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 » February 19th, 2014, 3:32 am

Thermistor Termination of NiMh Charger -- External Version (for battery compartment)
schematicexternal.jpg

[edit] TI's version LMV431 may be superior to that shown.

U.S. Sensors thermistor table at http://www.ussensor.com/do-35-standard- ... jg1k-table Limited home tests of the battery thermistor may not distinguish fully from similar "F" instead of "J" version http://www.ussensor.com/do-35-standard- ... fg1k-table but makes not difference to the circuit tested, given the large increment in temperature applied.


Less acute response than comparator of internal version but tested adequate to target 17.5v stoppage.
[edit] member Gordon supplied a simulation with engineering software. The transition is not as sharp as the comparator version shown below, but tested adequate. Much fewer components and connections when building.
TINAtransistor.jpg
TINA Spice Simulation of the Transistor version Neato Charger Adapter


[edit] This is important for the 1/10 charging rate topping off cycle Neato mentions, compensating for the small 100ma load of the parked vacuum computer running, and NiMh self-discharge. This was not tested, but with a 20C and more temperature increase added over even ten minutes, the rate would well exceed the 1C/minute standard mentioned. The fast charging rate was observed to increase at a .1v per minute in the logging graph previously posted (for NiMh batteries -- Lithium not tested). The .1v delta for switching the transistor would occur over just 2-3 minutes, not hours, and would seem to be fast enough.
[edit] this will be tested in due course.

schematicexternal2.jpg

TINAcomparator.jpg
TINA Simulation of Comparator version Neato Charger Adapter


Transistor version: Below target voltage on transistor base increases at small fractional rate of battery voltage. Above the shunt setting Base voltage increases at same rate as battery voltage, switching on over .1v increase (as shunt maintains a fixed drop). Large addition to sensed temperature exceeds delta-T 1C/minute Neato standard by large margin.

The shunt works to lower resistance as needed above the set voltage, not disconnect below that voltage, and draws as shown. This affects the setting of resistor R3 so as to be at the ~.55v transistor Base level at which the transistor starts to switch; it must be based on the combined resistance of the shunt setting divider and the shunt itself.

Can be calibrated with a 10K resistor and 2K 15 turn trimmer on two 9v batteries ~18-19v, to sweep the upper range at higher precision for input in place of the charger, and an LED for indication as transistor load, 20K limiting resistor across the full 18v. Then use combinations of fixed resistors, 1 per cent precision, for final assembly.
[edit] setting resistors for the transistor circuit cannot be done theoretically except for the ratio R1/R2 setting the voltage reference (keeping the total within around 180K for the minimum current requirement). R3 needs to be empirically set with a test trimmer with the circuit input at the maximum, trigger level because the reference IC resistance varies with voltage (even below the trigger level) and affects the current through R3 in a complex way. In contrast the comparator circuit with more parts and connections has theoretical correct simple voltage dividers. Once the values for the transistor version are selected, however, they can be used interchangeably between different units if in high precision like 0.1 per cent, around $1 each versus the usual 10 cents. R3 may be inexpensive 1 per cent as deviations appear to affect only the small slowly changing range of the transistor base voltage, which becomes swamped by greater changes once the trigger level is reached. This needs some more complete math.

[edit] Still needs testing with an actual Lithium battery to know the rate of increase in charging voltage during the constant current phase. Testing here was with an NiMh battery. Expected to be similar. Exact behavior at the CC-CV transisition point is not well documented in web materials seen.

[edit] An external NiMh charger for similar size packs uses a thermistor for charge termination of larger packs:
http://www.batteryspace.com/multi-current-universal-smart-charger-for-9.6v---18v-nimh/nicd-battery-pack.aspx
actually works on Neato's included thermistor besides the probe supplied.
So not surprising Neato uses this temperature method. Apparently the standard procedure.

This circuit can be added to balancing and protection modules used with Lithium batteries to terminate the Neato charger before triggering over-voltage protection and a Neato error when transitioning to the constant voltage charging phase of lithium charging, absent in Neato's NiMh charger constant current only. The maximum voltage, protection limit for 4-cell Li Ion is 16.8v, 18v for 5 cell LifePo4. Li Ion requires protection circuits for potential fire hazards when limits exceeded, though some cells such as the Samsung mentioned early in this thread appear very robust in overages.
Nevertheless as previously argued, the newer, much safer LiFePo4 is more appropriate for a cleaning appliance, and can be charged to 95 per cent of capacity with Neato's charger. Li Ion batteries will have to terminate not much over 50 per cent of capacity to avoid triggering an error, from the charging profile curves seen. These could vary between different cell brands. See prior posts re mounting LiFePo4 cells.

