Possible to convert to li-ion?

[Xamindar]

I have read several posts of people talking about using lithium ion batteries in their roomba. I am interested in building a li-ion pack using 8x18650 cells and a protection circuit for my 560.

My main question is, will the roomba be able to charge this type of pack? Does the roomba charger cut off at a specific voltage or does it cut off when the pack gets hot? Does anyone know how the roomba charger detects when the battery is fully charged?

Thanks for any help. I found one site talking about this but it is for the older roomba models and says you can't use the roomba to charge li-ion packs.

[sageman]

vic sells a li-ion pack

http://www.vic7767.com/li-ion-batteries ... xx-roomba/

[mfortuna]

Older 400 series work fine with Li-ION packs. 500 series get a chargiing error when end of charge occurs and they are on a dock. The battery is charged but the error may prevent schedules from running.

[Xamindar]

Any idea why the charging error? Can I still manually start the roomba or do I then have to remove the battery to "reset" it? That's not too bad, at least it charges it. Thanks.

I did see vic's site but all the li-ion batteries there were for the older models and he doesn't give any information on how to build them, just sells them. It seems like more of a sale site, not an info one.

[TechGuy]

At the end of charging, the Li-Ion internal controller disconnect the Li-Ion cells from the battery terminals. 500 Series Roomba detected as Err5 charging error.

You can used the charged battery in manual mode. But the automated schedule cannot be started. Roomba needs to be off the homebase or charger. The Err5 comes back within a minutes after you put back the Roomba with a fully charged Li-Ion battery on the homebase or charger. While it is not charging, the IR system will use some of the stored energy between the time you take the Roomba off the homebase or charger and the time you start the next mission.

I have a Li-Ion battery and I do not schedule my Roomba. I usually start my Roomba and go around the house to pickup stuff off my floor. I use Li-Ion 18650 cells. There are 2 cells with a controller in a package. I put two 2-cells-pack in series to get 14.4V. I put two 14.4V sets in parallel to get more mAh. Therefore, a total of 8x18650 cells.

IMPORTANT: Don't switched the polarity of the rebuilt battery. Otherwise, you will be kissing your Roomba good-bye.

[Xamindar]

So basically, the error 5 happens because the battery gets "disconnected" by the protection circuit kicking in. That doesn't sound too good as it means the charger will just charge forever if it doesn't detect certain characteristics of nimh batteries. I assume the roomba has a Delta V equip charger then.

I have a Li-Ion battery and I do not schedule my Roomba. I usually start my Roomba and go around the house to pickup stuff off my floor. I use Li-Ion 18650 cells. There are 2 cells with a controller in a package. I put two 2-cells-pack in series to get 14.4V. I put two 14.4V sets in parallel to get more mAh. Therefore, a total of 8x18650 cells.

Thanks, that's really good info. I was thinking of getting two of these and having two sets of 14.4 packs in parallel for more mAh. But I haven't measured the space yet to know whether it all will fit or not.

[mfortuna]

The theory is when the protection circuit kicks in the 500 series sees the open circuit voltage from the battery charger which is in the range of 22V. The home base sends out a pulse at regular intervals and thes pulse causes the voltage detected to drop. The roomba sees the drop and declares a charging error. If the power supply is plugged directly intot the roomba the pulse doesn't occur and the roomba will charge fine but cannot run a schedule.

The roomba charging circuit will shut down and not continue charging when it detects the error. It will not charge forever.

[vic7767]

I did see vic's site but all the li-ion batteries there were for the older models and he doesn't give any information on how to build them, just sells them. It seems like more of a sale site, not an info one.

That site is both. There are "How Do I" pages and also offers for goodies to enhance your iRobot robot.

There are LI-ion batteries offered by ProTechRobotics that do not cause the err 5 failure but certain criteria must be met to charge properly. You can check there for more info.

[Gordon]

I have read several posts of people talking about using lithium ion batteries in their roomba.

You ought to read this thread, starting about here: http://www.robotreviews.com/chat/viewto ... 933#p66933

I am interested in building a li-ion pack using 8x18650 cells and a protection circuit for my 560.

I won't say it can't be done, but I think using 18650 shaped cells makes the task more difficult!

... I found one site talking about this but it is for the older roomba models ...

The brandeis.edu illustrations show why it would be easier to use 18650 shaped cells in R2 Roomba -- they have a container to hold them.

