Hello All,

I need to interface a thermosistor, that will be operating in the range of about 40 - 60khm:, to an A/D converter. I'm sure this has been done many times before but I'm having trouble finding the right differential amplifier to use. Some of the parameters I think I'm looking for are: low common-mode voltage, single supply (5v), low gain. Most of what I find accept high common-mode voltages. I'm not using this for sensing current through high voltage motors or something like that. So I don't know if that would pose any kind of problem. As you can tell I'm new to the analog world.

The only part that I think comes close and that I can find avaliable in small qty is the AD22050 from Analog Devices. It is expensive and I'm not sure that it is the right part for the job. Anyone have any suggestions?

Thanks, Dave

 

Are you using a micro-controller?, if so you can read the thermistor directly using a simple capacitor charging technique - no A2D or opamp needed (check my joystick PIC tutorial for details) - it works the same way as reading a joystick on a PC.

__________________
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk

 

Thanks to the link to your code, but I really don't need code. I can handle that. I want to convert to a single ended analog voltage which I will read with an A/D converter but I need to maintain as much resolution as possible, hence, the diff. amp. first. Make sense?

Dave

 

I think you missed Nigel's point.
A capacitor and a resistor will charge to a certain voltage in a certain amount of time, it's a very basic electronics concept. If you simply switch the thermistor and cap using a switching transistor for a very brief period you're going to get a fixed voltage that's related to the thermistors resistance. Exactly how much precision do you really need? Thermistors aren't precise enough by themselves for most applications. A thermistor is not linear so resolution is relative to the extremes you need that resolution in and the thermistors resistance grade.

__________________

 

Yes, BUT are you using a micro-controller?, the capacitor charging method can give better resolution than most A2D's!. Why waste an opamp and an A2D for an inferior result?. Assuming you want to do it that way though?, you might check my analogue tutorial, which overcomes the rail to rail problems, and uses a cheap opamp!

__________________
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk

 

Ok, I think I'm getting the idea now. I must admit, I just didn't think of doing it that way, but I do agree it seems like a much better method. Much simpler, easier (for me), cheaper, quicker, more accurate, sounds great! As far as linearity goes, I think I can account for that in the mcu. Thanks, guys, for the help. I'm glad I asked!

Dave

 

Keep in mind, the primary source of inacuracy in reading a thermistor is the REST of the circuit. An accurate MCU clock, the capacitor itself and it's thermal drift, all of that needs to be taken into acount if you want any real degree of acuracy.

__________________

 

Like I said, it's how the joystick ports on a PC work, but the idea is a great deal older than that - I can remember seeing it in the early micro-processor days.

__________________
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk

 

No, it doesn't make sense. An ADC does not need (or generally want) a differential voltage. It needs a low impedance (<2.5Kohm for most PICs) for an accurate reading.

Your thermistor with a pullup resistor may have a voltage of like 0.5v to 4.5v, at least for most of its range (consider how much range you actually will use, do you need to go down to -20F or whatever?).

So did you want to expand the 0.5v-to-4.5v range to 0v to 5v? That would require an instrumentation amplifier configuration. You can build one with op-amps but there are offset issues to consider. Worse still, op amps have limits as far as how close the voltage can be to the rails for input and output and it would be hard to get a benefit from this.

The prob with using the resistor-and-cap method is it's not just only as accurate as the oscillator, but only as accurate as the cap. Caps are typically very wide tolerance parts in both initial capacitance value and variation with temperature. There are specialty caps you can get with lower tolerances, but still it's hard to get a really high degree of absolute accuracy.

Actually start from the beginning. What are you trying to do? How accurately do you need to measure this temp? A temp sensor is often a better choice than a thermistor as long as you don't need to go very cold (below freezing). A DS18S20 is both accurate and wide-range, with a digital output.

__________________
I thought what I'd do was I'd pretend I was one of those deaf-mutes.

 

Oznog, I think you just went about 300 meters over didbirds head =)

__________________

 

"Leaps tall building in a single bound!"

__________________
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk

 

Well, he did mention accurate and as much resolution as he could get. Oznog did give a darn good answer.

FYI: I used an LDR with a cap on a PIC project to detect light levels, it worked great. Still in use for that matter (running on batteries at 32Khz and SLEEPing as much as it can.). Thanks to Nigel, and it was in assembler too

 

300 meters? No, not really. Without getting into all the details, I am limited to this one thermosistor in a water proof probe that I took from an indoor-outdoor thermometer. I'm using it because I need to measure the temperature of a liquid. I've used the Dallas 1-wire temp sensors and they're nice and simple for other things, but not for this. This thermosistor varies widely over its full range, but for my purposes, it will vary only from around 40k to 60khm: . I was going to use the diff amp to convert a voltage difference across the thermoristor, in series and between two other resistors, to a single ended analog voltage which would be read by an A/D converter. The method suggested by Nigel seems more sensible to me particularly because the design issues which have previously been mentioned are more within my area of expertise, or experience, shall we say. Also, this design will not have to be duplicated and mass produced; it is simply for my own use. So, no, he didn't necessarily go over my head, what he did was miss the point all together. But that's OK, I must assume he was only trying to help and just didn't have all the facts, as they were all not available in this thread.

Dave

 

The original poster has three posts to his credit before the start of this thread... They're obviously in school, and looking for actual help rather than the typical school work trolls of late. Nigel's suggestion was right on track, and my warning about 'the rest of the circuit' was probably right up there with the posters ability to comprehend the intracacies that can develop with analog and digital circuitry. People spend huge amounts of time on their doctorates going over the various details of circuit compensation methods and there are as many people that are sure they have the one true answer as there are one true answers. Any further discussion past that is flogging a dead horse.

__________________

 

This is why you're supposed to provide as much information in the first post as possible... These entire side conversation don't exist if you just say what you mean.

__________________