Final Test of the ASI2600

M63 - Comparison ASI2600 vs. ASI1600

As a final test of the ASI2600 I decided to compare the results I obtained imaging M63, the Sunflower galaxy using the ASI2600 with the same object using the ASI1600.  Both images had pretty much the same number of subs, all taken with 60sec exposures. Each was calibrated and processed using the same workflow. 

First, a comparison of the field of view (FOV) of each camera. Note that both images were cropped a bit during postprocessing as needed to account for slight shifting of the masters among the multiple filters and to eliminate vignetting at the corners. As reported in my previous post, I was pleased that the vignetting due to the use of 31mm filters was minimal. The ASI2600 has a wider FOV, mainly in the long axis. The images below are not exactly to scale; the ASI1600 had more cropping than the ASI2600, so the change in FOV is a little bit exaggerated. The actual sensor sizes are:

ASI1600:  4656x3520
ASI2600:  6248x4176

Next, with some cropping of the ASI2600 image to bring the scale of each closer to 1:1 we can examine each for actual image details. Please note that the sky conditions were not completely identical in each session and so there may be some differences due to quality of subs, not the actual difference in the sensors.

However, some of the 'promised' improvements in the ASI2600 over the ASI1600 can be seen. First, the artifacts due to the microlens light scattering in the ASI1600 are completely gone in the ASI2600 (see bright star just to the right and above the galaxy). Second, the overall brightness of the galaxy is greater in the ASI2600, likely due to the greater quantum efficiency of the 2600 over the 1600 (91% vs. 60%).

In addition, but not obvious in these quick comparisons, is the much lower noise in the ASI2600. Dark frames were consistently measured as 500ADU, across all typical durations (30sec - 1200sec)! And, the total absence of any amp glow makes post-processing even easier.

One thing that is noticeable in the 2600 image is the size of the stars - they are bigger than the ones in the 1600 image. This might be due to slightly different post-processing. Another contributing factor may be the lower full well capacity of the ASI2600 at gain of 100.

Overall, I am very pleased with the performance of this new camera. 

ASI2600 Testing Continues

Test of new OAG spacing

You may recall in my last post I had discussed the plan to reset the spacing of the ZWO OAG on the scope to help with the strong vignette of the guide scope camera image. I moved the 15mm spacer in front of the OAG, mounting the OAG flush with the filter wheel. A thin 0.5 mm spacer was used between the nose piece and the 15mm spacer to act as a stop when the assembly was attached to the Moonlite 2" adapter. As expected, the pick-off prism is now rotated a bit from the center of the sensor's long axis.

Although this did improve the guide image, there is still an obvious vignetting present. The ZWO prism and light guide are just very small compared to the Celestron OAG.

Gain Settings

Early Friday morning I imaged the Sunflower galaxy, M63, so that I could compare the results of the ASI2600 with the image I took with the ASI1600 back in March. I ran a similar set of LRGB subs, 60sec exposure, but I used Gain 0. It turns out that these subs are essentially useless as I needed either a longer exposure time, or higher gain setting. The images did look rather nice, but they couldn't be stretched in postprocessing since they lacked sufficient dynamic range.

The darks at Gain 0 on the ASI2600 produce frames of 500ADU average at -10C. The 60sec lights of M63 came in at 520-560ADU. This means that once the dark frame data is subtracted from the light frames the net result will be only 20-60ADU. This is extremely low signal content, especially with a camera that has a 16bit ADC (0-65535). Any amount of stretching to bring out the faint details will produce a final image that is extremely 'posterized'.

To illustrate the point, I took a few subs of the Blackeye galaxy, M64, last evening during a very short period of clear skies in that direction. Clouds interfered with the 60sec gain100 sub so I couldn't do a comparison with the 60sec subs.

Three subs, 60secGain0, 120secGain0 and 120secGain100, are shown in the image below. A quick stretch of each was made to bring out the details. As you can see, the images themselves seem pretty nice. But this is misleading as each stretch is a different relative strength, normalized so that they all appear with the same relative level of brightness and contrast. When a background model (ABE) is created for each sub the underlying problem is made apparent. Notice the model of the 60 sec gain0 sub. There are only 11 levels of dynamic range in this image. Once stretched and processed the result will be very blotchy and posterized - there simply isn't enough data in the sub to work with. Note that one of the reasons I purchased this camera was to get higher dynamic range, 16bit vs the 12bit of the ASI1600. At gain0, 60sec, the result is worse than that of the 1600.

However, when setting the gain to 100 and increasing the exposure time to 120sec the dynamic range is greatly improved. At Gain100_120sec the mean ADU of the image is 1071, twice that of the comparison dark frame.

Reviewing the specs of the ASI2600 suggests that Gain100 should be the optimum setting as there is lower read noise at gain 100 than at gain 0, and the dynamic range is almost the same. The down side is that the full well capacity is greatly reduced (16000 down from 49000). So, for long exposures, the lower gain provides some safety in protecting from blown out star images as the camera can hold far more electrons that at gain 100. Since my general process is to take more, shorter exposure subs, the higher gain is the obvious first choice. 

Since my earlier M63 subs suffer from the low dynamic range, I will need to retake that series once the weather clears and the moon is out of the sky - probably next month - and then do the comparison. Stay tuned!

More ASI2600 Testing

The ZWO OAG finally arrived this past Monday and I was able to do some initial testing of the 2600 on the EdgeHD11.

Getting the spacing just right was a challenge but I had ample spacers in my shop, and ZWO provided a good variety with their cameras. 

My final configuration is as shown in the pic below; from left to right - nosepiece to mount to the Moonlite Focuser on the EdgeHD, ZWO OAG, 15mm spacer, ZWO Filter Wheel, 2mm adapter and ZWOASI2600. The placement of the 15mm spacer behind the OAG was chosen to allow the guide camera mounting to clear the filter wheel so that it could be rotated to get the pick-off prism to be centered over the long axis of the sensor. 

Initial test exposures yielded very good results, as can be seen in the following images. Since these are single, unprocessed images (except for stretching), the dust motes and vignetting are obvious.

Starfield 300sec

M82 in Ha 300sec

I was very pleased that the f/10 focal length of the Edge produced even less vignetting at the corners than with the f/5.7 WO GT102. Since I had already verified that proper post processing with good flats eliminated the vignetting with the GT102, I should have no problem with the edge images!

However, the placement of the spacer behind the OAG caused serious vignetting in the guide camera view, so much so that finding guide stars became a big problem.

What I had forgot to account for was the longer path of the light cone entering the guide camera. I needed to add 24mm of spacer to the camera. Now the ZWO's pick-off prism is fairly small, and coupled with the guide camera's sensor distance being a bit excessive the effect is like 'looking through a straw'. This needs to be corrected. I decided to move the 15mm spacer in front of the OAG, mounting the OAG flush with the filter wheel. A thin 0.5 mm spacer was used between the nose piece and the 15mm spacer to act as a stop when the assembly is attached to the Moonlite's 2" adapter. The pick-off prism will be a bit rotated from the center of the sensor's long axis but should still be out of the light cone. Now all I have to do is wait for clear skies to test this out.