hillzofvalp
100 kW
Yeah one was like 8.6K at room temp while the other was 9.35K or so. I have reason to believe I damaged it.. I have another in my controller I needed to readjust anyways so I'll just pop it out and use it in the motor for now.
ArduEbike - Arduino Ebike projects
- Thread starter Alan B
- Start date
grindz145
1 MW
Bummer. Maybe it's a good time to order some RTDs. They aught to hold up to a little more abuse.
I made the mistake of designing a thermistor into a military grade power supply once. When the systems engineer did the MTBF analysis, I got a big time slap on the wrist.
I made the mistake of designing a thermistor into a military grade power supply once. When the systems engineer did the MTBF analysis, I got a big time slap on the wrist.
hillzofvalp
100 kW
When I was initially testing my code a few months ago, I touched the thermistor with a 500C soldering iron! i think I'll just get a replacement thermistor... BUT I will def try again when the motor has cooled to determine if the "good" thermistor is really "good".
Try a type K thermocouple and a MAX6675 amp chip. Reads out digitally. Good to 1023.5 C (the MAX6674 has twice the resolution, but only reads to 128C).
I am building a device that is slightly oriented towards high-power LED measurements, but should have lots of uses for the ebike world. It measures the input voltage and current, driver output voltage and current, two thermocouples, two light sensors, color sensor, etc. The input power is switched to the driver with a very low Rds FET. The processor is a ATMEGA328 that is Arduino compatible.
Some explanation here: http://budgetlightforum.com/node/12296
Image of control program screen with plots here: http://budgetlightforum.com/node/12296#comment-225832
I am building a device that is slightly oriented towards high-power LED measurements, but should have lots of uses for the ebike world. It measures the input voltage and current, driver output voltage and current, two thermocouples, two light sensors, color sensor, etc. The input power is switched to the driver with a very low Rds FET. The processor is a ATMEGA328 that is Arduino compatible.
Some explanation here: http://budgetlightforum.com/node/12296
Image of control program screen with plots here: http://budgetlightforum.com/node/12296#comment-225832
grindz145
1 MW
Do type K thermocouples still need a relative thermistor reading at the cold junction?
hillzofvalp
100 kW
The reason I chose this thermistor is that it is directly compatible with the the CA v3.... Digital is something to look into.. I can always use a thermistor for CA and a digital solution for logging. Also, I'm sampling 511 times for 2 thermistor outputs... In order to keep things simple. A digital system would probably cost more, right (edit: yes: MAx chip is $15)? And to et the use out of it I would want to take more samples which my board doesn't have memory for,
hillzofvalp
100 kW
would you think that your suggested setup would really be worth it for what I'm doing? Seeing the relationship between stator ID steel temperatures and those of the windings? Would you suggest using a multiplexer to switch thermocouples and use a single chip or buying two chips at $15/each. Could be a nice project...
do you think that calibrating to each thermocouple would be easier than with thermistors? OR same thing?
I kind of like this thermistor:
http://www.ussensor.com/sites/default/files/downloads/PT103J2%20REV%20NONE.pdf
do you think that calibrating to each thermocouple would be easier than with thermistors? OR same thing?
I kind of like this thermistor:
http://www.ussensor.com/sites/default/files/downloads/PT103J2%20REV%20NONE.pdf
Alan B
100 GW
Thermocouples are a difficult way to go.
They require special wire and connectors. The special metal of each conductor cannot be changed all the way to the electronics.
The signals are very small and noise sensitive.
They are best for large temperature ranges and high temperatures.
They require complex / expensive signal processing on each channel (can't mux them before that).
Cost becomes a factor for a multichannel system, and four channels would be about right here for ambient, controller and two for the motor, or one for battery and one for motor. Perhaps even 6 channels.
They require special wire and connectors. The special metal of each conductor cannot be changed all the way to the electronics.
The signals are very small and noise sensitive.
They are best for large temperature ranges and high temperatures.
They require complex / expensive signal processing on each channel (can't mux them before that).
Cost becomes a factor for a multichannel system, and four channels would be about right here for ambient, controller and two for the motor, or one for battery and one for motor. Perhaps even 6 channels.
Alan B
100 GW
hillzofvalp said:
That one states for best life only good to 120C. I would want something rated to 200C.
I am tempted to try the 1N4148 solution.
