Analog val reading not what its 'meant' to be

Troubleshooting
21

Why again, do we need an op amp?
What is the thermistor’s resistance at the ranges you expect to use it?
I normally take the resistance about mid range, and use a resistor of that value in a voltage divider format.
Then I calibrate with a good thermometer. Use the map() command, and you come out pretty close.

22

Hi @Jack,

This particular thermometer (maverick et-732 for bbq grills) isn’t the finest piece of engineering. It easily hits 1.8M ohm, and the particle/spark analog inputs are only good up to roughly 40k. Bko’s suggestion was to use a voltage divider to remove the impedance problem. It eventually worked.

My findings (which are probably based on weird assumptions, but may make it easier for those that follow):

  • You can’t run the 3v3* to the opamp v+. It only outputs a fraction of that input voltage, wondering if its current/etc/things I don’t understand. My experience was it would output 2.06v max. Personally, I progressed some scaling maths (ie times the particle aval by 3.32/metered max) but ultimately the numbers were not quite right so I don’t know if this is a good path to chase.
  • You do need another power supply for v+ if you’re expecting analog values close to max (4095) which require the particle to see 3.32v.
  • Run the thermistor (variable resistor) from 3v3*. 3v3* to fixed resistor (i used 40k) to a common row of resistor, thermistor and opamp input(+)
  • I used a 0.01 cap, but bko advised it wasn’t necessary so have removed it.

Image is a thousand words:

23

I think the 3v3* is not designed to provide much current, just a reference voltage (and that is no longer available on the photon).
In any case, your voltage divider should consist of two resistances of near equal value. If your thermistor is in the 1.8 meg range, then the second resistor, should also be in that range, (not 40k).
Maybe I don’t understand the difference of a thermometer and a thermistor.

What temperature range are you wanting to check? BBQ pits can get pretty hot.

What do you think?

24

It’s a solid idea, but the problem is that the thermistor impedance is huge. 200k fixed resistor results in:

  • 1.17M Ohm thermistor value (bad)
  • analog val of 3500 (good)
  • formula output of 21Celsius (good)

Desired temp range is at least 50F - 500F (10C to 260C). I think these probes are rated to 600F.

Edit: overall I’m happy with the $2 opamp, it works!

1 Like
25

Great, if it is working. Good project.

26

Hi @mterrill and @Jack

Let’s see if I can clear up some of the questions here:

  • The op-amp is for impedance matching. The Core inputs are fairly low impedance (around 40k ohm at the normal sample rate) and high-resistance thermistor does not work with that low of an input impedance. The op-amp has essentially infinite input impedance with input currents measured in nanoamps.

  • The thermistor is roughly 1Mohm at room temp with a very high NTC such that it is around 25kohm at the “working” BBQ temperature. I am assuming you have the usual third-order fitted equation with R^3 somewhere either in the core or the spreadsheet.

  • The 3.3V* supply is a filtered version of 3.3V. There is a series ferrite bead and two capacitors. It is possible that the op-amp didn’t like the low supply voltage which is near its minimum.

  • Voltage dividers can be set anywhere between the two extremes, not just at 50%. You pick values that are convenient for what you are doing. With the op-amp providing high input impedance, you are free to choose convenient values less than say 10Mohm. With the “working” temperature resistance being the kilo-ohm range, that is a convenient choice.

  • There is some danger using a 5V op-amp supply in a 3.3V system if the thermistor is going to be removable (like with a plug). You want to make sure you can’t overdrive the analog input. A 3.6 zener diode might be a good idea but it looks like the sensor is on pull-up side anyway.

  • It might be safer to tie the (-) input of the unused second op-amp in the package to the second output, or tie both the inputs (+) and (-) to ground, but I don’t think it will oscillate anyway. And you do have a second channel for another probe if you need it.

2 Likes
27

Thanks for that detailed clarification, it helps sort out my rough guesses with solid knowledge.

Yes, I’m intending on using the second input set for a food probe. Eventually I’ll find a 4 way package when I locate some replacement probes, or well documented ones via alibaba, for a reasonable price.

1 Like
28

Thanks for the detailed information. I didn't realize the input had such a low impedance. That could be why some of my readings came up with unexpected values

One more question: is the low impedance only when using analog inputs? For a digital input, it is still high impedance?

29

Yes the digital inputs are normal high impedance CMOS inputs.

1 Like
30

EDIT: removed the post as I solved the problem with both analog inputs seeing roughly the same value even though only one thermistor was plugged in. It was being powered via usb. Changed to the 5v wall wart and the op amp was happy again.