Temperature Calibration needed or something else wrong?

[Hect0rG]

I have had incorrect temp settings for a bit and decided to take a look. I believe I have it all set up correctly but I find the calibration values need to be adjusted much more than I am expecting.

I am using a Sequent Home Automation V4 board which I know is not one of the suggested boards but so far its been working well for my needs ( except fotr the temps).

Here is my setup.

Notice my calibration value of -17.2 ( this seems like too much of an offset to be correct)

[tagyoureit]

How do you have the thermistor wired? The Home Automation v4 has an on-board 15kOhm resistor (manual - page 12). Our Temperature Sensor Wiki has instructions for adding a 10kOhm resistor for the dedicated ADC's but those don't have a resistor on-board already.

You could add a 30kOhm (check my math, I'm not an expert) resistor in parallel with the 15kOhm on-board resistor to effectively achieve a 10kOhm reading. See page 13 in the manual.

(Images also pasted here since the manual locations may change over the years)

[Hect0rG]

Oh bummer, that explains it. I don't have an extra resistor on there, I wish I would have noticed that earlier. I just thought these sequent boards would be plug and play.

I took a look at the calculation and you are correct. I used this website to make sure in case anyone bumps into this in the future.

https://www.allaboutcircuits.com/tools/parallel-resistance-calculator/

[Hect0rG]

@tagyoureit I had purchased the 30k 0805 resistors and when I received them and unmounted the board, I realized that the manual on the sequent board was wrong. It indicates on the manual that there is a slot available to add the resistor but it is incorrect. I reached out to sequent and this is what they said.

"Because of a lack of interest and to save board space, we removed the empty resistors footprints in the last version. The 15K pull-up resistors are 0402, hard to replace without the right tools"

"Unfortunately, V4 does not have the 0805 uninstalled resistors. The pull-ups are 15K, highlighted in the picture below.
You might consider reading your lower value thermistors with the board as is.
You will have a loss of resolution, but 12 bits resolution means better than 0.01%, and the precision is anyway limited to 1% by the components.
I am also assuming 1% precision for temperature measurement should be enough.

The alternate solution is to solder lower resistors on top of the ones shown in the picture."

Now these resistors are even smaller than the 0805, they are the 0402 smd resistors that are to me microscopic meaning that I will likely need special equipment to solder this and I would prefer not to try to do that. Is there a possibility I could modify something in the code to account for the different value of the resistance?

[marthoc]

@Hect0rG You will recall I was working on getting the Home Automation V4 board supported in REM last fall, but unfortunately the cooler fall temperatures in Ontario, Canada and the need to close my pool for the winter ground my efforts to a halt. Given that the reopening of my pool is coming up on my agenda, I have been back poking around at the REM code.

The REM code calculates temperature assuming that the voltage divider constant resistor is 10k. As you have pointed out, on this board the constant resistor is actually 15k. The question is, what difference does it make, and is it worth changing the code to accommodate?

I did some calculations to figure out the impact that using 10k rather than 15k has on the temperature calculation given that our board has a 15k ohm constant resistor. I am only a tinkerer, not a mathematician, but I think the actual impact is negligible. In the range of temperatures we are working within (say, 60F to 100F) the difference between the temperatures calculated with 10k and 15k is approximately 25 degrees - in other words, as a result of the inaccuracy the temperature using a 10k thermistor on this board reads about 25 degrees hotter than it should. I say "about" as I am using generic values for the calculations so depending on your actual thermistor's values, the magnitude of the difference will be different, but the important point is that across the range of temperatures it is virtually constant (differing by about 1.5 degrees across the range from 60 to 100).

The takeaway of all this is that you can adequately accommodate for the inaccuracy by doing what you have done - obtaining a reference temperature (using a regular thermometer), calculating the difference between that temperature and temperature reading in REM, and using the difference as the calibration offset.

As a result, I don't think it is worth the effort to amend the code, though in a perfect world, you could.

[Hect0rG]

@marthoc I did try that but the fluctuation in temperature was pretty off ( 22 degrees) which is why I opened this thread initially.

Since your last post, I tried something new. I had a few 15k 5% resistors here so I put two in series to create a 30K resistor. This will allow me to get the equivalent of the 10K resistor that REM is expecting when put in parallel. To test, I put one end of the resistor on the 3V3 pin of the raspberry pi header and the other end on the A1 input to the sequent card. This basically mimics putting the resistor in parallel. The problem is I am still seeing fluctuations. For example last night I calibrated the ambient temp sensor with the new resistors in place and both it and another temperature sensor (Govee H5072) were reading aprox 68 degrees Fahrenheit.

Here you can see how both temperatures differ although they come back together when it gets to about 68 degrees which was the temp when I calibrated them.

Now that I have the correct 10K resistor, I still think there is some issue with the calculations

[Hect0rG]

ok got a question that I can't find the answer to on. There seem to be two different places to calibrate. The first being from REM

and this other calibration from the pool controller itself. Could I just be adjusting in the wrong place?

What led me to this is the very last paragraph on the Temperature Sensors section below. It doesnt mention calibrating in REM so maybe I am calibrating the wrong thing? Ill try tonight but is there a difference?

[marthoc]

@Hect0rG just playing the Devil's Advocate here, but how do we know which temperature curve is correct once you get above 80 degrees? What is to say the Govee is correct and the 10k probe isn't?

[Hect0rG]

@marthoc you have a 100% valid point. I was thinking the same thing last night so I am going to try a second temperature sensor today when I get a chance. To make sure I will try to find an old school mercury thermometer :)

Was my method of calibrating incorrect and now correct when doing it from njspc?
This is how I have it now which does seem to make more sense. I just dont understand why two different places to calibrate.

[Hect0rG]

@marthoc hate to admit it but the govee was in fact not reading correctly. My pool ambient probe is just under the while box for my chlorinator. I put all three of the thermometers I had in hand and notice the govee is off by 3 degrees. The govee meat thermometer on the left had both probes measuring 77. The Acurite on the right was locked on to the meat thermometer as was the pool controller.

When I look at them while in front with the exception of the govee in the middle the temps seem to stay locked within .5 degrees which is acceptable to me. I do see some fluctuations during the day when looking through homeassistant but those I feel are due to the meat thermometer not having good connection via Bluetooth and only updating every 30 - 60 minutes. When inside it updates every minute or so.

I ordered a mercury thermometer so I can do that final test soon. I'll report back but I think this was my issue.

[Hect0rG]

This is what I see for the past 4 hours. I eliminated the original Govee Temp probe. These two are the meat thermometer probes and the pool ambient temp. Weird how they cross each other. I would expect them to be in sync.