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 10th, 2014, 11:22 am

Thanks to Neato Guest for charger operation details. Looks like a sophisticated program.

The additional test of the LM3420 on thermistor modification could not be performed as the LM3420 has been damaged. I have no plans to obtain another at present, having enough batteries to last most of the year. More useful information may also be acquired over time, such as specifics of Neato's thermistors. [edit] turned out to be just a wrong connection; but waiting on more interesting voltage reference IC's maybe simpler to use (nice to get the absolute minimum of componenets).

An additional charging behavior was observed, however, because the LM3420 was locked in the "ON" state.
With the test performed with the battery more fully depleted, with an orange indicator, the high temperature set on the high side battery pack sensor did not terminate charging at a low charging voltage around 15v. Instead, it appeared to slow the charging to a low rate, perhaps the same as the lower rate Neato just mentioned in regard to topping off procedures. Restoring the temperature to its actual level accelerated the rate.

Voltage on the thermistor was also observed higher than previously due to low environment temperature, consistent with Neato containing a resistor in series with the thermistor and processing the voltage as the input, similar a meter operation. The data sheet when found could specify some of the parameters more completely.
I'd say the voltage increased 2 tenths over a drop of 10 degrees fahrenheit in casual observations, down to maybe 60 degrees. No exact measurements are available, but the order of magnitude is indicated.
The thermistor itself can have a non-linear response to temperature, again the data sheet is needed for more precision.


The LM3420 control of the thermistor tested is shown below. The test circuit simply adds a fixed additional load in parallel with the thermistor, lowering the combined resistance.
schematictemp2.jpg

The opamp is needed for the 150 microamp observed load of the LM3420 when active (85 off), and a high impedance over-all. High precision divider resistors are needed to avoid trimming individual assemblies. Different battery types, such as LiFePo4 and Li Ion, require different ratios to set just before the transition to constant voltage charging phases.

Since the effect of interest is some increased resistance of the combined thermistor and added loads, representing a shift upward in temperature sufficient to reach a high range, adding a fixed resistance only works over a certain range of initial actual temperatures and an addition adequate to bring up to a high range some interval of possible environment temps. A more flexible circuit which can target a particular temperature with a feedback control loop created by an opamp is shown here. No details have been worked out or tested; only the method is illustrated here.
schematictemp.jpg



[edit] the 16.8v version of LM3420 does not appear available except in large quantity orders. All references at Digi-Key say "call for pricing" and Octoparts says no stocking distributors. So the voltage divider to 8.4v and associated op amp appear necessary with this part. With a different division for LiFePo4 with a 17.5 v 5 series peak a divider would be needed anyway, so nothing much lost.
Last edited by glnc222 on February 14th, 2014, 1:33 am, edited 1 time in total.
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » February 10th, 2014, 10:34 pm

Alternative Parts

The LM3420 discussed is made for constructing battery chargers, and only happens to be used in the Neato control application here. Its advantage for chargers is a built-in voltage reference preset to just the particular level needed by Li Ion chargers, saving components, but as seen above that level does not directly fit the Neato application since discontinuing the 16.8v version, nor LiFePo4 charging. Since an op amp had to be used with LM3420 above, the same op amp might be controlled by more general types of power management IC's, namely Voltage Reference IC's. The come adjustable with attached resistors, for any application.

I don't know whether it could work, but it looks like these can be used to drive the op amp or even directly a transistor instead of driving the charger IC with the op amp. They have very low power consumption similar to the charger controller.
For example, TLVH431BQLP Adjustable Precision Shunt Power Management IC, 0.5 per cent precision, only 95 cents qty1 [33 cents volume] and with through-hole transistor-style packaging for hand assembly besides all the surface mount styles. Data Sheet http://www.ti.com/lit/ds/symlink/tlvh431a.pdf

I gather "shunt" means it is put in series with a voltage source and a load, and only conducts when the voltage across it exceeds the adjustable reference value fixed by resistors, but do not have an official definition. So just position between the Neato battery and transistors or op amps modifying the thermistor signal in the example shown previously -- if all the details work out.

I think I'd look at these before getting any more LM3420's. Another part to play with anyway, and cheaper ( so can have more replacements for those experimental accidents).
Last edited by glnc222 on February 11th, 2014, 6:17 pm, edited 3 times in total.
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » February 11th, 2014, 5:37 am

Direct Thermistor Resistance Measures

Unused aftermarket batteries placed with a thermometer: one in refrigerator, another room temp, then a cooler room, and under a Vornado regulated room heating fan. Ranges from 35 to 90 fahrenheit, 71 room temp.
Neato batteries
Temp ... Resistance K ohms ... Vishay 10K@25C (interpolating tables at 5C intervals)
------ .... --------------------- ... -----------------
1.67C ......... 21.32K ............ 23.46K
19.72C ........ 12.39K ............ 11.94K
21.67C ........ 11.50K ............ 11.21K
32.22C ......... 7.65K ............. 7.95K --not the Neato part, also different packaging.
[edit] As Gordon mentions previous page, the Neato thermistor is most likely U.S. Sensors 103JG1K with very close matches.

