AnsweredAssumed Answered

ADM4073  accuracy

Question asked by MarkS on Jun 23, 2015
Latest reply on Jul 9, 2015 by MarkS

Hi Analog Devices,

 

I request help understanding the error sums applicable to ADM4073.

 

We're using ADM4073F (gain = 50) with a rather small value shunt, 4 milliohm, and only light current, up to 3A.  This means that we're dealing with tiny input signals, so I should expect trouble, and indeed we're seeing large percentage variation unit to unit as I'm testing.

 

We're also powering the IC from 5V, but metering the current of a power rail up at +24V.  Given that the typical CMMR for the ADM4073 is given as 90dB, I expect a typical voltage error out of the device due to input common mode voltage of 0.03795V.  Scaling for the shunt of 4 milliohm, the metered current error resulting would typically be 0.18974A.

 

There is also in the datasheet a value of 1mV given for the typical value of "extrapolated input offset, Vos".  This sounds as though this should produce a direct metering error inversely related to the shunt resistance.  So in our case, as it would take 0.25A to produce 1mV across the shunt, our current metering error band due to input voltage offset is +/-0.25A.

 

If I'm right about input offset voltage effect, I don't understand the datasheet's Figure 10, which shows total output error for ADM4073F as less than 0.2%.  Figure 10 is drawn with Vsense = 6.25mV.  This input signal level should mean that input offset is typically one part in 6.25 of the signal level, or 16%, dwarfing the 0.2% accuracy performance shown.  This interpretation makes sense in the light of our scatter of results.  What have I misunderstood about Figure 10?

 

I notice that the datasheet Specifications table shows accuracy figures with the condition that Vcc and Vrs are the same.  Is this a good thing to do?  We could do this.  Under "Features" on page 1 of the datasheet it says "Wide common-mode range: 2 V to 28 V independent of supply voltage", however I can't see any subsequent information about the accuracy implications of exercising this independence as we are doing by powering from 5V and metering at 24V.

 

Any explanations of my (mis)understandings, confirmation of my CMMR sum, or other hints appreciated.

 

Thanks and best regards,

 

Mark

Outcomes