Alan B said:
Another tip, I haven't see this one elsewhere yet: if you need a higher temperature constant, you can solder 2 diodes in series and thus have twice the voltage change per degree of temperature
, at the expense of having twice the "common mode" voltage. You can actually put several in series, but I guess in normal conditions this is only feasible with up to 2, which can be soldered very very close, because otherwise it's harder to ensure they both always stay at the same temperature.
hillzofvalp
100 kW
It says for long term use store it under 120C.. I think it's 150C rated. I like that it is 0.2C accuracy but only $7
Njay, if you have two 200C thermistors and you want to put them in series to measure 300C, what is the point? The only reason to pure them in series is to take an average in their designed temperature range that peaks at 200C
Njay, if you have two 200C thermistors and you want to put them in series to measure 300C, what is the point? The only reason to pure them in series is to take an average in their designed temperature range that peaks at 200C
Not talking about thermistors nor about measuring a higher temperature, hillz. Diodes have ~ -2.3mV/ºC temperature coefficient on their Vfw; if you put 2 in series very, very close to each other, you have a "device" with a ~ -4.6mV/ºC temperature coefficient on Vfw, easier to measure than a -2.3mV/ºC.
Alan B
100 GW
hillzofvalp said:
It says for stability to store and operate under 120C. Basically over 120C they change their calibration permanently, so I would consider their upper limit to be 120C. I suspect the epoxy is the problem, glass encapsulated parts are likely a better choice.
Series connecting two sensors creates a larger signal, for greater resolution, not a change in peak temperature capability. Makes it easier to read them out. Might be worthwhile.
I've worked with Platinum RTDs which are great for accuracy and calibration but require more wires (3 or 4 conductors) and amplifiers as the signals are pretty small. They cost a bit more too. Probably not worthwhile for this application.
Another way to measure the winding temperature is to use the copper wire in the motor itself. Easy to make a measurement when the controller is off and the motor is stationary. So it requires stopping for a few seconds to take a measurement. But it does not require opening the motor to install sensors. It requires some circuitry and perhaps a relay to disconnect from the motor during operation.
Alan B
100 GW
Copper as Resistance Sensor
Copper changes resistance at 0.393 % per Centigrade degree. So about 40% resistance increase in "normal" operation from room temperature to 125C.
Here is a nice example in Justin's measurements of this:
http://endless-sphere.com/forums/viewtopic.php?f=2&t=14494
In this case a 9C 2805 went from 0.096 ohms at room temperature to 0.128 ohms after dyno testing. This represents a 33.3% increase in resistance which is about 85 Centigrade degrees hotter.
To make this measurement requires some gain, and probably a differential measurement. In this case to get a 1V signal for the ADC would require a 10 Amp excitation which is a bit much for a small measurement device. A more practical 100mA excitation would require a gain of about 100 to produce 1V. An FET used to switch the excitation on and off, and taking measurements both ways would allow compensating for drift in the offsets. The motor would have to be still to avoid back EMF fouling the readings.
That plus a relay to disconnect when not taking readings.
edit - fixed typo
Copper changes resistance at 0.393 % per Centigrade degree. So about 40% resistance increase in "normal" operation from room temperature to 125C.
Here is a nice example in Justin's measurements of this:
http://endless-sphere.com/forums/viewtopic.php?f=2&t=14494
In this case a 9C 2805 went from 0.096 ohms at room temperature to 0.128 ohms after dyno testing. This represents a 33.3% increase in resistance which is about 85 Centigrade degrees hotter.
To make this measurement requires some gain, and probably a differential measurement. In this case to get a 1V signal for the ADC would require a 10 Amp excitation which is a bit much for a small measurement device. A more practical 100mA excitation would require a gain of about 100 to produce 1V. An FET used to switch the excitation on and off, and taking measurements both ways would allow compensating for drift in the offsets. The motor would have to be still to avoid back EMF fouling the readings.
That plus a relay to disconnect when not taking readings.
edit - fixed typo
hillzofvalp
100 kW
Yeah.. That's interesting. I would need to have a comparable measurement technique for the stator.. So I would like to use some type of sensor.
However, this is a great method of calibrating the thermistors at high temperature!
However, this is a great method of calibrating the thermistors at high temperature!
Alan B
100 GW
Why the interest in the stator temperature?
hillzofvalp
100 kW
Because that is what contacts my cooling inserts. I just want to see a little bit of what thermal transfer relationship is happening between the stator steel ID And the windings.
Alan B said:
The MAX6675 makes using thermocouples VERY easy. They chips can be had for $4 on Ebay. I get very good readings at room temp. Thermocouples can be had for a couple of bucks each. You don't need special connectors, you can solder the wires or pay 2 bucks for the connector. And you can extend them with regular wire as long as each lead has the same wire combos. (still it is best to use thermocouple wire).
New Maxim chip for thermocouples:
http://app.ic.maximintegrated.com/e/es.aspx?s=248385305&e=80
http://www.maximintegrated.com/datasheet/index.mvp/id/7273
http://datasheets.maximintegrated.com/en/ds/MAX31855.pdf
http://app.ic.maximintegrated.com/e/es.aspx?s=248385305&e=80
http://www.maximintegrated.com/datasheet/index.mvp/id/7273
http://datasheets.maximintegrated.com/en/ds/MAX31855.pdf
I'd like to get into programming Arduinos and thought the ebike might be an excellent place to start.
I'm not really interested in data logging.
