Sunday, January 18, 2015

Review: Starlight XPress SX-36 CCD camera

"Can a SX lady cause a financial crisis?"

The past and today

Elegance and power
Over a decade ago, as  I was about to enter the astrophotography scene, two companies were enjoying a high standing and a good reputation worldwide: SBIG and Starlight XPress. SBIGs were the Kodak gorillas having the power for doing everything. The Starlights were the beautiful ladies doing the same things but in a silent, elegant way. Most of the ladies were hosting a Sony sensor. If you wanted a serious camera to capture the stars, you had to choose between them, either a gorilla or a lady. End of the story. It was like the funny battle between PC and Mac in the '90s.

Her Majesty, the SX-35
Since then, two cameras dominate the upper end of the Starlight XPress product portfolio. Both of them offer full frame Kodak sensors. The heart of the first one is a legend. The KAI-11000 sensor of the Trius SX-35 camera, has been controversially discussed. Its low resolution, high dark signal, high read noise, relatively low Hα sensitivity, and its interline architecture have been criticized. Nevertheless, nothing stopped this 11MP sensor from getting the most breath-taking space photos in the last decade. Nothing is wrong on a good calibrated image from a KAI-11000 based astro camera.

Her Royal Highness, the SX-36
The second one, the "Thirty-Six" did not share the same popularity grade as her sister. For years, the 16MP camera segment was dominated by the KAI-16000 sensor. This full frame chip had a higher resolution but the same high read noise and a lower Full Well Capacity (FWC) than the practice proven KAI-11000 chip. Nothing can defy the law of physics. In that case, the price was a lower dynamic range. But even so, technology moves on and there are good news.

The "Thirty-Six" is back, and now she looks to have some say in the advanced astrophotography league. Her new heart, the modern Truesense  KAI-16070 sensor has lower read noise, higher FWC, and slightly improved sensitivity. These benefits result to a higher dynamic range. Her key advantage is still her higher resolution compared to the SX-35, making her a better match for smaller instruments. The SX-36 offers 16MP x 7.4uM square pixels vs. 11MP x 9uM of the SX-35.

Facts and features
Design is everything

However, the biggest step forward is the new Trius camera design. This is the point where both her Majesty and her Royal Highness profit of. This compact design (118mm in diameter x 102mm long) is Hyperstar capable. Light weight was always a virtue of Starlight Xpress cameras. Weighting only 1100g, the SX is light weight enough to ride atop on the focuser of a Takahashi Baby-Q without tilting it.

CCD chamber
The argon filled chamber prevents moisture under wet climatic conditions. People in northern countries look for features like that when a camera upgrade is in sight. These cameras look like a work of art. Beauty is simplicity.

CCD sensor
Unlike the KAF-16803 based beasts (even its FWC and sensitivity is a bit higher compared to the ladies), the full frame chip of both Starlight ladies is still illuminable with standard 2" filters. That means, you save thousands of Euros when small 2" filters are also doing the job. IMHO, if a 36.3x24.2mm pixel array is not large enough for your application, I wonder whether a squared 36mm array will finally satisfy you. Especially, if you look at the modern TVs, computer displays or iPads. Nothing nowadays is square.

Furthermore, if you want to use an off-axis guider this smart aspect ratio of the SX-35/36 offers you the free space you need to place an off-axis mirror above the sensor.

According to the manufacturer, dark frames are not necessary for most of the brighter deep sky objects, due to the low dark signal of the "Thirty-Six". Anyway, you can judge her by examining the master dark frame I am providing you here.

The sensor offers an effective anti-blooming (ABG) with a minimal effect on linearity and no lost active area. Although ABG architecture lacks a bit sensitivity, amateurs like me might not like to miss it particularly when having experience with light sensitive NABG (non anti-blooming) chips like the narrowband beasts Kodak KAF-402 ME und KAF-1603.

Download times
The electronic shutter of both SX sisters can take very short exposures, making them capable for solar and moon photography, although the built-in rustic USB 2.0 interface needs approx. 15 seconds to download each frame. Let be honest. Asides from marketing discussions, very fast downloads, i.e. high read-out frequencies result to unwanted high read noise. Believe me, you will want to read out your CCD chip slowly, if your are doing serious astro imaging. Hence, 15 seconds is a fair value. It's worth to download a good image slowly than getting a bad image quickly. Advanced birthday photography is not the application area of the SX. Especially, if you see her good QE (quantum efficiency): 52% at peak (green light), 31% at Hα.

