24 Eneloop AA batteries Roomba 581 modification

Inside the Roomba and Scooba and more, Cool mods, Repair and Upgrades - including the all new iRobot Create Kit. Let's void that warranty baby!

24 Eneloop AA batteries Roomba 581 modification

Postby Atlantis » August 16th, 2016, 7:57 pm

Hello,
I recently bought second hand Roomba 581. It is in excellent condition - I assume somebody didn't use it much because everything is clean and only with minor scratches on bumper. I even recieved replacement brushes and I assume the installed brushes are originals (never replaced).
Everything is good but battery - I tried everything but 2cells are dead (0Ah capacity), two of them are good (2.5Ah) and the rest is at 0.5Ah.
I was thinking about these options:
1) Buy genuine battery
2) Buy aftermarket battery
3) build battery from good cells.

I decided to go with 3) bucause I don't like aftermarket batteries (you don't know what cells are inside) and genuine battery is no fun :lol:.
Here I could just buy 12 SC Panasonic 3000mAh batteries and be done but in many my projects I use Eneloop batteries so I decided to go with them. Roomba has relatively low discharge current (2A max?) compared to cordless drills or even handheld vacuum cleaners - they go up to 15A from battery and you really need high current cells for them which Eneloops are not as far as I know. But for Roomba they are absolutely fine.

The only problem is AA cells are longer than original cells and I could squeeze only 12of them resulting in only 1900mAh capacity which is not so great. So I decided to go little further and modify roomba battery housing to fit 24 of those AA Eneloops resulting in capacity around 3800mAh.

First I desoldered battery spring contacts for +, - and thermistor. I tried several orientation how to place the batteries to fit and the best one seems to be 3 rows of 8 in lying position. But I was missing around 1-2mm of space. So I decided to thinner the plastic wall between the battery compartment and front of the robot. I used dremel with abrasive disc and also brush and paper wipes with acetone to dissolve the plastic away. Actually the acetone gave it nice glossy finish so it looks like it is factory made :lol:

IMG_20160813_193803.jpg
Thinner plastic wall


Next I was going to make the battery pack - I placed 8 eneloops in row and gleud them together with neutral silicone
IMG_20160811_222151.jpg


After these 3 packs cured I glued them together to form one big pack
IMG_20160812_020852.jpg
One big pack


Then soldering comes in. I know that soldering is supposed to be bad for batteries but I think I can do it safely without any damage to the cells and I have done it many times (even with lithium batteries) and the batteries worked fine for many years. Trick is to use good flux and big soldering iron and keep the time it is heating to minimum. I use flux for pipes (agressive! must be washed away!) and I am done with soldering in 3seconds.
IMG_20160817_013935.jpg
Soldered pack

The connecting bars were tinned only from one side to keep the thickness to minimum - this is why it looks like it is not soldered properly. But trust me - it is.

I then Installed thermistor in the middle of the pack. I was thinking about using the original one from old battery pack but I had another one with long wires already around. I checked they have very similar resistance at various temperatures. I simply sticked it between the batteries in the middle of the pack.

This is how it looks like inside the robot:
IMG_20160817_014409.jpg
Completed pack inside Roomba


Oh and final adjustment had to be dome to bottom cover - I had to remove these pieces of plastic to allow more room for batteries:
IMG_20160817_014601.jpg
Modified cover


After full charge and full discharge I get 3562mAh (read from serial port in roomba during charge). Before this pack it read max 2696mAh (something around this number).

Summary - I get better capacity for even lower price compared to genuine battery (at least where I can locally buy it)

PS: I know there is topic about battery rebuilding but I think mine modification is different and deserves its own topic because it is not just about battery but about roomba mod.
Atlantis
 
Posts: 7
Joined: August 16th, 2016, 7:15 pm

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby vic7767 » August 16th, 2016, 8:37 pm

Good idea and nice mod of the battery well. Post back with run times and let us know if performance improves.
Roomba and Neato Mods, come visit: http://www.vic7767.com/

http://www.Robot-Doc.com/
User avatar
vic7767
Robot Master
 
Posts: 14300
Joined: January 14th, 2006, 8:31 pm
Location: Haughton Louisiana - USA

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby a1robotrepair » August 16th, 2016, 8:56 pm

Never thought of using them that way. The batteries I use for my remotes and flashlights. Panasonic makes great batteries.
Roses are red Violets are blue Poems are hard Flowers and Chocolate
http://www.daycounter.com/Calculators/
a1robotrepair
Robot Master
 
