Problem with internal calibration in AD7124-8

Hi epdg ,

Can you confirm if you are calibrating on a single channel only? 

Please also note that the AD7124-8 is factory calibrated at a gain of 1, and the resulting gain coefficient is the default gain coefficient on the device. The device does not support further internal full-scale calibrations at a gain of 1.

Can you also please expound this statement?

epdg said:

And subsequently the measurement in continuous read mode does not make sense. Why does my Gain get cleared to 0?

The AD7124-8 installed on the evaluation board is a re-designed silicon.

There is a silicon revision on the AD7124. Changes on the device include the difference on ID register. The AD7124-8 standard part has an ID of 0x14 and the AD7124-8 B-grade and W-grade have an ID value of 0x16. The re-designed die for the standard part, B-grade or W-grade has an ID of 0x17. You may refer to this FAQ to determine the differences: (+) AD7124-4/8 Standard Silicon, B-grade, and W-grade - Documents - Precision ADCs - EngineerZone. 

When a product is re-designed or production flow (fabrication site, assembly site) is changed, a Product/Process Change Notice (PCN) is issued which documents the changes done to the product. PCN links are also available on the FAQ link above. 

Thanks and regards,

Rod

Hi epdg ,

Can you confirm if you are calibrating on a single channel only? 

Please also note that the AD7124-8 is factory calibrated at a gain of 1, and the resulting gain coefficient is the default gain coefficient on the device. The device does not support further internal full-scale calibrations at a gain of 1.

Can you also please expound this statement?

epdg said:

And subsequently the measurement in continuous read mode does not make sense. Why does my Gain get cleared to 0?

The AD7124-8 installed on the evaluation board is a re-designed silicon.

There is a silicon revision on the AD7124. Changes on the device include the difference on ID register. The AD7124-8 standard part has an ID of 0x14 and the AD7124-8 B-grade and W-grade have an ID value of 0x16. The re-designed die for the standard part, B-grade or W-grade has an ID of 0x17. You may refer to this FAQ to determine the differences: (+) AD7124-4/8 Standard Silicon, B-grade, and W-grade - Documents - Precision ADCs - EngineerZone. 

When a product is re-designed or production flow (fabrication site, assembly site) is changed, a Product/Process Change Notice (PCN) is issued which documents the changes done to the product. PCN links are also available on the FAQ link above. 

Thanks and regards,

Rod

Hi Rod,

Thanks a lot for your reply.

Regarding your questions:

1. I only activate one channel, yes (see below)
2. The measurements after calibration are 0x000000 for any measurement (GND, V_Ref, 1.6 V), which I guess makes sense with a gain of 0

See here my complete interaction with the board that leads to the behaviour:

     Write Register 0x00: ff ff ff ff ff ff ff ff // reset
     Read Register 0x45: 14
     Write Register 0x09: 80 01
     Write Register 0x19: 00 10
     Write Register 0x21: 01 01 80

// read registers to verify state
     Read Register 0x59: 00 10
     Read Register 0x61: 01 01 80
     Read Register 0x69: 80 00 00
     Read Register 0x71: 55 54 d1  // factory gain
     Read Register 0x49: 80 01
     Read Register 0x4a: 00 01
-----
     Write Register 0x09: 80 01  // activate Channel0
     Write Register 0x01: 05 50 // idle mode
     Write Register 0x29: 80 00 00  // set offset to 0x800000
     Write Register 0x01: 05 58 // internal fullscale claibration, followed by 500 ms delay
     Write Register 0x01: 05 54 // internal zero calibration, followed by 500 ms delay
// End of calibration, read results
     Read Register 0x69: 7f ff d9
     Read Register 0x71: 00 00 00  // gain 0 0 0  (?)
     Write Register 0x01: 05 c0
     Write Register 0x09: 80 01
     Write Register 0x01: 0d c0 // start continuous read mode

What am I doing wrong?

I made an even more minimal example which shows the same problem. This uses all default configurations and then starts an internal calibration which returns a gain of 0. Here are the registers I write and read:

Write Register 0x00: ff ff ff ff ff ff ff ff // reset

// read out default values after reset
Read Register 0x69: 80 00 00  //OFFSET_0_REG
GAIN_0_REG:     Read Register 0x71: 55 52 89  // GAIN_0_REG

Write Register 0x01: 05 58 // internal full-scale calibration
// waiting 500 ms
Write Register 0x01: 05 54 // internal zero-scale calibration
// waiting 500 ms
// read values after calibration:
Read Register 0x69: 7f ff c7 // OFFSET_0_REG
Read Register 0x71: 00 00 00  // GAIN_0_REG

I don't understand why this sets the gain to 0.

Hi epdg ,

You are trying to perform internal full scale calibrations at gain = 1 and the device does not support further internal full-scale calibrations at a gain of 1.

The channel being calibrated has a Gain = 1 and full-scale calibrations cannot be performed at a gain of 1. This is why your gain_0 register reads 0x0. Please ensure that the Gain > 1 when performing internal full-scale calibration.

Thanks and regards,

Rod

Hi Charles,

Thanks for the clarification!

Is there any benefit in

- doing a full-scale calibration at higher gains for measurements at gain 1 or are calibrations only valid for one gain?

- doing a zero-scale calibration without the full-scale calibration? (I remember the datasheet saying that a fullscale should be done before zero-scale)

Hi,

Apologies for missing the thread. 

epdg said:

- doing a full-scale calibration at higher gains for measurements at gain 1 or are calibrations only valid for one gain?

• Each time that the gain is changed, an internal FS calibration should be performed to calibrate the gain error due to the PGA.

epdg said:

- doing a zero-scale calibration without the full-scale calibration? (I remember the datasheet saying that a fullscale should be done before zero-scale)

• Performing a zero-scale calibration without doing a full-scale calibration will only remove zero-scale errors of the device.
• It is still recommended to perform both internal full-scale then zero-scale calibrations for better device accuracy. 

Thanks and regards,

Rod