Neato lithium ion battery revisited

News and information about the Neato XV-11 Robotic Vacuum. All discussion and troubleshooting questions go here.

Re: Neato lithium ion battery revisited

Postby glnc222 » March 12th, 2014, 9:35 pm

There is no commercial supplier as yet. This stuff was for anyone else who might be trying to build such a battery, to maybe save them some time, or show possibilities not thought of. The battery has not been tested yet, though all the parts have been separately. I don't expect to be doing it myself for maybe a year.

I think it has been almost completely proven, but there is a risk if you try it yourself.
I can give you any more detailed instructions required privately; you would have to assemble circuits on perfboard. Depends on your experience and skills, hopefully better than mine when it comes to the materials themselves.

It is really supposed to be for getting commercial and custom battery suppliers to take an interest and do the rest of the work.

The smaller size of the Neato battery market compared to iRobot deters larger battery suppliers, but at the same time, creates an opportunity for smaller players. Early bird gets the worm.
Yet there is always the risk that Neato Robotics itself might offer a new type of battery. Everything they do, however, is bound to be high priced.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » April 8th, 2014, 10:09 pm

More Lithium Neato Batteries in Germany

14 in all. Several are LiMn (Lithium Manganese Dioxide, one of the safer alternatives to LiIOn)
Some use a Carson brand balancer
([edit] 4/27/2014 I am informed Carson Model Sports in Germany has discontinued production of their 8-cell balancer (listed in the German catalog only, not the english language site).
The heat from balancer bypass resistors can trigger thermistor control of charging termination, though a roundabout way of doing it.

A German homebrew balancer and battery: 4 LiMn 18650 size per side. 70 minutes run time. Improvements in progress, electronic thermistor control, protection circuits. Balancer schematic

Roboter-Forum reports: thread ... ade/page10



Last edited by glnc222 on April 27th, 2014, 3:24 pm, edited 1 time in total.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » April 11th, 2014, 5:41 pm

Sony Laptops Recalled For Battery Fire Hazard
4/11/2014 Wall St. Journal
TOKYO— Sony Corp. 6758.TO -0.16% said Friday there was a risk that batteries made by Panasonic Corp. 6752.TO -1.55% in nearly 26,000 of Sony's newest Vaio personal computers could overheat and catch fire, the latest product glitch for the Japanese electronics and entertainment giant.

As a result, Sony is asking customers to stop using its Vaio Fit 11A laptop as soon as possible.

Sony said it had received three reports of batteries overheating causing partial burns to Vaio computers. The first incident was in Japan on March 19, followed by similar incidents on March 30 in Hong Kong and April 8 in China. The company stopped selling the product at the beginning of this month.

Why use LiMn (Lithium Manganese Dioxide) or LiFePO4 (Lithium Iron Phosphate) types -- inherently safe like NiMh.
[edit] Also equipment in use which has proved safe is probably not at risk. These are mfg defect problems.
The safe chemistry types are insurance against those for consequential damages. Cell phones and laptops can be handled when there is a problem. Vacuums running or charging while asleep are something else.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » April 26th, 2014, 2:40 am

Completed Tests Of A LiFePO4 Battery

The charging adapter circuit charges the battery on Neato with some limitations noted below.
The battery cells proved to be of much less capacity than expected, measuring only half the theoretical size.

Two Charging Regimes

Neato's charger has two major divisions, with the indicator blinking orange or green. The manual describes these two states in terms of accumulated charge. Neato has charge for a short run when blinking green and cannot clean in the orange state. Full charge is indicated by solid green. How Neato measures the charge present on a newly connected battery remains a puzzle.

When a fully charged NiMh battery is re-connected after a shut down, Neato charges in the orange state briefly and then switches to the green. When a LiFePO4 battery with similar voltage is connected, Neato remains in the orange state and will not transition to the green.

The LifePO4 battery must be charged from a substantially discharged state. If nothing else this allows Neato to record an accumulation of charge.

When in the green state Neato completes charging instantly on a sensed rise in battery temperature. The rate of increase in temperature is used to distinguish battery heating from the background environment temperature changing with location and season.

One difference between NIMh and Lithium charging involves the gap between the charging voltage and mostly unloaded discharge voltage when the charger pauses. A difference as large as two volts can be observed on a full NiMh battery, while the difference was not more than one volt on LiFePO4. Whether Neato responds to this difference is unknown.

External Charger

An external SmartCharger for 18v NiMh packs had a simpler operation. It completed charging immediately on the first rise in temperature. No indication of accumulated charge similar to Neato's color coded light is provided, so the simplicity is not surprising.