What I would like to to for a first project is readouts of various sensors, basically making my own bicycle computer, to show speed, pedal cadence, ambient temperature etc.
That might be simple enough to start. Then maybe could follow readouts on my android phone, throttle control and other things.
Could anyone maybe recommend the hardware to get started so I have everything together to build something that'll work.
If I get a display, I'd like to be able to weatherproff it and keep it mounted to the bike.
I'm open for any suggestions and pointers.
I'm not really interested in data logging.
What I would like to to for a first project is readouts of various sensors, basically making my own bicycle computer, to show speed, pedal cadence, ambient temperature etc.
That might be simple enough to start. Then maybe could follow readouts on my android phone, throttle control and other things.
Could anyone maybe recommend the hardware to get started so I have everything together to build something that'll work.
If I get a display, I'd like to be able to weatherproff it and keep it mounted to the bike.
I'm open for any suggestions and pointers.
While not arduino, the hardware bits here:
http://www.tricorderproject.org/tricorder-mark2.html
and here:
http://www.tricorderproject.org/tricorder-mark1.html
might help with sensor setup.
http://www.tricorderproject.org/tricorder-mark2.html
and here:
http://www.tricorderproject.org/tricorder-mark1.html
might help with sensor setup.
Alan B
100 GW
Thread Wakeup!
What Arduino goodies for eBikes have people been doing lately?
What Arduino goodies for eBikes have people been doing lately?
hillzofvalp
100 kW
I'm going to be getting my thermal management for pack going soon for winter storage/charging (originslly installed a 10- sensor array of digital one wire sensor.. Each has its own identifier in code). I may make a cut-off for charging....over/under temp. Ttinking of mapping the peak temperature of differential to the curve of the thermistors suitsble for ca, then having a switch on dash to flip between motor and battery.
Also adding RFID for enabling bike and enabling charging.
For 2.5kWh A123 pack on motorcycle
Also adding RFID for enabling bike and enabling charging.
For 2.5kWh A123 pack on motorcycle
Alan B
100 GW
There are more Arduino based ebike projects out there now.
https://endless-sphere.com/forums/viewtopic.php?f=14&t=78428 Arduino BMS with Charge Shuffling
https://endless-sphere.com/forums/viewtopic.php?f=1&t=78682 Cell Welder
https://endless-sphere.com/forums/viewtopic.php?f=2&t=78626 Current Limiter
https://endless-sphere.com/forums/viewtopic.php?f=4&t=76168 hub based bike dyno
https://endless-sphere.com/forums/viewtopic.php?f=3&t=75339 ebike control
There are more, I just searched first posts for the last six months or so and skipped quite a few where they weren't ebike or they were undeveloped proposals. Feel free to add links here to other Arduino threads, this can be a resource to find other ebike Arduino projects.
I used to do Atmel AVR based projects before Arduino came along, but I find it is generally easier to start with the Arduino stuff these days and expand it as needed. I'm also moving away from the big boards to the small boards like the Teensy, Trinket or Nano. Lately I've had more time to re-start some of my stalled projects. One is a throttle manager that will convert a PWM controller to a torque throttle, and handle two motors for AWD. Another is a graphical display for the handlebars that I can customize. I did some mockups, here's one with lousy fonts but it gives some idea of where this might go.
This is the test setup, it will NOT be the actual display. This is a TFT that would not be good enough in sunlight:
Here is a quick hack on a display layout with a primitive graphics library. The final one will use a much better library and improved layout:
https://endless-sphere.com/forums/viewtopic.php?f=14&t=78428 Arduino BMS with Charge Shuffling
https://endless-sphere.com/forums/viewtopic.php?f=1&t=78682 Cell Welder
https://endless-sphere.com/forums/viewtopic.php?f=2&t=78626 Current Limiter
https://endless-sphere.com/forums/viewtopic.php?f=4&t=76168 hub based bike dyno
https://endless-sphere.com/forums/viewtopic.php?f=3&t=75339 ebike control
There are more, I just searched first posts for the last six months or so and skipped quite a few where they weren't ebike or they were undeveloped proposals. Feel free to add links here to other Arduino threads, this can be a resource to find other ebike Arduino projects.
I used to do Atmel AVR based projects before Arduino came along, but I find it is generally easier to start with the Arduino stuff these days and expand it as needed. I'm also moving away from the big boards to the small boards like the Teensy, Trinket or Nano. Lately I've had more time to re-start some of my stalled projects. One is a throttle manager that will convert a PWM controller to a torque throttle, and handle two motors for AWD. Another is a graphical display for the handlebars that I can customize. I did some mockups, here's one with lousy fonts but it gives some idea of where this might go.
This is the test setup, it will NOT be the actual display. This is a TFT that would not be good enough in sunlight:
Here is a quick hack on a display layout with a primitive graphics library. The final one will use a much better library and improved layout:
