[FDS] FDS noise - revisited

[Borries Demeler]

Hi Glen,

thanks for your ideas. See my comments below.

> This is a wacko idea.  If the linear slide were rough then the radial
> motion might also result in vertical motion, resulting in the focus
> position to move up and down.  This could cause the optical sample volume
> to up and down in the glass window, resulting in the oscillations seen.
> I would expect it to be greater at smaller radii due to the linear
> slide being further extended, which is contrary to what you show.
> If this motion is occurring then the oscillation should be seen on
> other samples in the same run, assuming their vertical positions are
> all exactly the same.  Is this true?

I don't know because that was the only sample in the experiment.
We need to rerun this with the same sample in all cells and channels to
most accurately reproduce this. Then I can answer this question.
Seems plausible.

> I'm assuming the oscillations reoccur with subsequent scans with the
> same pattern.  Is this true?  Are the oscillations truly a function of
> the radius and not time?

This is definitely a radius function. There are 8 hours worth of scans
averaged out in the fit. Rotor speed was 3000 rpm. Could it be a 
rotor speed dependence? We could run the same sample at several
speeds to see if this is reproducible.

> Another possibility is that the laser (or some other electrical component)
> is oscillating. 

Keep in mind that the same pattern essentially is observed in every scan, 
and it has a radial dependence. This noise is definitely time-invariant.

> If this is the case then the next scan in the series
> should should show the oscillation amplitude continuing from where it
> left off at from the previous scan (high).  If the initial oscillation

I didn't fit the entire scan, the bottom data is useless. So the oscillation
may keep going.

> amplitude is small like it was with the previous scan then the implication
> is something mechanical is occurring.  Which is the case?
> 
> What is the amplitude of the oscillation compared to the total intensity?

The amplitude of the entire scan was > 3500 counts. The amplitude
of the oscillation was only a few counts.

> It appears the frequency / waveform is unlike sedimentation, so the
> noise may be more of a visual than fitting impediment.

THat's correct, and I can pull it out easily with the TI noise fitting
in UltraScan, despite the fact that it is less than 1 % of the total
signal. However, it is remarkably reproducible over all scans.

Virgil: can you do the following experiment:

take some smaller DNA in TRIS/200 mM NaCl and label it with SYBRgreen
and put it in 3 cells/channels and measure it (50-100 scans each). Then
we can find out how reproducible the error is. Shoot for 3000-4000 counts
of signal.

I would be happy to analyze someone else's data if you could do the same
experiment to see if your machine has the same problem.

Thanks, -Borries