Orange State Charging

The orange light state has a complex pattern with the LiFePO4 battery (when already charged). Each time the temperature rises Neato pauses charging, waiting for the battery to cool. Charging resumes when the temperature falls. The delay between a temperature rise and the pause in charging increases as charge-and-pause cycles are repeated. ([edit] or the rate of voltage increase accelerates, did not observe closely enough to tell.) The voltage at which the Neato pauses increases slowly across cycles as a result. After twenty minutes and numerous cycles, the over-voltage protection on the battery is triggered. The protection used here is disconnection of the ground circuit to the rear docking contacts, simulating a withdrawal of Neato from the dock. In that event Neato performs its wiggling motion attempting to improve the connection, giving up after a few seconds. For tabletop work the Neato and dock are raised on blocks.

When the LiFePO4 battery is mostly discharged, the orange state transitions to green and the charging completes immediately on the first rise in temperature.

Reports from users of Lithium Ion or Lithium Manganese batteries lack detail needed to determine if the empty battery requirement applies only to LiFePO4. [edit] I do not have complete information on German examples. I hear some charge well when full, others require unusually long times. More info may become available. There are language problems.

[edit] Lithium ion differs from LiFePO4 in having a lower 16.8v maximum voltage and reaching that level at smaller fraction of the capacity. Though why that would matter to Neato's algorithm is unknown. I wonder if what Neato is doing is trying to loosen up NiMh batteries which are "sticky" from excessive shelf-life, by repeating heating cycles at increasing charge voltage. The simple external charger clearly does nothing like the elaborate procedure observed.

Charger Control Method

There was some difference in the temperature control method in the fully charged perpetual orange state case and the successful charging. The repeated cycling used a temporary temperature rise held until a few seconds after the voltage falls from the charger pausing. The empty battery charging held the temperature high until manually released a couple minutes after charge completion. An NiMh battery might take half an hour to cool off in the enclosed compartment. The rapid termination on the temperature rise suggests holding the temperature is probably not a determining factor. Only more detailed tests would be definitive.

Adapter Circuit

The temperature manipulation circuit used is based on the ICL7665 dual voltage monitor IC, which is easier to use than the LMV431 voltage reference in previously shown circuits. This IC is made to implement under/over voltage monitoring of power supplies and batteries. The LMV431's and transistors used before got damaged several times from unknown causes, and the ICL7665 is a socketed IC with no soldering problems. Plus only one IC is needed compared to several transistor packages. The 16-18v power supply limit of these IC's require application to the low side pack (larger of the two used) instead of to the entire battery.


[edit]This simplified version without delay proved adequate:

[edit] small capacitors shown prevent power-on surges damaging the sensitive IC, especially needed in battery circuits compared to line power supplies. See datasheet.

For these experiments the adapter is built externally on solderless breadboard, and connected to the battery only for charging. A satisfactory battery is needed before a permanent construction is warranted.
These circuits can be adapted to other types of lithium cells with different resistors governing the activation voltage levels. On the NiMh charger lithium can be charged to the point where the maximum cell voltage is reached, a point with only fifty percent or so of the capacity used on other types of lithium (beyond which lithium chargers use a constant voltage method not available on NiMh chargers). However, the much larger capacity in which those types are supplied can equalize the effective capacity across types. Neato will utilize only about 2ah of charge under programmed limit. Increasing the battery size does not affect run time.

Some refinements to the circuit can reduce the quiescent load, which is still small enough; high precision resistors were on hand in the sizes used.

Over-Volt Protection

The second half of the ICL7665 is used for over-voltage protection with the mosfet circuit in prior posts.
The disconnect circuit is a separate small circuit board installed at the docking contacts connector in the end of the wheel well. A single thin wire wrap 30 gauge wire connects to the battery compartment under the fan and battery covers, where space exists without modification. The adapter circuit detects the over-voltage condition.


The method here also insures Neato's charger does not damage Neato's system board, which occurred in previous experiments where the charger reached 21v. There is a risk of this happening with any sort of defective battery. The exact conditions creating the effect are unknown, so how to avoid them remains a problem.
In these tests charge termination was set at 17.9v for LiFePO4 and over-voltage at 18.25v, same as the cell specifications and commercial protection modules.

Note on commercial protection boards: an inexpensive small protection board for two-cell LiFePO4 was initially used and failed when needed. Subsequent testing with instruments showed no over-voltage protection claimed, and a refund was obtained from the seller on ebay. There are quality issues in such components. Perhaps these boards were actually Li Ion boards which have higher voltage limits, and were sold as LiFePO4 boards. I suggest any such products obtained be tested. I used resistors in place of cells and a variable voltage supply from a potentiometer and flashlight batteries.
Had this circuit worked a Neato system board would not have been damaged.

The Battery Cells

The battery was assembled from ten 18650 size cells similar to Neato's normal battery. The cells are connected in five parallel pairs in series (5S2P configuration). The largest cells available with pre-welded tabs for soldering were labeled 1500mah, for a supposedly 3000mah total. This type of lithium can be charged to 95 per cent capacity with the constant current method used by NiMh chargers. The battery acts more like only half the rated capacity, as though it were only single cells in series. There are commercial LiFePO4 packs using such parallel pair arrangement, so it is supposedly practical.