The author of that brandeis.edu page admits to not being an expert, and while I too am not expert in these matters I was able to spot what I think is an error in his connection tactics. I am not certain how much hazard there may be, but I am fairly certain that cells in his pack will not provide anywhere near their rated capacity.

First, we should get a grasp on how his cell protection circuit functions. Take a look at this schematic, I think it fairly represents the circuit on the PCA pictured on the brandeis.edu page:

At far right, you can see four cells connected in series (4S), which is the configuration this PCA was designed to serve.

At bottom right corner the is a resistance named R_sense. It serves to tell the controller how much current (charging or discharging) is passing through it.

Above and to the left are two power MOSFET switches. Only one at a time is set into channel conduction. For example if discharging is occurring, the right FETs channel would be conducting, and the current path through the left FET would deviate to pass through its bulk-diode and on out to the load. When charging, current 'flow' reverses and the two FETs exchange roles.

Since there is only a single set of these protection switches (FETs), when the controller chip (shown as "S-8254A") senses that any single cell has reached the specified limit voltage (either HI or LO) the appropriate FET switches OPEN and current flow stops. Current stops whether untapped energy remains in the other cells, or whether other cells have not completed charging.

Now, let's take a look at what brandeis.edu said he did. Next I show the same picture, but with another four cells connected in // with the original four -- giving what I call a 4S(2P) configuration:

In this case, the controller chip treats the // cells' data as one datum -- it can not discriminate between either of the two cells, so the battery continues to 'look like' a 4S pack.

Cells will, in general, not be exactly alike. As this pack discharges, I think it will be possible for a stronger (higher capacity) // partner to hold voltage a bit higher than its weak mate (but, not a huge voltage difference can occur before the strong one starts charging the weak one). This could be a latent hazard, or simply be a source of lost (not provided to the working load) capacity.

The most obvious discharge issue with this 4S(2P) pack is when voltage across one of the pairs falls to the LO-limit, poof!!, the entire pack goes off line!

Charging the 4S(2P) will tend to run into the an overall capacity loss. Whichever cell (//)pair that reaches the HI-limit cut off first will dictate how much charge has been stored in the pack.

I think TechGuy talked about building Li-ion packs that use 2P(4S) wiring, and with each 4S string being connected to its own 4S protection circuit.

Other than doing that, it would be feasible to wire eight cells which have their own IC, such as cellular phone cells, in the 4S(2P) configuration.

[Xamindar]

That site is both. There are "How Do I" pages and also offers for goodies to enhance your iRobot robot.

There are LI-ion batteries offered by ProTechRobotics that do not cause the err 5 failure but certain criteria must be met to charge properly. You can check there for more info.

Your site is pretty good for other things and all the items you sell on there seem to be priced pretty reasonable.

I did see the ProTechRobotics li-ion batteries. But at the price they charge I can buy 5 nimh batteries for my roomba. It wouldn't even be close to cost effective to go that route. Their site has no details on how they got it to work without throwing an Err5 so it doesn't help much.

I have a bunch of 18650 cells here already and new ones in this size are cheap and easy to get so I would consider this a more worth while project. Plus I like to do things myself and learn from it. I just sized the roomba battery compartment and 8 of them will fit with plenty of space to spare for the protection circuit. The only problem is finding a way to build the pack so it fits snug in there.

[Xamindar]

I think TechGuy talked about building Li-ion packs that use 2P(4S) wiring, and with each 4S string being connected to its own 4S protection circuit.

Other than doing that, it would be feasible to wire eight cells which have their own IC, such as cellular phone cells, in the 4S(2P) configuration.

What you just explained is an issue I am thinking about. With all the sells in one protection circuit the weakest one will determine the capacity of the pack.

If I get a bunch of single cell protection circuits like these and then connect the cells in 4S(2P) wont the whole series set (4 cells) stop charging the moment one of the cell's protection circuits kicks in?

[vic7767]

There are two issues that have to be visited when attempting to build a LI-ion solution for a 500 model. When using single cells and a PCB on each all the cells need to be balanced and have similar characteristics during charging and discharging since as mentioned earlier one cell will be the weakest link.

The second issue will be providing electrical connections to the spring contacts mounted on the Roomba PCA for both power and temperature and a way to hold it all together.