Posts: 770
Joined: October 14th, 2013, 8:40 am

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby glnc222 » August 17th, 2016, 6:06 pm

A Neato forum thread is devoted to the experiment of making Neato batteries out of AA Eneloop and equivalent cells such as EBL. http://www.robotreviews.com/chat/viewtopic.php?f=20&t=17994
The cells fit, even in a Botvac. They can run the robot a few times, but the drain rate supported by these cells does not appear to match the application of driving the motors and the load around three amps, sometimes 4-5 amps when climbing or what not. So the cells wear out quickly. Maybe get a few weeks or months use depending on usage.

I built a battery cycling circuit to duplicate the typical robot run and charge cycle, compared to standard testers which completely discharge the batteries, using an external charger and a resistor load, with relays and logic circuits (details in that thread). Possible long term testing was cut short when a number of cells showed significant deterioration after only a few cycles.
The 65mm NiMh cells used in the standard robot packs appear to be made for higher drain rates and do not come with the low self-discharge features attractive in the AA size cells. Whatever sub-C size used in Botvac's is not the most common inexpensive cells sold online, usually without detailed specs.
Some literature found mentions higher drain types of AA size cells, but which are not Eneloop types.

There are many ways to make the same size battery. Hybrid cars use NiMh cells costing $25 each supporting over a 100 amp drains.

The Eneloop advance since duplicated by all the major brands, was reducing the self-discharge rate and shelf-life problems of rechargeable NiMh batteries, so they can compete with the single use types better. The technique may involve a trade-off with maximum drain rate supported.

Lithium battery chemistry has also advanced, with the more safe LiNMC types now used in power tool packs and after-market batteries for Roomba and Neato. The Lithium Ion cells in computer packs have a fire hazard aspect when incorrectly made or used. In the newest aircraft switching from hydraulics to electrics, they are installed in fire proof vaults. Air transport of lithium batteries has been restricted since a few years for the hazards emerging from increased use.

The most interesting battery chemistry is LiFePo4, used in electric bicycles etc., which supposedly has much longer life than other types, but has lower charge density and capacity, like other "safe" variants of lithium ion. There is one experiment with Botvac in the Neato forum, still awaiting long term results. They seem to work in the lower 12v system of the Botvac, but did not work well for me in XV 14.7v systems (I used 2-parrallel when 3 was needed, but would not fit in the compartments alone). They support high drains but slow charging rates. Just no such thing as a "battery" anymore, all sorts of different technologies. One university research project reported is potential replacement of lithium with aluminum in the future, which could be very significant economically. Nothing heard recently.

[edit] Adapter circuit for LiFePo4 on NiMh charger DIY http://www.robotreviews.com/chat/viewtopic.php?p=124772#p124772
Last edited by glnc222 on August 17th, 2016, 10:56 pm, edited 1 time in total.
glnc222
Robot Master
 
Posts: 4409
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby Atlantis » August 17th, 2016, 7:30 pm

I consider Eneloop cells to be really tough. Not just as low self discharge but also speaking of cycle life and overall service life. Also if you visit candlelightforums, everyone is praising eneloops to be great even after X years of active use and I have the same experience (but well - not in roomba yet). Only time will tell....

Also I run my roomba with those batteries and got like 2+hours of runtime on carpet. Probably it ran longer, i had to go out. 2hours of discharge = on average 0.5C discharge is low current. Yes, there can be peaks when climbing some door sill etc. but I don't think it is significant. I will read the thread you posted throughly (too lazy to do it now :D).

about LiFePO4 - those look like perfect batteries for roomba. I was really thinking about lifepo4 mod. Unfortunatelly I had no idea how to make charging electronics to be compatibile with roomba - meaning it should charge without reporting any errors etc. Because nimh charge termination is different compared to lifepo of course and just disconnecting battery pack at full charge would get me some error I guess....maybe with some advanced electronics like emulating thermistor to fool romba that the battery is rapidly heating to stop charging? Maybe...I don't know. It would need some further research. It is really hard to reverse engineer what current/voltage will firmware diagnose as fault etc.... Maybe later if those eneloops die I will look into lifepo mod ;)

