Author Topic: Mesuring tempco of Nichrome  (Read 2682 times)

0 Members and 1 Guest are viewing this topic.

741

• Regular Contributor
• Posts: 223
• Country:
Mesuring tempco of Nichrome
« on: September 03, 2020, 10:38:49 am »
A while back I got some MQ2 gas sensors. This post is mainly about the internal heater (not the sensor). I had imagined it would exhibit a marked tempco, and stabilise due to the increased resistance at elevated temperature. I decided to measure this, but I seem to see a really tiny tempco.

The data sheet says to run the heater coil at 4.9...5.1V, and it has a nominal resistance of 33R.

I found a PHd thesis saying an Sn02 sensor needs to operate at 200 - 800C. I will guess that 5V gives about 300C, and the coil (which is inside a tube of ceramic, over which the sensor coating is applied) must be somewhat above that, maybe 330C. Also, I found various tempco (alpha) figures for Nichrome, based on which I will assume a value 0.00015 (150ppm).

As per the attached XLS, I was unable to measure any tempco (I only used a bench PSU and a meter though). My calculations suggest I should see a change in R of about 1 Ohm.

UPDATE:
After 30mins... at about 5V, I do see the expected magnitude of change in resistance (1.8 R) but I see a decrease (which must be wrong)
At 0.22V supply, Rheater was 32.33. After leaving the unit at 5V for about 30min, I see 30.51R, a fall of 1.82 Ohms.

https://assets.omega.com/pdf/cable-and-wire/heating-wire/NI80.pdf
Nickel-Chromium Alloy: 80% Nickel / 20% Chromium temperature coefficient of resistance much less than that of Nickel-Chrome 60.

Factor by Which Resistance at Room Temperature Is to Be Multiplied to Obtain Resistance at Indicated Temperatures

Temp °C     20           93         204         315         427         538         649         760         871         982       1093  °C
Factor    1.000      1.016      1.037      1.054      1.066      1.070      1.064      1.062      1.066      1.072      1.078

« Last Edit: September 03, 2020, 01:31:20 pm by 741 »

SilverSolder

• Super Contributor
• Posts: 3411
• Country:
Re: Mesuring tempco of Nichrome
« Reply #1 on: September 03, 2020, 02:01:42 pm »

Are thermal EMFs messing with your tests?

741

• Regular Contributor
• Posts: 223
• Country:
Re: Mesuring tempco of Nichrome
« Reply #2 on: September 03, 2020, 03:21:18 pm »
From what I recall, thermal emfs are usually 'small'. But then, so is my signal. I'm predicting say 1.5R change in 33R, that is in series with the sense resistor (1R). So
5*1/(33+1) = 0.147 Volts
5*1/(34.5+1) = 0.141 Volts

This is a 6mV signal I'm predicting by moving from no power to full power.

There is the connection to the heater itself, inside the module. The main heat transfer would be conductive, along the heater wire to the internal connection with the sensor pins. Could this heat difference be generating emfs I'd notice, (of the order of say 1mV)?

There is a good teardown showing this sensor here - https://lastminuteengineers.com/mq2-gas-senser-arduino-tutorial/

( In the circuit (see xls), I have the 33R heater in series with 1R. I tried holding the 1R in case it's tempco mattered, but it never gets at all warm ).
« Last Edit: September 03, 2020, 03:28:40 pm by 741 »

SilverSolder

• Super Contributor
• Posts: 3411
• Country:
Re: Mesuring tempco of Nichrome
« Reply #3 on: September 03, 2020, 03:25:12 pm »

Maybe try measuring the voltage coming out of the thing when it is unpowered (after letting it heat up first)?

741

• Regular Contributor
• Posts: 223
• Country:
Re: Mesuring tempco of Nichrome
« Reply #4 on: September 03, 2020, 04:31:26 pm »
Because I'm looking at just the heater coil (I could equally have used any other 'filament' of Nichrome, but this is convenient), there is no output, other than Johnson noise.

There is a zero-power resistance, but that was rather 'off' (seemed to be 38R), presumably because however the meter measures Ohms is not the "same system" as I have to use when the coil is connected to a PSU.

So I have omitted a zero power direct resistance reading due to inconsistencies in measurement method. Rather, I am keeping to the formula shown on the XLS to derive heater resistance given the voltage across the 1R sense resistor.