Vishay NTCLE100E3103JB0 http://www.vishay.com/doc?29049 -- detailed explanation.

These are NTC type, lowering resistance with rising temperature.
I do not see any listings for the 25C ratings used to describe these things anywhere differing so slightly from 10K, indicating the Neato version is one of those 10K products. They all vary in curve slopes.

Comparison of catalog thermistors rated 10K@25C
Temp ... Vishay ........... Epco
------- ... ------- ............. ------
0C......... 27.9K ............. 24.3K
20C ....... 12.1K ............. 11.8K
25C ....... 10K ................ 10K [added, forgot this line; the lines cross at this specification]
30C ........ 8.3K ............... 8.5K
40C ........ 5.8K ............... 6.2K -- different slopes

Vishay gives an exponential formula for the resistance vs temperature function, with several parameters which can vary across parts. The differences between models is greater at low temperatures and higher resistances; lines squeeze together all approaching zero.

Using the curve of the Vishay fairly similar to the Neato to interpolate observations, the Neato resistance would be 18.7K at 40 fahrenheit -- below which Neato might not be useful.
When a 7.5K resistor is added in parallel to this the indicated temperature would be around 43C used in the test previously posted terminating Neato charging. At any higher environment temperature the indicated temperature would be even higher, ranging to 60C and above. The point is this single added resistor in parallel appears to be able to boost the apparent temperature to a level for terminating charging (43C in test reported, and assuming Neato responds that way), for all temperatures down to 40f. It may be possible then, subject to testing, to use simple switching adding a single fixed resistance in parallel to the thermistor, whatever the current temperature (contra targeting a specific high temperature with feedback loops needing more components). One high temp is as good as another above a certain level.

Incidentally, in the circuit switching with a transistor, the difference in voltage across the thermistor and the added resistor (the drop across the transistor), was observed on meter at only 15mv, too small a factor to affect resistance selection.

A possible rationale for Neato not terminating on high temp at low starting voltages as observed, has occurred to me. The battery may become heated from discharge when running, an aspect of NiMh batteries, so the charging algorithm might assume the battery will cool off during charging and responding to high temp right after returning to base might not work well. HIgh temps at the end of charging could indicate something wrong with the battery or charger, so would be more likely treated as a terminating condition. The actual procedure is not disclosed.

[edit] Gordon had found best fit of Neato thermistors to U.S. Sensors 103JG1K.
Substituting varied resistors for thermistor for USB temp readings, best fit is slightly different 103FG1K:
13C Neato 14.89K; 103FG1K 16.150 (103JG1K 17.255K)
32C Neato 7.462K; 103FG1K 7.677K (103JG1K 7.402K)
48C Neato 4.243K; 103FG1K 4.357K (103JG1K 3.888K)
69C Neato 2.216K; 103FG1 2.226K (103JG1K 1.813K)
-- battery shutdown alert issued (references cite common limit 60C; assume Neato uses same)
Last edited by glnc222 on March 1st, 2014, 6:31 pm, edited 3 times in total.
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby Gordon » February 11th, 2014, 1:21 pm

glnc222 wrote:Direct Thermistor Resistance Measures

Unused aftermarket batteries placed with a thermometer: one in refrigerator, another room temp, then a cooler room, and under a Vornado regulated room heating fan. Ranges from 35 to 90 fahrenheit, 71 room temp.
If you do this sort of experiment again, be sure to indicate either soak-time-at-temperature, or state that you periodically checked thermistor resistance until it no longer changed.

I handed you leads to a good U.S. Sensor's thermistor (my post on previous page), and you ignore and veer off looking at Vishay and Enco products! What gives?

Take a look at the U.S.Sensors' thermistor data as I sub in approx temperatures in place of your Vishay resistance column:
Neato's battery-thermistor's R(T) estimates:
Temp ... Resistance K ohms ... U.S.Sensors' 10K@25C (corresponding T,(C)
------ .... --------------------- ... -----------------
1.67C ......... 21.32K ............ |7 to 8 degC| {ignore this one on basis soak-time (?) too short}
19.72C ........ 12.39K ............ |19 to 20 degC|{fantastic temperature match!}
21.67C ........ 11.50K ............ |22 degC|{fantastic temperature match!}
32.22C ......... 7.65K ............. |31 to 32 degC|{fantastic temperature match!}
And the U.S.Sensors' thermistor package == Neato's thermistor package.