USB hub
If cable salad is not your favorite dish, there are good news: the new SX Trius line has a built-in 3 port powered USB hub. That means you can directly connect your lodestar and your Starlight filter wheel on the SX-35/36. That results to a single USB cable control between your laptop and the SX for all features. Since we use an Astroholgi filter drawer on our SX-36 we don't need all three mini USB ports.

Both SX-35 and SX-36 cameras have a three stage Peltier cooler for efficient CCD cooling. Two ventilators, one on the camera and the other inside it blow out the hot air. At room temperature we achieved -15°C to make the calibration frames available here. We think, a delta of -37°C is enough for the night use, especially in the winter and if you consider its low power consumption. It less than 1A at 12V DC - 12W at 110V/240V AC. In my opinion, these are key arguments for the field operation.

Both SX have a back focal distance 17mm +/-1mm. That's why I was previously talking about their useful aspect ratio. You may want to mount an off-axis guider (and adding weight) on the ladies. Especially, if you consider their acceptable weight. The circle is complete.
The input thread (72mm) is compatible with the Takahashi FSQ telescopes. We use a "WideT to 72mm" adapter to connect our SX-36 to our FSQ reducer (see images).

The tilt angle of both SX is adjustable. Although, this is a mandatory feature for a large format camera, you will probably not need it. AFAIK all their cameras are precisely adjusted with laser measuring instruments before leaving the factory. However, if you still think, you new SX has a tilt, you should first check your focuser, before trying to adjust your (probably perfectly adjusted) SX. If the focus point is far outside of the tube, this could be a possible reason for focuser tilt.

Both sisters are coming with the standard Starlight XPress software package. Hence everything you need to get started is inside the box. My calibration frames are made using this software. Although, it seems to work, addicted MaximDL users like me may finally prefer the Diffraction Limited software to control the lady. Now hold on tight. Diffraction Limited, the creator of MaximDL, has recently bought SBIG!

How to use the SX-36

Mono cameras have a different handling than OSC (one-shot-color) cameras. Whenever a DSLR camera needs only one exposure to get a color image, the mono cam needs three of them => red, green and blue through a R, G, B filter respectively. See the following examples:
  • RGB image from a consumer DSLR camera Canon 7D: RGB single exposure at ISO6400 without filters (JPG).
  • Tricolor image using the SX36: Stacked exposure 10x200ms balanced 1-1-1 using Baader RGB filters (JPG).
Calibration files

Astronomical CCD cameras can be precisely calibrated. This step is mandatory in order to get the most out of the image data. You need calibrations files to do so. Please find attached the calibration files of the SX-36 we reviewed (SNR-007).

Flat fields remove artifacts caused by inhomogeneities of the CCD sensor, dust particles on both the sensor and the optics, and vignetting. You need separate Flats for each channel.

Flat field of the SX-36 on a Takahashi FSQ-85EDX with reducer f/3.9 at -15°C=> (JPG/FITS)
Dark frames remove the dark signal of the sensor. They are taken at the same binning, same temperature and at a equal time length like the light frames. A dark frame always contain a Bias frame.

Dark frame of the SX-36, 20 minutes at -15°C => (JPG/FITS)
Bias frames remove the noise of the electronics. They are useful for the calibration of the short-length L, R, G, B Flats.

Bias frame of the SX-36 at -15°C => (JPG/FITS)

FFT evaluation
The 2D Fast Fourier Transformation (FFT) of the master BIAS averaged of 30 Bias gives information about the electronics. Periodical patterns indicate badly shielded electronics. A point in the middle of the image with perfect noise around it without any pattern is a very good sign.

Read noise FFT of our SX-36 (FITS).


Read noise evaluation
Having a Bias and Master Bias frame and knowing the gain of your camera, you can try to calculate the read noise of your CCD according to the instructions in

Read noise frame (FITS) of the SX-36.
Do it yourself:
Bias and Flats of our SX-36 (FITS)

Flat fields, darks, and Bias should be combined with the Median or Sigma Clip filter in MaximDL to build a master file (see screenshot). The Average filter does not remove the cosmics effectively.

