OPTA Analog Input voltage offset

Official Hardware Opta
1

Hi, I'm using the OPTA (wifi) main unit and EMR expansions to read analog voltage from a sensor outputing 1-5V. The reading is not accurate, I'm currently getting an offset of about 0.2V. For this specific issue I know it's not a problem to correct this offset but I'm worried that as I connect more sensors this offset would change. Can someone please explain why this happens and if it's possible that in the event I will be connecting more sensors to additional analog inputs, the offsets/scaling would change?

Thanks

2

Well it would take a long time to explain it, but this is not uncommon that an internal A/D converter has an offset. It is basically what you get when you try and get an analogue device to operate in a closely coupled digital environment.

It should stay constant and will not change when you read more sensors. Providing of course that the sensors have an impedance range in the 10K to 50K region.

3

Thanks for your response. My sensor's analog output have the following spec:

image
image1689Ɨ102 16.9 KB

Can this 1K impedance present a problem?

4

Yes

So you will need to compensate for both offset and gain errors. However, it has such poor accuracy and repeatability it may not matter too much

image
image1062Ɨ523 54.1 KB

5

Thanks - I'm fine with preforming a calibration curve that I can rely on at +-2% repeatability. Can inputs be affected by each other in this scenario?

6

I'm not an Opta expert and I can't say with 100% certainty that there is no interaction between ADC channels, although I would be surprised if there was.

Maybe an expert will see your post and respond.

7

Hi to @dotan_qd-sol, @Grumpy_Mike , @jim-p ! the opta analog input stage features a very basic design and therefore, the input impedance is relatively low.

Interaction between ADC inputs is an undesirable but possible effect named 'crosstalk'. fine tuning of the analog RC inputs networks and ADC sampling time can solve it.

Crosstalk effect is the result of residual charge (between conversions) at the ADC internal Sample&Hold capacitor. That means the SH cap does not reach the voltage value of your analog input.

imagen
imagen831Ɨ261 54.2 KB

However, there is a very simple solution involving software:

If you need to connect two analog sensors, double every connection. The second input reading should be better than the first one, as it is more near the real value of your analog input.

A1=sensor 1
A2=sensor 1 << more precise
A3=sensor 2
A4=sensor 2 << more precise

To probe the crosstalk effect, the nightmare scenario is:

A1=GND
A2=VREF (maximum ADC input voltage)
A3=GND
A4=VREF
...

performing succesive conversions, only crostalk-free desings can reach clean and repetible conversions covering the ADC full span.

On certains scenarios and voltage levels, crosstalk can lead to offset voltages and even signal overlap.

Hope this helps. If not, may be follower mode op-amps.

8

The OP is using the EMR expansion module for analog input.

9

No this is the standard technique, known about for decades on this forum.

A1=sensor 1
A1=sensor 1 // more precise

and even better will be :-

A1=sensor 1
delay(some value)
A1=sensor 1 // more precise
10

Yeah, good luck!

11

Bye bye.