Full Well, ADU, and Exposure Length

[JayTee]

Here's a screen shot from AvisFV 2.0. It seems to be a very capable AP image analyzer.


This program is far superior to ASI Studio Fits Viewer

Cheers,
JT

[Juno16]

Here's a screen shot from AvisFV 2.0. It seems to be a very capable AP image analyzer.


This program is far superior to ASI Studio Fits Viewer

Cheers,
JT

Thanks JT! That looks pretty cool!

[STEVE333]

I am just going to post these links in case anyone wants to take a look

Two light frames from a few nights ago. Image stats are from AP Lab.

East Veil light frame. 60 second exposure at a gain of 104 and offset of 70. 97% moon.



Thanks for all of the help!

Hi Jim - If my thinking is correct, and your mean e- value is for one channel (R or G or B), your Veil data shows [(615.4/4) - 70] = 84 e- as the mean value collected across the image. Because the dark current is so low it is not a factor. Thus you would be achieving

sqrt[84/(84 + 1.5 * 1.5)] = 0.987

which means you would be achieving 98.7% of the best Signal-to-Noise possible with your 60 sec exposures. Thus, from a S/N perspective 60 sec exposures are long enough. The other factor to consider is your dithering "dead time", i.e., the time from the end of one image till the start of the next image. My system vearie between 20 - 40 sec "dead time" which isn't too bad. With short image exposures like 60 sec, the dithering time starts to significantly reduce the amount of actual imaging time, so, longer exposures will lead to improved S/N in the stacked image. Of course, longer exposures will also cause more stars to be saturated. ALWAYS A TRADEOFF!!

I'm still unclear if the mean value of 615.4 e- is for one color channel or the sum of all of them or an average of the three. Can you help me with this.

If you are interested I would be glad to model your sensor to show the S/N performance (higher S/N means a better image) as a function of the Exposure Time and the Dither "dead time". Just FYI.

Cheers - Steve

Hey Steve @STEVE333 ,

In post 31 above, there are download links to the light frames if you have a better way to extract the image statistics.

Hi Jim -
I downloaded the E. Veil Light .fit frame and opened it in PixInsight. PI treats it as a mono-image with a mean value of 555 which agrees with the APT Pixel Aid value that you found. So, not sure that is real useful. Your ability to get the mean value for each color channel is much better.

The picture below shows the results of my model for your camera. I assumed Gain = 100. The Vertical axis shows the relative Signal-to-Noise for your stacked images, so, larger is better. The Horizontal axis shows the length of your individual Exposures. I've used an average dither time of 20 sec (your dither time was 60 sec but you only dither once every three images). I've included the results for three different Sky Glow flux levels. The Sky Glow flux level is found as follows:
1) Measure the mean ADU (16 bit) value for the entire frame for one color channel (for example Green).
2) Convert mean value to e- using e- = [(mean/4) - 70] where 70 is your Offset. Using your measured value for the Green channel from above: e- = [(615/4) - 70] = 84 which is the number of collected electrons.
3) Divide e- by the length of the exposure (60 sec for your images). Sky Glow = 84/60 = 1.4 e-/sec for the Green channel.
4) Repeat steps (1)-(3) for all three color channels.

For your Green channel you could use the _._._._ curve in the model (Sky Glow = 1.5 e-/sec) to estimate your performance. Longer exposures would slightly increase the quality of the final stacked image, but, at the expense of saturating more stars. ALWAYS A TRADEOFF! With you Gain of 100 (or 104) an Exposure somewhere in the 60 - 120 sec range looks pretty reasonable to me. 60 sec will have less saturated stars and only slightly more noise.

ASI533MC Model by STEPHEN KING, on Flickr

I hope this makes some sense. If not, feel free to ask any questions.

Steve

[Juno16]

Hi Jim -
I downloaded the E. Veil Light .fit frame and opened it in PixInsight. PI treats it as a mono-image with a mean value of 555 which agrees with the APT Pixel Aid value that you found. So, not sure that is real useful. Your ability to get the mean value for each color channel is much better.