How to create your calibration files

You need calibration files i.e. a master dark, a master bias, and several master flats (one for each channel) to calibrate your astro images. It is recommended to create new calibration files periodically, i.e.:
  • Darks: once a year. Like everything on this planet, CCD sensors change their characteristics over the time.
    Example: Master dark SX-36, 20min, -15°C (JPEG/FITS).
  • Bias: once a year.
    Example: Master bias SX-36, -15°C (JPEG/FITS).
  • Flats: each time you turn or mount/dismount your camera on the scope. As the time goes by, new dust particles sit on the optical path. Even in stationary setups it is advisable to periodically (once a year) make flats.
    Example: Master flat SX-36, FSQ-85(f/3.9), -15°C (JPEG/FITS). 
This procedure can be automated to a certain degree with a computer-controlled filter wheel. Download here my MaximLE sequence files on my site, to find out how I do it.
TIP: enable the check box "Group By Slot"when making flats in MaximDL to minimize filter wheel rotation!
Make your flats at the same temperature as your darks and at the same focus position as you captured the stars. Your are advised to make the flats immediately after ending your photo shooting session. Even if you can reproduce the CCD sensor temperature at the next day, the scope (especially a fast one) may not have exactly the same tube length (!) or optical characteristics (color correction, focus point, etc.) when it is operated at room temperature at the next day.

Flat field panel
Gerd Neumann's Aurora flatfield panel is a perfect tool to make your flats. Do not exceed the 2 hours time when using it. R,G,B flats can be calibrated by subtracting a Bias frame. Flats for narrowband (NB) images take longer time to expose. Hence, you must calibrate them with master darkflats having the equal exposure time as your NB flats.

The final step is to calibrate your your master flats. Download my check list. Use the Median or Sigma-Clipping filter for stacking to create the master files. For those fellows making scientific work on CCD sensors without any commercial interest, I am making available the dark, flat and bias calibration frames of our new SX-36 camera (Serial No. 007, KAI-16070 sensor): frames in JPEG for previewframes in FITS format for scientific purposes

How to process your images:

Process your RGB images strictly in the following way in MaximDL:
  1. Acquisition of red images with this monochrome camera => RRAW
  2. image calibration RRAW (using the master dark, the master bias, and a R calibrated* master flat) => RCAL 
  3. remove gradients (i.e. the light pollution) in R channel using MaximDL's "Auto remove gradient" function => Rclean
  4. align all Rclean images using MaximDL manual star alignment algorithm => Raligned 
  5. stack all Raligned  images using Sigma-Clip method => Rsum
  6. ... repeat steps 1..5 for GRAW (green) images
  7. ... repeat steps 1..5 for BRAW (blue) images
  8. align the three Rsum, Gsum, Bsum images => Rfinal, Gfinal, Bfinal
  9. color combine Rfinal, Gfinal, Bfinal => RGBfinal)
Your RGBfinal image:
  • has an improved S/N ratio as a stack of single exposures.
  • is free of numerous artifacts due to calibration with flat fields (step 2):
    • inhomogeneities of the CCD sensor.
    • has no artifacts due to vignetting of the optics.
    • has no artifacts due to dust particles on the telescope optics or the CCD sensor.
  • is free of the bias signal (step 2).
  • is free of thermal current (step 2).
  • has no light pollution or moon glow gradients (step 3).
  • has no satellite traces, due to Sigma-Clip stacking method (step 5).
  • is in color
  • has tighter stars with accurate colors
Having now an excellent RGBfinal image your post-processing with Pixelmator, DxO OpticsPro or Photoshop is easier and leads to better results. You can better stretch your image, do good white-balance on it, etc.


The SX-36 addresses advanced astrophotographers needing a large format astro camera, embodying a sophisticated concept with modern electronics. It nicely matches portable flat field telescopes without sacrificing the image quality in terms of resolution, dynamics and noise. Its low power consumption and its adequate weight make it suitable for field use. Its ability to interwork with numerous other components of the manufacturer´s product portfolio makes it a future-proof investment.

The Thirty-Six offers a well balanced architecture implemented in a comprehensible, upgradable product. It is made by Europeans having astronomy in mind. Good investments like this one never cause financial crises. The beauty of simplicity is timeless.

Thanks for reading.

Panagiotis Xipteras

Special thanks to Holger Weber for all his support during the review phase.


Pavel Pech said...


nice write up, but I am missing the important information here, what actually is the readout noise of that CCD camera??? at a given temperature and binning? Did you try to ever calculate it? Do you have some raw FITs files like at least 10 bias frames and couple of flats to share?

thank you

xipteras said...

Hi Pavel,

thank you very much for your feedback. Here are the Bias and Flats you have asked for:

You may proceed with your read noise calculations and publish your results in the comments area of this thread.

Thanks a lot.


Anonymous said...

So when will you have some pictures to share with us?

xipteras said...

I have already asked the camera owner to provide me some images, when the weather conditions in his location permit it.