Twenty five minutes discharged to nearly empty on a 2 amp lamp load. The battery charges at Neato's 2 amps in similar time. Neato cleaned 28 minutes on carpet with this battery (which would be fifty per cent higher on hard floors), about half the capacity of a Neato's NiMh battery, of which only 2ah is used. Whether the tab soldering in parallel is faulty or the cells merely over-labeled is unknown.

These cells were brandless Chinese make with a "UN Certification", referring to some sort of standards program for marketing. Still, a 100 per cent over-labeling is unusual. A higher rated 1800mah cell is sold but without soldering tabs. A $120 spot welder would be needed to assemble packs.

The cells are arranged with three pairs in the low side compartment and two pairs in the high side compartment, leaving room for adapter electronics. Assembly is simple using crazy glue ([edit] epoxy may be needed in some cases) to fasten pairs with tabs aligned for soldering, and then into stacks of pairs. When soldering small bar clamps were placed over the ends as protective heat sinks, though the necessity is unknown. Scotch heavy duty packaging tape makes an excellent wrapping, and thin for the tight fitting packs, compared to thicker duck tape. The older Neato's have an aluminum heat sink siding to the battery compartments which pinches the sides in around 1/16", making some packs a very tight fit.



Lithium Ion and LIthium Manganese are available in higher capacities than LiFePO4 with its lower charge density. LiMn shares with LiFePO4 the same inherent safety as NiMh, avoiding dependence on the highest quality protection circuits to avoid hazards. The advantage of LiFePO4 is longer lifetime over cycles several times that of other types. Without cells with the requisite capacity, or you can use a less than full run time, this longer lived type does not provide a satisfactory power supply for Neato. Perhaps the assembly is faulty, but it takes a lot of stuff to analyze. Perhaps other cells will become available over time. There has already been a substantial increase in the capacity of NiMh cells offered, and new cells keep appearing.

[edit] Possible Next Step

An unusual possibility is noticed to install five 26650 size cells, rated 3300mah. The fifth cell can go in the side of the dust bin. They are 1 inch wide by 2.6" long, fitting all the way to the front, with plenty of space behind towards the filter. There is space to either side of the bin, a gap in the compartment width, for wire loops and D-pin butt connectors, and the fan compartment connects to the battery compartments at the bottom cover.

[edit] needs checking with actual cells.

[edit] another way

[edit] Neither cut out shown works after receiving an actual cell to check. Just a tad too big. I will show the way it could be done if the cells prove to have a useful run time. The cell would go in the top up front aside the intake, and be pulled to the side onto the Neato top when opening the bin. This allows room for thickness of Neato's fan-fold filters etc. in the back. Testing in progress.
Last edited by glnc222 on May 4th, 2014, 11:11 pm, edited 6 times in total.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby Methos » April 27th, 2014, 7:50 am


Why not used 18650 cells with protection, the right cells will have a pcb at one end which disconnects the battery at the correct voltage during charge and the same during discharge.

End result (I think) is that there would be no need for any electronics. The only reason I can think of why this might not work is if the neato needs the temp sensor to register some set values.

Posts: 1
Joined: March 20th, 2012, 3:25 am

Re: Neato lithium ion battery revisited

Postby mfortuna » April 27th, 2014, 7:57 am

Hi Mike. That is how this thread got started. Neato's algorithm does not like the disconnect that occurs during charging which causes a failure.
Reds x 3, Dirt Dog, Disco (now a parts bot), Create, Scooba 350, and Security Dawg
Evolution Mint
Neato XV-11
User avatar
Robot Master
Posts: 5767
Joined: February 5th, 2006, 10:35 am
Location: NH

Re: Neato lithium ion battery revisited

Postby glnc222 » April 27th, 2014, 12:38 pm

LiFePO4 cells have less need for balancing circuits than other lithium types. With aging of cells balancers extend longevity of lithium ion batteries. The protection circuits enclosed in lithium cells -- made for flashlights -- do not balance. So when a board must be added for balancing, some protection circuits are included as well, in some versions but not always. Also the over-voltage protection is relevant to charging only, so is included in chargers already, instead of the battery. The price of cell-placed end cap circuit batteries is also higher. Flashlights are also capable of over-discharge use so under-voltage protection is included in end caps. Neato includes this in its own internal circuits refusing to run on too weak batteries.