Don't you agree, the U.S.Sensors' thermistor does well in this cold region?

Note that these cold data are less important than tracking hot data well.
Incidentally, in the circuit switching with a transistor, the difference in voltage across the thermistor and the added resistor (the drop across the transistor), was observed on meter at only 15mv, too small a factor to affect resistance selection.
Yes, also too small to be believed!
... HIgh temps at the end of charging could indicate something wrong with the battery ...
Right! Like good cells turning into weak cells--happens all the time, therefore a charging controller must be ready for NiMH cells to go way HOT during fast charge!
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Neato lithium ion battery revisited

Postby Gordon » February 11th, 2014, 2:21 pm

glnc222 wrote:Alternative Parts

The LM3420 discussed is made for constructing battery chargers, and only happens to be used in the Neato control application here. Its advantage for chargers is a built-in voltage reference preset to just the particular level needed by Li Ion chargers, .... Since an op amp had to be used with LM3420 above, the same op amp might be controlled by more general types of power management IC's, namely Voltage Reference IC's. The come adjustable with attached resistors, for any application.
The referent device (TL*431) is NOT a "power management" device, nor does it "come ... with attached resistors"! Yes, it can participate in a power management system, as well as millions of other components.

The TL*431's major attribute is the capability to serve as an adjustable Zener diode. The new version's P/N that you have uncovered here, is discriminated by the asterisk I have been using, viz.: * = "VH" in the part number.
I don't know whether it could work, but it looks like these can be used to drive the op amp or even directly a transistor instead of driving the charger IC with the op amp. They have very low power consumption similar to the charger controller.
"Similar to" current draws, but greater than LM3420's--I think you will find.
For example, TLVH431BQLPR Adjustable Precision Shunt Power Management IC, 0.5 per cent precision, only 33 cents and with through-hole transistor-style packaging for hand assembly besides all the surface mount styles. ...
Right about the TO-92 package, but WRONG about its price. You see that tail-end "R" in your above P/N? It stands for "reel", a reel of 2000 units! Buy that item and you will have an extra 1999, or so TL*431 to play with. Buying only one, will be just under one US dollar, not including S&H.
I gather "shunt" means it is put in series with a voltage source and a load, and only conducts when the voltage across it exceeds the adjustable reference value fixed by resistors, but do not have an official definition. ...
"Shunt means you will use this in the same way that you use a fixed Vz Zener to regulate incoming voltage. The TL*431 will shunt current to GND as it establishes its Vz at cathode. There will be a series, voltage dropping (ballast) resistance between cathode and some power source at the input end of the resistance. You will pass Vz off to whatever circuit element you need the Vz to operate.

Go into the datasheet, find & study the RECOMMENDED OPERATING CONDITIONS table; and jump down to Figure 2. That shows what I just described. Now your work is to figure out how to size the resistance values of R1 & R2 to make that TL*431 bugger do what you desire!

On the very day of your post about this device I was also looking at the TL431 to serve as part of the voltage divider that senses a specified battery voltage, and to use Vz to flip the binary state of a non-inverting Schmitt Trigger IC, the output of which would begin charging capacitor "C" in an R-C integrator to develop a voltage function conforming to neato's advice: dT/dt = 1degC/minute. That voltage ramp connects directly to the Gate of a low-power, N-channel MOSFET whose Drain & Source terminals straddle one of Neato's thermistors. Gate voltage rises, and Rds lowers. If one wished, a 2k ohm resistor could be inserted between the Drain and thermistor connections.
I think I'd look at these before getting any more LM3420's. ...
This final mention of the LM3420 made me look at its 'block diagram' while pondering my message. This figure:Image
Suddenly, I saw the possibility of constructing your own simulation of that IC! Granted, your circuit could not match accuracy and tight tempco specs of the LM3420, but it might be an interesting experiment! My idea involves use of the TL*431 as the "1,23V" band-gap voltage reference. See Figure-1 in the TL's datasheet. It is that wiring--REF-pin connected to cathode, and no R1, R2 needed, that would be used in your dummied up 'LM3420. Yours would use metal-film resistors for the circuits' voltage divider, and provide ratios different (or same as) than any available in a stock chip! The onus would be on you to select a satisfactory op-amp (don't plan on finding it at Radio Shack). Selection of its output transistor is the easiest portion of the task. Does that sound like fun?!
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 11th, 2014, 3:02 pm

corrected price reference previously made, and some schematic graphics here and there (graphic typos...).

temp readings -- no "soaking" in refrigerator; left in 35f airy frig an hour, so 1.67C instead of 32f/0C (could have left outside for that this winter day, but concerned with damage possible at actual freezing -- don't know the effect so avoid).
Not using freezer near zero f could damage cells, though thermistor reads that.
Similar high 90f reading under sophisticated Vornado regulated heating fan, slow blow through a 1ftx1tftx18"x18" shelf stand.
The oven only goes down to 170f and previously found melts thermoplastic, so avoiding that. Too tedious to modulate with opening the door.

