ebikessf's 2k thermal sensor?

[auraslip]

Ok, how does this work? It's a one wire sensor?

http://www.ebikessf.com/node/202 Only $300 for the H series motor too. Best deal I could find.


here's a picture showing the wiring harness. Yes they use bullets here. me gusta. (still have to upgrade the wiring though. Looks like 16 awg)

[wojtek]

oh yes Ilia is a legend!!! great service and best price

you can email him and ask: ibrouk@hotmail.com

[GCinDC]

anyone know how to wire the sensor then?

[migueralliart]

Correct me if I'm wrong but the thermistor is a 2 wire device. So my guess is that one wire is grounded and the other connected to the controller fail switch. When the temp is normal the thermistor is passing ground thru the connection and when it heats up is is essentially a switch that opens (cuts) the ground connection.

[ebikes-sf]

Hey guys, the built-in thermistor model is KTY81 210 , one lead connected to hall sensors ground (black wire) the other lead is a loose white wire.

2000 ohms @ 25 °C
2417 ohms @ 50 °C
3392 ohms @ 100 °C
4280 ohms @ 150 °C

Dataseet: http://www.nxp.com/documents/data_sheet/KTY81_SER.pdf

Have been busy and wasn't able to find a way to utilize it, but first thing I was gonna try is use CA input voltage feature like outlined in CA manual, section 9.2, page 19
http://www.nxp.com/documents/data_sheet/KTY81_SER.pdf

May be with CA v3 it will be eazy to utilize.

[Joe Perez]

When the temp is normal the thermistor is passing ground thru the connection and when it heats up is is essentially a switch that opens (cuts) the ground connection.

I think that we may be mixing metaphors here, so to speak.

There are thermoswitches out there which behave as you've described, passing a closure below a certain temperature and then opening above it, or vise-versa. Functionally, these are a lot like an old mechanical thermostat in your home, and were commonly used in automobiles from the 1970s through the 1990s to control the radiator fans. These devices, however, tend to be comparatively large and bulky, as they contain moving mechanical parts (a set of contact points and a bimetallic strip.)

There are three other major types of temperature-sensing devices in common use, and I suspect that that you have one of them.

The first, which is the only device properly named a thermistor, is a variable resistor whose resistance changes with temperature, either increasing as temperture rises (PTC, or Positive Temperature Coefficient) or falling as temperature rises (NTC, or Negative Temperature Coefficient.) These devices have two wires, of which one is usually grounded and the other has a pullup voltage applied through a fixed resistance, thus forming a voltage-divider.

The second is essentially a zener diode whose reverse breakdown voltage changes with temperature. These are also two-wire devices, and are used in a similar manner to thermistors. An example of this would be the Texas Instruments LM335.

Finally, there's a class of devices which are internally similar to the above, but which contain active circuitry and output a positive voltage proportional to temperature. These are three-wire devices, and an example would be the TI LM35.


Now, I don't know which one the aforementioned motor includes, so that's worth a call to ebikessf. But the fact that only one wire is exposed does not tell you anything about it. Internally, the other leg (or legs) are probably connected to the ground (and +5, if applicable) of the Hall Sensor harness, in order to minimize the number of discrete wires passing through the axle. This is how Cell_Man's MAC motors are wired, for example. They use LM35s internally, where the +5 and GND are internally cross-connected to the Hall board, so that you only have one wire coming out which is specific to the temp sensor.




EDIT: Looks like the answer beat me here. That's a PTC thermistor.

[migueralliart]

Like I said it was a guess.... without the specs(description) of what he used it's impossible to tell.

[GCinDC]

thanks, ilia.

i'll take some pics once i open the motor. before i slather it with jb weld. don't worry, i won't be asking about warranty...

rather than hook it up to CA to limit current/speed just yet, i'd like to use it to monitor temp. i assume it will not work hooked up to HK temp monitor (along w/ some current to power the display)?

if not, anyone know of any cheap consumer displays/alarms that could be hooked up (if only a tiny dmm to display resistance - an analog mm dial would be cool but doubt it would survive long on board)? or, a simple circuit with perhaps an LED with variable brightness/color? or a couple alarms, one to go at 80, another at 100, and a third at 120?

[spinningmagnets]

This is a smart move for any manufacturer/retailer. The probe by itself is very cheap. This allows the end-user to add a digital read-out or a Cycle-Analyst when funding allows, without having to disassemble the hub. Does anybody have a part number of a probe that is Cycle-Analyst ready, like this one?

edit: just saw the posted number above "the built-in thermistor model is KTY81 210"
Heres a link from Digikey...$1 EACH
http://parts.digikey.com/1/parts/954667-ic-temperature-sensor-sod70-kty81-210-112.html

I've worked in the bay area several times, and it makes sense that an E-bike retailer from there would be one of the first to start doing this, since they are famous for having some very hilly regions...

[Joe Perez]

Just as an FYI, the software on the CA3, at preset, natively expects either a 10k NTC thermistor or a linear, zener-type device. The device listed above will work, you'll just have to treat it as a linear device and come up with the proper calibration for it. I just did a quick plot of the signal you'd get from feeding it a 5v pullup through 1k, and it's around 0v = 2.245v, plus 0.9v / 100°.