It would be nice if someone speaking German were to inform the German robotics forum, from which the possibility of adapting lithium batteries to Neato's NiMh charger originated, making this entire investigation possible, that a substitute for the discontinued LBA-1 balancer they used has been found, although not manufactured but still inexpensive and easy to make. The dragon is slayed and Siegfried can have his Lithium batteries.
Last edited by glnc222 on March 10th, 2014, 3:22 pm, edited 14 times in total.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby Gordon » February 19th, 2014, 1:38 pm

glnc222 wrote:Thermistor Termination of NiMh Charger -- External Version (for battery compartment)...previously posted (for NiMh batteries -- Lithium not tested). ...The .1v delta for switching the transistor would occur over just 2-3 minutes, not hours, and would seem to be fast enough.
Yes, "fast enough" for NiMH application. However, when applied to Li-x cells...
... if someone speaking German were to inform the German robotics forum, from which the possibility of adapting lithium batteries to Neato's NiMh charger originated, ..., that a substitute for the discontinued LBA-1 balancer they used has been found, ... The dragon is slayed and Siegfried can have his Lithium batteries.
    1) No slayed dragon is yet evident.
    2) This termination tactic does not also act to achieve cells' voltage balancing. And,
    3) "just 2-3 minutes" will not be quick enough to limit Li-x overcharge--unless you can always place that 2-3 mins dt interval ahead of each cell's Hi-limit voltage.
Just my opinion.
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Neato lithium ion battery revisited

Postby glnc222 » February 19th, 2014, 1:55 pm

If the transistor alone is not fast enough, the response can be increased some by lowering R4 expanding the temperature jump. Even more is possible switching to the comparator method using the voltage reference IC (in its more standard configuration). Would not take much more space, just a bit more power. An op amp can also be configured as a comparator. Everything has to be finalized working with the intended battery type.

Found an 1800mah LiFePo4 18650 on ebay claiming an 18mm diameter (contra slightly thicker) will be checking with one cell -- would allow a six cell pack to fit for a nominal 3600mah battery. I expect ratings are exagerated, but likely adequate since the NiMh standard pack is 3200, and the discharge curve is better. Close enough for government work.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » February 19th, 2014, 2:26 pm

2) This termination tactic does not also act to achieve cells' voltage balancing.


Of course not. Balancers galore on ebay. There is no practical way to convert Neato's NiMh constant current charger (or any other one internal to a machine without radical internal changes) into a CC-CV charger for Lithium. The best that can be done is preventing the error termination by stopping soon enough. With LifePo4 this is 95 per cent of capacity.

[edit] an NTE1434 comparator at Radio Shack (mail order and ebay) is rated for the voltage range, but could draw a couple milliamps.
Actually not out of the question given Neato's 100ma load parked. Only 5-pin sip, more compact than 8-pin DIP.
Plenty of choices.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby Gordon » February 19th, 2014, 2:51 pm

glnc222 wrote:
Gordon wrote:2) This termination tactic does not also act to achieve cells' voltage balancing.
In that list element (2) I was merely invalidating your claim of:
glnc222 wrote:..., that a substitute for the discontinued LBA-1 balancer they used has been found, ...
so don't please do not bother weasel wording your way around it -- sounding somewhat like a refrigerator salesman.
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Neato lithium ion battery revisited

Postby glnc222 » February 19th, 2014, 4:20 pm

The replacement for the LBA-1 is in respect of terminating the Neato charger, not performing balancing. (Recall also Gordon never accepted the LBA-1 terminated the Neato at all, as I understand it anyway.) As I said, the terminating circuit is to be added to whatever balancing and protection circuit is used. The terminator does NOT replace balancing circuits and I did not claim it did. Buy them on ebay. [edit] Quotes need to be placed in a context to be meaningful. Press is full of that sort of thing.

Some attention may be needed to any thermistor inputs on balancer and protection boards. They may have to be fooled in the opposite way as the Neato, insulating from the imposed temperature increase. [edit] just supply its own separate thermistor -- not too expensive. I would not be using any such boards with LiFePo4 initially anyway. I'm keeping it as simple as possible.

New facts of interest: the high side pack return wire (black) is the positive terminal of the low side pack, around 8v (compared to the system ground on the thermistor return wire). So the power supply for the low power TLC393 comparator is available in the high side battery compartment, superior to NTE1434 measured at 1.9ma quiescent load.

Image
[edit] TI's LMV431 may be superior to the version shown.

The comparator replaces the transistor in the External Version Terminator as in the Internal version. The voltage reference IC replaces the 5v line for reference. Couple more parts but less further testing...

As to the latter, the PNP transistor there is not needed to protect the IC when the 5v power is removed; the 4v input when unpowered, applied through resistance, was found insufficient to have any effect. The PNP transistor disconnect for Neato's Shut Down mode is just an extra frill to reduce the quiescent load with the batteries installed, and may not be worth the trouble, as there is a disconnect switch for the battery on newer Neato's, in the dust bin compartment.

The Internal Version post will be shortly revised with added notes -- and note additions there of improved circuits courtesy member Gordon, including simulations of operation on engineering software.

