The question:
Can the individual temp sensors be calibrated? Can I somehow fix one that reads off? Or did one of mine (or the board channel for it) go bad and needs to be replaced?
The story:
After I put my VX-1 NiMh battery together (replaced a couple of cells), one of my temperature sensors is now off compared to the rest. It is not the actual cell block temperature that is off, only the sensor's reading of it. The censor was fine before I took the battery apart.
It is not completely dead, the reading kind of rises and goes down slowly (not jumping) by a few degrees over a charge and with use, but it lags a lot from the rest of the sensors, which all seem to show reasonable temperatures vs. ambient.
Apparently, having more than a few degree difference between the several temp. censor readings messes-up the charging process. The "bad" sensor reading is around 20C (as lo as 19.x to as high as 22.x is what I have observed over the past few days, and it seems to mirror the temp change of the other sensors - higher when they are warmer too). Unfortunately, when the other sensors get to something like 30C (it is warm 'round here), this one stays at around 20C and that apparently shuts off my charging - it only wants to do EC charge and quits the CP stage almost immediately...
I almost got stranded because of that today! I did notice the temp difference in my garage when I put the pack together a few days ago. But it is cool there (23C ambient, so I suppose the charges have been completing fine and were not bothered by one group of cells showing as 20C while the rest were around 23-25C).
So, getting normal charges over the past few days in the garage, I assumed all is well. Wrong, as I feared. This morning I noticed the charge had completed sooner than expected (the garage had warmed up the previous day to 29C or so). So this morning I only got 143V during an overnight charge (on a timer) instead of the 146V it would normally get after a full charge. I started my trip at 143V and rode it to work with the idea I'd charge there as I normally do. But that did not work out as planned - outside temp was around 30C, but the "cool" temp sensor stayed around 20C. And the charger would not charge other than do a 5 minute EC count down at low amps. Basically, I had to make the return trip on the remaining less than 1/2 charge in my battery (was not a full charge to begin with). I made it home OK, it is a 24 miles round-trip so I can make it on one charge, but the return trip had to be at lower speeds and at the end had to slow down to 30mph on the flats to keep the voltage above 120-122V. I had at one point the teltale hot battery and battery light-up, even though my battery temp was only 29C in use and had not risen during the ride from where it was earlier in the morning- it was at ambient, more or less. I stopped, tried to plug in again, did not work (just 5 min of EC charge, which I did not wait to complete), so I went on my way. The lights came off and did not come back for the rest of the trip. Nor did the temp rise any more - stayed at 29C until the end (I rode gently!).
Long story short, do you have any ideas on what may be wrong? The temp is not jumping around (so the connectors to the temp sensor boards are probably fine), the temp reading from that one group is not some unreasonable number and actually fluctuates a bit up and down. Could the thermistor be damaged somehow but not enough to throw off a wild reading?
I will check the connectors tonight, but if that's not it, getting to that one sensor in the bottom of the front pack will mean I need to take apart the entire battery, again...
Is there a way to calibrate the sensors? Or to ignore one of them?
Thanks!
The sensor was working fine, you detach it from the battery, and now it reads 20C high.
Before finding a way to calibrate it, test de sensor again attached to the battery again, with the original, expected, voltage jumps. I guess it will work again.
Everything *is* attached to where it is supposed to be: all sensors are screwed in the packs, the two sensor boards are installed.
I have opened the bike-up again to troubleshoot - the problem is with the front temp sensors. When I disconnect the control board's plug, they all read 60C (as expected), plug back-in and the one that's off continues to be off, showing around 20C with the rest at about 25C.
I will see if I can measure the resistance of the thermistors via the wire leads at the control board end first, so this way I don't have to dissasemble the battery - as I said, the faulty reading comes from a sensor on the bottom of the front pack, unfortunately. This means I have to take out the rear half of the battery, then take out the front battery and disassemble it to get to the bottom layer...
