much better battery, internal resistance is 17mOhm as compared to 25mOhm. 1000 cycles to 80% rough duty to 800. I have it now, can't find a swing 4400, now they're 5300, so in it goes.
http://liionbms.com/pdf/bostonpower/sonata4400.pdf
Also notice how LPI packs cells are assembled correctly, only one place in the middle of the pack where there is a chance of unwanted contact as opposed to two you'll find in other packs. While reassembling the pack I added a 10A fuse link between B1 & B2.

http://www.analog.com/media/en/technical-documentation/data-sheets/ADM1062.pdf

[Peter101]

This is my second battery rebuild. The first time I recycled all the parts. This time I bought a 1 meter nickel strip from China for $1 which reduced the rebuild hassle considerably. I took some photos for other would be re-builders. The most important spot welding detail is a few pounds of force on the spot welding electrodes. My electrodes were fashioned out of 12 AWG copper wires that were stripped and flattened. The curved copper wire creates a stiff spring so both electrodes firmly push the metals together. A good spot weld will tear the nickel strip at the weld if pulled apart. Practice and adjust the weld energy (or heat) on an old battery. Making a hole in a battery is easier than one would expect. The original cells were glued together but I use tape. My re-build is slightly larger than the original but fits in fine.

-Peter

[a1robotrepair]

I won 2 of these lithium batterie packs on an eBay auction. Time to hack 1. Not recommended unless you're like me and are going to use the case and balance/protection board for building your own. I'd show you the first one I built from scratch using a cut down R2 battery case but I don't have photos of it. EDIT, I woke up at 4:30 and was told if all I was going to do was toss and turn, get up! I took the battery back apart to start rebuilding it. DO NOT USE THESE BATTERY PACKS! THEY ARE ICRs WITHOUT INDIVIDUAL PROTECTION AND NO BALLANCE CIRCUIT. THEY ARE SERIES WIRED TOGETHER WITH JUST A CHARGING BOARD AND A CIRCUIT TO FOOL THE THERMISTOR CIRCUIT. IT'S JUST A MATTER OF TIME BEFORE SOMEONE BURNS THEIR HOUSE DOWN WITH ONE OF THESE! The eBay seller is http://www.ebay.com/usr/alpowcity?_trks ... 2749.l2754. They are also sold on Amazon by INNOPOW

Faceplate rating of 3Ah and they fit Roomba nicely.