I would appreciate any elaboration of the terms and symbols used in Neato's description of their algorithm in their post, especially whether lower case "c" is different than upper case "C". It is especially surprising they make some reference to the rate of change in temperature. It appears to be a very complicated process they manage.

On parts I was previously struck by the MAXIM offerings with their MAX4211 flexibility with built in voltage reference and two comparators for any use, and metering op amps to do just about anything -- but in one of those microscopic packages for mass production. Costs $30 with shipping and services to get installed on a prototyping adapter.

[addendum] looked at a 1W zener diode in Radio Shack (only type there,5.1 and 12v), and it seems to not do much without a substantial current through it over 50ma -- conducts at voltages below the clamping level too much for low current uses, if I've interpreted it correctly.

[addendum] interesting op amps one dual pkg 24v rail-to-rail compared to 741: LM6132BIN/NOPB
Digi-Key http://www.digikey.com/product-detail/en/LM6132BIN%2FNOPB/LM6132BIN%2FNOPB-ND/120154
and a precision metering one: CA3420
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » February 11th, 2014, 4:17 pm

I fail to fathom what drives your mind! I handed you leads to a good U.S. Sensor's thermistor (my post on previous page), and you ignore and veer off looking at Vishay and Enco products! What gives?

Apologies but I have a number of different things to think about and get distracted by some parts of posts and leave the rest to examine in more detail later. But all detailed information is appreciated. I just may not respond at once to every point. Also, I was not looking for thermistors to buy as much as getting the ranges of their different properties, and just pick on what I happen to be looking at some moment, which can sometimes do the job.

These communications are casual and elliptical. We are not all focused on the same specific points and there are many issues being examined simultaneously. The language is also not polished to precision, and can have much ambiguity requiring interpretation. e.g. "attached" does not have to mean some technical part included in a component, but just something you would attach externally for a purpose. We are not writing data sheets.
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby Gordon » February 11th, 2014, 5:25 pm

glnc222 wrote:corrected price reference previously made, ...
So noted. A wise additional edit to do is to delete the "R" suffix from the TL*431's P/N.
...I would appreciate any elaboration of the terms and symbols used in Neato's description of their algorithm in their post, especially whether lower case "c" is different than upper case "C".
Yes, it is, and will elaborate shortly. I thought I should first suggest that only a teensy fraction of Neato's fast-charge termination algorithm was likely revealed to us by neato.
It is especially surprising they make some reference to the rate of change in temperature.
Not to me, having seen the necessity for that high-temperature termination in early Roomba/Scooba models' hi-rate charging sessions.
It appears to be a very complicated process they manage.
We will never know the full extent of all feasible termination possibilities that have been programmed into Neato's F/W.

Now, this is what neato wrote for us (split into two parts to aid discussion):
neato wrote:...The first phase is a fast charge at 2A current limit until dt/dT of 1c per minute is reached.
There are three elements to consider in this one sentence: "current limit", "battery-temperature slopes, 2ea., (near full charge)", and "what happens following that one-minute of slope survey?". IMHO, I say these things about those topics:
    1) Current Limit of 2A -- tells me that "fast-charge" will not exceed 2 amperes at any time, but leaves the option that fast charge may be moderated to less current, as required; and does reveal there to be a reduction in applied charging voltage (applied voltage being on the order of one volt above instantaneous battery terminal voltage for fast charging at the 2A current) to accommodate a battery's state of charge depletion.
    2) Increasing Temperature vs Time Slope near EoC -- is determined by Neato's frequent (could be every second, say as Roombas do) sampling of each battery's temperature sensor. Neato computes dT/dt,(deg Celsius/minute) (I am certain that neato will agree to typos causing his inverse letter "t" upper-/lower-cases) every 60s, and when it discovers the value of dT/dt equals, or exceeds one deg-C, (Celsius), a flag is set by the CPU (w/in the MCU).
    3) Now, what happens following that one-minute of slope survey? -- when the F/W detects the set-flag?
      * Might Neato immediately halt fast charge?, or, avoiding premature termination it...
      * Might it double check slope status by processing another 60s of samples?, else what...
      * Might Neato do?
    An interested investigator will have to determine answers to those Qs by measuring Neato's thermistor resistances as f(t).
neato wrote:The optional second phase is a timed top up charge at ~0.1C current.
Possibly, neato makes the assumption that anyone messing with Neato's batteries and charging system will be able to discern his use of capital "C" unit for top-up current is a fraction ("~0.1") of battery Capacity--a typical, but annoying convention! Neato's standard battery capacity is 3200mAh, so (C/10)_top_up ~ 3200mAh * 0.1 = 320mAh. To shed the "h" unit, we could divide by several assumed top-up phase periods (thus bringing in the "timed" adjective to give such possibilities as: 320mA for 1 hr, or 160mA for 2 hrs, or 80mA for 4 hrs, etc. Once again, "timed" is a TBD datum for the investigator to evaluate empirically.
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Neato lithium ion battery revisited