Luckily, that won't matter too much soon, because I have already bought the Leaf cells for Li conversion, so the whole 9 yards of battery and censors are coming out of the VX-1's guts for good pretty soon :). I was just hoping to use it for a few more days until the upgrade begins. As it is now, I can still use it if I bring it inside at home, where the ambient with the A/C going on is close enough to the faulty reading, so I can charge normally there - just can't charge in the warm weather outside...
You may want to check this: http://visforvoltage.org/forum/13558-looking-some-temperature-sensors
Thanks! From that post I see replacing a faulty sensor (or fooling the temperature circuit) would be easy.
Measured the resistances of the 6 thermistors on the front half of the battery lack today: all but one are about 9.5 KOhm, with one being about 11.5 KOhm. That's measured at the front temp. control board (at the 3 "common wires", one per layer of batteries, and 6 wires that go to the 6 individual sensors, two per layer). Obviously, one sensor is measuring 1/4 higher resistance than the others (at the ambient temperature in my garage), which translates in a reading that is off at the computer.
Incidentally, the charts in that post match almost exactly my readings: I get around 20C for the 12KOhm sensor and 25C for the working 9.5KOhm ones.
Suggestions? Is there a calibration procedure? I suppose I can replace the faulty sensor, but I'm curious if there is a provision to deal with such issues by calibration...
Took apart the battery to get to the sensors. The "bad" one, of course, had to be on the bottom layer :( Measured its resistance again, at a different temperature, and it was again off by about 1.5KOhm vs. the other 5 thermistors. So it is working, just it is off by some value.
My theory is that it must have been damaged by heat from that one hot cell that vented - it was mounted right there on top of that cell.
Edit: the $0.05 solution I think will be to solder a 40KOhm or so resistor in parallel to the sensor leads (at the board or somewhere else, doesn't matter). That it seems would give me a similar total resistance as the "good" sensors...
So, does anyone know if there is a calibration provision for the temp. sensors?
The question is mainly of informational interest at this point, as I will not be using these sensors any more. Plus I can repair the bad one easily if I do need them or want to sell them to someone someday...
I suppose no one knows if there is a calibration procedure. I'm sure it can be done in the source code...
Regardless, FYI, I replaced my one faulty thermistor today. Now readings are all as they should be (before, the bad one was off by 1-2kOhm, depending on temperature, which is what the Vectrix considers a significant number, resulting in about 5 degree or so false reading, which was enough to light-up the red lights and stop charging).
Pics attached. To do this repair the battery needs to be disassembled, unfortunately, unless your faulty one is on the top of the three layers of cells, in which case you don't even need to take the battery out. Mine, of course, was at the bottom layer of the front half of the battery, so I had to take everything out and apart...
.
Damaged thermistor. The little dark dot:
A strip with new replacement thermistors. Just one of these is needed, I ordered more as they are cheap (in case I damage or lose them - they are tiny!):
The part number and model of the replacement thermistors (from Digi-key). These are a bit smaller than the original ones, but should have a similar thermo-resistance curve. Need food eyesight and a steady hand to remove the old one and solder a new one:
I am sure many other people have a more intimate knowledge of these
temp sensor assemblies than I but someone needs to do it.... please
correct any of what follows if wrong..... here goes:
The "thermistors" as they have been called by reference are not really
thermistors..... they are but they are not... if that makes any sense.
What the little chip is: LM45 search for lm45 datasheet and you will
see that it is a tiny integrated circuit.
The way these temp sensors work (as I understand it):
each of the little sensor boards has two wires connected to it: one is
a "power supply" and the other is the "sense line". The power supply
line/wire is pulsed by the processor, while the pulse is active the
processor reads the sense line for that sensor, and, converts the
feedback voltage level into a digital code. That code is what is
sent onto the can bus (otherwise known as raw A to D conversion data).
