temperature sensors

I want to install a slew of temperature sensors on my HVAC unit.
discharge (cold) air, return air, temperature of the freon between the compressor and the coil, the temperature of the air as it exits the evaporator coil (detect frozen coil), etc

my unit froze up once over the last heatwave and I am curious as to what the thing is doing.

accuracy is not as important as durability. my HVAC unit has an old compressor so it shakes.

I looked at the MAX6675 thermocouple that says it has a resolution of 0.25 degrees with a K-type thermocouple.

The [u]LM34/LM35[/u] is cheap, accurate, and easy to use. (Or, there are a couple of other semiconductor temperature sensors.)

Thermocouples can handle higher temperatures and the temperature-related voltage-difference between dissimilar metals is fixed, so in theory they can be very accurate (especially with a reference thermocouple in a real ice-bath).

But, the votages are low which means you need an amplifier. That makes them a pain to use, and amplifiers are imperfect and subject to noise and in the real-world they are usually not as accurate as the theory would suggest.

my HVAC unit has an old compressor so it shakes.

That shouldn’t be a problem with any temperature sensor as long as your connections don’t come loose. And of course, it’s best if you can mount the Arduino and other electronics to something solid that doesn’t vibrate, and protected from temperature extremes or high humidity.

Where necessary, mount your sensor directly to the surface being monitored and use RTV or Silicone-Seal around the sensor to hold it down, where your environment is vibrating.
The best way to catch the system freezing up is to use differential mass air-flow readings from both sides of the evaporator coil (the one inside).
Temp rises on the compressor and condenser coil is not necessarily an indicator of a problem, just that the system is working hard.

A humidity sensor and some check that the drain is working may also help.

Good luck.

any idea how long of a cord you can put on a DS18B20 ?

dave-in-nj:
any idea how long of a cord you can put on a DS18B20 ?

I just went thru a freeze-up on a 16 yesr old Bryant 410A heat-pump. I built this for the evaporator temp.

I can sit upstairs and watch the temp on my tablet.

But, you could use a 328 and implement bunches of temps using separate thermistors. Then spit it out on a LCD… or WiFi it like I do with a GPS sentence containing time-date.

PS: my problem was an “oil logged” coil… over time, oil would build up in the evaporator because the low side pressure could not cycle it due to low suction. (worn out compressor)

Ray

dave-in-nj:
any idea how long of a cord you can put on a DS18B20 ?

They come with 5m but I understand you can go further if you don't go cheap on cable and perhaps reduce the 4k7 pullup to 2k2. People use CAT5

I see the DS18B20 with 1m lengths. not see the 5m ones (yet) but that is great news.

I will be doing/copying Ray's project, but I want multiple temperatures.

supply/discharge air from the coil
coil surface temp, check for frozen coil
outside air
return air
then on the refrigerant lines:
high side temp
low side temp
temp after the expansion valve (before the evap coil)
maybe temp on both sides of the condenser coil to see how well it is working on a hot/humid day
humidity of outside air
RH of Return air

although I will be adding a ESP8266, this is too much analog in for that device.
then, since all it is doing is reading, I thought about watching power.
current on each of 3 legs
voltage on each of 3 legs

differential pressure across the air filters

I like that the temperature input can be serial, 1-Wire® so that means all those do not us any analog pins.
the RH is 1 wire serial, but not the same protocol as the temp, so that means 1 digital pin for each RH sensor.

that frees up all the Analog in for the electrical bits.
even allows for an I2C display

http://www.ebay.com.au/itm/2-15M-Waterproof-Digital-Thermal-Probe-Temperature-Sensor-DS18B20-Arduino-Sensor-/161951924940?var=&hash=item25b51642cc:m:mmEyqS6_jUkgzFiMG43czuA

I came here to suggest this :

https://www.aliexpress.com/item/Stainless-steel-package-Waterproof-DS18b20-temperature-probe-temperature-sensor-18B20-For-Arduino/32265871189.html?spm=2114.01010208.8.79.7RydcZ

but looks like I'm too late :smiley:

