Raspberry pi 3 --> adc accuracy (0.01)

[pnaven03]

Hi Team,

We are using raspberry pi 3 and kernel version Linux raspberrypi 4.4.11-v7+ #888 SMP Mon May 23 20:10:33 BST 2016 armv7l GNU/Linux . In board three calibration resistors (1000,850,680) are connected, first reading differential voltage (AIN0 -AIN1) second reading differential current (AIN1 - AIN2) using these two values calculating resistance.when am reading resistance
1000 output value is 10007 like some hardware error getting , so how to remove the error using calibration resistors.
please help on this.

Regards,
Nv

[jamesh]

Sorry, really not sure what you are doing.

There is no ADC on the Pi, so how are you measuring the resistance? What additional hardware do you have attached and how?

[pnaven03]

we are integrated adc (ads1247) with raspberry pi 3 via spi communication.
reading PT1000 sensor data.

[Gavinmc42]

Unless someone here has used these chips before then I suggest reading the datasheet for this section

Self and System Calibration

Section 9.4.5 Calibration?

24bit ADC powered from 3V3 digital bus?
Hope you paid good attention to the power supply.

Doing a running average helps filter the noise and gets a few more bits.
I usually start with an average of 16 readings. It is sort of how the chip works.

[pnaven03]

Thanks for the reply,

Power supply and all are fine , when am reading calibration resistor some error values are getting like
1000 -> 1007 ohms
820 -> 828 ohms
650 ->657 ohms
above values are expected some leakage in hardware side,using calibration resistors remove hardware error using application.
so any methods to remove hardware error using calibration resistors.

Regards,
Nv

[Gavinmc42]

That is what the calibration registers are for, you change them until the values match.
They all read high so start there and see how to get lower readings.

[pnaven03]

How can i change the values ?
calibration resistors are fixed hardware side,
any formula's to remove error ?

[jamesh]

How can i change the values ?
calibration resistors are fixed hardware side,
any formula's to remove error ?

As gavinmc says, there are registers in the device for calibrate it. You need to set those registers, probably by trial and error (a binary search would be quick), to do the calibration. Registers are set over the SPI bus.

[pnaven03]

Thanks for the reply,

I tried to set registers but same problem.

any idea using Newton Divided-Difference to sensor resistance to remove hardware error ?


Regards,
Nv

[texy]

I would concentrate on getting the calibration right, not adding additional circuitry. If it has 'the same problem' after you have set the registers them maybe it wasn't done correctly.
There's lots of hits via google on this, eg
http://e2e.ti.com/support/data_converte ... 337/903844
Texy

[pnaven03]

Thanks to all for support ,

This issue resolved by to apply newton divided difference method in application code and using fixed calibration resistors.
removed the hardware resistance.

Regards,
Nv