Photometry with an SXVR H18

[Tony Rodda]

(I've also posted this thread in "Cameras" but thought anyone involved in Photometry might have a view too).

I have an option of using an SXVR H18 for photometry (I currently use an H9) but need a little advice on the SX statement 

"Anti-blooming with minimal effect on linearity".

Would I have to limit the maximum pixel count to avoid an (upper range) non-linearity?

And, if so, what would that limit be?

I'd really like the wider FOV of the H18 but can't risk using a camera not suited to photometry.

Any comments/views welcome.

Regards

Tony

[Oddsocks]

Hi Tony,

The question you are asking is not that straight forward to answer as there are too may variables for the CCD manufacturer and camera implementer to specify where the ABG gate effect begins.

Your current H9 also has a ABG and is no different from the SXVR-H18 in that respect so if you are happy with the H9 there is no reason why the bigger H18 will not work for you just as well.

The SONY ICX814ALG CCD chip, used in the SXVR H18  outputs an analogue voltage, proportional to pixel well electron count, and then the subsequent analogue to digital converter circuitry added by the camera designer (SX) produces the digital equivalent for the analogue signal. The actual ADU count will depend on how the A to D amplifier/converter has been adjusted by the camera manufacturer during production and this may vary depending on operating time, handling, temperature and oxidation of external trimmer potentiometers used for the bias stage in the A to D amplifier/converter.

For these reasons there is no specification sheet you can refer to that will give you an upper ADU value where the anti-blooming gate begins to move the analogue to digital conversion into the non-linear phase, you have to determine this for your particular camera by producing your own calibration chart.

You also need to take into account that the ICX814ALG has a large difference in spectral sensitivity, there is a 25% reduction in sensitivity in the blue region compared to red and green therefore when evaluating your observational data you need to calibrate by comparison to known spectral types similar to the star under observation in order to calculate the absolute photometric value.

To calibrate your camera and find where the linear portion of the output signal ends (assuming the camera manufacturer has set the A to D to respond to the full well signal and not clipped it early) set up a halogen light source at around two mtr distance from the camera in a darkened room, place a diffuser of opaque or frosted glass over the camera aperture, the camera should not be on the telescope, just standing alone pointing at the light source, then begin a series of light frame and dark frame exposures starting at the shortest exposure time that the camera acquisition software will allow and then increase in linear exposure time steps, say 10ms longer each time, until saturation is reached.

Set the desired camera cooling temperature and take each image, subtract each corresponding dark frame and measure the actual ADU at at least three points , I use the measuring tool in P.I. but you can use whatever image processing application you have, average the three measurements and plot them either in a manual hand-drawn graph of exposure time against average ADU or easier, enter them in a EXCEL spread sheet and use the graph plot function in EXCEL to generate the output graph.

From the graph you will then either see a flat line up to max saturation, this is what I used to see with my old SX H9 since the A to D was clipped early before the anti-blooming gate effect was reached, or a gradual curve at the fall-off point as you approach the anti-bloom gate effect, from your chart it is then easy to determine what the maximum ADU you can use before the non-linear portion of the camera output is reached.

It is a lot of work to produce the calibration graph and it really needs to be done regularly if seeking to produce an accurate data set from long term observations, unfortunately there is no easy fix or work around, at least it can be done indoors when the weather is too bad to be out gathering real data to work on.

I still use a 12V halogen lamp with a 240V-12V step down transformer from a kitchen cupboard lighting kit for my calibration set-up but I see you can now buy direct mains voltage halogen lamps in BC/SES/SBC format from most DIY stores so no need to faff about with wiring and transformers, just pop one of the direct mains bulbs into a table lamp and your good to go.

Once you have the upper ADU linear limit determined for your particular camera then you can acquire your data set ensuring the target object does not exceed that ADU value, I try to achieve a maximum value of around 80% of the upper limit though for me it is easier because around four years ago I bought a QSI 532 camera with the KAF 3200ME chip, which is the non anti-blooming gate model, and is linear right up the maximum saturation ADU, so I just adjust the exposure time to not exceed the 80% max value, even though I still have to calibrate on a similar stellar type to the observed object because like the H9 and H18 the spectral response is not linear across full visible spectrum.

I do still check my camera every six months though, even though there is no ABG on the QSI it still has internal bias adjusters and power supplies that could drift and produce a non-linear response.

William.

[Tony Rodda]

William, 

Many thanks for such a great response.  It's crystallised my thoughts immensely.

I had started to capture a couple of open clusters in B, V & R to generate something similar to an H-R cut-off diagram.

The idea being to examine a broad range of individual stars' registering an ADU/count and plotted against known catalogue magnitudes.  

Where the relationship breaks down for the brighter stars should then be obvious from the departure from a linear count vs magnitude graph.

That way I get multiple light sources at multiple data points in the same frame (and at the same temp/dark calibration).  OK, I'm assuming there's no gradient!

(I got the idea from the AAVSO calibration technique for transformed magnitudes).

Also, your comments about the H9 and it's ABG help considerably.  I've probably been operating under the same restrictions for years its just that the H9 is linear 'further up the slope' than the, more obvious, H18.

OK, it's more noisy and less QE 'sensitive' but it's got it's redeeming features.  I'll persevere a while longer and see how things go.

Regards

Tony