Nick_Pyner:
http://www.ebay.com.au/itm/2-15M-Waterproof-Digital-Thermal-Probe-Temperature-Sensor-DS18B20-Arduino-Sensor-/161951924940?var=&hash=item25b51642cc:m:mmEyqS6_jUkgzFiMG43czuA

I have 4 or 5 of these Chinese units… have found them “not all that great.” Personally, a 1% resistor and a quality NTC thermistor is preferable IMO. From this project:

 Schematic:
   x-----------------/\/\/\/\/\/\/\/---------------x---/\/\/\/\/\/\/\----xGND
   |                NTC 10K Thermistor             |  10K 1% Resistor
   |                                               |
   |              ATTINY85 / ARDUINO               |
   |                     +-\/-+                    |
   |    Ain0 (D 5) PB5  1|    |8  Vcc              |
   x--- Ain3 (D 3) PB3  2|    |7  PB2 (D 2) Ain1 --- 
(Blue)- Ain2 (D 4) PB4  3|    |6  PB1 (D 1) pwm1  (Green) ----------|<---x
   |               GND  4|    |5  PB0 (D 0) pwm0  (Red) ------------|<---x
   |                     +----+                                          |    100 Ohm
   x----------------------------------------------------------------|<---x---/\/\/\/\/---Vcc 3.3V

Depending on the required accuracy, the A_in from the voltage-divider can be utilized in a number of algorithms: Steinhart-Hart, table-lookup, or table + approximations.

Ray

darn... I was liking the idea of a 1-wire and the ESP8266 as a way of trying them out together.
I have purchased a few, not used one yet.
I would need to use a NANO to use the thermistor with 7 channels. cheaper than using an ADC.

dave-in-nj:
darn... I was liking the idea of a 1-wire and the ESP8266 as a way of trying them out together.
I have purchased a few, not used one yet.
I would need to use a NANO to use the thermistor with 7 channels. cheaper than using an ADC.

For your use, Dave, the one-wire device may work well.

I have been considering analog multiplexers (TI has a fair offering with reasonable pricing and on-resistance under 20 Ohm ... a higher quality has 3Ohm or less) for use with the ESP8266. Because the ESP8266 AD is 1Volt the series resistor is a high value, so long wires of a decent gauge znd less than 20 Ohms switch resistance should have little impact on Steinhart-Hart.

Ray

dave-in-nj:
darn... I was liking the idea of a 1-wire

As indeed do most other people. And since you have already bought them, using them sounds like a really good idea.

Until proven otherwise, there are three reasons for not using DS18B20s:

  1. You are operating outside their practical range of -25>125C, which is possible with furnaces etc.

  2. Your readings need to be faster than they can deliver, which is a lot less likely.

  3. It is essential that they operate wet, or you haven't heard about thermowells yet. DS 18B20s are available as a screw-in fitting for wet operation but the price is so eye-watering that nobody in their right mind could come at buying them, which I guess goes some way to explain why they are so hard to find.

It depends on what your real and practical requirements are, and what you describe doesn't sound that spectacular. I have a swag of these things. They operate with in conjunction with commercial heat meters, which have all the compliance marks from the Eurocracy, and upon which a lot of Europeans base their heating bills. I haven't seen anything to suggest that DS18B20 couldn't do the same job, and it's possible that Englemann actually use them in their devices.

It appears that the job you are doing is not very different from what I'm doing, and I don't think you have anything to worry about.

mrburnette:
I have 4 or 5 of these Chinese units... have found them "not all that great." Personally, a 1% resistor and a quality NTC thermistor is preferable IMO. From this project:

wanna that that to the next level ?
you can put in an LED in the water stream. red for hot, blue for cold and whatever in-between

Nick_Pyner:
As indeed do most other people. And since you have already bought them, using them sounds like a really good idea.

Until proven otherwise, there are three reasons for not using DS18B20s:

  1. You are operating outside their practical range of -25>125C, which is possible with furnaces etc.

  2. Your readings need to be faster than they can deliver, which is a lot less likely.

  3. It is essential that they operate wet, or you haven't heard about thermowells yet. DS 18B20s are available as a screw-in fitting for wet operation but the price is so eye-watering that nobody in their right mind could come at buying them, which I guess goes some way to explain why they are so hard to find.