Fafner has gone home after the show to run his Neato, having used the Roomba for fire breathing practice.
Last edited by glnc222 on February 24th, 2014, 4:15 am, edited 7 times in total.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby mfortuna » February 19th, 2014, 4:41 pm

Gordon wrote:
glnc222 wrote:
Gordon wrote:2) This termination tactic does not also act to achieve cells' voltage balancing.
In that list element (2) I was merely invalidating your claim of:
glnc222 wrote:..., that a substitute for the discontinued LBA-1 balancer they used has been found, ...
so don't please do not bother weasel wording your way around it -- sounding somewhat like a refrigerator salesman.


There is really no reason or use for this type of statement. Please keep it cordial or I will start editing posts.

While balancing is useful, please note that roomba users have been using Li-ION cells with protection circuits but no balancing. I got 5 years out of one Li-ION pack and 3 years out of another before failures occurred. This is better than any NIMH pack I have owned.
Mike
Reds x 3, Dirt Dog, Disco (now a parts bot), Create, Scooba 350, and Security Dawg
Evolution Mint
Neato XV-11
User avatar
mfortuna
Robot Master
 
Posts: 5767
Joined: February 5th, 2006, 10:35 am
Location: NH

Re: Neato lithium ion battery revisited

Postby glnc222 » February 19th, 2014, 7:56 pm

Added comparator to external version above. Should eliminate any response speed concerns, best could be done. If that doesn't work, forget it.
The target voltage setting should be slightly below the maximum lithium voltage, which is a guide to the setting only.
[edit] I have posted the circuit diagrams and report of research to the German robotics forum lithium battery thread, using google translator. Incidentally, they have found another balancing product there to replace the LBA-1 use with Neato and have resumed constructing Lithium Ion battery packs for the robot, with some limitations and difficulties explained by the findings here. The only way balancing circuits can adapt lithium cells to the Neato charger is by heating the battery compartment, a very inefficient and unreliable method in general, requiring very particular products working by accident.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Add-On Circuit Halts Charging At Specific Voltage

Postby Gordon » February 22nd, 2014, 5:51 pm

Simple Add-On Circuit Can Cause Neato To Terminate Fast-Charging At Specific Charging Voltage

And, adding to the heading: Although the voltage cut-of point is specified by the user/implementer, the termination act is the result of a simulated too-hot battery that makes Neato take action! (same as what glnc222 has been talking about in these recent pages)

@glnc222: You might like to breadboard this version using the adjustable Zener diode TLVH431B to the subject service.

How it works:

A circuit assembly composed of a TLVH431 IC and three, or four, resistors are arranged to monitor Neato's battery charging voltage, and while doing so a reduced level sample is presented to the IC's "REF" pin. When the sample voltage rises to 1.24 volts the IC switches (fairly rapidly) into conduction mode and conducts reverse current from its cathode, (K), to its anode, (A).

In the following schematic, a reader can see the IC's K-pin connects to Neato's thermistor- (part of each battery) bias resistor junction through R4 (4.7k). Neato's bias resistor is modeled as a 21.6k ohm resistor -- a value stemming from glnc222's recent thermistor R vs. T measurements.
Schematic+cut_offData.png
The IC's turn-ON action pulls cathode voltage down very close to anode voltage (zero volts) as R4 quickly parallels the thermistor's temporal resistance (indicated to be "4k" for a warm NiMH battery approaching end-of-charge). The quickly obtained net // resistance of ~2160 ohms tells Neato its battery is so hot that it must shut off Fast-Charging current.

glnc222's testing has shown him that Neato does exactly that. Charging has been terminated, because Neato wanted it to be done. No battery charging issue was encountered from that quick turn-off.

In the above figure a "DC results" window has been superimposed to illustrate changes in three principal voltages as battery-charging begins at an assumed charge-depleted level of "12V" and heads to a maximum of "18V".
    1) The bottom curve shows "e_cow" starting at ~0.9V and increasing toward ~1.4V.
    2) The top curve shows essentially that same thing, but ranging from 12V to 18V--a destination it will not reach, due to what we see in the middle curve!
    3) The middle curve illustrates the simulated voltage across the Right-Battery's thermistor--shown constant in this simulation. In a real charging case this curve would be lowering as a NiMH battery approaches full charge. However, the simulation holds thermistor resistance fixed at a warm level (just more realistic than allowing it to remain at 10k to 12k for a cool robot at room-ambient temperature). But, pay attention to the cut-off action where "ThermsTap" voltage suddenly falls way low! That is when R4 'switches' into // connection with "Rt-Thermistor", and it is that reduced voltage that is being monitored by Neato's charging controller system. If the // resistance is sufficiently small Neato will cut off charging voltage and Fast-Charge will be over. The (RED) cruciform marker, (a)x,y reveals "x" = 16.5,(V), and "y" = 1.24,(V) to be the specified point of action.
Great! What then happens after charging stops and the add-on circuit is then battery powered--not that it then serves any useful function? That is a good question to ask, because one does not want some parasitic load connected to a battery that may quickly discharge it.