[Gordon]

...If I get a bunch of single cell protection circuits like these and then connect the cells in 4S(2P) wont the whole series set (4 cells) stop charging the moment one of the cell's protection circuits kicks in?

I think not, if you select and wire them to avoid early switching open. Notice the 4S(2P) pack has eight cells in it, not the "(4 cells)", i.e., 4S. The weakest cell in a 4S pack would dictate the shut off in a stand alone load test. However, Roomba will still be thinking NiXX chemistry and will shut off at the 12-volt crossing. A 4S string of IC-protected Li-ion cells, all capacities being equal, would disconnect when the battery's terminal voltage reached V_co = 4*2.4V = 9.6V, but when powering Roomba, each cell would be at a three volts level when the robot calls halt.

Assuming you do wire eight individually protected cells by paralleling four pairs, followed by wiring those four sub-packs in series (to make the 4S(2P) battery), and if you were careful to not pair the two weakest cells in the lot of eight, then the weakest cell would switch off line first (thus throwing the full load current onto its // partner), and if that partner happened to have more reserve capacity than the 2nd weakest cell, the battery could continue powering Roomba beyond the point where the 2nd weakest cell went open circuit.

From that point on, full current would be drawn from two cells, and they would be giving up their charge faster than the remaining two pairs. To predict which cell would next go open requires knowledge of individual cell characteristics.

If you don't obtain cell characteristics using your own empirical tactics -- so you have, at least, some idea of how the eight should go together -- you should do intra-wiring of cells in such a way that you can easily modify the electrical groupings following some trial Roomba missions.

BTW, Batteryspace does not provide enough information about that single-cell protective PCA. Its photo is lo-rez and I can't rev-engr its circuit, so I can get a little more comfort that it can do what is needed. Keep in mind that that battery supplier is the outfit that (ca., 2005) sold 4XXX Roomba drop-in packs with wiring instructions that reversed the battery polarity, Quite a number of Roombas got killed, and there was little to no customer satisfaction!

You need to be aware that:
a) Roomba's PSU will current limit up around 1.4A when it starts filling an empty battery.
b) Until one cell's PCA opens its switch, the high current will be shared between // (4S(2P)) cells, but after one cell opens at HI-limit, the full 1.4A will then pass into the // partner cell. Therefore, you would like to be charging a 1400mAh capacity, or greater, cell capacity -- if you intend to adhere to a 1C max charging input.
c) Some attention to "temperature" is required, while some could be ignored.
d) Roomba detects presence of a connected battery by measuring resistance of the battery's thermistor. Therefore, your Li-ion pack should include a dummy thermistor, viz., a 10k ohms resistor. Roomba will run missions w/o a thermistor, but it won't start charging.
e) If 1C, or less, charging a Li-ion cell heats the cell, there is something wrong with the cell.
f) The batteryspace single-cell PCA appears to have a "T" solder pad on it, which could be for connecting some sort of temperature sensor that has been bonded to the cell it is protecting. Rev-engrg the PCA seems to be necessary to figure out how the "T" pad may be used.

[Xamindar]

A 4S string of IC-protected Li-ion cells, all capacities being equal, would disconnect when the battery's terminal voltage reached V_co = 4*2.4V = 9.6V, but when powering Roomba, each cell would be at a three volts level when the robot calls halt.

So they really should never cut out as a result of overdrain while running in the roomba unless there are two really weak cells put together.

Assuming you do wire eight individually protected cells by paralleling four pairs, followed by wiring those four sub-packs in series (to make the 4S(2P) battery), and if you were careful to not pair the two weakest cells in the lot of eight, then the weakest cell would switch off line first (thus throwing the full load current onto its // partner), and if that partner happened to have more reserve capacity than the 2nd weakest cell, the battery could continue powering Roomba beyond the point where the 2nd weakest cell went open circuit.

Ok, I understand it now. Thanks for the clear explanation.

From that point on, full current would be drawn from two cells, and they would be giving up their charge faster than the remaining two pairs. To predict which cell would next go open requires knowledge of individual cell characteristics.

If you don't obtain cell characteristics using your own empirical tactics -- so you have, at least, some idea of how the eight should go together -- you should do intra-wiring of cells in such a way that you can easily modify the electrical groupings following some trial Roomba missions.

I have used a hobby charger to charge all my 18650 cells and label the capacity the charger reported. They all range between 1200mAh and 1600mAh. I will probably do it again to each cell when I get to the point of building the pack just to make sure. With your explanation, it seems the best way to arrange them would be to put a stronger cell with a weak cell and not to have two weak cells paired up together.

