Rebuilding 400 series Roomba battery with li-ion cells

[bcm00re]

So I have two 400 series Roombas that I would like to get running again. One is grey & silver and the other one is red. They worked well last time I used them 7+ years ago. Needless to say all the old battery packs I have for them are no longer functional.

I was thinking of rebuilding the battery packs replacing the 12 sub-C cells with eight 18650 li-ion cells (in a 4S2P config) but closer inspection has revealed that the cell holders that I intended to use will not fit within Roomba battery case. So my new plan is to use four 21700 cells in series -- more specifically these Eizfan IMR21700 3750mAh: https://lygte-info.dk/review/batteries2 ... %20UK.html . Based on the review they are really more like 3500mAh cells, but I believe that is at least as much as the original 400 series batteries. I am also looking at these LG 5000mAh 21700 cells: https://lygte-info.dk/review/batteries2 ... %20UK.html . I plan to use one of these two cell holders:
A) https://www.ebay.com/itm/174070336887
B) https://www.ebay.com/itm/183495449573
as well as the following protection circuit:
https://www.ebay.com/itm/123984120079 . I believe another user here has used that PCB with good success -- even allowing him to use the stock setup for charging.

Any tips/suggestions/comments would be appreciated!

[vic7767]

Cell holders have an inherent artificial high resistance when using the Lithium cells. This leads to poor performance.

A fix is available by purchasing the 4400 mAh Li-NMC pack for the 500 -900 Roomba pack, remove the top of the battery pack and lift out the PCa and cells. Dremel down the sides of the PCA and insert the complete assembly (PCA & Lithium cells) into the R2 (400 model Roomba).

Re-wire the R2 battery leads on to the R3 Lithium PCA + and - pads, install the top of the R2 battery, install into your R2 Roomba, place it in the charger and you're good to go. Don't forget to wire the existing R2 thermistor circuit also.

[bcm00re]

Thanks for the suggestion. Can anyone some quality aftermarket ones?

Will that increased resistence -- if I use cell holders with lithium batteries -- only lead to more losses and thus degraded performance? I have several 18650 flashlights that just use spring connections. I can build a couple batteries my way for about $30 each using high quality 5000mAh LG cells (mentioned above) and when the cells do need replacing it'd be half that price to replace them -- or probably even less.

[vic7767]

Cell holders will cause poor performance with any cells NiMH, or Lithium. The reason is compounded when the robot requires 2 amps to start-up and run its motors. Here is a picture of a cell pack with control board being installed into an R2 battery pack. It will give you an idea of what's involved.

The Li-NMC pack can be ordered on Amazon for about $65. If you have the time this may be the way to build your own Lithium pack for your 4XX Roomba.

I sent you an email with info on purchasing an R2 Lithium pack from my website.

[bcm00re]

Thanks for the info. It looks like yours (and maybe the Roomba one) uses the same cells as the battery offered by Lithium Power Inc. It looks like those same Boston Power Swing 4400 Lithium cells can be had on ebay for $12.50 for two (and like my plan four are needed). Those Boston Power cells look to actually be just a pair of 18650 cells linked and bundled together.

I already bought four of the LG 21700 cells for another project (as well as a couple holders) so I think I am going to do a quick proof-of-concept test with them in one of my Roombas. I guess worst case I could get some low temperature solder and connect them them that way, but I have flashlights that pull a lot more than 2A (again with just spring loaded connectors). I'm not yet entirely convinced the holders won't work; even if the added resistance loads (from the holders) cause a 10% loss in capacity then those LG cells would still be good for 4500mAh. If I have to go the solder route then I might get these cells: https://www.aliexpress.com/item/4000557285333.html because they have built in solder tabs.

[vic7767]

That's good if you can solder cells together instead of clipping them into cell holders. Less loss is better. And yes the cells I use are from the Lithium Power vendor. What I like about their power control board is that it also performs cell balancing.

[bcm00re]

So one of my BMS circuits finally arrived. It is the 2S one for my Eureka stick vac. I assembled the new battery pack using two LG 50T 21700 cells to replace five sub-C NiCD (or maybe they are NiMH) cells. I connected everything properly but no power. I then "woke" the BMS circuit by applying 8V briefly then I read power on the input/output terminals, but when I try to turn on the vacuum it only very briefly turns the motor then it shuts down and there is no voltage. When I bypassed the BMS and the vacuum starts and runs just fine on just the two 21700 cells. The BMS is rated to do 8A with 15A cut-off, and I measured the vac pulling around 5A when running and a bit higher during start-up. So I don't know what is going on. I then hooked the new battery pack (with the BMS) to an old 12V drill and it runs that fine. I even applied some pressure resistance to the drill while using my clamp meter and confirmed the pack could supply peaks of over 7A and 6A sustained (kind of briefly). Anyone have any idea why the new battery pack and BMS won't run the vacuum?

FYI, this was a smaller test case I had hoped to do before ordering more LG 50T cells to runs my Roombas (with 4 cells each and a 4S BMS circuit). The original Eureka battery contained only five sub-C cells nothing else at all -- and it only has two connections/terminals.

[vic7767]

Is that Eureka expecting a battery temperature from the stock pack?

[bcm00re]

No thermistor in the original battery -- literally just five sub-C cells. On another forum someone pointed out that a DC motor that pulls 4-5A continuous should pull much more than 6A at startup; so it appears my meter wasn't catching the peak startup current correctly. I confirmed this by attempting to power the vacuum with my variable power supply (that I found after a little research is rated to supply 15A with peaks of 18A) and found it cannot start the vacuum either! I was able to use the new battery pack and my power supply together to get the vac started, then confirmed just the new battery runs it just fine. So I think I just need a much, much larger BMS circuit.

[bcm00re]

If anyone is interested, I did some research on DC motors and found the inrush current that occurs at start-up because when the motor is at rest its resistance is very low. So I got to thinking the BMS circuit might be thinking it is detecting a short -- so it might not be a current issue. With some research and testing, I found a couple ways to make my existing BMS circuit work. One way is to put a DC-DC converter between the BMS output and the motor. The converters that I had on hand only handle 3A so I wired two in parallel. Their small LED lights dim significantly when I switch on the vacuum, but they recover quickly and the vac run just fine. The second way was to add a 4ohm power resistor is series between the BMS output and the vac. It limits the vacuum to only pull about 2A but once the motor is running I can short/remove the resistor then the motor gets up to full speed. This vacuum actually has two modes; one runs just the suction and the other also turns on a small motor that turns a beater brush. I always run it in the latter mode, so I could install the resistor on only the first mode then it would become just for start-up. I have also read about using a NTC thermistor that start with some resistance that then goes away as the temperature rises so that might an option too.

I haven't decided what to go with yet, but I am leaning towards the DC-DC converter route because I can drop the voltage output of the battery down to 7.5V which is the maximum voltage that could be seen from the original NiMH battery. The lower voltage should also yield a lower current pull too -- which means less stress on the vacuum and less stress on the battery.

[bcm00re]

I ordered a larger BMS circuit so the above ideas -- for my Eureka vacuum -- are really just back-up plans should it not work. I will probably order some 2.5ohm NTC thermistors just in case. I decided that having a DC-DC converter and a BMS is just overkill, and any consumption savings I'd get from running the vacuum at a slightly lower voltage (and thus lower current) would likely mostly get eaten-up by the inefficiency of the converter anyway. I ran the vacuum this weekend directly off the 21700 batteries (but was sure they were fully charged and only ran them for about 20 minutes) and it worked just fine.

I am still waiting for the Roomba BMS circuit to arrive...

[bcm00re]

To circle back on the Eureka, the first larger BMS still didn't work so I also added a NTC thermistor -- but even that did not work for long. On the first use a couple of the transistors got so hot their solder melted and they moved (then things stopped). I think the seller mispresented what that BMS could handle! I finally found a better one (with more transistors) and that one has been working fine for a couple months. This one also doesn't require applying some power to the outputs to "wake up" the circuit. To be clear, this vacuum has 2 modes and I added the thermistor to the one I don't use (except to get the unit started) because the simple way I did it it wouldn't be good to have the thermistor inline while running the vacuum.

I still haven't gotten around to building the Roomba battery packs yet -- maybe over the holidays.

[bcm00re]

Finally built one of these battery packs. The four 21700 li-ion cells ran the Roomba for over 2 hours and it still had energy to spare! I just used the spring loaded holders and everything seems to work fine. The protection circuit I added (within the battery itself) for the li-ions cells appears to allow the stock charging system to charge my battery. Now I just need to build another one for my other Roomba.

[vic7767]

You can watch the entire charging system process by monitoring the SCI channel during charging.

[bcm00re]

Thanks, but I don't have the hardware/understanding (at the moment) to do that. Here is the protection circuit that I used:
https://www.ebay.com/itm/4S-20A-14-8V-1 ... 2573311525
I will take some pictures to post when I build the next battery.

[bcm00re]

As promised here are a couple pictures (see attached).

[bcm00re]

My new, modified battery weighs almost half as much as an original battery (14oz versus 27oz) despite having about 50% more power. Over the weekend my Roomba ran on carpet for 1.5 hours and still had juice left. For reference I have been using my red Roomba 4300 for all testing so far. I also have a silver/grey 4220 model that I plan to start using soon.

[bcm00re]

Another update:
I have done a few runs with both my old Roombas using my new batteries, and I am seeing about a 2 hour and 45 minute run time. This is with the units running on hard flooring (a mix of hardwoods and ceramic tile). So far I am happy with the results!

[Roombaworker]

So i have upgraded with a 500 series battery. I've charged them. Or atleast tried. They won't charge up and only blink red. So i put older battery in nothing. I got new batteries and now it won't charge or power on. However the new batteries work in another machine. Any ideas on how to repair?

[vic7767]

The R2 Roomba uses two charging MOSFETs, U2 and U4. One or both of them may be damaged due to extreme charging current during the charging cycle. Monitoring the charging reports that are reported over the SCI channel may give a clue as to the state of the charging system. The solution to this condition is to replace the two charging MOSFETs.