A response to the question is given in the next image in the form of 'quiescent' current being drawn by the circuit, over and above the robot's nominal standby current drain.
PostCutOff+leakData.png
Values of two variables have been modified to represent changes that can be expected following cut-off of charging voltage. Battery temperature will eventually head back to the cooler room-ambient temperature, so thermistor resistance is increased to 6k ohms (around T = 37ºC); and battery voltage will immediately fall a few tenths of a volt upon removal of charging voltage, so I show it to be "16.2,(V)". An ammeter has been inserted between the battery voltage and the principal load path, viz., R1, P1, & R2. The ammeter reads ~ 90 micro-amperes. So, not to worry.

If your battery has a capacity of 3200mAh the time to bleed off that capacity (barring any other loss, such as "self-discharge") would be:

t ~= 3.2,(Ah) / 90,(µA) = 35556,(hours) ==> 1481,(days) ==> 4,(yrs).

The battery will have been re-cycled before charge depletion!

As glnc222 has indicated with his similar circuits there is the associated pita work required to connect a finished circuit board (containing the above circuit) to Neato's battery voltage terminals (VBAT(+), and VBAT(-)); and to a particular "ThermsTap", all accessible w/o dismantling Neato.

Let the reader be informed that the circuit shown in the images above has not been turned into real hardware, and that analytic results shown above are outputs from a circuit simulation application.
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Neato lithium ion battery revisited

Postby mfortuna » February 22nd, 2014, 6:55 pm

One thing to mention is the Neato (at least mine) seems to periodically charge itself while docked, unlike the roomba which go into trickle charge mode and happily stay there even if the battery drops to were a fast charge would be required.

So it appears the Neato monitors the battery and will charge it if it drops in voltage. I don't know if the Neato constantly trickle charges, I don't think so.
Mike
Reds x 3, Dirt Dog, Disco (now a parts bot), Create, Scooba 350, and Security Dawg
Evolution Mint
Neato XV-11
User avatar
mfortuna
Robot Master
 
Posts: 5767
Joined: February 5th, 2006, 10:35 am
Location: NH

Re: Neato lithium ion battery revisited

Postby Gordon » February 22nd, 2014, 8:40 pm

mfortuna wrote:One thing to mention is the Neato (at least mine) seems to periodically charge itself while docked, unlike the roomba which go into trickle charge mode and happily stay there even if the battery drops to were a fast charge would be required.

So it appears the Neato monitors the battery and will charge it if it drops in voltage. I don't know if the Neato constantly trickle charges, I don't think so.
Yes, a couple pages back, neato dropped a hint that Neato's top-off charging 'mode' would take care of its battery, and made no mention of a trickle charge routine.

Thus, relative to any concern about parasitic loading that I mentioned above, that concern becomes moot when top-off is available. There is also no concern if a user has de-mated at least one battery from the robot.

A third case looms into view, in which Neato is neither connected to a charging source, nor with battery disconnected. Then, it is likely that the operating system will discharge the battery way faster than would a 90uA parasite. That Neato's owner faces a learning experience.
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Neato lithium ion battery revisited

Postby vic7767 » February 22nd, 2014, 9:00 pm

Another tidbit of info - the new Roomba LI-ion pack also creates a false thermistor report to the charging firmware that in turn stops the full charging condition.
Roomba and Neato Mods, come visit: http://www.vic7767.com/

http://www.Robot-Doc.com/
User avatar
vic7767
Robot Master
 
Posts: 14371
Joined: January 14th, 2006, 8:31 pm
Location: Haughton Louisiana - USA

Re: Neato lithium ion battery revisited

Postby glnc222 » February 22nd, 2014, 9:21 pm

Testing above circuit momentarily -- just happened to be setting up for an overnight test of the periodic slower topping off cycles. Just the ticket eliminating the transistor to simplify any manual assembly and costs (will require extra high precision resistors I think, 0.1% $1 each, to achieve interchangeable parts without trimmers).

Having observed the Neato for a couple years any trickle charging makes no sense, as one can see the periodic topping off cycles as the light goes blinking every few hours, and the system load has been measured a bit less than 100ma, consistent with running the battery down to around 15v found needed to initiate charging, over just that time frame. As well as Neato's statement here.
[edit] I admit the oddity, though, of Neato not being able to power its system off the dock without charging; guess they saved on power routing control parts, mosfets.
[edit]This needs a better, external measurement since the load I recall was over USB and may be from the dock, not the battery.

