HELP: SET - SPAN/ZERO for Gas transmitter

[abicash]

Hello

I am designing a LPG transmitter to transmit a 4-20mA proportional output to sensed parameter.
I found the MQ6 to be a good candidate for this purpose.It measures a 300ppm to 10,000ppm of LPG change and converts to a 1.7V (normal condition) to 5V(max) of input ppm sensed.
For conversion to 4-20mA i am using a XTR111 from TI.

I saw few commercial transmitters and they have a procedure for calibrating the sensor out by setting zero and span.
Here is a section on one of the calibration procedures

1 Check  the  input voltage to the transmitter, which should be 24V DC.
2 Check  the  head  voltage of detector. Set it to 2V DC.
3 Insert the probes of the multi meter in to the Test pits (T.P). Check the milli volt reading. Set it
to 40mV (if required) with a tolerance of + / -3 mV using ‘ZERO’ trimpot.
4 Expose  the  detector  with  a  known  concentration  of  gas (flammable)  through  an  adaptor 
and tube at  constant  flow (i.e. 0.5 / 1  litre per  minute) and allow the reading to get stabilized.
5 With respect to the  concentration of the gas adjust the reading (if required), using SPAN  or
‘S’trimpot with the probes still in the Test pits.
6 Stop the gas flow, remove the adaptor and allow the detector to settle down to zero i.e 40 mV
with a tolerance of + / -3mV.
7 If  it does not settle down to zero, set it to ZERO  approx  40mV + / -3 mV using  ‘Z’ trimpot.
8 Repeat  the  above  procedure (step 4 to 7)  till  proper  calibration  is  achieved.

Can some learned member guide me on educating me how this can be applied to my proposed design?
Please help even if its thinking out loud..
Thanks and regards

 

[KeepItSimpleStupid]

I was involved in the setup, installation, calibration and maintenance of a Hydrogen alarm which used a Catalytic sensor.

A few things might be missing from your understanding. The datasheet says to use a 1000 ppm calibration gas. The calibration gas requires a flowmeter and regulator. The detectors I used had metal cones around the sensors. The middle of the cone there was a fixed tube where a calibration gas could be introduced. Think something like a 1/4" tube an inch or so away from the sensor.

Sensors in ducts required the same sort of thing. A way to shower the sensor with a gas with the blower off.

Unstable zeros meant it's likely to replace the detector.

% LEL (Lower Explosive Limit) is probably a better display than PPM. See https://en.wikipedia.org/wiki/Flammability_limit

 

[abicash]

Hi and thanks for replying
I have understood the procedure which would require a flowmeter,regulator and known concentration of gas.
The display unit can be calibrated to show %LEL or ppm since i would be using a microcontroller which reads a 4-20mA output for measured concentration of the gas.
My question was regarding the transmitter side, since it wont have intelligence.The calibration will be carried out with analog circuits.
The steps i have understood are as follows
1)The MQ6 has a usable range from 1.7vdc to 4.7vdc give or take.
2)The XTR111 needs a range from 1v-5vdc to convert to 4-20mA out.
3)Calibration would require to set a zero at cleanair (XTR111 can be adjusted to 1v at 0v input)
4)Say a 10,000 ppm of gas is thrown at the sensor, this value should be spanned to 5v.

Currently i am not sure whether the sensor is acting good.
The datasheet says that with a RL = 4.7k , the clean air reading should be 1.7vdc , but it keeps on falling up to 0.7vdc, from a reset value of 3v or so.Initial ramp down is quite fast, but when it goes below 1v, it slows down, say 100mv reduction per minute at the output.
I read somewhere that the sensor requires a 24 hour to 48 hour 'burn-in' period so that readings are stable, but i don't find that on the datasheet.
So is there a better but expensive naked sensor for LPG measurement out there?i couldn't find one .

Provided the sensors zero value settles, then i thought of adding a SET-ZERO & SET-SPAN analog front end to the XTR111 as shown below

Source : TI Application Note SLYT015

The gain in my case would be very small, close to 1.5 (not as shown in the app note)
R2B sets ZERO and RFB sets SPAN
In my OP i have mentioned of the existing procedure for one of the manufacturers.Is it on the same lines?
I thought i would require a rail-to-rail opamp and found the AD8541 and am building the schematic around it.

Am i on the right track?

 

[KeepItSimpleStupid]

I read somewhere that the sensor requires a 24 hour to 48 hour 'burn-in' period so that readings are stable, but i don't find that on the datasheet.

It's there in the datasheet. Do a control-F in your browser and search for "24".

You'll never calibrate at full span. In a system i created once that involved calibration, I used "Nominal" values and a "multiplier". So if the output was supposed to be 0-5, I used 5. then had a multiplier of say 1.012 In another system, I switched in a 5V reference to set Full scale. That full scale value was real close. real calibrations could later nail it.

2)The XTR111 needs a range from 1v-5vdc to convert to 4-20mA out.

5V = 20 mA will be IMPOSSIBLE to achieve with a 5V supply. Single supply OP amps have a LOT of nuances to deal with.

You may find that the sensor really is current out. I can't tell from the datasheet. So the load resistor can be at the measurement side and the 5V supply at the sensor.

The +-0.1V may mean that a simple 5V 3-terminal regulator won;t work.

The mounting of the sensor might be critical as well. e.g. if your using clip leads now. The leads will probably need to be a thermal sink.

It looks like for really accurate readings, you would want a temperature/humidity sensor at the sensor as well.

The sensors i used did not output 4 to 20 mA. I believe there was a heater supply and a low voltage output because the sensor was a bridge.

 

[abicash]

Hello

You'll never calibrate at full span. In a system i created once that involved calibration, I used "Nominal" values and a "multiplier". So if the output was supposed to be 0-5, I used 5. then had a multiplier of say 1.012 In another system, I switched in a 5V reference to set Full scale. That full scale value was real close. real calibrations could later nail it.

I guess you are right. The opamp is not a true rail to rail device.
The 5V reference will be created by the XTR111 , and it is fairly stable
Vreg output :
vs temp : 30ppm/°C
vs supply : 0.1mV/V

Effectively a 4.8v would be output at the Vin at XTR111? Is that what you are saying?
Which would mean that i would need to set a Vcc >5v at the opamp?

5V = 20 mA will be IMPOSSIBLE to achieve with a 5V supply. Single supply OP amps have a LOT of nuances to deal with.

I am supplying XTR111 with 24Vdc and the 1-5v is the input to Vin pin, which converts to an equivalent 4-20mA.
The Load resistor is of course at the receiver side.

 

[alec_t]

If you want a 4.8V/5V output from the opamp shown in post #3, even if it's nominally rail-to-rail, I'd suggest a +ve supply >5V. You have 24V available, so that (or a regulated lower voltage derived from it) would do.

 

[KeepItSimpleStupid]

Effectively a 4.8v would be output at the Vin at XTR111? Is that what you are saying?
Which would mean that i would need to set a Vcc >5v at the opamp?

]

Yep. Also look at the range of the inputs and outputs. You might look at the "Over the top" OP Ap series from Linear Technology too. They will not die if the inputs drastically exceed the supply voltage.

BTW: KISS works for the long name.

 

[KeepItSimpleStupid]

BTW, the last issue of Nuts and Volts magazine that I got has an article on a multiple gas sensor project and they will continue next month.

 

[abicash]

Which month?

 

[KeepItSimpleStupid]

Aug and Sept (Not published yet) 2015 See: https://www.nutsvolts.com/magazine/article/August2015_Kibalo