By sheer coincidence some balancers also heat up duplicating the thermistor control for NiMh batteries. This is a separate function not the primary purpose of the balancers.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » April 27th, 2014, 4:38 pm

Charging Rate Issue

I've seen nothing in the literature to suggest this, but am not an expert in batteries. It is a complex field. So I wonder if it is possible the LiFePO4 cells were charged only half way because of the 2 amp charging rate, and if charged at a slower rate would reach the 95 per cent level. Batteries are often specified for slow charging, with fast charging a sort of footnote.
Such cells are normally charged with a CCCV two phase lithium type charger. The second CV stage is supposedly only reached at 95 percent of capacity for LiFePO4, much higher than other lithium types. What if that is not always true, at least in fast charging? At the high amperage it might go to constant voltage phase earlier, and then get fully charged.
Each cell had rated capacity of 1500mah and was charged at 1 amp, in parallel pairs drawing 2 amps. So that is less than 1C charging rate, not exceptionally fast. This type of cell is rated for very high currents, why it is used as lead acid substitutes.
To test this I would need a single cell LiFePO4 charger not on hand. [edit] further research is underway but will take time.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » April 29th, 2014, 2:02 pm

Balancer Data From Germany

The Carson Model Sports eight-cell balancer has just been discontinued (a second such after the LBA-1 some months ago, because of revised offerings from component makers). German Roboter-Forum member King_Draki posted data logging of the Neato charging process using lithium ion cells with this balancer. There is a partly english dashboard option when registering, but posts are in German.
Google translation with added editing:

Here is shown timing of a charging process with Lipo and Carson balancer, Thermistor directly on the heat sink of the balancer.

The Lipo reached 3.8 V per cell.

Minute 20-51: charging 2A, battery voltage rises to 8.2 V
Minute 52: The first cell exceeds 4.2 V and the balancer begins to operate. The Temperature Sensor kicks off (gray curve). Phase 1 completed. [not the orange and green light phases, some sub-division defined elsewhere]

Minute 50-85: bypassing current 0.5 Amp. More cells are full and the balancer operates. At thermistor Temp of 42 ° C, the charging is paused. Until the balancer cools somewhat.

Minute 85-165: This little game is repeated until the charging is finished. The cells are then almost perfectly balanced. Maximum difference of 0.03 V between the cells was measured.

Conclusion: ideally plan on a 0.5 amp balancer current.

Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » May 7th, 2014, 4:58 pm

Another LiFePO4 Incompatibility Found

The maximum charging rate specified for the LiFePO4 cells turns out wrong. The ones obtained cannot be charged at the 2 amp rate of Neato's charger. Tests with a 26650 3300mah pack went over-voltage very quickly, while it charges smoothly on a .9 amp external charger.
The "standard charge" rate specified is 1.65 amps with maximum 3.3 amps. It appears the actual limit is much smaller, probably more like the standard rate. The standard rate is about 0.5C compared to Neato's nominal 1C.

Some experiments are being done with adding a constant current 0.5 amp bypass to the Neato charger, made of a voltage regulator with some controls. The 18650 cells previously found only half the rated capacity were recharged on Neato with this modification and charged perfectly in twice the time, to the same one half capacity.
I have a different set 1800mah to test eventually, but the incompatibility of charging rates would have to be solved.

Even if a bypass circuit works to charge the cells, it is not a convenient solution because of the heat dissipation. 10 watts makes resistors and heat sinks too hot to touch. For single run use it is simple to plug in external hot components when charging, but automatic recharging would involve fairly radical constructions such as large heat sink plates on the bottom or extra contacts for the dock.

An interesting side effect is that were external hot components used, Neato's charger could be converted to a full CCCV lithium charger and get the full capacity out of LiIon or LiMn cells, instead of the half charge possible terminating the NiMh constant current charger. That is assuming the fine details of Neato's charging can be emulated sufficiently to satisfy the software. There appears to be more to it than simply terminating NiMh charging, with the troublesome transition from orange state to green state.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » May 9th, 2014, 4:06 am

Full Run with LiFePO4

3300mah 5-cell 26650 size, fifth cell in the dust bin.
Constant .5 amp bypass applied to match lower cell charging rate, with external hot components.
Initial charge from 1/3 full 1hr 15 minutes, switching from orange to green after 35 minutes.

Run time on carpet 45 minutes, limited by space, not capacity. Still a recharge is requested before running further.
The NiMh cells could usually run at most 15 minutes more, but also request a recharge after 45 minutes.

Recharge time similar 48 minutes, switching from orange to green after 25 minutes. Very similar to NiMh recharge time.

Recharging was at a fairly high voltage level on the discharge curve, so it might work on short runs, but this remains a question. A lot of testing is required to know all the characteristics.

Repeated runs will be needed to know whether all factors have been accounted for. Circuits were only temporary on external breadboard for this test.

circuit revisions
Additional circuit details found: the low side pack voltage cannot be used to measure the whole battery voltage due to cell imbalance among the cells. The ICL7665 inputs must come from the total battery voltage, while the IC power supply was from the low side pack, in a temporary test circuit. The IC power and the inputs come from different pins on the Neato connector, so when inserting the plug unstable conditions can occur. To protect the IC either power must be obtained from a low power voltage regulator on the total battery voltage, or some resistors and diode connect the high voltage to the IC power when the regular supply is absent. Plus capacitor delays on the inputs.