I just happened to observe a wrinkle on the New Battery option: when set with the unit docked and fully charged, it immediately initiated a charging cycle, which had not been noticed before setting it off the dock. Could have been a fluke; needs more experimentation.
Last edited by glnc222 on February 23rd, 2014, 1:42 am, edited 1 time in total.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » February 22nd, 2014, 10:46 pm

I cannot get the Vref Shunt to work as shown above, with similar but not exactly the same resistances. There could always be some mistake, but I don't see it yet; I am reading varied voltages consistent with the setup.

I use a 4.7K + 2K Trimmer on a 19v supply to sweep simulated battery voltages over 13-19v, and much higher resistances in the test circuit. A separate 5v A/C adapter is used as well. I have 141.5K between a variable 12-20v supply and ground, with division to the REF input for 1.24v at 17.5v being used for targeting (~133K/10K, trimmed with small resistors).

With a 5v supply through 20K to a 10k "thermistor" imitation, and a 4.7K bypass to the shunt cathode, anode grounded, I get different behavior.

The shunt is always conducting on the cathode line, with resistance smoothly changing from low REF voltages below 1v to voltages up to 1.4v, crossing the 1.24 internal reference spec 0.5% precision this version (I see 1.22 in your diagram, but that's also been covered in the sweep). There is no sharp switching observed as I found with the transistor configuration. And the thermistor reading is always deviated from actual by the bypass, in substantial degree.
There the cathode was on the 17v supply instead of on the 1.5v line from the thermistor.
These are both seemingly non-standard configurations for using these shunts. The standard way seems to be as shown with the comparator circuit, where the shunt is used to make an upper limit on increasing voltage to grounded anode. Conduction below the maximum doesn't matter then.

I wonder if I am observing the consequence previously considered of the device not cutting off below the set voltage configured by the divider, but just imposing a maximum. I tracked the resistance over low REF voltages from .39 to 1.1v and found a non-linear curve (cathodes in the 12v up interval divided down), with not fixed ratios to anything; current is similarly non-linear.
I don't know whether the simulation shown incorporates this behavior.

For the moment the transistor is back in style. I am testing whether the faster comparator version is needed for the supposedly slower charging topping-off cycles.

[edit] I expect now the transistor is needed because the low .5-1.5 voltage across the the thermistor circuit, at the cathode, is close to, or below the 1.24v reference of the shunt IC, and it is not made to manipulate resistance at the small size involved, when the REF input is below the reference to boot. The use is out of range of the design, which is to divide higher voltages down to the built-in reference. Even when a maximum voltage is imposed at these levels, there is not enough room to maneuver for a big shift at the set point, and it draws to much at the lower voltages, in empirical tests. If something else is the case hopefully it will come to our attention.
Last edited by glnc222 on February 23rd, 2014, 1:58 pm, edited 2 times in total.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » February 23rd, 2014, 3:10 am

Revelations in NiMh Charging
Very detailed but concise info on NiMh battery charging methods at Battery University site
http://batteryuniversity.com/learn/article/charging_nickel_metal_hydride
Gives a good context for Neato's specification of their charging procedure previously given here. This is the charger to which the lithium batteries must be adapted.

Of special interest:
Many chargers include a 30-minute topping charge of 0.1C to add a few percentage points of extra charge.

and
It is difficult, if not impossible, to slow-charge a NiMH battery. At a C‑rate of 0.1 to 0.3C, the voltage and temperature profiles fail to exhibit defined characteristics to measure the full-charge state accurately and the charger must depend on a timer. Harmful overcharge will occur if a fixed timer controls the charge. This is especially apparent when charging partially or fully charged batteries.


So much for trickle charging. More remarks on that at the site.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » February 26th, 2014, 2:50 am

Cost Comparison Li Ion and LiFePO4

Not too much difference; if anything, some favor to LiFePO4.
For LiFePO4 properties see above http://www.robotreviews.com/chat/viewtopic.php?p=121285#p121285

At a minimum, there is no premium attached to the inherent safety of those cells equivalent to NiMh. There was some previous impression such a premium applied.

Li Ion
Given how Li Ion cells may only be chargeable around half full with Neato's NiMh charger (lacking the second constant voltage phase), an equivalent to Neato's 3200mah NiMh battery might be four 26650 size Li Ion cells, two per compartment, found offered in an unusually large 7200mah size for $38 delivered as in ebay listing http://www.ebay.com/itm/4PCS-Red-7200mAH-High-Power-26650-3-7V-Li-ion-Lithium-Rechargeble-Battery-Cell-/161129575597?pt=US_Rechargeable_Batteries&hash=item25841234ad
assuming 3600mah usable.
(There is no point to higher capacity as the run time is fixed in software.[edit] Neato also uses about 2ah of the 3.2ah preventing damaging deep discharge. The high rate of NiMh wear probably stems from using fast charging and discharge, inherent in the application.)

