Notes on test plan for Resonon units delivered summer 2009 August 24, 2009

Notes on test plan for Resonon units delivered summer 2009

from email on 8/24/2009 10:45 AM

Report on Blue and Red Imaging Spectrometers for MOBY

Report on Blue and Red Imaging Spectrometers for MOBY

from email on 8/24/2009 10:45 AM

From: parr [parr@comcast.net] Sent: Thursday, September 03, 2009 11:37 PM To: Johnson, Carol Cc: Dennis Clark Subject: image shape with fiber

Carol,

Chirs indicated that the group at Boulder had not done any work in looking at image quality in multimode fibers which are being used as a light pipe.  The communication folks only care about the single mode issues in general and don't worry about problems like we have.  He suspects we will have to characterize the fibers ourselves if we need to know the quality of the image.  This may not be hard and in fact it can probably be deduced from the SIRCUS calibration data by looking at the image using the CCD information.  

What concerns me is that the image shape will change if the fiber is moved or bent in differing ways. I suppose the units are sealed and the fibers do not flex or otherwise move during operation but if there is a configuration change due to pressure or temperature, it could have an effect upon the image. I suggest that some time be spent looking at the image using an integrating sphere source illuminated with a laser. this would give a flat field for the input and we could examine the output of the fibers and see how the output may change with movement.  

I would hope the fibers are mechanically stable and don't change from calibration to deployment and use and hence they shouldn't introduce any significant error problems.

I'll keep an eye open for more information but I haven't found the topic covered using the web search engines.  I suspect there is no universal answer and the image shape will depend upon the details of an individual light pipe.

Best wishes,

Al

Carol Johnson

Send 11/18/2009 12:33 PM

Hello Mark

This configuration will be

The 14 fiber by XX m long Romak bundle (I’ll state the core diameter, but we could add a part number etc;

The blue in line spectrograph as delivered in 2009

The Pixis blue camera

Bare fiber mounted in front of the sircus sphere

Do you have part numbers or model/serial numbers you want in the record? What other details for the record? I can look on the camera myself but I think you’ve named the spectrograph

Thanks, Carol

They should have a serial number on one of the pigtails

It turns out these are 800um core/880um clad.1100um buffer with PVC jacket, Low OH.  (Not what we will use in the future - I think we decided on the High OH fiber in the past)

My Assembly number for the blue spec is: 8B09-100-000,  Blue CCD drawing 8B09-101-000.

The CCD part and serial numbers should be on the instruments.  I guess we use the 8B09-100-000 for the blue spec part number and use #1 for the serial number.

I can find no Resonon model numbers other than Blue and Red

(The Red systems drawing numbers: Red CCD 8R09-100-000 and Red Spec 8R09-101-000)

FYI - the "8" means the eighth radiometer system I have designed, R and B are blue and red,  09 is the birth year of the design.  The rest of the numbers are to keep the assembly and part drawings organized.

Mark

Resonon tasks #1

from email on 11/24/2009 1:06 AM

1. what distinguishes track 7 from track 8?
2. is the apparent baseline on the ADU data of about 600 the background and readout error uncorrected?
3. the apparent dynamic range of the ADU data is about 2 orders of magnitude (base to peak) and I am wondering how well we know the corrections to the data to generate corrected data that has an apparent dynamic range of 4 or 5 orders of magnitude?
4. Is there a readout  error correction made? If so, what is it?

Hey Steph, the pixels in this camera are 13 um square. The specs say there is no dead space. So what is the data telling us about the size of the image? If the spectrograph  imaging is 1:1, then the tracks should be 800 um wide or 61.538 rows. Is this about right from the data? Thanks, Carol

Maybe we will want to plot log in the aduAnd I understand the took saturated and unsaturated and we will have to normalize by the monitor photodiode but I don't know where these values are, ask Bob or Steve.  Once they are normalized we would plot the same wavelength/track together. Another thing we could do is do the along trackbinning
      

Al, maybe on Fri we can all have a telcon on this data. Mark is also a good one to ask. The CCD is so cold (-70C) that the darks don’t depend much on exposure time, they are running about 600 ADU. As for dynamic range, this is a 16 bit system just like the others so the full range will be had with varying exposure times and bin factors (here will be on chip sub pixel binning with different values for each track). What is a readout error correction? No, we have not corrected for anything – temperature, “shutter” delay, linearity, etc.  To be characterized.

Hey Steph, the pixels in this camera are 13 um square. The specs say there is no dead space. So what is the data telling us about the size of the image? If the spectrograph  imaging is 1:1, then the tracks should be 800 um wide or 61.538 rows. Is this about right from the data? Thanks, Carol

Hi Steve Sounds interesting. Yes there should be no problem having the instrument back on SIRCUS next week.  I will be doing a quick test today and then it should be free.  Did Mike Lin get the website information where Steph is posting the data? If not can you forward it to him.   I would like to see this artifact, can you and Mike join an GoTo meeting later this afternoon? Your input on how we are presenting and analyzing the data is welcome and appreciated. Steph remind me what time works best for you today? Carol

From: Brown, Steven W. Sent: Thursday, December 03, 2009 11:15 AM To: Johnson, Carol Cc: Saunders, Robert D. Subject: RE: monday telcon?

  Carol, For some reason the Octopus has been removed from SIRCUS. You may/probably have other needs for it at the moment, but .. Preliminary data, if the ambient is properly subtracted, appear to show a baseline scattered light level ~10(5). I would like to confirm this for all channels if possible. Also, the averaged data – when we took multiple scans at once – showed an artifact in the data Mike analyzed. Not sure if Stephanie observed the same artifact, but it is something I would like to show Bob and we might explore a bit further. For the above reasons, I think it is worthwhile to have the Octopus back on SIRCUS next week for a day or two. If the MOBY team concurs, let me know and I will coordinate with Bob. Cordially yours, Steve

I am thinking more about this apparent change in bias along the wavelength axis.  This shows up in the blowups of the dark data. If the signal is created by dark accumulation, it would be related to readout rate and show up in both X and Y axes of the full images.  It does not, it is only in the wavelength axis.  Also we have shown that dark current is next to zero.  So, I think dark current and temperature is not at issue.  It seems this is happening in the CCD serial readout register or the readout amplifier.  It is a constant (at least under these lab conditions) and should correct out fine with the dark subtraction of the light images but it would be nice to understand the underlying source of this signal artifact.  

Do we have any 2mhz background (internal shuttered) data yet to compare with the 100khz data?  If the bias slope changes with readout rate it is likely something going on in the CCD serial register.  If not, it has something to do with the readout amp and circuitry.  

Let me know when we have comparable background data at 100khz and 2mhz and I can present it to PI for their read on the issue.

We need to watch this issue when we get the environmental chamber data.  If this signal's origin is in the readout circuitry, its slope may be affected by ambient temperature and could be an issue with dark correcting long integration time exposures.

Mark

Mark

Hi Mark! I am losing track of all that is going on! Been tied up today on an IPO project, I will be getting funding to support IPO and VIIRS and the Aeronet Ocean color project that Bob Arnone etc are working on.  Mike Ondrusek is also slated for funding. I think this is good for NIST to be involved. Anyway. We just took some “wiggle test” data and yes the magnitude of the signal changes. This is a WAG test, as I can’t swear to the stability of the source.  We took rapid sequential w/o moving fiber to test this. Zhigang is trying to send the data to Steph but the web site is not cooperating. Steph, if you don’t get them call me at home and I will log in remotely and send file wise via email. The first thing to look at would be comparing the average for the track using the track definition you found earlier --- you will have to normalize to the average of the sub sets (hopefully goes by base file name) because the brightness changed etc. I did not mess with darks. Probably the ones with more counts are better to look at.  The second thing would be to compare the saddle shapes as the fiber is bent – maybe make a movie - I havent’ had time to follow the artifact discussion. Mark, We/I have questions on the tests you want done, so I will call you when I get home. Carol

