EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: SArepairman on April 08, 2014, 06:10:38 pm
-
Is it appropriate to test the linearity of a DACs LSB by putting say 5V ref into the DAC and then summing the output of the DAC with a negative DC signal to bring down the DC by a known amount so that a more high resolution range on the multimeter could be used for categorizing the step size?
At your meters 100mV full scale you get a much better idea of what is going on then on the 10V full scale, so I'd rather measures (4.9-4.85=0.05) 50mV volts on the 100mV then 4.9 volts on the 10V range.
I feel with a precision op-amp, a foil offset adjustment pot (or stable wire wound) and some good negative reference would make this method pretty useful and not noisy.
is this a method commonly used with source meters?
**could be a use for the precision adjustable resistance standards that dave jones compelled me to buy.
it seems like a difficult method for exact measurement but it should allow a precise categorization of step size, even without precision addition adjustment because the process is additive.
-
I would think a dynamic process would be better. Generate a linear count, highpass filter it and amplify the error. If it's perfectly smooth, each step should be visible with the same difference. The steps will contribute an overall high frequency spikiness, harmonics of the sample rate. Errors will show up as sidebands of the spikes, but mostly as lower frequency rumble. If the harmonics are filtered out, the RMS noise remaining should very much resemble the DNL (also RMS). Note that DNL is usually periodic with hiccups around certain code transitions; this will look like every Nth sample spike being taller than usual.
Tim
-
What do you mean by dynamic process?
Measuring the voltage with an op amp network/dac configured to track the input voltage so that the net voltage is never outside of bounds for the lowest multimeter range?
How could you set that up? ADC/MCU to adjust a DAC? I don't get what you mean by high pass filtering, I am thinking about DC measurements.
How about using two dacs to "comb out" error?
-
Analog Devices has published a lot of good material on this, e.g. http://www.analog.com/library/analogdialogue/archives/39-06/Chapter%205%20Testing%20Converters%20F.pdf (http://www.analog.com/library/analogdialogue/archives/39-06/Chapter%205%20Testing%20Converters%20F.pdf)
-
How much resolution are you dealing with? The simple way to test static DAC linearity is to sum the output with a better DAC or use an ADC to measure it directly. Alternatively I might use harmonic distortion measurement of a low frequency sine wave as a proxy for linearity.