Making accurate low frequency measurements with your scope

[joeqsmith]

    Need to see if it really is the INL or something else.  If it is, I plan to model it with a polynomial. 

[joeqsmith]

Looks repeatable.  I ran 3 sweeps on each channel.   Shown is INL in counts for channel 0 in counts (again, after oversample).

[joeqsmith]

Notice how that graph looked very similar to the single sweep I had made the other day....

Showing INL for channel 1.   

[joeqsmith]

Imagine trying to fit that stupid INL curve...  Table lookup.   Green,  least squares.  White is raw (after 256X oversample). Red is compensated.     Some hope ....  almost.....

[joeqsmith]

Shown is INL before (green) and after (red) compensation.  Raw INL was filtered with IIR (0-phase) using Butterworth to create the lookup table.   Starts out really good, then all hell breaks loose.   Suspect better filter could remove some of it but really, there are bigger problems with it...

[joeqsmith]

And here is the money shot.  It would be very interesting to know more about the design of this DSO.   That's not the input signal doing that.  Why DNL goes in the crapper at a volt, no idea.  Good question for LeCroy designers.

So, in the end, there are gains to be had.  This will help out in my power measurement program but if you want good data, don't use a scope...

Hope it was helpful.