Steve Brown

Hi Stephanie,Yeah, I want to re-do a couple of tests so we get the ambient level correct.I think that may happen next week.Anyway, if you have nothing to do on a rainy day, one thing I might suggest would be to see if you can create a stray light correction matrix for Track 7 using the current data set to see how things look? I think you can stitch unsat and sat data together at a tie point - say the 0.1 % unsat level. (Not totally sure I trust the linearity of the monitor yet - this is a kHz laser system and we are looking at the DC signal ...and you see some of that in your track 7 laser data).Then you might try to convince Bob/Carol/Zhigang to measure validation sources with track 7 alone. This lets you evaluate the stray light correction independent of cross-track coupling effects, which might be a good benchmark.If I understand the track to track stray light graphics, tracks not illuminated divided by in-band area is a negligible number, on the order of 10(-4) (ADU/sec), compared with in-band signals on the order of 10(5). It is not clear I understand the graphs completely, because that seems like an awfully small number. 1. you are also binning the other tracks and 2. can we look at normalized saturated data sets? Another way to look at this might be to take a white light single track spectrum and see how much light is measured by the neighboring tracks, in percentage of the illuminated track in-band signal.Nb: if I am butting in where I don't belong, just let me know and I will shut up.Steve

Mark and I discussed these tests on the phone last night. I think I captured the discussion correctly.  Yesterday Mark asked about darks at 100kHz vs 2Mhz camera readout.  I don’t see this in the initial task list, so I made a comment in there to this effect. Mark, give one of us a call so we can clarify. Thanks, Carol

Attached file: Resonon tasks_cjcomments.doc

I am interested in Resonon's take on the apparent etaloning effect (the interference induced intensity wiggles) we see at the short wavelengths.  I wonder if it is an interference pattern created in the CCD structures, the grating itself or due to thin structures in the PGP?  If it is in the CCD perhaps a different detector would eliminate the problem just as with the Red spec at long wavelengths.  These structures would be so small I think it is in the grating itself or perhaps a coating?

Mark

I am interested in Resonon's take on the apparent etaloning effect (the interference induced intensity wiggles) we see at the short wavelengths.  I wonder if it is an interference pattern created in the CCD structures, the grating itself or due to thin structures in the PGP?  If it is in the CCD perhaps a different detector would eliminate the problem just as with the Red spec at long wavelengths.  These structures would be so small I think it is in the grating itself or perhaps a coating?

Mark

Mark

This is the link showing the QE hit we would take at shorter wavelengths by changing to a deep depletion CCD in the blue spec.

The Green line is the CCD we have now and the Red line is the Deep depletion CCD

http://www.princetoninstruments.com/products/imcam/pixis/dsheet.aspx

Mark

I finally had a second to glance over things. I am pretty impressed, so much information is there with analysis and documentation, and the system's performance seems promising. The wiggle test is interesting -- except for the extreme bends the variability in the cross track shape appears to be not much worse than the repeats we measured on the 20th (see track to track straylight). The along track shape / level we may have a problem. But I think a more robust test would be useful, especially in defining what is an expected fiber "bend" - magnituded and frequency. The spectral and spatial stray light look good - aside from the 400 nm "artifact" there are no features I see, the behaviours are smooth, and the levels low - if I am looking at things correctly etc. What conclusions do other see? Steph, for the along track wiggle test shape, can you normalize so we can see in percent the variability? Throw out the outliers where there were extreme bends etc.

Steve, when it is back on SIRCUS is it possible to take some off ccd data to the red and possibly the blue side? There is an artifact from the white light data sets that looks similar to the artifact pointed out by Steve and Mike L, see http://data.moby.mlml.calstate.edu/moby2_testing/091120_Carol_Bob/offtrackslc_.html Carol