BTW, Batteryspace does not provide enough information about that single-cell protective PCA. Its photo is lo-rez and I can't rev-engr its circuit, so I can get a little more comfort that it can do what is needed. Keep in mind that that battery supplier is the outfit that (ca., 2005) sold 4XXX Roomba drop-in packs with wiring instructions that reversed the battery polarity, Quite a number of Roombas got killed, and there was little to no customer satisfaction!

I remember reading about that a while back, good to know these are the people who caused it. Unfortunately I haven't been very successful in finding alternate suppliers. It would be nice if someone made a 4S(2P) all in one circuit that functioned in the way you described. I'll keep looking.

You need to be aware that:
a) Roomba's PSU will current limit up around 1.4A when it starts filling an empty battery.
b) Until one cell's PCA opens its switch, the high current will be shared between // (4S(2P)) cells, but after one cell opens at HI-limit, the full 1.4A will then pass into the // partner cell. Therefore, you would like to be charging a 1400mAh capacity, or greater, cell capacity -- if you intend to adhere to a 1C max charging input.

With my current cells it looks like it might be close. But if this pack turns out to work well then I will probably buy larger capacity cells for it.

c) Some attention to "temperature" is required, while some could be ignored.
d) Roomba detects presence of a connected battery by measuring resistance of the battery's thermistor. Therefore, your Li-ion pack should include a dummy thermistor, viz., a 10k ohms resistor. Roomba will run missions w/o a thermistor, but it won't start charging.

What about just ripping the old thermistor out of the old battery pack and reusing it along with the terminal plate?

So far:
-At a good price I found these holders. They look like they would work well but I haven't checked the measurements yet. They would allow me to change out the batteries whenever I want.
-For protection I am so far looking at this one as it seems it would fit on the side of those holders better. Of course, if there was one PCB that did what these 8 little ones do then it would be better.

I'm still researching at this moment. So far this is pretty fun and thanks for all the help.

[Gordon]

A 4S string of IC-protected Li-ion cells, all capacities being equal, would disconnect when the battery's terminal voltage reached V_co = 4*2.4V = 9.6V, but when powering Roomba, each cell would be at a three volts level when the robot calls halt.

So they really should never cut out as a result of overdrain while running in the roomba unless there are two really weak cells put together.

I can't reply to a question that seems to have a parallel cells basis, but refers to a series connected string.

...Ok, I understand it now. Thanks for the clear explanation. ...
I have used a hobby charger to charge all my 18650 cells and label the capacity the charger reported. ... With your explanation, it seems the best way to arrange them would be to put a stronger cell with a weak cell and not to have two weak cells paired up together.

That's what I'd try.

...c) Some attention to "temperature" is required, ....

What about just ripping the old thermistor out of the old battery pack and reusing it along with the terminal plate?

Yep. Cheap and easy!

...At a good price I found these holders. They look like they would work well but I haven't checked the measurements yet. They would allow me to change out the batteries whenever I want.

Yes, they provide that sort of assistance, but probably at an operational price you can't afford to pay. I am thinking about contact resistance and the voltage drop that will result. Let's say each cell-terminal to spring contact develops 0.2 ohms of contact resistance. For simplicity in estimating voltage loss I will refer to a 4S pack which would have eight contacts, giving 8*0.2 ohms = 1.6 ohms total contact resistance. When Roomba draws between 1.5A to 2A from that battery the series resistance would result in dV ~ 1.7A * 1.6 ohms = 2.7V reduction at battery terminals, which puts you just that much closer to the 12-volt shut off level.

Use welded or soldered connections between cells.

Even w/o that problem, the mechanical constraints of fitting four of those holders into a 5XX-Roomba's battery pocket will probably prohibit their use.

-For protection I am so far looking at this one ...

They appear to use same parts as the circular PCAs, just a different layout.

[Xamindar]

There are LI-ion batteries offered by ProTechRobotics that do not cause the err 5 failure but certain criteria must be met to charge properly. You can check there for more info.

Funny coincidence with this. A couple of days ago I found a non-charging 560 on craigslist that came with some nice accessories and a li-ion from protech robotics for a nice price. I was able to repair the 560 thanks to your guys on these forums.

