[evlatests] 3bit sampler state counts

[Ken Sowinski]

On Tue, 31 Aug 2010, Barry Clark wrote:

> Attached is a plot of state counts as a function of input
> power for sampler 15A, in steps of one dB.  As expected, the
> distribution becomes wider as the power increases, but the
> pattern also shifts left.
>
> For a while, I was hopeful that that could be explained by a
> non-uniform step size of the digitizer, with steps falling
> closer together on the left than on the right.  But in fact
> this figure cannot be explained that way.  There appears to
> be a real shift to the left (that is, a DC offset) as power
> increases.

I have seen this behavior in several of the three bit samplers.
In the extreme as the sampler is overdriven by many dB, the
distribution is approximately uniform except for a large a large
excess in the most-negatice-sample bin.  We expect a large excess
in both the lowest and highest bins.