LiFePO4
Compare to a 10 cell LiFePo4 18650 size 1800mah used in 3600mah parallel pairs, but chargeable to near 95 per cent or 3400mah. Offered around $42 delivered on ebay http://www.ebay.com/itm/4PCS-Red-7200mAH-High-Power-26650-3-7V-Li-ion-Lithium-Rechargeble-Battery-Cell-/161129575597?pt=US_Rechargeable_Batteries&hash=item25841234ad
(I happen to be trying to assemble a pack from these.)

Added Costs
In addition, though, the Li Ion requires full individual cell protection circuits because of the well known hazards of this chemistry, fairly safe with proper circuitry, but never used without it. This can at least be reduced to around $8 using low power versions such as ebay listing http://www.ebay.com/itm/171246255825?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649
This brings the Li Ion cost to $50 near a quarter higher than LiFePO. (The protections boards, though, might be reused over cell replacements.)
No such protection is needed with the LiFePo4 cells, which are equivalent to NiMh in safety respects.

A balancing function is also often mentioned with Li Ion batteries, while I have seen LiFePo4 described as self-balancing. Worst case same cost for both. Protection is combined with balancing, so could make the Li Ion closer to the LiFePO4 (lithium iron phosphate).
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » March 10th, 2014, 5:38 pm

CAUTION: a prelimary test with some bad cells appears to have caused some system board fault just from charging. Everything works over USB, but a mysterious "internal error, press to reset" msg appears whenever the brush is about to start, and there is a red ring light. The only thing unusual in this test was a 21v charging voltage at one point, with no messages at that time. Insufficient data to identify the fault, but there are all sorts of sensors and regulators on the system board.

[edit] the improved design of this adapter, with a different type of IC, is below at http://www.robotreviews.com/chat/viewtopic.php?p=124772#p124772

[edit] The circuit shown proves insufficient for terminating Lithium battery charging; additional features such as hysteresis are being tested. The initial tests with an NiMh battery used a thermistor and there is some effect of the actual NiMh heating missing from the simplification tried here.

Final Designs for 18650 LiFePO4 Packs
(not yet tested with constructed packs, being well supplied with NiMh batteries at the moment)

The remaining question with these cells is the higher voltage profile using 5 in series, nominally 16v vs NiMh 14.4v, with charger voltages 17.7-18v vs 18.4v observed when charging NiMh. An over-charged NiMh was observed to initially run over 17v minimally loaded, so may not be a problem. NiMh typically charges to a 16.5v level minimally loaded on the resting Neato. The charging voltage clearly goes through Neato's system at higher levels, with internal voltage regulators for the circuits. So there is likelihood it will work but remains to be tested.

The adapter circuit (tested only with NiMh battery, but Roomba lithium batteries use a similar technique)
schematicLiFePO4.jpg

layoutadapter2.jpg

A commercial protection board can be added with additional 3-hole space at the bottom, but may be relevant only for initial testing of this battery type. Neato's ordinary operation and the charger adapter should provide sufficient protection, and only the battery is at risk -- not consequential damages like Lithium Ion fire hazards.

[edit] Alternative IC An alternative to the single reference voltage IC shown here has been brought to my attention by members of the German Roboter forum: dual voltage monitor ICL7665, available in 8-pin through-hole DIP, specific to just the type of application here. This would combine the transistor and voltage reference, as well as supply a second comparison for use as over-voltage protection (as controlling an FDP8880 mosfet, excess capacity but available through-hole). A complete circuit for this monitor IC may be posted in future. A German lithium battery user was using this for protection.

layoutadapter3.jpg

Tight fit with 18mm standard 18650 size; thicker ones labeled this size won't fit.

Protective Fuse
A polyfuse (PTC resettable fuse), 50 cents, is normally included in battery packs for their large capacity, where like a car battery any faulty wiring or rare shorting failures in devices have fire hazards from arcing and heat. A permanent 5 amp fuse can be used instead, but takes more room and fittings. A polyfuse used in some Neato packs is
6.3 amp trip, 3.5 amp continuous SRP350F, though there are also "strap" form ones and disk forms. Neato's running load was observed at about 2.6 amps with occasional peaks at 3.5 amps (when wheels get stalled).
It is very unlikely the fuse will blow in normal Neato operations and common failures or defects; they are used because of the fire hazard, or to preserve the battery.

Polyfuses and thermistors in discarded worn NiMh Neato batteries can be reused.
Neato has an "emergency shutdown" response to 66C temperature reading. The thermistor in one pack on hand matched U.S. Sensors 103FG1K best, but only the slightly different 103JG1K is stocked (would trip at 60C). Under $1. These temperatures are enough to warp thermoplastic.
Lithium batteries do not produce heat like NiMH, and the thermistor is included for controlling the charger more than any heat problem (external chargers for such packs include thermistor probes for charging). Roomba lithium batteries replace the thermistor with fixed resistors. The charging adapter circuit works with either fixed or sensor.

