Shop OBEX P1 Docs P2 Docs Learn Events
Solved-- I just soldered together a temperature probe that was supposed to meas — Parallax Forums

Solved-- I just soldered together a temperature probe that was supposed to meas

photomankcphotomankc Posts: 943
edited 2009-08-17 16:57 in General Discussion
cry.gif·Ugh,

I am working on a circuit to measure voltage of a battery.· Following a idea I found online I constructed an R/C circuit to be measured by using one of the R/C time objects.· It uses 660K Ohms and a 100K POT that attach to the battery voltage and feed one side of the cap.· The other side of the cap is grounded.· The positive side then is fed to the prop through a 3.3K resistor.· Things seemed fine on the breadboard.· It was a little hard to hold calibration but I thought it just might be a noisy breadboard.·

Now that I have it all soldered on a board it has become apparent that the 'noise' was temperature.· Watching it over time it becam apparent that would slowly decrease the sensed time as it warmed up.· My horror came when I simply cupped my hand over it and watched it drop like a count-down timer.· Bad news, this will be enclosed and within a box where a heat-sinked regulator will be providing heavy 5v output.· It's going to get warm inside.· The really bad news is I need .1v resolution which it will do AT ONE TEMP, but if it's going to drift by .5v depending on the load on the regulator I'm sunk.· I was not expecting that kind of drift.· The parts are standard 5% resistors and some small ceramic caps.· If I keep air flowing over it then it stabilizes, but having a fan running is a non option.

Are there parts I could swap that would be better for this?· Is the design just doomed?· Should I go for bigger caps and smaller resistors?

|3.3K|
|660K|---|100K pot|
BATT V
····················· |
····················· |
····················----···
····················5 nf x 2
····················----
····················· |
····················· |
··················· GND

Post Edited (photomankc) : 8/17/2009 3:14:27 AM GMT

Comments

  • Greg LaPollaGreg LaPolla Posts: 324
    edited 2009-08-17 00:48
    You don't mention what kind of battery or how much voltage, but you might look into a high precision battery monitor like a ds2760 or similar.

    here is a search from digikey, you can select the parameters and narrow down to what you need. Most of these chips use 1 wire which is very simple to do with the propeller.


    Greg
  • photomankcphotomankc Posts: 943
    edited 2009-08-17 01:10
    6AA NiMH for a total between 8.4 and 5.8V or so.· In testing I'm just using a standard wall-wart going to an adjustable LM317 to feed the voltage to circuit and allow me to test.· I'm going to need to solve this using components if possible.· I don't have space for another chip and most all of them are surface mount so I can't use them on the proto-board anyway.

    Post Edited (photomankc) : 8/17/2009 1:16:17 AM GMT
  • photomankcphotomankc Posts: 943
    edited 2009-08-17 02:07
    Solved!· It was the ceramic caps.· I had a few of the little .01uf caps I purchased off this site, not sure of the type, but they matched well with the two .005uf caps I was using in the circuit.· When I tested·them with the meter and warmed them with my fingers they showed very tiny changes in value.· The ceramic type I was using prior would show an easy 10% shift.· Didn't even think to look at that before.· Something new added to my knowledge.· I have to re calibrate my numbers now but that's much easier when the lightbulb in the room isn't providing enough heat to affect the reading!!!· lol.gif
  • Beau SchwabeBeau Schwabe Posts: 6,568
    edited 2009-08-17 15:47
    photomankc,

    'Some' ceramic caps can even be used as a microphone pickup.

    Just as an FYI to explain one method... There are many times that temperature or other unwanted characteristics can be introduced to a circuit. One effective method for designing a circuit that is temperature compensated is to build it in duplicate and apply the returned values differentially to one another. This way ambient changes cancel out and your left with an increased signal to noise ratio of what you are trying to detect. In your case one of the dual circuits would have a 'fixed' voltage reference. You take the known reading of the fixed value and compare the amount of error due to temperature. Apply the same amount of 'error' in reverse to the second reading where you are looking for an unknown voltage. You may need to characterize the temperature curve, but once done, it should be consistent.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.
  • photomankcphotomankc Posts: 943
    edited 2009-08-17 16:57
    Good tip. Yes, it was very consistent. With my finger touching the cap it would drift the apparent voltage upwards by right at 0.6V before it leveled off. If I put it in the airstream of a little fan it would drop by about 0.3V every single time. With the poly cap installed warming it with my fingers for several minutes deflects the indicated voltage by about 0.01V and turning the fan on it did little to change the time reading. Very stable for my needs in this circuit. I'll have to keep the dual circuit idea in mind. I was actually wondering how on earth I could correct for the totally unknown temperature that would be generated by widely varied loads on the 5V regulator. I liked that the poly cap was a desolder-and-swap solution because I was out of board space!
Sign In or Register to comment.