Latest attempt is a bridge, with other side 68R to 0V then (500R preset with 2k) to Vcc. This means I can trim to zero mV differential output at (say) 0.5V. I am seeing possibly better results, but lots of wild rapid swings, maybe an RC would help.

graybeard

• Frequent Contributor
• Posts: 253
• Country:
• Senior III-V RF/mixed signal/device engineer
Re: Mesuring tempco of Nichrome
« Reply #5 on: September 03, 2020, 06:21:18 pm »
Are you doing a Kelvin (4-point) resistance measurement?

SilverSolder

• Super Contributor
• Posts: 3411
• Country:
Re: Mesuring tempco of Nichrome
« Reply #6 on: September 03, 2020, 11:37:24 pm »
Because I'm looking at just the heater coil (I could equally have used any other 'filament' of Nichrome, but this is convenient), there is no output, other than Johnson noise.
[...]

So you are seeing exactly zero volts with the system hot and powered off (no current in the heater)?

741

• Regular Contributor
• Posts: 223
• Country:
Re: Mesuring tempco of Nichrome
« Reply #7 on: September 04, 2020, 07:34:42 am »
Both good points.
• I had not thought of Kelvin connections, that might make a big difference.
• I've just now checked the situation for no power, both with the PSU off, and with the supply leads shorted. In both cases I have the same result, which is a reading (across the 1R sense resistor) of 0.0mV.
Just realised - I think that to use a Kelvin connection, I'd need to create a variable current source (0..160mA). Also, I then no longer have the tracking reference I have with the existing "Wheatstone" setup. At present, when I vary VSupply "REF" moves in sympathy. Both sides have about 33:1 divider ratio, the RHS is trimmed to match LHS for some low supply say 0.5V.
« Last Edit: September 04, 2020, 07:54:38 am by 741 »

741

• Regular Contributor
• Posts: 223
• Country:
Re: Mesuring tempco of Nichrome
« Reply #8 on: September 04, 2020, 08:44:58 am »
Added a x100 diffamp. Now I see significant changes in reading witrh temperature. Tried with a big 100W, 33R resistor in place of the heater - the measurement acroos the 1R is then "rock steady". Its resistance increases if I apply a hot blast from a hairdrier, and (despite its size), I can even see a small increase in resistance as I increase supply from 0.5 to 5V.

With the heater: I zeroed the diff-amp at supply = 2v5. Decreasing supply makes the resisance fall in a believable / sensible & repeatable way. However, when I increase the supply, something hard to explain occurs -

Things are sensible up to around 3v7 - that is, the heater resistance rises with supply voltage. After 3v7 though, there a fast (every second), huge changes in resistance, both up and down.

If there really is a coil inside the sensor tube, and it's fine enough, maybe some loops are expanding & shorting?

Of note is when using these generic MQ2 sensors (looking at the gas detector resistance, not the heater resistance) a very similar effect is seen. I have tried about 6 samples of sensor as gas detector, and they all jump around hugely just indoors in clean air - and even after a long "burn-in" time. That has meant that I needed very wide hysteresis to make a useful sensor.

I am now wondering whether the heater's unexplained behaviour above 3v7 gives rapid temperature swings, this would affect the sensor output similarly. I am also wondering if a branded sensor would behave in this manner.
« Last Edit: September 05, 2020, 03:55:31 pm by 741 »

741

• Regular Contributor
• Posts: 223
• Country:
Re: Mesuring tempco of Nichrome
« Reply #9 on: October 27, 2020, 05:01:26 pm »
I re-did this as per images. I used a simple "one op-amp diff amp" across (x, Ref) using 10k inputs, 100k feedbacks. Diff-amp circuit is floating (battery powered). Heater jig is powered from a bench PSU.
Heater itself now firmly fixed via some hard-sprung contacts (previously, the pin header/jumpers made reading erratic because they were in series with the heater and seemed to change by as much as 0.5R when touched). Heater/sensor module is insulted with grey foam.

I double-checked the results by paralleling a 1 k resistor across the heater wire at V=0.5 and the computed resistance reduced by almost precisely the expected amount (brown text).

Experimental procedure was to adjust RTrim at "V(heater circuit jig)=2.5" such that diff amp output is zero. Then measure test jig from 'x' to GND. This is the value "Ref at 2.5".
Now I can calculate what "Ref" is as V(jig) varies, and I know how much (x - Ref) changes via diff-amp.

The problem is I am seeing a negative tempco. At present I can't see the error source, have checked polarity of diff-amp etc.

Smf