Current sensed by the .06 ohm shunt measured very low, for whatever reason. It is difficult to trace faults.
The op amp had to be expanded to include two stages using both halves of the IC package, for the very low sensor input voltage. There is a limit to how low an input the op amp can use, regardless of the multiplication ratio set.
Perhaps a comparator would work better, but they are typically specified to be unreliable near zero. I may test one in future, if it looks like the parts count could be reduced this way.

commercial balancer
A five cell balancer circuit with protection features -- the only five cell one listed -- is available for $25 on ebay. This would eliminate need for my own protection circuit -- if it worked properly. The cheap one that failed was only $5, without balancing.
The balancing circuit is large, 2.5 in. square, but could fit to the side of 26650's. No telling till actually tried, with the additional thermistor controls. I used my own circuit mainly for testing, with a commercial part worthwhile only once all the details of LiFePO4 charging were worked out.

different cell sizes
While the 26650's require a section of the dust bin, they are rated for 2000 cycles compared to 1000 for the 18650's (nearly ten times longer life than any other cell type).

connecting external parts
To connect external hot bypass components:
On old Neato's with a side plug for the charger brick, the side plug can be used for charging and the unused dock contacts used to connect the external parts mounted on the dock. This would allow only single charge runs, with manual charging.

For multiple charge runs with automatic recharging, it may be possible to add additional docking contacts on the bottom:


[added: a simpler possibility]

I have another, 3600mah set of 18650's to test, but they may not be in working condition after various accidents, attempts to solder them etc. The attraction of fitting within the battery compartments is not as great given that additional external components are found needed because of the heat dissipation involved. Plus the shorter rated life. [edit] the ebay listing for these flat top 1800mah 18650's actually says 2000 cycles, if you can believe that.
Why some advertise such different specs is a mystery. Can they really vary so much, in the same size? The capacities do come in a wide range for the same size cell, but the durability?


Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » May 11th, 2014, 3:23 am

18650's Charge Without Bypass

Finally soldering the flat top 1800mah 18650's with a 140W gun iron, these packs charged much better than the 1500's with tabs tried before. No bypass complications necessary. The 1500's seem to have been defective.

Haven't run yet, but charged for an hour with a constant .02V per minute rate of increase until the last .1v at twice that rate, just like the profile graphs. Neato turns from orange to green half way through, very much like the NiMh cells. Splitting the 2 amp current to halves of the parallel pairs puts the charging rate near the 1/2C requiring a bypass on the 26650's tried, without that complication.

The cells cannot be recharged once filled without an over-volt condition. Some bypass would be needed for that, and still could fail. Neato over charges NiMh in that case. The cells must be discharged some if disconnected in a fully charged state causing Neato to demand an unnecessary recharge. The fan could be run over USB for about a 1 amp load. How far down it would have to go would be tedious to discover.

Running and recharging still remain to be tested.

Reproducing more fine details of the NIMh cell profile would be complicated both to discover and to replicate. Something in there gets Neato to transition from orange to green fairly quickly, but there are many variables.

These cells present no fire hazard soldering like Lithium Ion. 22 AWG solid wire solders in 3 seconds, with solder applied in advance to the narrow gun tip for heating through melted solder.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » May 13th, 2014, 12:55 am

LiFePO4 18650 Run Time

18650 LiFePO4 1800mah, in 3600mah parallel pairs. Charge to 95 per cent capacity. Five pair series, 18v max.

After initial charge from a low level, run time on carpet 68 minutes (by usual 50% increase on hard floors, 102 minutes; plush carpet imposes a large traction load). Similar to regular NiMh batteries. Recharged in 43 minutes, which would not seem to be the same as the charge usage during the full run.

The unloaded voltage after charge might be a little lower, 16v, than when initially charged. The discharge voltage is lower than the charging voltage reached, similar to NiMh. A second run would likely be shorter.
Neato's Pet filter can increase fan load substantially and reduce run time. This test with an equivalent to Neato's original filter for about 1 amp fan load, and the blade brush with lower drag load.

There is only one source of these 1800mah unusually large cells, on ebay around $50 delivered for the ten used:
Most cells, with tabs, are 1500mah tops, twenty per cent less. My problem with those may been poor soldering of the tabs together, instead of connecting with wires soldered to the tabs. Retesting will depend on time available.
Obviously not the best capacity.

There are still details of charging electronics to work out for recharging from a high level, where over-volt conditions appeared. Some possibilities of current bypass may be explored before finalizing the design. This is not the same as the bypass created by lithium ion balancers, used to heat the thermistor. The thermistor signal here is directly controlled electronically. The balancer bypass current, however, may help explain why those systems terminate the charger better than thermistor effects alone. Perhaps a much smaller bypass might suffice for merely shaping the charging profile, compared to heating up the thermistor. It is all trial and error (mostly the latter).