EDIT: This data is wrong. The pullup in the CA3 is 5k, not 1k. I'll re-compute and fix the kittygraph.

[spinningmagnets]

Do you have an idea of a part number that would not need any special calibrating...something just plug-and-play for the not-very-technical?

[Joe Perez]

Well, setting up the sensor from above is not that big a deal- you just tell the CA3 it's looking at a linear sensor and plug in the calibration data from the kittygraph.

If you want the simplest device possible, then get a 10K NTC thermistor with a beta constant in the 3900k vicinity, such as this one: http://search.digikey.com/us/en/product ... -ND/739889


If you're looking for a zener device rather than a thermistor, the LM235 is what you want: http://search.digikey.com/us/en/products/LM235Z/497-7324-5-ND/1038647 This device gives you exactly 1v / 100°C up to 125° continuous, and since 0°K = 0v, 0°C = 2.73v, which would be your zero cal point.

[auraslip]

Can anyone wax eloquently on the difference between a two wire thermals sensor and a three wire thermal sensor?

[Joe Perez]

Can anyone wax eloquently on the difference between a two wire thermals sensor and a three wire thermal sensor?

There are three basic categories of sensor that are of interest to us here. Two of them are two-wire devices, but internally they are very different.

1: Themistors (both PTC and NTC) are essentially just resistors whose value changes with temperature. These devices have two wires, and are non-polarized. To use them, you arrange the circuit such that the thermistor forms "R2" in a standard resistor-divider network, where R1 is the internal pullup resistor inside the CA3, for instance:



So arranged, the voltage at the center-point between the two resistors can be predicted as Vout = (R2 / (R1 + R2)) * Vin. You then have a table of resistance curves for the thermistor, so from the fact that you know Vin, R1 and Vout, you can calculate what R2 must be, and then from that, derive what temperature would cause the device to be at R2.


2: Zener-type devices such as the LM 135/235/335. These are also two-wire devices, but they are not resistive. Zener diodes are kind of unique in that, when a voltage is applied across them in what would nominally be considered the reverse direction, they will drop that voltage to a very specific and precise level. (They are commonly used as precision voltage references in electronic devices.)

Sensors of this nature take advantage of the fact that, when properly constructed, the Zener voltage changes with temperature in a very predictable way, and contain some internal circuitry to calibrate this voltage to some desired scale. Typically, these devices will have a zener voltage ratio of 10mv per degree K, from absolute zero (theoretically) all the way up to 125 or 150 degrees C. They are wired identically to a thermistor (with a pullup resistor) and read the same way. From a practical standpoint, the main difference is that their output is perfectly linear and follows a published chart, so it doesn't matter what the value of the pullup resistor is, the voltage will always follow what the data table says it should.


3: The last type is the three-wire devices such as the LM35. Internally, these are similar to #2 above, however they require +5 and GND, and then output a voltage on their own, so they do not require a pullup. The LM35, as an example, is calibrated in degrees C rather than degrees K, at 10mv / degree, so that the output can be read directly on a cheap voltmeter (.50 volts = 50 degrees, 1.10 volts = 110 degrees, etc.

[GCinDC]

no cat background?

seriously, thanks, this is very helpful! (tho more help may still be needed. )

[Joe Perez]

no cat background?

Heh. Sorry, I'm typing this from my tiny laptop while sitting on the couch waiting for some DP420 to dry, and I don't have PaintShop installed on this machine.

(I should probably rectify that oversight so that I can include more kittens in my data in the future.)

[auraslip]

Awesome post.

I'm currently using an lm35 powered temp sensor.

Is there any advantage to it? Perhaps easier to create a digital read out with it? Something tells me it should be fairly easy to do. I have the HK-VT. It works well and runs off 12v power, but it constantly flashes between voltage and temperature. Slightly annoying.

Ideally, I'd want a single display that will switch between controller temperature and motor temperature. I'm honestly surprised someone here hasn't made a circuit for that and is instead relying on BBQ meters.

[Joe Perez]

I'm currently using an lm35 powered temp sensor.

Is there any advantage to it? Perhaps easier to create a digital read out with it? Something tells me it should be fairly easy to do.

Yes, that's the primary advantage of the LM35 over the LM135/235/335. Its output is exactly 10mv per degree C, with 0v = 0º C. So the LM35's output voltage is always .01 * temperature, and can thus be directly displayed on any cheap voltmeter capable of accurately displaying small voltages from 0-2v, such as:
http://www.ebay.com/itm/3-DC-2V-Blue-LC ... 330wt_1169
http://www.ebay.com/itm/Class-2-5-Analo ... 1527wt_952

With any such meter, a temperature of 80º, for instance, will show up as ".80"

The LM135/235/335 is similar, but with one major difference. That device is calibrated for 0v = 0º Kelvin, which is -273º C. Handy for scientific purposes, but requires conversion to produce a meaningful display in ºC.