It depends on what your real and practical requirements are, and what you describe doesn't sound that spectacular. I have a swag of these things. They operate with in conjunction with commercial heat meters, which have all the compliance marks from the Eurocracy, and upon which a lot of Europeans base their heating bills. I haven't seen anything to suggest that DS18B20 couldn't do the same job, and it's possible that Englemann actually use them in their devices.

It appears that the job you are doing is not very different from what I'm doing, and I don't think you have anything to worry about.

I know much more about thermowels that I need to, but exclusivly for K type.
for HVAC, the range of air temperature from the space is about 60F to about 85F, the discharge into the space is from about 45F to about 105F, maybe a little more.
for OA, it goes from -30F to 110F, so a huge swing there.

you are correct that speed is of no importance. once per minute scan time would make me happy.

as for wetting, that brings up another point. I can easily make my own thermowells for HVAC, but what about the contact between teh sensors and the case ? some sort of heat-transfer grease ?

Nick_Pyner:
It depends on what your real and practical requirements are,

in my case, I will align all my sensors to one and then use that point as the reference.

when cooling, one will show colder, one warmer.

Once the thermistors and DS18B20's arrive, I will test then for freezing and boiling, that is more than I need, but it will be nice to get that data.

dave-in-nj:
I know much more about thermowels that I need to, but exclusivly for K type.
for HVAC, the range of air temperature from the space is about 60F to about 85F, the discharge into the space is from about 45F to about 105F, maybe a little more.
for OA, it goes from -30F to 110F, so a huge swing there.

you are correct that speed is of no importance. once per minute scan time would make me happy.

as for wetting, that brings up another point. I can easily make my own thermowells for HVAC, but what about the contact between teh sensors and the case ? some sort of heat-transfer grease ?

-30F sounds seriously cold to me, and may be outside the scope of the DS18B20. A PT500 may be a better choice for that.

I could not find a DS18B20 thermowell off the shelf and made my own. These are simply turned up brass plugs in a copper T-fitting. I used used computer heat sink paste Arctic Cooler. For all that, I have not installed the last one here. See reply #3. The paste is also used here but I understand ordinary silicon goo is fine. Anything is better than air.

Nick_Pyner:
-30F sounds seriously cold to me, and may be outside the scope of the DS18B20. A PT500 may be a better choice for that.

I could not find a DS18B20 thermowell off the shelf and made my own. These are simply turned up brass plugs in a copper T-fitting. I used used computer heat sink paste Arctic Cooler. For all that, I have not installed the last one here. See reply #3. The paste is also used here but I understand ordinary silicon goo is fine. Anything is better than air.

whatever the coldest outside air temperature is. I thought there was some place in north Dakota that got to -30 for a month every year ? in 2011, Oklahoma recorded -31 .... I really do not care if I am off by 15 degrees at that extreme,
and, there is no reason that a speciality sensor cannot be used on that one point.

my expectations are a 'normal' winter of something above 0 F and a normal summer of less than 120F. maybe a bit more on a hot roof

I use compression fittings with 1/4" copper tube. make as long as one likes, but what to fill it with is the question.thermal conductivity , but not electrical.

dave-in-nj:
wanna that that to the next level ?
you can put in an LED in the water stream. red for hot, blue for cold and whatever in-between

It already is...
It's a tri-colored LED, so the color is mapped... easy to mod.

 // 261 = 32F, 447 = 64F, 537 = 75F, 575 = 82F
  b = map(ADCcount, 261,  447, 100, 255 );
  g = map(ADCcount, 435,  574, 250, 100);  // overlap green & blue
  r = map(ADCcount, 575, 1023, 250,  50);

Ray

mrburnette:
It already is...
It's a tri-colored LED, so the color is mapped... easy to mod.

 // 261 = 32F, 447 = 64F, 537 = 75F, 575 = 82F

b = map(ADCcount, 261,  447, 100, 255 );
 g = map(ADCcount, 435,  574, 250, 100);  // overlap green & blue
 r = map(ADCcount, 575, 1023, 250,  50);




Ray

now, how to get that LED into the water stream......