[edit] All these tests are done with an older model Neato with 2.4A charger incompatible with Neato's on the newer 1.67A charger. I will not be testing anything on a different Neato; up to someone else.
Supposedly the brick revisions do not change the NiMh battery charging, just something in the component combinations for efficiency and cost. The battery was not changed, so how it charges has to be similar.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » May 15th, 2014, 4:32 am

LiFePO4 Shutdown Recharge Tests From HIgh Charge Level

When Neato is shut down through the menu or the battery disconnected, Neato will not run without first charging the battery some to determine its charge state. Apparently the battery voltage is insufficient for this purpose.
It turns out the thermal signal termination of charging for the LiFePO4 battery is adequate when recharging from an already mostly charged state. A fully charged state was not tested. As NIMh batteries were observed being over charged in this case, I would always discharge a battery some before recharging anyway.

The slower charging 26650 single series cells would not charge properly in this case. The parallel pairs of 18650 size cells which divide the charging current between them proved adequate. A test was performed by first fully charging the battery and then discharging 1/4 ah after shutdown, using a parallel pair of 50 ohm 10 watt resistors for about a 1/2 amp load. The battery recharged in five minutes or so, depending on the precise depth of discharge, similar to the NiMh battery, starting in the orange state and immediately terminating in solid green upon the temperature increase.

(A related case is interrupting charging near the end. There it was observed that when Neato has entered the blinking green phase, it resumes charging just where it left off when reconnected.)

The terminating voltage was lowered to 17.8v to allow more room for Neato to over-shoot were that to occur. The voltage rises so rapidly from there to the maximum 18v there is no significant loss of additional charge delivered.

Just in case the sort of repeated temperature cycling observed with the 26650 cells would happen a bypass circuit was also tested, using a power transistor controlled by an LMV431 adjustable voltage reference, analogous to a zener diode. This made no difference to the result and fortunately this extra complication to the adapter electronics is unnecessary.

A small refinement remains to be made consisting of simplifying the over-volt protection circuit with a P-channel mosfet instead of the N-channel previously used. The P-channel can be directly controlled by the ICL7665 comparator without additional transistors etc. Awaiting delivery of more parts.

As the battery is no longer charging with any over-volt conditions it is already usable with the ICL7665 based adapter alone. A number of repeated runs is needed to detect any remaining undetected problems. The "lifetime" Neato XV battery (good for 2000 charging cycles) costing little more than the NiMH batteries lasting only a year or so, seems to have arrived. With the NiMh batteries lasting only a year in heavy use, or even less with multiple recharge runs when cleaning, the savings involved over a few years can be as much as $300. Of course, the only way to know the precise life of these cells is to use them several years and see how much the longer life is over stated in claims. Yet the characteristics of the chemistry are widely reported and not limited to advertisements.
LiFePO4 batteries have been in use several years (I don't know the exact history), so why robot makers do not supply them is a question. Perhaps cost of these products is not an issue in luxury markets they serve. Yet there seem to be a lot of bargain hunters on these forums anyway.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » June 4th, 2014, 6:12 pm

Completed LiFePO4 Charging Adapter

[edit]adapter works but cells not yet proven -- don't build this yet
see later post

The charging circuit is completed but accidents may have damaged the cells, preventing full evaluation. Assorted shorting outs, over-voltage charging, over-discharging etc. Another set of cells will be tested in time. The capacity appears to have been reduced around 40 per cent with charging and run times reduced to 35 minutes, from previously observed full run times. Surprising it still works at all.

One final factor was discovered: when the charger terminates normally, there is a switching transient over-voltage momentary spike needing filtering by the over-voltage protection circuit to prevent false responses. The filtering used may be over-kill, but was found usable by trial and error and is easy to add. (Over-voltage detection experiments were prone to the damaging accidents, a bit ironic.)

[edit]per later results the termination voltage level was raised 1v from original post. It is now proven in practice.


Voltage divider resistors R1-R5 must be trimmed with fillers (R2, R5) for the precise termination voltage at SET1 in the 18.7-9v range. 5% tolerance resistors are usually smaller than the spec, error on the downside. The exact values used happened to require no trimming:
R1 1.292M
R2 82K nominal (late addition not measured)
R3 4.99K
R4 81.3K
R5 14.92K
[edit] variation exists among ICL7665 units, so adjustment is likely necessary. There is room to build the circuit with 15-turn adjustable trimmers if preferred.

parts list
resistors -- 1/8W 5% tolerance, for size; capacitors ceramic 10v 10% tolerance
device .... description .......... Digi-Key Part#
ICL7665... power supervisor.... ICL7665SCPAZ-ND
BC557 .... switching transistor . BC557BTACT-ND
1N4148 ... switching diode ..... 1N4148TACT-ND
FDP8880 .. power mosfet ....... FDP8880FS-ND
Fuse holder clip 5 x 20 mm ..... 283-4046-3-ND
Neato Plug Molex .062 pins ..... WM1232-ND
...female pins .................... WM9118TR-ND
(The removal tool is expensive; drilling the open end with a 3/32" bit was found to work without damaging the
Neato connector wire: 22 AWG stranded hook up wire for flexibility (20 gauge tends to break where bent under the compartment covers).
30 AWG solid wire wrap wire makes good on-board jumpers except 2 amp charging connections. Requires a wire-wrap stripping tool for so thin.