PS: I was just thinking about changing somehow voltage measurement calibration (tracing voltage divider and tweaking the resistors) - fooling roomba into thinking it has slightly lower battery voltage. This would result in shallower discharge of the battery and probably longer life. Also with 12cells in series there is serious problem with unbalace in all 12cells at the end of discharge (mainly at older cells) meaning if you discharge the pack to 1V/cell (12V), some cells can be at 1.1V while others can be at 0.8V or even lower! You can have 11 cells at 1.1V and one weaker cell at -0.1V and ruined! Voltage during discharge is relatively flat until it shoots down like a brick around those 1V/cell - meaning difference between safe 1.0V and dangerous 0.5V can be like...50 mAh maybe even less :roll: Discharging whole pack to 13V would mitigate this problem as it stops you before you reach this "brick" region :lol:

Also I think Eneloops have sligher higher voltage when charged compared to regular NiMH.
Atlantis
 
Posts: 7
Joined: August 16th, 2016, 7:15 pm

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby TechGuy » August 17th, 2016, 8:47 pm

Unbalance cells is the problem the robotic vacuum since the original Roomba. Now, we have Li-Ion battery pack with the cell balance feature.

The idea of using AA low discharge NiMH cells is a good idea. It is even a better idea if we can find room on Roomba to install individual AA cells so that the AA cells can be removed and charged individually by AA charger such as MAHA MH-C9000 charger which can also re-condition AA NiMH cells.
Charging battery directly: 400 Series, 500 Series,. How to Desolder
For Sale: Black CHM 2 Ball Bearings set
User avatar
TechGuy
Robot Addict
 
Posts: 1489
Joined: February 17th, 2009, 11:52 pm

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby LazyRoomba » August 17th, 2016, 9:48 pm

How does the battery weight affect the springs within the wheels?
LazyRoomba
Robot Addict
 
Posts: 179
Joined: May 9th, 2013, 6:22 am

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby glnc222 » August 18th, 2016, 3:06 am

I consider Eneloop cells to be really tough. Not just as low self discharge but also speaking of cycle life and overall service life.

True only for the intended load and charging rates for the cells. Motors discharging cells in an hour are not mentioned in Eneloop marketing, about electronic gadgets, cameras etc. There is a reason to use a single fat cell instead of two small cells in parallel. Even if a cell can deliver a high drain the longevity can be based on smaller loads. The LIFePo4 difference is an entire order of magnitude. They are also described as self-balancing. Just too bulky for the most compact vacuums, and maybe too expensive as well. 1100mah capacity, 3.2v. Nice in solar power installations, bikes etc.
glnc222
Robot Master
 
Posts: 4409
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby Atlantis » August 18th, 2016, 2:26 pm

LazyRoomba wrote:How does the battery weight affect the springs within the wheels?

Not at all, everything works fine. 24xAA is just slightly heavier (in respect to robots total weight) compared to original 12x SC.


glnc222 wrote:
I consider Eneloop cells to be really tough. Not just as low self discharge but also speaking of cycle life and overall service life.

Motors discharging cells in an hour are not mentioned in Eneloop marketing, about electronic gadgets, cameras etc.....

But that is it - it is not discharging in hour. More like 3hours. I am currently measuring current and it draws around 1.1A on hard surface and 1.5A on carpet. For 2 cells in parallel it is really NOT a high current.
Your findings concerning low life expectancy must be caused by something else - maybe your robot (and your tests) draws really high current (assuming your topic "5600mAH, 100minutes run time=3,3A), uses high current for charging (roomba only 1.2A)or you don't terminate charging properly. Or maybe you discharged your celly really low and disbalance caused excessive wear on the weaker ones.

Bottom line - I am really curious to real life expectancy in roomba.


PS: I am quite disgusted that roomba draws some current from batteries even when off and even when it is off because of low battery = it doesn't have any low voltage cutoff. Or has it? I only tried it few minutes on my bench power supply.
Model scenario: roomba misses dock and discharges completly. But I am on vacation etc and find it after 7days and the battery is discharged to almost zero and ruined...
I am thinking about making low voltage cutoff circuit to completely disconnect battery in case the voltage drops below some threshold.
Atlantis
 
Posts: 7
Joined: August 16th, 2016, 7:15 pm

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby vic7767 » August 18th, 2016, 2:47 pm

The Roomba, Scooba, Neato, Botvac, Dyson, 360, and Samsung cleaning robots all present a small load on the batteries used to power them constantly. One reason that this happens is that ALL the bots continually monitor for remote control devices whether they are on a home base or not. None of them have a low voltage cutoff circuit.
Roomba and Neato Mods, come visit: http://www.vic7767.com/

http://www.Robot-Doc.com/
User avatar
vic7767
Robot Master
 
Posts: 14300
Joined: January 14th, 2006, 8:31 pm
Location: Haughton Louisiana - USA

