Group Actions
Engagement
  • Author Author: fmilburn
  • Date Created: Date Created
  • Views 5722 views
  • Likes 10 likes
  • Comments 14 comments
Related
Recommended

Testing the Microchip MCP3421 18-bit ADC

fmilburn

I have what seems to be a continuously ongoing project to measure low resistances with milliohm precision.  The latest version had an inexpensive voltmeter that used the Microchip MCP3421MCP3421 in the display.  The chip costs $2.39 in small quantities from Newark and several were ordered with the idea of installing it directly on the milliohm meter PCB.  One was soldered to a SOT-23-6 DIP adapter for testing with an Arduino MKR1000.

image

The MCP3421 is an 18-bit delta-sigma ADC in a SOT-23-6 package with differential inputs.  It self calibrates internally for offset and gain on each conversion.  The internal voltage reference has a 2.048V accuracy of +/- 0.05% with drift of 15 ppm / degree C.  The input range is -2.048V to +2.048V.  It uses an I2C serial interface for communication with a microcontroller and operates with a supply voltage of 2.7V to 5.5V.  Data rate options include 12-bit (240 samples / second), 14-bit (60 samples / second), 16-bit (15 samples / second), and 18-bit (3.75 samples / second).  Here I'll be connecting it to an Arduino MKR1000 and using the 3.3V supply from the microcontroller.  Sampling will be done at 18-bits.

 

Test Setup

 

An easy to use library named MCP342X for Arduino was found and modified slightly.  A bench power supply was gradually increased from 0 to 2.05V and the measurements from the MCP3421 were recorded along with measurements from a Tenma 72-102072-1020 bench multimeter.  The test equipment was allowed to warm up for 20 minutes and the ambient temperature during the test was 21 degrees C.

image

 

Results

 

The results are tabulated below.

image

 

As shown in plot below, the difference between the MCP3421 and the bench multimeter never exceeded 1.1 mV.

image

The percent of full scale difference did not exceed 0.055 %.

image

The test results as compared to the datasheet are within the accuracy of the instrument that I am using for comparison.

 

Summary

 

I really like this chip.  The SOT-23-6 is a bit of a pain for me to hand solder but doable.  It is easy to use and has the resolution and accuracy that I need for my particular application. 

 

 

 

 

 

 

  • in reply to fmilburn

    If you figure out what the normalise() method does I'd be interested to know as well.

  • in reply to Andrew J

    Thanks for posting this though - I plan to write my own code for a non-Arduino but haven't gotten too deep into it yet.

  • in reply to jc2048

    Dammit.  I missed that - thanks Jon.

  • in reply to Andrew J

    Section 5.4 (page 24) of the datasheet explains how the General Call aspect works.

  • Thanks Frank - actually, I wanted to pick your brains if you were using the same one - which you are!  I was trying to understand how these three methods worked:

     

    uint8_t MCP342x::generalCallReset(void)

    {

    Wire.beginTransmission(0x00);

    Wire.write(0x06);

    return Wire.endTransmission();

    }


    uint8_t MCP342x::generalCallLatch(void)

    {

    Wire.beginTransmission(0x00);

    Wire.write(0x04);

    return Wire.endTransmission();

    }


    uint8_t MCP342x::generalCallConversion(void)

    {

    Wire.beginTransmission(0x00);

    Wire.write(0x08);

    return Wire.endTransmission();

    }

     

    Transmitting to address 0 is a broadcast: i.e. to all devices on the bus.  I would assume this code is written on the basis that there is nothing but one or more MCP342X devices on the bus.  In each case it follows with a byte which sets the configuration register for the device:

    Page 18 of the datasheet

     

    image

    So, taking generalCallReset() as an example, it writes 0x06 or 00000110 so sets S0 and G1 (60SPS/14-bits and x2 Gain).  S1/S0 are the Sample Rate Selection bits and G1 and G0 are the Gain Selection bits. I'm not sure that makes sense unless the 'default' configuration is meant to be that - all the examples call this method at the start.  That doesn't really make sense and neither do the bytes used in the other two methods. 

    generalCallLatch() sends byte 00000100 (Sets S0 so 60SPS)

    generalCallConversion() sends byte 00001000 (Sets S1 so 15SPS)

     

    There's no documentation/comments for these methods so I was wondering what I was misunderstanding.   I have no need or intention of broadcasting but I'd still like to understand these methods.

     

    Any ideas?