Postby Gordon » February 11th, 2014, 6:18 pm

glnc222 wrote:...These communications are casual and elliptical.
That is a really bad omen! "Elliptical is just a squished circle, and one may waste a lot of time running around in circles!
We are not all focused on the same specific points and there are many issues being examined simultaneously.
Here too, "...many issues being examined simultaneously..." is like riding off in all directions at once!

Good grief! Get a handle on your major task of interest, and plot out a course to follow. I presume you wish to come up with a magic circuit that makes Neato accept using some Li-x celled batteries in place of its design-point NiMHs. There are many things for you to get a handle on before sketching full schematic diagrams that you think might work. Why can't you see how futile that is for you to do the major task in one fell swoop? I know that I am baffled by such a task!

Why is it not obvious that you must learn all that you can about Neato's stock behavior (with its NiMH batteries) before attempting any electronic changes? By starting a quest to learn all that you need to know, you can list a series of baby-step measurements that satisfy sub-goals. Many would be easy to do, but require test accessories to safely perform, such as the batteries' break-out cable mentioned in a prior post.

Put your a/c USB scope on the shelf for now, and buy an inexpensive USB DSO to use. I'm sure you will be more satisfied with such an instrument.
The language is also not polished to precision, and can have much ambiguity requiring interpretation. ...
If you are doing engineering work, then use standard science & engineering speak. "Polishing" is not needed, and reduction of ambiguity & interpretation is up to an author to clean up. Avoid unclear writing by writing nothing, if that's what it takes.
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 11th, 2014, 8:39 pm

The advice for pursuing professional aims may be of help to those with that purpose. Mine is more of a hobby here, with more limited objectives. You can find my more serious stuff elsewhere on the web, not in engineering. The topic here is more technology details of use with the Neato more than practices and organizing engineering activity.
That can be saved for the office. [edit] or perhaps in tutorial materials of which much is found on the web, and could well be of interest here to some readers. I've found some very useful. The audience isn't particular individuals, whose interests can vary widely.

Testing Vref Shunt IC's end of the week.
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » February 12th, 2014, 1:18 pm

Suddenly, I saw the possibility of constructing your own simulation of that IC! Granted, your circuit could not match accuracy and tight tempco specs of the LM3420, but it might be an interesting experiment!

(re: LM3420 charging controller, post http://www.robotreviews.com/chat/viewtopic.php?p=122175#p122175)

Actually I tried doing that with an op amp and the forward voltage drop across a few switching diodes 1N4148 as a voltage reference, but it did not work well enough. The forward voltage did not seem very constant at low currents and voltages. It worked as a kind of upper limit on the voltage drop but not a fixed drop at all voltages (needed for a very high multiplication with the op amp). Perhaps if I used a larger current voltage divider, but that seemed to be an unsuitable drain on the battery (besides which the part is not made for precision in the drop over units). So the LM3420 was needed, or an alternative Vref IC now being tried, to get a voltage reference with the needed properties matching the application.

I used a 7805 5v power supply voltage regulator for reference in a circuit being used to monitor the air flow through the Neato filter (a function of the fan current with its constant rpm control). http://www.robotreviews.com/chat/viewtopic.php?f=20&t=17188 A crude ammeter was installed with led readouts, each led controlled by op amps or comparators. The current draw is more than I would like on a battery monitoring device, and used in a 1amp loaded operating condition. The drain from a 7805 even unloaded is a few milliamps. There is no particular cost penalty in using the better suited part, maybe just the opposite.
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby Gordon » February 12th, 2014, 2:34 pm

