ADC linearity

naseerak · Joined: 08 Nov 2008 Posts: 138

Hello all
I want to make a milli voltmeter to measure the voltage across a shunt, I think I will use vref of 1 volt by divider resistors(1%) but have some concerns about the linearity of ADC module in Atmega8

(BASCOM-AVR version : 2.0.7.6 )

i.dobson · Posted: Thu Jun 06, 2013 7:48 pm Post subject:

Hi,

Read the pdf specification for the chip. and use a opamp to amplify the Signal.

Regards
Ian Dobson

naseerak · Joined: 08 Nov 2008 Posts: 138

adding opamp will betray the accuracy actually this is an milliohm meter

AdrianJ · Joined: 16 Jan 2006 Posts: 2483 Location: Queensland

Maybe we should take this discussion to forum Various, as it has nothing to do with Bascom.

The linearity of any reasonable op-amp is at least several orders of magnitude better than the ADC in an AVR chip. And those ADCS are already within +/- 2 counts or so of full scale ( 4096 counts ) in 12 bit mode. What exactly do you mean by linearity ?

<edit>
Sorry, my mistake, the AVR ADCs are 10 bit, but still well linear in that range.
And very good for measuring differential inputs with settable gains on the input pairs. Makes millivolt measurements quite easy.

My favourite ADC if you want more accuracy is the Texas TCL2543 or its variations. 12 bit, 11 single ended inputs.
Use with something like LTC1013 opamps for differential low level inputs. Been making current shunt measurements like that for years.

Paulvk · Joined: 28 Jul 2006 Posts: 1257 Location: SYDNEY

Hello naseerak

I have tried to make a milli voltmeter to do this same job but the AVR ADCs are not good enough you need 12bits as a minimum 14bits to get good results. I tried all the methods oversampling ect but it did not work so I have been working with 12bit I2C ADCs you can find the code here for one of them http://www.mcselec.com/index2.php?option=com_forum&Itemid=59&page=viewtopic&t=11430
I have also 14bit and 16bit I2C ADCs that I am going to get working. With the meter I am building I will use the a ACS712ELC this is an IC that puts out a voltage in proportion to the amps flowing through it also it is isolated from the current it is measuring. In the tests so far with the device in the watering system blog I have had stable 5 digit readings from the 3volt back up battery for the real time clock. The other advantage of the device is its inbuilt high precision voltage reference.

Regards Paul

naseerak · Joined: 08 Nov 2008 Posts: 138

Hi
Current measurement is not a problem here. I want to make some DIY shunts from iron wires which have relatively increased resistance compared to copper and aluminum wires, though 75mv shunts are available here but are expensive for the products which have to be mass produced.

AdrianJ · Joined: 16 Jan 2006 Posts: 2483 Location: Queensland

Pure metal shunts are a bit of a problem. The temperature coefficient of most pure metals is far too high for accurate work. Thick film resistors using somethig like nichrome or magnanin are avaialable, and relatively cheap.
See something like:
http://au.element14.com/meggitt/mr-tl3ar047f/resistor-nw/dp/9973354
But of course it depends on your application.