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby Atlantis » August 18th, 2016, 2:52 pm

Yeah but it looks like a really bad idea - there should be some low voltage cutoff instead just stupid error "please charge roomba". Or is it intentional to promote replacement batteries sales? Also I can imagine people storing roomba with battery connected for longer periods of time(yeah my relatives did this when moving from one house to another and this probably ruined the battery). And yes, i know it is in users manual :twisted:
Atlantis
 
Posts: 7
Joined: August 16th, 2016, 7:15 pm

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby LazyRoomba » August 18th, 2016, 3:32 pm

Roomba does not monitor its battery, unless in a charging state. I noticed this when my arduino was plugged into the battery whilst Roomba sat on the home base. The battery discharged whilst current was trickling into the battery.
LazyRoomba
Robot Addict
 
Posts: 179
Joined: May 9th, 2013, 6:22 am

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby glnc222 » August 18th, 2016, 3:52 pm

The Roomba is indeed lower powered than Neato's and might get longer life out of AA cells, but could still be less than the standard batteries. It takes a long time to test this unless accelerating the process with cycling instruments running 24/7. By all means give it a try. We can always use more data.

I saw mention of a "black pro" Eneloop model, but have no details. Could be just higher capacity. Capacity increases have been a feature of NiMh offerings over years with refined techniques. Maximum drain rate is another matter. Just because a battery can run at a certain max drain does not mean cycle life is the same in such use. They don't give you all the detailed graphs, unless maybe you are an industrial customer.
NiMh pack prices have fallen over years and the cost advantage of AA cells is not clear.

Neato's have under-voltage shut down routines, I think to protect the system more than the battery. Such features could change with new models of robots, newer Roomba series. Maybe it is in all of them.
The typical problem is the robot has run almost to the end of charge and gets stuck somewhere, with the computer still running in an error condition, with a small drain. Left for hours the battery gets fully discharged and a shutdown is needed. Samsung robots are finicky and shut down on any error when cleaning anticipating this. Annoying because the map is discarded and manual intervention cannot continue the job. Putting in a time delay might interfere with programmer's coffee breaks.

It is common with lithium batteries to include protection circuits in the battery pack, with mosfet cutouts for both under voltage and over-voltage, more critical with this type of chemistry. Such circuits can be made with the ICL7665 IC and FDP880 mosfets I used in Neato experiments. Lot of work, and probably not compact enough for packs. I do not know if there are NiMh protection boards, try ebay and Aliexpress.
glnc222
Robot Master
 
Posts: 4409
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby a1robotrepair » August 18th, 2016, 4:08 pm

The R3 rev8 and below do not have low cutout. The 700 series and above do but it is only software shutting all systems down, there is still leakage current thru out the PCA. The only way around this is with a mechanical switch. The current battery setup doesn't allow this. You could easily add a switch with your setup.
Note: Have not checked R3 rev 9 or 10.
Note: eneloop rated @ 1C discharge and 1C charge max. For you that's 3.8Ah, no problem in a Roomba.
Roses are red Violets are blue Poems are hard Flowers and Chocolate
http://www.daycounter.com/Calculators/
a1robotrepair
Robot Master
 
Posts: 770
Joined: October 14th, 2013, 8:40 am

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby LazyRoomba » August 18th, 2016, 4:11 pm

A good AA battery can hold upto 2000 mAh.. 12 are required for the 14.4 volts. 2 packs would give a theoretical capacity of around 4000 mAh.
LazyRoomba
Robot Addict
 
Posts: 179
Joined: May 9th, 2013, 6:22 am

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby glnc222 » August 18th, 2016, 5:11 pm

About low voltage: Neato's preserve about a 30 per cent capacity buffer against deep discharge which shortens cell life. Roomba ought to do the same, standard practice, used in hybrid cars etc. Returns to base or "needs charging" error, way before full discharge and the minimum voltage spec. Minimum voltage cut offs of the battery are for extreme conditions only, failures in the systems.
When test cycling batteries caution is needed not to damage cells by over-discharge. I used another circuit just for discharging packs to get full charging cycle data.
When soldering cells precautions are needed against accidental shorts, covering connections etc.
glnc222
Robot Master
 
Posts: 4409
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby Atlantis » August 18th, 2016, 8:29 pm

glnc222 wrote:Roomba ought to do the same, standard practice, used in hybrid cars etc. Returns to base or "needs charging" error, way before full discharge and the minimum voltage spec.