Pre-charging
Neato doesn't like fully discharged batteries so must be turned on only when attached to the charger initially.
It could help to pre-charge new lithium cells with an external charger but not strictly necessary. Things to test...
A charger for Neato NiMh packs is the Multi-Current charger http://www.batteryspace.com/multi-current-universal-smart-charger-for-9.6v---18v-nimh/nicd-battery-pack.aspx
sometimes cheaper on ebay. I've heard this can use the same thermistor as Neato instead of its included probe, so could also be used with the adapter circuit. Regular lithium LiFePO4 chargers for 18650 cells are also on ebay $10 http://www.ebay.com/itm/18650-LiFePO4-LiFePO-Charger-Charges-2-Batteries-3-65V-1A-/171211475906?pt=Battery_Chargers&hash=item27dcffcbc2
These will fully charge beyond the 95 per cent level used with Neato -- and possibly present a charging problem when moved to the Neato charger; best discharge them some there before docking.

[edit] It is best to put neither fully discharged nor fully charged batteries onto the Neato charger. Packs tend to come partly charged, but individual cells could be full discharged. When removing a fully charged battery, it should be run down a bit in the Neato after re-installing before charging on the dock. The button must be pressed to restart, off the dock, and four hours at the minimal .1amp load should do. Otherwise the charger will over-charge attempting to determine the state, putting wear on NiMh batteries, and possible over-voltage conditions damaging the Neato itself with others (see later post for added protection).

[edit] Parts List

10 LiFePO4 cells 18650 size 1800mah ebay
http://www.ebay.com/itm/171246255825?ss ... 1439.l2649
[edit] flat topped cells -- cold soldering attemps poor, need tabbed cells for soldering.

External charger:
http://www.ebay.com/itm/18650-LiFePO4-L ... 27dcffcbc2

12v 25W bulb for load testing
http://www.ebay.com/itm/Ancor-Standard- ... df&vxp=mtr
some auto bulbs similar; scooter headlight bulb. Halogen desk lamp bulbs (on transformers -- very hot, not easiest)

Digi-Key.com parts (additional parts will be needed for a protection circuit being finalized, inexpensive).
Qty | Digi-Key# | Description
1 | 985-1047-1-ND | 2.49K 0.1%
1 | 10KADCT-ND | 10K 0.1%
1 | 56KADCT-ND | 56K 0.1%
1 | 75KADCT-ND | 75K 0.1%
1 | RNF18FTD10K0CT-ND | 10K 1%
1 | S5.6KCACT-ND | 5.6K 1%
1 | CF18JT150RCT-ND | 150 ohm 5%
1 | LMV431AIZ/NOPB-ND | shunt regulator
1 | BC547BTACT-ND | NPN transistor
10 | WM1003CT-ND | female pin
2 | WM1232-ND | connector plug shell
2 | SRP350F-ND | ptc resettable fuse (polyfuse)

1 | 3006P-202LF-ND | 2K 15 turn trimmer for testing
1 | CF14JT4K70CT-ND | 4.7K 5%

Digi-Key note: best register account, personal. Bug in emailing: when checking out,
when the shipping address is displayed, see that the email address is correct and edit in that form (not the profile) if needed.

Packaging MIscellany

1. Tight fit sideways in old aluminum model compartments; thin covering needed. Scotch shipping tape thinner than duck tape.

2. A paint can lid opener tool, with a flat hook on the end, helps remove tight fitting packs.

3. Cells must be wrapped flat sided. When applying tape, a strip can be tacked with reversed pieces on a table stretched flat while cells are pressed onto it gripping only at tangents.

4. Factory tabbed cells, spot-welded, are more expensive but assist in connecting compared to flat tops, and may not be available in the best capacity. Offerings seem to be expanding over time. Soldering with regular tools ruins cells with excess heat. Without professional tools, cold soldering alternatives include conductive glue, carbon Wire Glue and silver infused epoxy, less resistive. Previously prohibitive in price a new silver product from Atom Adhesives http://www.amazon.com/gp/product/B00EPYD1YO/ref=oh_details_o00_s00_i00?ie=UTF8&psc=1
Drawback is single use packaging; not sure whether superior conductivity.
[edit] tests with 5/16" square copper sheet contact gave 0.06v drop across each connection, too much across ten series connections of five cells. Tabbed cells for soldering are needed.

Carbon products are too resistive on wire connections, but may work with large area contacts using thin copper sheet or foil; thin glue layer also decreases resistance. A 1/2" strip of 36 gauge .005" copper has the volume of three 22 gauge solid wires, for the current. http://www.ebay.com/itm/COPPER-Sheet-Metal-Tooling-Foil-6-x9-x-005-36gauge-5mil-axvz-ART-HOBBY-CRAFT-/251467401637?pt=LH_DefaultDomain_0&hash=item3a8ca009a5
Some copper sheet is also in Ace Hardware stores.

[edit]In the diagram above, U-shape around the four ends connected allows the terminal wire from the other side to pass through the gap between cells to the same side as the other terminal.