Unfortunately vic, the protech robotics li-ion battery is nothing special. The roomba still throws an err5 if you charge it on the dock. Works fine directly plugged in though. Protech seems to have non-existent customer service as I have been unsuccessful in receiving any replies from them. I am also finding tons of terrible customer experiences reported about them by various people on the web. Too bad, it seemed like the most hopeful option.

Regarding the err5, their site simply states: "If your power supply is below the manufacturers rated 22.5v needed you may experience an Err5 code that will not affect your roomba, a replacement power supply may be required to complete the charging sequence to 100% capacity and resume normal operation." My power supply IS 22.55v so it should meet. I assume the dock knocks the voltage down a little. If this statement is true, is it possible to connect a power supply to the dock that puts out something like 23 volts to compensate? Or would that damage it? Protech robotics have been no help at all with the vague statements on the site and 0% response to inquiries.

[Gordon]

... The roomba still throws an err5 if you charge it on the dock. Works fine directly plugged in though. ...

You might study, then try, the hocus pocus operation that Spirit Force writes about in this post: http://www.robotreviews.com/chat/viewto ... 901#p68901 , and his preceding posts in the same thread.

[Xamindar]

That thing Spirit Force wrote about didn't work. But I think I solved it. I took the home base apart and did some measurements. By the time the 22.5 volts goes through the home base it is down to about 18 volts going into the roomba. I have an old 24 volt power supply that happened to have the same connector as the irobot power supply and tried that instead. I then docked the roomba with the li-ion battery installed and let it charge. This time, after charging for a while the clean light went green as it should and no err5. I am now letting roomba run a full cleaning mission where it will dock on its own at the end to charge. We will see how that goes. I hope the 1.5 volt increase will not strain the home base too much.

[Gordon]

That thing Spirit Force wrote about didn't work.

Thanks for checking. I could not see why his process should work!

But I think I solved it. I took the home base apart and did some measurements. By the time the 22.5 volts goes through the home base it is down to about 18 volts going into the roomba.

There are two reasons for voltage to be low where you made that measurement: 1) dV dropped across the MOSFET SW and blocking diode within the HB, and 2) because charging current is (apparently) high enough to cause the PSU's regulator circuit to lower its output voltage. As charging progresses, and cells take on charge, battery voltage rises and reduces the current draw. Eventually, the current regulator regains control at 1.25A, and the place where you measured "18 volts" will be around 20 to 21V.

I have an old 24 volt power supply ... and tried that instead. I then docked the roomba with the li-ion battery installed and let it charge.

It would have been wise for you to have measured those "volts going into the roomba" as that recharge began; and also wise to have measured charging current. If you had those measurements I think current would be higher than preferred, and the voltage datum would also be too high.

This time, after charging for a while the clean light went green as it should and no err5.

Err5 occurs when "...the full charging current is below 400mA for more than 30 minutes.". For that to be true your Roomba must have commanded high-rate (1.3A) charging to cease before charging was terminated by either Roomba's controller, or by the Li-ion FETs switching open.

. I hope the 1.5 volt increase will not strain the home base too much.

"Hope" won't do it! Be sure to also worry about the robot's main PSA. You must know that the alternate (24Vdc) PSU has current control that is similar to the iRobot PSU. Even if its current control is similar the HB's FET and HB's power diode will be dissipating more power than planned; and, the charging FETs and series power diode on Roomba's main-PCA will all be dissipating more power than planned. To reduce dependency on "hope", you must get more quantitative and learn how the charging circuit functions (the power handling portion, not necessarily the firmware's instructions).

[Xamindar]

The 24V power supply is a 1A power supply which I figured is pretty close to roomba's 22.5v 1.25A power supply regarding power output.

It would have been wise for you to have measured those "volts going into the roomba" as that recharge began; and also wise to have measured charging current. If you had those measurements I think current would be higher than preferred, and the voltage datum would also be too high.

Ok, I may not be "wise" yet because I have never blown up an electronic device as a result of modifying it. I will probably learn my lesson some day. I finally got some time to measure voltage and current using the 24V psu. I took measurements where the docking tab cable plugs into the HB board.
While charging: 18.5 volts, 1.55 amps
While charged: 22.8 volts, 0.14 amps
This to me seems acceptable. It has been working the past 4 days perfectly and no more errors. I am of course open to any more insight if you feel it helps. Thanks again for the detailed help and explanations thus far.