Remote temp measurement

Simply build a phantom powered sensor logging module which sits at the far end with the sensor on a four wire (excitation and sensing pairs) measurements system and communicating the readings back to you over the digital link that shares the pair of wires with the power supply (AC or DC).

How far is the "far end"? Generally, you don't want to have a "naked" analog sensor at the end of a long leash of unknown properties, or in a noisy environment. Instead, you'll need a complete thermometer, so you can send the data in some more robust, usually digital format.

If your line isn't too long (< couple of meters), a Dallas/Maxim

A thermomistor in a Whetstone bridge will do it. A thermistor is merely a resistor that changes its resitance quite dramatically as the temperature changes (All resistors do that, which is something you don't really want in an electronic circuit, but a thermistor does it on purpose and fairly predictable). They are quite cheap, Radio Shack sells 'em for a buck or so. They are rather small and very sensitive. In order to convert the resistance to a voltage you need a socalled Whetstone bridge. Google if you want the theory behind the bridge.

Greetz

Waldemar

"hamilton" schreef in bericht news:41741695 snipped-for-privacy@omega.dimensional.com...

measurement.

AD590 if 1000 foot is far enough. Search for AN273 on analog devices site.

measurement.

Hi Hamilton,

The low tech and cheap solution would be a PTC or NTC resistor (resistive temp sensor) as Waldemar suggested. However, you don't need a Wheatstone bridge. Calibrate that sensor in house before putting it up there, unless it comes calibrated. Also, briefly short the two wires at the far end and have someone read the resistance if it's hard to get there and you don't want to do two trips. Alternatively you could short at the other end and measure from the sensor location but that means dragging a meter up there. I have a nice pocket meter for such jobs.

You could measure the sensor resistance with a dual slope conversion. Or sacrifice a multi-meter that has that and a display built in. Some even come with a PC interface. Don't forget to lop off the cable resistance. Another good idea is to place a cap across the line where you connect your meter to muffle RF noise.

Regards, Joerg

Or use four wires. Two wires will carry the (constant) loop current through the sensor and the remaining two wires will connect to the sensor terminal at the sensor end and using high impedance voltage sensing on the other end. Since practically no current will be flowing in the voltage sense lines, there will be no voltage drop in these lines.

With a constant current source driving the loop will produce a voltage on the sense lines directly proportional to the resistance. Alternatively, you could use feedback to keep the voltage at the voltage sense line constant by altering the loop current and measuring it, so that the resistance is inversely proportional to the loop current.

With resistive sensors, watch out for sensor self heating due to the loop current, thus a constant dissipation system would be better.

Use twisted pairs for the loop and voltage sense lines, use ferrites around the wires close to the display unit and use UHF rated SMD bypass capacitors across the sense wires just inside the display unit to keep the RFI out (including any cellular phone signals), which otherwise could disrupt the measurements by self-rectification inside the display unit.

Paul

What temperature range? What accuracy and resolution? How far away is it? How often do you need to read it?

Without those details, I would suggest that a 0.5 degree accuracy, 0.1 degree resolution, solution would be a Dallas 18B20, an MSP430F1120, PIC or whatever, a TC55 3V3 regulator, and a transistor across the two wires. Supply at say 6V supplied through a resistor at the receiving end, use a comparator to sense the volatge across this resistor. Send slow NRZ, it should be happy over several hundred metres. One processor could deal with several sensors.

Paul Burke

All these "analog" answers from a bunch of talented embedded (digital) engineers!

Asan old analog layabout I'd go for

martin

Serious error. All shortcuts have disappeared. Screen. Mind. Both are blank.

Hi Martin,

May I be excused from this? I am an analog engineer...

Sometimes the analog path is lower in cost. In this case it might even be more reliable since as far as I understand Hamilton wants to place the probe quite remotely. That means things could be fried when a thunderstorm rolls through. A plain old resistive temp sensor at the far end and a cap at the near end would be pretty hard to beat here. You could even roll the cable through a toriod numerous times, ground it, whatever, to make it most reliable.

Regards, Joerg

Yes, the expected temperature range needs to be known. My knee-jerk answer to OP was Dallas 1-wire thermo too since I've used them with very good results for several years. "Solar collector" set off warning bells here because certain parts of solar collectors get nasty hot, hot enough to make toast out of a DS1820, for example. On the other hand, I've had excellent results "talking" to Dallas 1-wire thermos through twisted wire-wrap wire or telephone extension cord at distances ranging from less than 1 inch to about 17 meters.