ICL7665 precautions: requires capacitor on the power connection when attached to a battery, to prevent damaging surges severe with batteries compared to transformer power supplies -- sensitive component.

[added] One circuit addition may be useful, a reset button for the over-voltage protection. When the protection is engaged, the Neato can still be run to discharge the battery, but the protection prevents connection to the charger.
Grounding the junction of R3,4,7,8, with many possible attachments, performs a reset. This saves disconnecting the battery to reset, which would prevent running the Neato until fully discharged and recharged with this battery's requirements. Neato demands a recharge whenever the battery is disconnected. Depending on the state, interrupting charging may not always require recharge before running. The charging from an already near full state seemed to work that way, compared to interrupting the initial charge after disconnecting the battery.
The ability of NiMh cells to bypass the entire charging current into heat gives a lot of leeway for oddities in Neato's algorithm, which become a problem with less forgiving types of cells.
See later post

This wiring matches the photos but may be a little less than optimal from late additions.

[added] Alternate layout with adjustable trimmers simplifies calibration. Set R2 for temperature switch point, R3 for over-voltage. Change R4 to 75K. R2 100K, R3 10K 15-turn trimmers. The 5 amp fuse is moved to the low side pack; make compact inline holder with perfboard strip if available inline holders too bulky.
wiring with adjustable trimmers for calibration


some resistors shown are 1/4W but only 1/8W needed; smaller.


D-Sub connector pins for attachment


over-volt disconnect mosfet in rear of wheel well


[edit]The 18 gauge wire fitting for the charging contacts connector above had to be improved. The above fits into the back of the connector. Works better through the pin holes. Note the pins must be kept straight as the receptacle is much larger than the pins, unlike Molex connectors etc. It relies upon the vertical tabs securing the plug. The stiff wire can be taped around the plug when inserted.

paint can opener helps remove tight fitting cells (with aluminum compartments old models)
(superseded by added tape pull tabs)

External Charger Usage
The circuits work with the external NiMh SmartCharger for packs
by connecting the dock inputs to the mosfet switch to the charging outputs of the SmartCharger; the battery being charged opens the switch. Thermistor inputs to the Neato can be plugged into the charger thermistor cable.
[edit] Note the charger thermistor return goes to the charger ground wire (and seems to be already connected internally), while the battery thermistor return goes on the battery side of the mosfet. Otherwise when the mosfet opens, there would still be a ground path back to the charger, through that thermistor line.

[edit] cells used

The cells were listed on ebay as 1800mah, from only a couple vendors. Many more vendors carry 1500mah labeled cells, with UN standard certification, now required for air shipment. I am not convinced these 1800mah cells are as large as labeled. I do not have 1500mah cells to compare, and measurements are difficult anyway. The 1800's charge from a very low level in a little over an hour on the Neato, which would not seem long enough for a 3ah charge. Yet they run the full programmed time. The unused reserve may be less than with NiMh cells. The higher voltage profile may play a role, but it is too hard to test. I leave refinements to any commercial supplier.

[edit] A belated test shows the advertised 1800mah cells are only around 1100, similar to the largest size from major brand A123. The more common 1500's advertised are unlikely less, and available with tabs.
As I use this running 45-50 minutes on low pile carpet, estimated usage is 1.875ah, leaving a small margin of the total 2200mah capacity. It is difficult to make an exact measurement at this time. Longer running is possible and tested up to 60min. (which translates to 90 on hard floors with smaller loading). There is some wear from deep discharging batteries, but this chemistry is very different than the others and I do not have any specific data on the effect. Only time will tell.

discharge curve

discharge curve
Last edited by glnc222 on August 6th, 2014, 9:32 pm, edited 15 times in total.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby lostinstx » June 9th, 2014, 12:21 am

As much as I love our Neato, I'm fed up with the short run time and short life of the current Neato batteries. So, I installed two 7.4 volt li-ion battery packs into my Neato. I spliced the red and black wires from the battery into the existing plug. I keep the yellow wire and inserted it into the sides of the new batteries like it was in the old batteries. My problem is the Neato now gives a low battery warning even though both of the new li-ion battery packs are fully charged. It won't run. I've checked the current at the plug, and it's okay. Any suggestions on why the Neato won't recognize the batteries? I understand there might be issues with charging. I had planned on just using the Li-ion charger and manually plugging it in. The new battery packs only have 4 batteries each. However, they are 5.2 Ah.
Posts: 1
Joined: June 8th, 2014, 11:59 pm