glnc222 wrote:
Gordon wrote:Suddenly, I saw the possibility of constructing your own simulation of that IC! Granted, your circuit could not match accuracy and tight tempco specs of the LM3420, but it might be an interesting experiment!
...Actually I tried doing that with an op amp and the forward voltage drop across a few switching diodes 1N4148 as a voltage reference, but it did not work well enough.
Au contraire! By your own admission you did not do the work expressed in this paragraph: Suddenly, I saw the possibility of constructing your own simulation of that IC! Granted, your circuit could not match accuracy and tight tempco specs of the LM3420, but it might be an interesting experiment! My idea involves use of the TL*431 as the "1,23V" band-gap voltage reference. See Figure-1 in the TL's datasheet. It is that wiring--REF-pin connected to cathode, and no R1, R2 needed, that would be used in your dummied up 'LM3420. Yours would use metal-film resistors for the circuits' voltage divider, and provide ratios different (or same as) than any available in a stock chip! The onus would be on you to select a satisfactory op-amp (don't plan on finding it at Radio Shack). Selection of its output transistor is the easiest portion of the task.
You accomplished only a SIMILAR, unsatisfactory trial!
... There is no particular cost penalty in using the better suited part, maybe just the opposite.
You can no longer purchase the "better suited" LM3420-4.2 IC, you could never purchase an 'LM3420-v.v' IC to work with your favored LiFePo4's (whose hi-limit "v.v,(V)" was unknown when the LM3420 family was designed). Being able to create the LM3420's function for any reasonable voltage trigger level is perhaps the main point in my thinking. You need a specific trigger voltage w/in a three to four volts range to support balancing a series pack of specific Li-x cells. When the trigger device can not be purchased, the cost of building the quantity of devices needed is not particularly critical to a driven DIY'er.
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 12th, 2014, 3:11 pm

I meant the cost of 7805 voltage regulators, not LM3420's. Some V-ref parts arriving in a couple days. I expect that is the last thing I would try as any simple possible solution. The more complex I leave to commercial developers and vacuum factory support for different batteries. It's not like some new functionality is obtained here, and it is merely a matter of battery cost over long periods. There's only so much that can be worth. Any really high value would likely promote commercial solutions.
I suspect a major reason vacuum suppliers do not use Lithium batteries is the safety and product liability issues mentioned before. Another opinion has been that Lithium is more expensive and found more on luxury brands. That does not seem to fit the cost of components found. Power tools adopted LIthium when safer LiFePo4 was developed, and iRobot supplies LIthium for it's gutter cleaner only, used outdoors and more like a power tool with multiple plug in power packs. The pace of investment in the vacuums appears very slow, so adapting to the new battery tech could understandably take longer than the power tool industry.
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » February 12th, 2014, 4:52 pm

you need a specific trigger voltage w/in a three to four volts range to support balancing a series pack of specific Li-x cells.

I envisage a commercial balancing and protection board being used as readily available; I would not think of developing a balancing device. LiFePo4 is also described as self-balancing, but does need protection given the non-standard charging situation. So just a small additional thing to terminate Neato's NiMh charging at 95 per cent capacity, around 17.5v, is needed. Will see how simple this can be with thermistor triggering.
I think going back to the infamous LBA-1 it has been analyzed here too much in terms of how it does its designed balancing job, when the particular problem with Neato is adapting to the incompatible NiMh charger. There are readily available balancing and protection boards for Li Ion, too. Vic reported they don't adapt to the charger and produce the expected error messages. I imagine when the constant voltage phase missing on NiMh charging is entered, the charger is forced to raise charging voltage to maintain constant current, tripping the over-voltage protection circuit and causing a battery disconnect error. If it reconnects right after charging ceases, Neato may then ignore the disconnection, which is a very severe error condition on Neato with an unusual red ring light requiring resetting the processor, and just show a charging error msg on the LCD. I saw a temporary condition report when fiddling with the connector.
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » February 13th, 2014, 8:42 pm

Neato's Thermistor Input circuit measures
For anyone designing circuits to modify the thermistor input:

Using currrent and voltage across a couple different bypass resistor sizes and known thermistor resistance at the room temperature, it can be determined that Neato is consistent with using the 5v logic supply through a 20K ohm resistor in series with the thermistor to produce analog voltage inputs for processing. [edit] refined to 21.6K by Gordon later post.

Several moving parts can then be involved in determining the virtual temperature indicated by transistor switching circuits for a given bypass resistor and varied actual temperatures, taking the transistor properties into account.
Simulating temperatures above 60C with total resistances in the 3K range yield voltages down close to the minimum voltage drop across bipolar junction switching transistors. Will have to see what that adds up to.