[GCinDC]

2000 ohms @ 25 °C
2417 ohms @ 50 °C
3392 ohms @ 100 °C
4280 ohms @ 150 °C

hooked white wire up to multimeter with other lead on hall GND to test resistance on the bench. read 1.930k ohms @ 72F ambient.

motor laced into heavy 19" M/C wheel/tire.

@80V WOT with no load, resistance rises to 2.6 on first steady throttle increase but then drops to 2k ohms when i release. then later, from WOT, i release, regen engages and it tops 3k and quickly drops back to 2k.

pulsing the throttle on and off, going quickly to WOT, then releasing and engaging regen, it bounces quickly between 2k ohms and 4k ohms.

i'm afraid axle is bending with only one end in vice, so will take a vid when i get it on the bike.

is that behavior to be expected?

[GCinDC]

interesting to see they added a 4th slot. surprised to see it attached to red, but forgot to verify what line that's on...


and of course the bearing stayed on the axle, and my gear puller claws won't fit, so i might trim them back to the lines i drew.

this is worth it right? the bearings won't seat perfectly otherwise, right??

[GCinDC]

interesting to see they added a 4th slot.

i was wrong: it's not a 4th slot - it's just the gap between the laminations, and the bamboo was cut short to accommodate it. here i'm replacing thermistor with LM35.

also, though the non-white wire is RED, i've confirmed it's connected to the hall GND, not the positive.

btw, the LM35DZ is a bit fat, but will it work fine? should attach the hall + to the +Vs and NOT the Vout?

lm35DZ.jpg (10.79 KiB) Viewed 473 times

i'd rather not toast it! thanks.

ps. i'm curious about your avatar. what's the story about it?

[Joe Perez]

btw, the LM35DZ is a bit fat, but will it work fine? should attach the hall + to the +Vs and NOT the Vout?

Assuming that you are borrowing power from the hall board, then yes. The hall's +5 wire (typically red) will go to the LM35's +Vs, and the hall's ground (typically black) goes to the LM35's GND. VOUT from the LM35 is a variable voltage output, and this should go straight to the (+) terminal of a voltmeter (or similar) with no pullup. Ideally, the voltmeter's (-) terminal will go back to the hall ground wire, both for safety and to minimize the opportunity for ground offset skewing your readings.

ps. i'm curious about your avatar. what's the story about it?

Well, it's a giant chicken dressed as Boba Fett from the Star Wars movies. Seems pretty self-explanatory to me.


I've been pretty heavily involved in the Mazda Miata (MX-5) aftermarket for a number of years, principally with regards to forced induction, ECUs and engine tuning. Back in 2007, a specific manufacturer of supercharger kits started making some false and unsupportable statements about a specific ignition-control device which he was offering, which a user on one of the Miata forums posted a link to. (It should be noted that a state of inter-forum cold war has also existed for many years between the forum in which said posting occurred and the rival MiataTurbo.net form, on which I'm a moderator. I routinely frequent both forums, however.)

An epic thread ensued, with forum members allying themselves both with and against the vendor in question, mostly reflective of the long-standing and nearly religious debate which has long waged between fans of supercharging vs. fans of turbocharging. The vendor in question started digging himself into a hole and throwing out a lot of obfuscation and pseudo-science in an effort to confuse the populace and thus draw attention away from the matter at hand. At some point during the debate, I used the "walks like a duck, quacks like a duck" analogy. Mr. pseudo-science then said (and I'm quoting directly) that "a duck, chicken and goos are all the same. Although the feces is completely different."

If you're particularly bored, you can read the whole thread here: http://forum.miata.net/vb/showthread.php?t=255415

At that point, the whole chicken / duck / goose concept virtually became a meme within the forum community. And at the suggestion of a user at MiataTurbo.net, I decided to adopt a poultry theme for my avatars, as both an inside joke and a persistent jab. Since then, I've used chickens and ducks exclusively as my avatar theme, both on the Miata forums and elsewhere, as it's quite memorable.

The fact that Ernie the Giant Chicken is wearing a Boba Fett costume in the current avatar is of no particular significance. I saw it in an episode of the TV show Family Guy, thought it was funny, and took a screen capture of it.

[amberwolf]

and of course the bearing stayed on the axle, and my gear puller claws won't fit, so i might trim them back to the lines i drew.

If you pull on that outer race the bearing will likely be damaged. You want to only pull on the inner race, because that's where it's connected to the motor. Pulling on the outer race is going to force that race against the bearings, which then force against the inner race.

[GCinDC]

If you pull on that outer race the bearing will likely be damaged. You want to only pull on the inner race, because that's where it's connected to the motor...

eek, grim thought. i've seen other wheel develop wobbles and i wonder if unseated bearings are the cause. i'd rather ruin the bearing and have to replace it. i carved up the ends of the gear puller and the bearing came off just fine:

i did notice a bit of rust developing underneath, so it'll be an area to keep an eye on..

[auraslip]

to be fair, a decent set of bearings is a good and cheap upgrade for an ebike. the ones that come with them usually suck, and USA made replacements can be had at a local bearing shop for $20.