Re: Neato lithium ion battery revisited

Postby vic7767 » June 9th, 2014, 12:28 am

The problem is that even if you were to put brand new charged up stock Neato packs into your Neato, it would not run a cleaning session until a charging session was accomplished. With the LI-ion packs the Neato will abort the charging session when the Power Control Board opens the charging path that is always done to signal end of charging.
Roomba and Neato Mods, come visit:
User avatar
Robot Master
Posts: 14371
Joined: January 14th, 2006, 8:31 pm
Location: Haughton Louisiana - USA

Re: Neato lithium ion battery revisited

Postby glnc222 » June 10th, 2014, 1:41 pm

To use lithium batteries in the Neato an adapter must be supplied to charge with Neato's charger made for NiMh batteries. Neato will demand to recharge the batteries anytime they are disconnected, or a menu ShutDown command selected. An external charger is not sufficient. Neato does this to verify the state of charge needed to know when to return to base with a sufficient reserve to travel back.

Lithium ion cells can be adapted to Neato with certain balancer circuits discussed above which will heat up in a manner resembling behavior of NiMh cells. NiMh chargers terminate by a thermistor signal given the unique heating effects of NiMh chemistry when batteries are full (the continued charging just goes into heat; lithium cells just stop charging and do not heat up). Adapting can also be performed for any type of cell with direct control of the thermistor signal as shown in the LiFePO4 battery, in addition to any balancing circuits. Such balancing is needed anyway with lithium ion cells, for longest life.

LiFePO4 needs less balancing than other lithium types, and can also be charged to 95 per cent of capacity on the NiMh charger. Other lithium types may only charge to around half way, as the remaining half requires a different procedure with constant voltage, instead of the constant current used for NiMh. Consequently very large capacity lithium cells are needed, but they do have higher density to supply that. Besides LiFePO4, LiMn (lithium manganese dioxide) is preferred for sharing the same fireproof inherent safety as NiMh; less dense than other lithium types but higher than LiFePO4. LiFePO4 has exhibited full Neato run time, which is programmed and cannot be increased by expanding the battery.

[edit] Vowerk Kobold's version of the Neato uses lithium ion batteries with the proper charging method programmed. I think the capacity used, being fully charged, is less than the NiMh battery in Neato, though similar run times are obtained. I've seen the figure but don't have it at hand this moment. Neato uses only 2/3 of its capacity, with a large buffer against damaging deep discharge, so there is wiggle room in capacities.

The incompatibility of the 2nd half of lithium charging results in over-voltage without the adaptation described to terminate properly, which produces charging errors one way or another depending on what protection circuits are present. In the worst case the Neato system board may be damaged by an over-voltage on the charger.

The Neato battery connector is an inexpensive Molex part available from Digi-Key (.062 size pins, reported here previously; Digi-Key part#'s WM1232-ND, M9118TR-ND etc.) Replacement battery cables can be made without even opening the Neato case to access the internal connections.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » June 14th, 2014, 2:58 pm

capacity issues

I am still exploring the effective capacity of the LiFePO4 cells. The one time full run observed has not been reproduced, and a few variables remain to be adjusted. There may still be a problem with the cell quality or something else.

The adapter, however, works well and can be applied to other lithium types, with or without balancers A crude charging profile was obtained with a very cheap data logging device and some graphics editing:


The charge from this run appears to be about 1ah charge confirmed by a 30 minute run time cleaning, with the battery back at the beginning state afterwards. Similar results from a timed discharge at .6 amps through a 25 ohm resistor in 1:45hr. The battery cannot be charged if much higher than this initial level, as the orange state cycling situation occurs where repeated thermistor trips lead to an over-voltage.


Analysis of the individual cell characteristics with the specific LiFePO4 CCCV charger is in order to get a full picture of the situation.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » June 15th, 2014, 2:16 am

maximum voltage issue

A single cell LiFePO4 charger was found to terminate at 3.8v per cell instead of the 3.6v used so far. 3.6v is mentioned in general descriptions, but the specs for particular cells is higher. The specs for cells say 3.8 or even 3.9v charging voltage. So some further experiments are possible with recalibration of the adapter switching levels.
3.8v per cell is 19v on the whole battery, higher than the typical 18.4v observed with NiMh Neato charging, compared to a lower 18v from 3.6/cell. Re-charging full NiMh cells can get the voltage up to 19v.

Results when available.

[edit] A full charge was obtained charging 54 minutes, by starting at lower level, balancing cells in advance, setting New Battery option, and raising the termination voltage to 18.2v. Run time tests forthcoming. The question will be how it recharges after running.
Robot Master
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.


Return to Neato Robotics

Who is online

Users browsing this forum: aarolar, Franzl, Snokia, teokakav and 412 guests