[edit] Appendix
Thermistor 71f room temp 11.5K ohms
Observation no bypass 1.74v
45.8K directly connected bypass (not switched), 1.7v, 9.8K bypass 1.126v.
Last edited by glnc222 on March 1st, 2014, 6:26 pm, edited 2 times in total.
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby Gordon » February 14th, 2014, 2:04 pm

glnc222 wrote:Neato's Thermistor Input circuit measures...Appendix
Thermistor 71f room temp 11.5K ohms
U.S.Sensors' R/T table for its 103JG1K thermistor brackets your 71degF ambient with 11419 to 11942 ohms values, thus close enough for this discussion.
Observation no bypass 1.74v
Means: 1.74 volts exists across the above "11.5k ohms" thermistor resistance. At that instant the bias current through the thermistor would be: I_bias ~= 1.74,(V) / 11.5k,(ohms) ~= 151 uA. I note that value is about half the 350uA current previously computed!

This 'new' I_bias result, plus the new knowledge that a "5V" supply powers a series resistance "Rs" (estimated as "20k ohms") in series with the thermistor, permits estimation of a more exact estimate for the 20k value: Rs ~= (5,(V) - 1.74,(V)) / I_bias = 21600 ohms.
45.8K directly connected bypass (not switched), 1.7v, 9.8K bypass 1.126v.
Two test cases are reported in which the 11.5k ohms thermistor resistance was trial shunted by two test resistors. An increasing I_bias is to be expected in this circuit, thus:
    1) With 45.8k // 11.5k, I'_bias ~= 185 uA, and
    2) with 9.8k // 11.5k, I''_bias ~=213 uA.
I assume the signal tap-point feeds a high-Z amplifier, making it not necessary to account for a fraction of bias current heading off in the signal processing direction.
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 16th, 2014, 6:00 pm

Charging Terminator Tested -- Internal 5v supply version

schematicinternal.jpg


[edit] The same comparator can be used with a voltage reference IC in place of the 5v line; see later posts.
Allows installation in the battery compartment with no additional internal wiring. This was really an experiment with parts on hand, with the best possible shown later.

Alternative with NPN instead of PNP transistor has been supplied by member Gordon, with engineering software (TINA) simulation of the circuit. Relies less on non-standard component features.
schematicgordon3.jpg

schematicgordon2.jpg



Terminated charging at 17.3v with 5 per cent resistors selected to target 17.5v.

Previous tests showed the high side battery pack thermistor must be used as the low side sensor does not appear to be monitored the same way in present software.

This version uses the regulated 5v supply for logic devices within Neato as a voltage reference for detecting a target battery voltage; accessible on the Wall Sensor and Cliff Sensor cables.
The 8-pin comparator IC is best installed in a socket for convenient soldering without damage.

A different version using battery supply only still waiting on parts delivery. A 7805 power supply voltage regulator could be used with this 5v version off the battery, but several milliamps quiescent current drain would be needed, besides bulk and needless complexity. This 5v version can be installed inside the Neato as well as in the battery compartment.

To install in the battery compartment the 5v supply must be wired into the compartment (even very thin 30 gauge wire wrap wire can be used). The circuit can be packed with the cells with a connector for the 5v line, and when the pack is removed the circuit is disconnected without the 5v supply, presenting no load on the removed battery.

All parts inexpensive in single quantity at Digi-Key.com.


[edit] Speculation re charging methods: the delta-V method found in charger controller IC's is used in small flashlight cell chargers where the cells are separated and their is no packaged thermistor. Ultra-fast 15 and 30 minute chargers can be observed to warm cells considerably from the fast rate used (it is not a defect; I've used them for years; only slow charging keeps the batteries cool). When assembled in multi-cell packs and fast charged, the temperature becomes more critical and the thermistor method is used. The delta-T rate of climb mentioned by Neato Guest here, compared to the level, serves to get ahead of the temperature curve, where NiMh chemistry produces heat after charging or discharging stops, from recombining produced gases. This was observed in a climb of several degrees C after charging stops. Li Ion batteries are supposed to generate less heat, so would not use thermistor charger control.
[edit] should have been obvious: the rate of change distinguishes the charging effect from the environment temperature level, which could be 60f in winter or 90-100f in summer. A battery might go from 60-90f in winter from charging, but be indistinguishable by level only from the environment. So the rate of change provides the distinction.


[edit] U.S. Sensors thermistor table at http://www.ussensor.com/do-35-standard-glass-encapsulated-thermistors-103jg1k-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-glass-encapsulated-thermistors-103fg1k-table but makes not difference to the circuit tested, given the large increment in temperature applied.

[edit] Note using the thermistor connector ground pin for the circuit ground disconnects its load when removing the battery, and on newer models the switch in the side of the dust bin compartment for the battery also disconnects the load on the battery (that switch appears to be on the low side pack ground connection). The thermistor pin connects to system ground within the unit, and on the dock there is power, but the battery won't be connected in that circuit with switch off. Gets a bit convoluted.

