Today, after a long wait, the PCBs and components for the revision finally arrived. I managed to find some time to populate the boards and now I am bringing you the results of the voltage accuracy test.
Now in the last project log I said this revision should have three voltage ranges - ±60 V, ±6 V and ±600 mV. To be honest, I kind of lied, because it has four voltage ranges, the three mentioned above and a ±60 mV one! I just wasn't sure how well will it perform, but according to my tests, it is totally usable. The only downside of this new AFE is that the lower ranges (especially the ±600 mV one) need to be calibrated with a simple offset. I'll try to remove the need for calibration in future revisions.
But enough talking, let's see the results:
The blue curve is the absolute error in volts, the orange curves are the maximal error margins for ±0,3 %, which is my target accuracy for this revision. Just for reference, ±0,3 % is the accuracy of the Brymen 805s, which is a 53 euro DMM! Now let's see a closeup of the sub one volt part:
I'd say this isn't bad at all.
And just to keep everybody informed - I hope that tomorrow I'll do tests with the new current stage and then I'll release the source. However, that will end the development of this revision and I will move onto 1.4, because this revision was intended mostly as a test for the new analog front end. Revision 1.4 will include this new AFE along with a much better processor - the SMT32F373, which offers a 16-bit delta-sigma ADC with differential inputs and external reference. This should vastly improve the accuracy, I am hoping for at least 0,1 %.