CCD charge transfer

On 01/06/2016 05:00, wrote:
On Friday, May 27, 2016 at 1:15:25 PM UTC-4, Tom Roberts wrote:
On 5/20/16 5/20/16 - 10:17 PM, wrote:
If a charge packet in a CCD pixel is accelerated
out of a pixel (in some direction) due to charge
transfer, I wish to know what would be the median
frequency of photons emitted from the pixel array
(simultaneously) by this acceleration; and what
then would be the corresponding median charge
packet velocity?

Most of the electromagnetic noise would be in the RF band at around the
readout clock frequency and odd harmonics give or take a bit of jitter.
The CCD should not notice it at all.

In a CCD being read out, the radiation emitted is completely
negligible, because the individual electrons are moving so slowly
and their acceleration is correspondingly very very small. That's
the microscopic explanation; macroscopically the currents are very
small as are the distances involved, so the radiation is (very
small) squared.


Tom Roberts

I appreciate the clarification.

I'm trying to account (in theory) for a source of evanescent waves
(fringes) across the pixel array, with frequencies quantized as
nu_n = n(nu_o), where n is integer quantum mode n=(NdP_n/h),
N is array dim in unit pixels, d is pixel dim. P_n = (h/Nd)n is
momentum, h is Planck's constant, nu_o is detector fundamental
frequency nu_o = (c/d)/N and c is light velocity in vacuum.

I think you need to a post a picture somewhere to demonstrate the effect
that you are seeing, but my first guess would be either some kind of
Moire fringing due to inadequate oversampling of a broadband image or an
interaction between the regular structure of the CCD pixel layout and
monochromatic light.

What could "diffract" during charge xfer/readout?

IR photons from the readout amplifier is one possibility typically
leading to a warm corner with faint radiating lines away from it.

Modern CCDs are very much better in this respect than early ones.

--
Regards,
Martin Brown

[Moderator's note: Apologies for posting this so late; I had overlooked
it before leaving for holiday. -P.H.]

On Sunday, July 31, 2016 at 3:31:35 PM UTC-4, Martin Brown wrote:
On 01/06/2016 05:00, wrote:
On Friday, May 27, 2016 at 1:15:25 PM UTC-4, Tom Roberts wrote:
On 5/20/16 5/20/16 - 10:17 PM, wrote:
If a charge packet in a CCD pixel is accelerated
out of a pixel (in some direction) due to charge
transfer, I wish to know what would be the median
frequency of photons emitted from the pixel array
(simultaneously) by this acceleration; and what
then would be the corresponding median charge
packet velocity?

Most of the electromagnetic noise would be in the RF band at around the
readout clock frequency and odd harmonics give or take a bit of jitter.
The CCD should not notice it at all.

In a CCD being read out, the radiation emitted is completely
negligible, because the individual electrons are moving so slowly
and their acceleration is correspondingly very very small. That's
the microscopic explanation; macroscopically the currents are very
small as are the distances involved, so the radiation is (very
small) squared.


Tom Roberts

I appreciate the clarification.

I'm trying to account (in theory) for a source of evanescent waves
(fringes) across the pixel array, with frequencies quantized as
nu_n = n(nu_o), where n is integer quantum mode n=(NdP_n/h),
N is array dim in unit pixels, d is pixel dim. P_n = (h/Nd)n is
momentum, h is Planck's constant, nu_o is detector fundamental
frequency nu_o = (c/d)/N and c is light velocity in vacuum.

I think you need to a post a picture somewhere to demonstrate the effect
that you are seeing, but my first guess would be either some kind of
Moire fringing due to inadequate oversampling of a broadband image or an
interaction between the regular structure of the CCD pixel layout and
monochromatic light.

What could "diffract" during charge xfer/readout?

IR photons from the readout amplifier is one possibility typically
leading to a warm corner with faint radiating lines away from it.

Modern CCDs are very much better in this respect than early ones.

--
Regards,
Martin Brown

160802

Thanks for the notes. Let me digress on this: since the effect
is not visible in the output of the CCD, but rather a result of
computational processing of an image, I'm a bit skeptical the
effect is "real," and if it is, that it's of practical use. My
experiment may be the noise. That said, I think this might be
of general interest, and independently reproducible.

With the input image dim. N fixed, N=512, I've run image data from
several detectors with pixel dim. d ranging 6.8 - 21.0 microns.
Overall, the frequencies, plotted versus counts, yield resonances
at 1.25 THz +/- 0.25 THz. So the fringes are sub-mm.

The average visibility is 0.90. I could send you images via email,
if you wish.

My best guess at this point (underscore guess), is that I'm "seeing"
plasmon modes, non-propagating typically; but I find that changes to
the primary source (point or extended), DO appear to couple, evident
in certain systematic changes to the statistics of observed phase
singularities / topological charges (orbital angular momentum) visible
as characterisitic forks in the fringes.

It's beyond me to know how conceptually misguided this might be.

Best,
mark jonathan horn