I need some help choosing a temp sensor for my project. Note that I am new to Arduino (and to electronics in general). I have looked at MAX31855/6675, DS18B20, LM35, TMP36, LM34. A quick summary:
Arduino UNO R3
Project is a peltier powered home bar cooler (using peltier as space is too small for compressor)
I will be using 4XTEC1-12715 peltier units, wired in series and driven with a constant current power supply (maybe HGV-480H54B) and controlled via PWM from the Arduino
I will need 5 sensors: one for the fridge to turn on the cooling, and four for the hot side heat sinks to turn off the system if it goes over temp. I think this rules out the MAX31855 as they each need 3 digital inputs.
Fans for heatsinks will be controlled through Arduino as well (one digital out for four fans), and will turn off after peltiers (not sure if set time after or after temp drops enough). Will also have another fan on the cold side controlled by Arduino (one digital out).
temp range for the sensors should be 0-80C (I think my Tmax will be ~70C)
The actual probes need to be thin wire that I can attach to the heatsinks via thermal cement - I think this also rules out the DS18B20 as all the probes I can find have large metal ends
I think one of my issues is trying to understand what probe type I can use for each sensor - the datasheets don't seem that clear (or I'm just thick). I know the MAX series uses K-type thermocouples but I'm not sure of the rest. Any advice would be appreciated.
The issue with the DS18B20 is that all the probes I could find were all large: I need a thin wire that I can cement to the heatsink - like a K thermocouple or thersistor.
Original has the DS18B20 TO92 package. Pro, you have digital output that need only 1 IO from the UNO. There are screw types too. All the same sensor in another enclosure.
Will you have it tiny? Use NTC or PTC, but that needs analog input for every sensor. The temperature curve isn't linear. The smallest approx the great of a rice grain.
Why PWM? Why not on/off whit a little hysteresis? I think you have more joy with PWM on the fan side.
The link for the screw type DS18B20 is super useful and may work. I also realized that your suggestions for a thermistor may also work - it is OK if the curve is not linear since I will be setting an upper/ lower limit only so one point calibration per sensor should work.
For PWM it is the standard method of controlling that power supply (there is a pwm input) to limit the current - note that the output of the power supply is not pwm but has been smoothed (at least that is what I see from their datasheet). I need to use this as I can't find a power supply with the exact voltage/ current requirements, so this will allow me to dial down the current to what is sustainable by the peltiers.
Yes, I did actually see that, so I'm hoping that my assumption about the power supply smoothing the signal is correct. See below for datasheet (my mistake - I forgot the hyphen in the name):
You should consider NTC resistors. (NTC = Negative Temperature Coefficient)
They are small and very useful over the range you need. Note each will require an analog input pin.
HLG-480H-54-B is 54V 8,9A. The TEC1-12715 is max 15A. The PWM voltage 10V. Isn't the HLG-600H-54 a better choice. Or 2 HLG-480H-30 with each 2 peltier parallel. 16A @ 94% = 15.04A.
So my understanding of these peltier chips is that they are much more efficient at lower powers (see datasheet) so my plan was to operate in series (since they like constant current) at 6-7 amps, which should be ~10V/chipX4=40V (note voltage will increase as deltaT increases).
Sounds like a pretty general statement to me.
In any case, LM 35 was one of the three mentioned.
You can't condemn analogue sensors just because you can't find one out of a very limited selection.
All sensors are fundamentally of analog types; where, the outputs are analog voltages which when digitized by internal ADCs are called Digital Sensors; else, they are analog sensors.
It is the question of price vs precision in the context of necessity. LM35 costs 60 Taka with a precision of 1-digit with a sensitivity of 0.1 oC; whereas, DS18B20 costs 150 Taka with a precision of 4-digit with a sensitivity of 0.0625 oC. If LM35 is good enough for a typical application, then why to think about DS18B20?
So I have decided to go with thermsistors, largely due to the small sensor size that I can easily cement to my heatsinks.
One question: I have 6 sensors, so I fill up all the analog pins, but was planning on using 2X I2C LCD displays. I am not clear from the pinout if I can use all analog pins (A0-A5) and still use the two pins above the AREF for I2C since they are labeled as D19 and D18 in the pinout and listed as "duplicate" on the data sheet.