Actually that is more complicated - Roomba use (at least) 2 ways to determine remaining capacity - It counts energy going in and from battery. So it knows roughly how much energy it has left. There is also "0% checkpoint" at fixed voltage=12V.

If the roomba discharges and reaches some %(didn't catch the moment) it changes color to orange, if it reaches 10% it changes to red. At this moment it should definetly head to dock (maybe even at orange). But if you have complicated house layout and it gets lost etc. it will continue co clean until it reaches 12V. After this it will say "please charge roomba" and it is done.
Also if you manually start roomba it will not head to dock and you can start it even with red light with battery at 0%.

btw - maybe you saw some instructions how to increase battery capacity by fully charging and letting roomba find dock without really having dock plugged in - this forces roomba to discharge to 12V no matter what capacity it has stored in memory. Doing this full cycle it writes new (higher) capacity value in memory. Because if for some reason it has in memory for example 2000mAh it will head to dock at (for example) 500mAh reamining and never fully discharge even if the battery has 3000mAh for example.
Also when I replaced my battery with higher capacity I had to do full discharge and I got red light for like 1hour :lol: when the voltage reached to 12V it died and changed battery capacity stored in memory from 2696mAh to 3600mAh.

This 12V voltage threshold can be changed by replacing one resistor forming voltage divider.
IMG_20160819_021735.jpg
Voltage divider for battery voltage measurement at roomba 581

One of them is connected to battery, the second one is to ground. In ther middle is voltage going to opamp and CPU probably.
I am going to change it to little higher voltage because down to 12V (1V/cell) there is no useful capacity and there is high risk of weaker cell overdischarge.
This change will affect also voltage measurement during charge but I hope it won't have any effect because it should terminate on voltage or temperature slope and not at fixed voltage. Will post update after few days

PS: Today my "Eneloop Roomba" cleaned 3.5hours (80%time=hard floor, 20%=carpet)
Atlantis
 
Posts: 7
Joined: August 16th, 2016, 7:15 pm

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby glnc222 » August 18th, 2016, 9:54 pm

Voltage should not matter much to charging with thermistor control. Charging voltage will be 1-2v higher than the discharge voltage at the end. Voltage rises as needed for a constant current.

Open chargers for individual cells use voltage termination where a tiny drop, .01v, in the charging voltage occurs when full, as the battery converts more current to heat. Thermistor control can involve a second phase after cooling at low current for a fixed time, to top off the charge. Lithium charging is simpler and based on reaching the maximum voltage, with no inherent heating. The maximum voltage must be strictly observed to avoid damage.
glnc222
Robot Master
 
Posts: 4409
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby a1robotrepair » August 18th, 2016, 9:57 pm

Changing R207 from 475K to 500K changes the lower threshold from 12V to just above 12.5V. This worked great for a Li protection board I used that tripped the lower protection at 12V, occasionally resetting the bot.
Ref: R207 66D 475K, R208 01D 100K... VBAT to R207 to U8-5 & R208 to R235//R257 to GND
SMD Resistor Codes: http://www.marsport.org.uk/smd/res.htm
Last edited by a1robotrepair on August 19th, 2016, 6:26 am, edited 1 time in total.
Roses are red Violets are blue Poems are hard Flowers and Chocolate
http://www.daycounter.com/Calculators/
a1robotrepair
Robot Master
 
Posts: 770
Joined: October 14th, 2013, 8:40 am

Re: 24 Eneloop AA batteries Roomba 581 modification

Postby glnc222 » August 18th, 2016, 11:18 pm

I tested Eneloop at 1 amp load, twice that of the old Roomba 500 per cell, so see how it lasts.

The maximum drain rate spec is for where the internal resistance can still support the spec voltage. Cycle longevity claims are based on smaller loads and slow charging. The cells used in Neato's support much higher drain rates than used by the bot. Still you are at a good fraction of the max and might well have good results (apart from the higher cost). Since the regular batteries last a couple hundred or more cycles, it could take months to be surprised -- or want another bot anyway.
glnc222
Robot Master
 
Posts: 4409
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Next

Return to Robotic Hacking

Who is online

Users browsing this forum: a1robotrepair and 346 guests