CO sensor

RE: CO sensor

gelmi — Mon, 17 Jun 2013 07:44:37 GMT

I would still like to know the value of CO concentration, even if the error would be maybe 10%, so I will use opamp with ADC every 60 seconds (that is the max time response from the sensor). Nevertheless, the use of comparator is not a bad idea. I have to think about it. It would be a good idea to start the alarm when the concentration rises to a dangerous level between these 60 readings. I have to check how to implement it and on the other hand not to use a lot of energy. Thanks!

RE: CO sensor

gelmi — Sun, 16 Jun 2013 12:44:28 GMT

Thank you for your insight. You have already answered to one of the things that bothers you. Yes, of course this 10^6 amplification value is the combination of both opamp and first stage resistor circuit. Sorry that my description was not clear enough. A for the bias currents I also could not find their values. But still I think that EFM32 chip's internal opamp does not have as large bias current as it would make the readings so untrue.

I realize that good opamps, like AD620 with input bias which is max 1nA are perfect for this job, but EFM32 cannot be that bad [View:/resized-image/__size/16x16/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-ae44790a-ece3-4cef-97b4-dc4b9dc17b68/contentimage_5F00_1.png:16:16]. I have checked the sensor with both analog front-ends (EFM32 and external opamp with little bias current) in similar CO gas concentration environment (jar) and Vout readings were similar, not equal but with resonable difference, less then 10%. Figaro sensor itself in this configuration has +/-5% error and my application has to detect hazardous concentration, so in my opinion the threshold level is more important than the error (of course in resonable margin). Plus, in almost CO free environment (ouside), the Voutput is about 0,2 mV.

Please , let me know what do you think about my point of view.

RE: CO sensor

vsluiter — Sat, 15 Jun 2013 20:38:23 GMT

Hello Gelmi!

Nice post, good to look at the opamp feature of the EFM32, I haven't been able to explore that. However, I do have a slight warning for you; the open loop gain of the opamp is only 100dB, which is a factor 10^(100/20) = 100000. You won't be able to get 10^6 gain out of that opamp. When these numbers are true (I'm a bit worried about the claim in the datasheet that the numbers are derived from simulations) you won't be able to get the full range of amplification you had wanted. With the high gain you ask, the bandwith also reduces considerably. I've made a very simple opamp model below with a 120 dB open loop gain and a pole at 60Hz, and you see that it takes quite a while to get to the desired output voltage.

[View:/resized-image/__size/699x640/__key/commentfiles/f7d226abd59f475c9d224a79e3f0ec07-ae44790a-ece3-4cef-97b4-dc4b9dc17b68/contentimage_5F00_178648.png:699:640]

The other problem I see in this circuit and in the specs of the opamp is that the bias current of the opamp is not specified. Unfortunately, a lot of opamps with rail-to-rail inputs have an highly input-voltage dependant input bias current, which could mess up your measurement.

I guess however that it is still possible to read out this sensor, but either with an external opamp with shutdown, or by measuring the current (first schematic in you post), but offsetting both the counter electrode and the '+' input of the opamp with an offset voltage from an DA converter.That will leave less room for amplification, but maybe you could do some oversampling?

Interesting sensors!

EDIT:

On a second look, I noticed that you take the first 60dB gain by using the resistor as a current->voltage converter, and then use the opamp for another 60dB, which is very well possible with 100dB open loop gain. Which defeats my complete simulation above, but does stress the influence of the input bias current.

EDIT2:

I just realized that my calculation in the LTSpice screenshot is off; I had used 380rad/s for 60Hz, whereas it should be the timeconstant, so it had to be 1-over-380.....

RE: CO sensor

DAB — Sat, 15 Jun 2013 18:42:24 GMT

Good post.

I like your solution to the dual voltage problem.

You might also consider going to a transistor or resistive bridge approach, but it looks like you are ready for your next step.

Look forward to seeing your sensor in action.

DAB