5. Wiring -- 22 gauge for the current. Stranded for flexibility, not putting strain on glued connections.

6. Charged battery precautions -- assembling a charged battery pack wiring involves live wires. Connectors to separate adapter circuits etc. may be warranted. Either headers, in-line connectors or individual line butt connectors. D-connector pins make small butt connectors with heat shrink covers, and the Neato connector pins a heavier duty version.

7. Neato connectors Molex 03-06-2042 inexpensive, or reuse ones in discarded NiMh packs.

The LiFePO4 Facts
http://www.hecobattery.com/china-customized_2ah_lifepo4_battery_cell_for_lithium_polymer_battery-1303775.html
Also Wikipedia http://en.wikipedia.org/wiki/Lithium_iron_phosphate_battery
[edit] additional details http://www.gebattery.co/geb/EN/ProductList.asp?SortID=138&SortPath=0,133,138,
LiFePO4lifetime.jpg

This could be a lifetime battery, lasting several years and equaling over $300 in NiMh batery replacements at current prices.

Profiles:
LiFePO4curves.jpg
Last edited by glnc222 on June 17th, 2014, 4:20 pm, edited 10 times in total.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » March 10th, 2014, 9:06 pm

CAUTION: a prelimary test with some bad cells appears to have caused some system board fault just from charging. Everything works over USB, but a mysterious "internal error, press to reset" msg appears whenever the brush is about to start, and there is a red ring light. The only thing unusual in this test was a 21v charging voltage at one point, with no messages at that time. Insufficient data to identify the fault, but there are all sorts of sensors and regulators on the system board.
This may all have been due to damaging a protection board soldering, and then charging cells damaged by testing without the protection, in a comedy of errors.
The charging voltage needs to be monitored while testing. I actually had a separate over-voltage detection circuit used in other parts of the work, and could have connected it to a relay for additional protection.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » March 12th, 2014, 2:07 am

Possible Dock Based Voltage Monitoring and Protection

I am considering adding a charging voltage monitor to prevent the sort of damage encountered with bad cells causing over-voltage by Neatio's charger. Besides any kind of battery pack protection circuit, which disconnects the battery, an even stronger protection is disconnecting the charger. This covers failure of Neato's charger to stop with an error on disconnection of the battery, with repeated probes by battery protection circuits after delays.

unused jack
Some Neato's have a charging jack on the back, unused in newer models, which could be reused to connect sensor lines. Perhaps the control circuit can be arranged so that when no sensor is plugged in, the protection is automatically disabled. Otherwise some switch on the control box.

The external trip level might be higher than the battery protection, for protecting the Neato instead of the battery. I've seen the charger voltage go to 19v without problems, and peak at 18.4 charging NiMh. Best make it adjustable with a trimmer, setting conservatively and tweak if any unwarranted interrupts occur. The charger will peak highest when reconnecting a disconnected fully charged battery, where the charger goes a few minutes to identify the state of the battery, repeating the full battery termination cycle.

I already have a 5v supply in the dock for a clock added on top, and can add another comparator circuit and a relay or mosfet switch to disconnect the brick from dock circuitry. The regulated 5v provides fixed reference voltages. There's a lot of space inside the dock. The internal dock additions use mating barrel connectors to insert into the brick output without cutting lines. Comparators can be configured to latch by connecting a diode between the output and the voltage reference input, pulling it to ground, so no flip-flop needed.

Some LED indicator is needed to show the protection was tripped.

Battery protection circuits use bi-directional mosfets in series with the battery, but only one direction needed on the dock power line. The easiest to use type of interrupter remains to be identified.

schematicNeatoProtect.jpg


[edit] It became apparent later that the adapter circuit, set to a higher voltage trigger, would perform the same function and be simpler. The transistor controlled by the voltage reference shunt would switch a relay or N-channel mosfet in the charging ground circuit, instead of modifying the thermistor. Some additions may be needed for assorted details. It also might be possible to install in the Neato as well as the dock, with the charging contacts connected through a plug under the bottom fan cover. The adapter circuit requires no power supply for a comparator IC. A possible mosfet replacement for the relay will be tested soon.
Last edited by glnc222 on March 18th, 2014, 5:12 am, edited 1 time in total.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby yonkoc » March 12th, 2014, 5:56 pm

Hello,

First things first. Hats off to glnc222, Gordon, vic7767 and the rest of the electrical wizards. My head is spinning from everything read so far and I am no better off from when I started. I have a very simple question.

After 12 pages of Li-Ion, Ni-MH and LiFePo4 tests is there a buildable LiFePo4 rechargeable battery pack for my XV-21? And I can assure you I cannot read through the electrical schematics.

Thanks
yonkoc
 
Posts: 2
Joined: March 12th, 2014, 5:51 pm

PreviousNext

Return to Neato Robotics

Who is online

Users browsing this forum: No registered users and 339 guests

cron