We see about 8kDN/V, so the rail-to-rail source-follower swing of 1.6V is worth 13kDN or ~450k electrons @ 35 e/DN . This would be an upper limit on well capacity, but some other factor will likely limit us to a smaller value.
The point of this measurement is to verify that we're operating in the linear range of the source follower, and that the changes we've made to remove the glow and odd/even effect have left us with a usable range. The changes to the operating point haven't changed anything much.
For reference, here is our present optimization of the array operating point:
3 comments:
450k electrons is much larger than I expected. Does it surprise anyone else?
I'm assuming the linearity measurements will come later. I'm thinking 1% linearizable should be our boundary (ie. once we have a linearization correction, we go to well depths where our linearization routine gets us to 1% of the actual well depth). Comments?
I'm pretty sure some other factor will limit us to a number much less than 450k electrons. Debias is the mostlikely, but there can also be issues with reset pulse feedthough.
We have data in the can to characterize linearity...they're next in line for analysis.
Thanks, Mark. Even with debias or something else limiting our well depth, does the 450K e- surprise you?
Looking forward to the linearity stuff. Do you think 1% is a reasonable level for defining our max well? If it ended up being 1e5 e-, the noise would be 0.3%, so the linearization error would be larger than our random noise. Just thinking out loud.
Post a Comment