DIY 2-Input Scanner for Met experiments

[enut11]

As often happens when playing the game of metrology, one thing leads to another...

I am working on a Hamon Divider project and this led to a Wheatstone Bridge to calibrate it.
Both the Hamon and Bridge circuits are constructed from low tempco precision resistors.
Both circuits also rely on a trimmer resistor to do the final calibration.
Now I am curious to see if the trimmers I have built have any effect on the temperature stability of the circuits.

This led to the need to heat and cool the trimmers while measuring both the resistance and temperature.
While I am sure there are many ways to do this, I took the simple, low cost approach and built my own 2-input scanner.

[enut11]

This scanner is based on a very low frequency oscillator driving a double pole relay.
The data inputs are connected to the NO and NC contacts.
Use your favourite data logging program.
The logging DMM, via the program, will output a composite file with 2 sets of data inter spaced by the time period of the scanner frequency.
XL is then used to separate and analyse the data.
The limitation of this method is that both data streams be the same function, eg both resistance or both voltage, etc.

[enut11]

The relay driver is based on an astable 555 timer circuit. The data sheet shows how to make it oscillate at a required frequency.
The 555 can drive a relay directly from pin 3, just needs a reverse diode across the relay coil to protect the output transistor.

[enut11]

This is what the data looks like when recording from my HP3456A DMM to an XL spreadsheet.
The time period on my scanner allows approx 2 data points from each input to be plotted in real time.
Column C on the left shows the captured data, mostly 2 at a time.
In the plot, the upper points are the temperature thermistor resistance and the lower points are the trimmer resistance.

[enut11]

On a longer time scale, this is what the captured data looks like.
The upper part of the green band shows a large variation in resistance and hence indicates the temperature change.
The lower part of the the green band, the DUT resistance, looks almost like a straight line. This has to be analysed in greater detail.

[enut11]

By expanding the scale to show the lower part of the green band in greater detail, what previously looked like a straight line now reveals the changes in trimmer resistance that I was looking for.
More analysis can be done by exporting the raw data into another XL sheet.

[enut11]

If your data capture program does not produce real time graphs, you can copy the data stream into an XL sheet, eg as in column B below. Using the XL '>' and '<' functions you can separate the data into two columns (D and E) and plot them as shown for further analysis.

The thermistor data is not very useful as-is and must be converted into temperature. I use downloaded thermistor tables. Over a small temperature range, I assume a linear relationship. Another way is to reverse engineer ohms vs temperature using a DMM. I do this with a pot connected to 2W Ohms, dial up the recorded resistance value and switch the DMM to Temperature to read the minimum and maximum results.
enut11

[enut11]

The Flexitimer kit from jaycar.com.au is perfect for this project as it allows programmable timing intervals for the relay.

[dietert1]

The 3456A does have a "voltmeter complete" logic output on its back. You could use that signal to step your scanner - with a flip-flop in the two channel case. Then the measurement data will be more regular and easier to process.

Regards, Dieter

[AlfBaz]

The thermistor data is not very useful as-is and must be converted into temperature. I use downloaded thermistor tables. Over a small temperature range, I assume a linear relationship.

Hi enut
I have been messing around with ntc thermistors recently with a view to build a multichannel temperature logger
I happen to have a couple of these hanging around from vishay
https://au.element14.com/vishay/ntcle100e3333jb0/thermistor-ntc-33k-5-radial/dp/1187032

I'm using labview to log data for initial tests to see what sort of resolution I can get out of these

Using the constants provided in the data sheet and the Steinhart equations also in the data sheet I seem to be getting good results
Because of the relative extreme resolution I am able to detect minute changes in temperature.

Accuracy, on the other hand is still an unknown, especially considering the formula given is more of a close fitting approximation rather than an actual measurement for the specific thermistor used.
For my purposes I am more after relative measurements over time rather than absolute accuracy

attached is the datasheet and an excel sheet with the constants for my particular thermistor along with the formula to calculate temperature in degrees celsius from the resistance

[enut11]

The 3456A does have a "voltmeter complete" logic output on its back. You could use that signal to step your scanner - with a flip-flop in the two channel case. Then the measurement data will be more regular and easier to process.

Regards, Dieter

Thanks @dieter1, an interesting concept to explore. Now I have to learn about flip-flops

[enut11]

The thermistor data is not very useful as-is and must be converted into temperature. I use downloaded thermistor tables. Over a small temperature range, I assume a linear relationship.

Hi enut
I have been messing around with ntc thermistors recently with a view to build a multichannel temperature logger
I happen to have a couple of these hanging around from vishay
https://au.element14.com/vishay/ntcle100e3333jb0/thermistor-ntc-33k-5-radial/dp/1187032

I'm using labview to log data for initial tests to see what sort of resolution I can get out of these

Using the constants provided in the data sheet and the Steinhart equations also in the data sheet I seem to be getting good results
Because of the relative extreme resolution I am able to detect minute changes in temperature.

Accuracy, on the other hand is still an unknown, especially considering the formula given is more of a close fitting approximation rather than an actual measurement for the specific thermistor used.
For my purposes I am more after relative measurements over time rather than absolute accuracy

attached is the datasheet and an excel sheet with the constants for my particular thermistor along with the formula to calculate temperature in degrees celsius from the resistance

Thanks Alf.  By plotting your data, I believe you entered your workshop around 7am?
The Steinhart method is good providing you have all the thermistor constants, which are missing for mine, that is why I reverse-engineer the DMM algorithm with a pot.
I only need a max and min value for this analysis and the range is about 5-10 degrees.
enut11