There is no "calibration procedure" for the sensor processor boards as
the raw data is sent to the Charger, and then gets converted into a
temperature value there (via a look up table). Now, as best as I can
tell, it appears that the code for the sensor processors was never
completely finished...(for whatever reason: not enough memory, conversion
speed, etc.) and an interim decision was reached to just send the unconverted
data to the charger for further processing. If I remember correctly the
signaling protocol for the LM45's is not straight forward, so while you
might be able to install a regular thermistor it is doubtful that it
will work correctly - the charger look up tables are calibrated to the LM45
characteristics.... meaning: the LM45 has a linear response to temperature
whereas a NTC/PTC thermistor may not necessarily react identically; so,
the conversion data will never be accurate compared to the other sensors; thus,
creating errors..... your best bet is to replace a faulty sensor with
a LM45. Hopefully this answers some questions and fills some of the
information gaps.
Galago
Good information! I have not tried the "repaired" sensor back in the bike, so I don't know if it works or not. You are probably right and it won't produce the expected values compared to the original sensors. The resistance value at room temp is the same as on the original sensors (I measured), but if the voltage reading is taken on the "sense" wire and is not the same as the simple voltage drop over the intermal resistance of the sensor's thermistor, then of course, it won't work...
As for the calibration, while the sensor may not be possible to calibrate, the interpretation of its value could be off-set in the software accordingly. I'm guesing that the charger code can be modified to offset any value that is fed to it.
So, ... we should be buying the LM45 modules and not plain old thermistors then. LM45 at digi-key looks different though: http://www.digikey.com/catalog/en/partgroup/lm45/23064?mpart=LM45CIM3%2FNOPB&vendor=296&WT.z_ref_page_type=PS&WT.z_ref_page_sub_...
I think it should be replaced with a component that is an acceptable match for the failed component, which might not necessarily be the LM45 in all cases.
There are at least two different types of sensors. One has a three legged chip and uses three wires, one of which appears to be connected to the battery terminal connection, which makes me think it might be for voltage measurement which presumably is used by the BMS on the later Lithium packs?
Unless my eyesight is mistaken, the middle leg does not even appear to be soldered to anything on the circuit board:
The other type only has the two wires connected across the thermistor (unless the original chip that has been replaced was a three legged one which also connected to the battery terminal connection):
If the original chip only had the same two connections, I reckon it is just a simple thermistor, in which case it should work OK when you refit it.
There is only one way to find out for sure, so please let us know the outcome after you have refitted it and tested it.
Alan
I agree there only appear to be 2 solder points for these little surface mount thingies. The big hole in the board is for a screw to attach the sensor to one of the terminals on a battery cell on the NiMh bikes. The schematic for the LM45 shows three connectors as mentioned: 1. Ground, 2. Vs (power supply voltage), and 3. Vo (for output voltage).
So, because there is no 3rd wire in my application, I don't think these are LM45 ICs. I don't think an LM 45 type device will work without all 3 being connected, right? I also think that most likely that 3rd wire shown on the other type of sensor is for a BMS on the Li bikes (or for the never done one on the NiMh ones) and not for the sensor's own use. Which leads me to suspect these are indeed simple thermistors, not LM45-like ICs...
I won't have an occasion to open the bike soon (hopefully, it's working fine now without the temp sensors with modified software). But when I do open it, I will reinstall the boards and will update this post. I've been meaning to put the sensors back in the bike to know what the battery temp is, even though my bike does not care.
So, am I missing something with my logic above?
The Vectrix sensors are simple thermistors which form a voltage divider with another resistor on the mother PCB. They are accurate to within about +-2degC. These are 10K thermistors at 25degC.
Hi all,
I can't start a new topic so I try to revive this one with my question.
Does anyone have an idea of what can cause this?
Apart from the first line, there are no temp readings at all....
Tamas
I am guessing this means your temp boards are disconnected or not reporting back correctly. I'm guessing it defaults to 60C in this case