[edit] To make small adjustments to common resistor sizes, large resistors in parallel can reduce by small amounts while small resistors in series increase. For a desired resistance Rd, initial resistance Ri, and parallel resistance Rp:
Rp = (Rd x Rg)/(Rg-Rd)
Last edited by glnc222 on February 23rd, 2014, 4:10 am, edited 8 times in total.
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby Gordon » February 17th, 2014, 2:19 am

glnc222 wrote:... When assembled in multi-cell packs and fast charged, the temperature becomes more critical and the thermistor method is used. The delta-T rate of climb mentioned by Neato Guest here,..., serves to get ahead of the temperature curve, where NiMh chemistry produces heat after charging or discharging stops, from recombining produced gases. This was observed in a climb of several degrees C after charging stops. ...
Perhaps, but take into account that no more energy is being added to that equilibration process. Exothermic reactions would be required to generate heat, following charging voltage cut-off.

A more obvious process to cause those increasing temperatures that follow charge termination is that of thermal conduction from one or more relatively hot cells, none of which have direct contact to the embedded thermistor temperature sensor, to where the thermistor is located. Think about it. There are six to twelve cells packed together in the batteries discussed on this board, but there is only one thermistor; and that thermistor can only contact one or two cells. Cells in a pack that are remote to the temperature sensor may become very hot, but that heat energy has to pass through a series of card-stock sleeves and through neighboring cells before it heats the thermistor! The latent process of adding heat to a battery's thermistor continues for quite a few minutes.

BTW, gl, I think your posted circuit does function as indicated, however, there is one correction required, and a couple adjustments to consider. I will PM you about them.
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 17th, 2014, 3:33 am

I can imagine the conditions for cutting off the circuit may not be as simple as appeared in simple measurements, but may not matter to usage in the end, removing batteries when not used for long periods.

One feature possible I failed to mention: the ground connection for the sensing circuits can be the thermistor ground connection pin on the battery plug, so that if mounted with the battery which is unplugged the circuit is broken at the ground, disconnected, eliminating load. This can be applied as well to circuits using only the battery voltage and voltage reference IC's instead of the 5v line.

The 5v shutting off then is just for Neato's shut down mode with the battery plugged in. Neato already advises not to leave batteries in for long periods without use, so some small load when installed then is not overwhelmingly important, if this feature attempted doesn't work out. The NiMh's have self-discharge anyway. Not getting the perfect lack of self-discharge in LIthium batteries is no big deal.

The unusually low C-E voltage drop on the transistor would not seem to matter if different, in terms of being able to terminate charging. It just affects calibration of the resistors to use, whatever the drop is. Other transistors might be used, anyway. The circuit is mostly for individual DIY use, where calibration would be done on the unit with meters and potentiometers. The exact responses of the comparator are not known and there is the standard ambiguous crossing region where it oscillates etc. given slow moving inputs, with some random variation in exactly where it would trip.

There seemed no need in this application to apply hysteresis feedbacks I've seen as a standard technique and used before. There is a small range over which the charge termination will vary. Allowance would be made by setting a suitably low termination point for whatever battery type. It cannot be perfect.
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Neato lithium ion battery revisited

Postby glnc222 » February 17th, 2014, 3:55 am

There may be one problem in that circuit relying on the open collector comparator output to shut off the bypass.
[edit] checked that, no problem, forget it.

It is not an actual disconnect but a lower than full level driving state which can push current into the bypass when the voltage across the thermistor is too low, as in a high temperature operating environment. It can light an LED for example. I don't know the exact specifications of it. Just sometimes it can be treated as a disconnect.

So there may be need to insert a switching diode like 1N4148 between the comparator and the bypass load to achieve a full disconnect when not bypassing ([edit] or add the common pull up resistor). I think the voltage drop on a diode would restrict the highest temperature environment at which it can work, lowering the effective bypass load added to already low resistance thermistors. I will check a larger range of initial conditions to see whether there is a problem; it can be done with potentiometers. It is reasonable to specify Neato operability within certain temperature limits, and working in 100 degrees F can be reasonably excluded, IMHO.

[edit] For maximum sensitivity the kosher solution is control an NPN BC547 to the comparator, reversing the inputs, and switching the bypass on the perfectly cutting off transistor. A shortcut was used to reduce parts count, which may work. Testing...
glnc222
Robot Master
 
Posts: 4541
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

PreviousNext

Return to Neato Robotics

Who is online

Users browsing this forum: Brett, CherkyB, Franzl, moradi2, susantx3, Wing Nut and 336 guests