The picture below shows the results of my model for your camera. I assumed Gain = 100. The Vertical axis shows the relative Signal-to-Noise for your stacked images, so, larger is better. The Horizontal axis shows the length of your individual Exposures. I've used an average dither time of 20 sec (your dither time was 60 sec but you only dither once every three images). I've included the results for three different Sky Glow flux levels. The Sky Glow flux level is found as follows:
1) Measure the mean ADU (16 bit) value for the entire frame for one color channel (for example Green).
2) Convert mean value to e- using e- = [(mean/4) - 70] where 70 is your Offset. Using your measured value for the Green channel from above: e- = [(615/4) - 70] = 84 which is the number of collected electrons.
3) Divide e- by the length of the exposure (60 sec for your images). Sky Glow = 84/60 = 1.4 e-/sec for the Green channel.
4) Repeat steps (1)-(3) for all three color channels.

For your Green channel you could use the _._._._ curve in the model (Sky Glow = 1.5 e-/sec) to estimate your performance. Longer exposures would slightly increase the quality of the final stacked image, but, at the expense of saturating more stars. ALWAYS A TRADEOFF! With you Gain of 100 (or 104) an Exposure somewhere in the 60 - 120 sec range looks pretty reasonable to me. 60 sec will have less saturated stars and only slightly more noise.

ASI533MC Model by STEPHEN KING, on Flickr

I hope this makes some sense. If not, feel free to ask any questions.

Steve

Hi Steve,

Thank you very much for that! Extremely valuable information!

I do have one blip in my supplied data. I noticed in the AvisFV 2.0 Header that JT posted that my offset was 10! I know that I had it set to 70 (default) in the ascom driver, but APT must have a default setting where it changes the gain and offset. I have been trying out the Autofocus, Live View, and Plate Solving tools in APT and there must be a setting enabled to allow APT to modify the gain/offset to work better with these tools. I checked the ascom driver last night and it is still set at 104/70.
I found that this happened with gain the first time that I was out imaging the Crescent Nebula, so I inserted a script command to set the gain at 104 for step 1 of the imaging plan. I guess that I will have to insert a step 2 to set the offset also at least for now until I get my feet wet!

With a gain of 10 for the image that you modeled in your post, things will change. In your calculation, this will change the number of collected electrons to 144. Since I plan to use 50-70 as my gain in the imaging plan, would this still keep my exposure range in the 60-120 second range? My thinking is that it would be closer to 120.

Sorry about the confusion. After setting the gain in the imaging plan, I checked it in the fits header, but not the offset.

For now, I will insert the script steps in the imaging plan to set the ccdgain and ccdoffset until I get a better handle on Autofocus, Live View.
Plate Solving works very well (ASPS solves in 4-5 seconds).

Thanks a bunch Steve, and if you get a chance, let me know if my exposure time thinking is correct for using a "real" offset of 50-70 instead of the 10 that I was using in the E. Veil light frame (unknowingly!).

Thanks again!

[STEVE333]

Hi Jim - The fact that your Offset was at 10 rather than 70 doesn't change anything for my curves. The only thing it changes is the calculation for the Sky Glow. Because your camera Read Noise is so low, it has almost no effect on the shape of the curves. Instead, the shape of the curves is dominated by the Dither Time. In fact, all three curves have almost the identical shape, and, only the amplitude of the curves is different.

I still think an exposure time in the 60 - 120 sec range is a good choice. 90 sec might be a good compromise to reduce the noise a bit while not sacrificing too much dynamic range.

Good luck.

[Juno16]

Hi Jim - The fact that your Offset was at 10 rather than 70 doesn't change anything for my curves. The only thing it changes is the calculation for the Sky Glow. Because your camera Read Noise is so low, it has almost no effect on the shape of the curves. Instead, the shape of the curves is dominated by the Dither Time. In fact, all three curves have almost the identical shape, and, only the amplitude of the curves is different.

I still think an exposure time in the 60 - 120 sec range is a good choice. 90 sec might be a good compromise to reduce the noise a bit while not sacrificing too much dynamic range.

Good luck.

Excellent! Thanks Steve!

I learned a lot about Offset today. I am making a set of darks with an Offset of 50. The median ADU runs about 2000 compared to 500 with an Offset of 10. The default in ascom is 70, but it looks like I have a good ADU cushion at 50.
The next time that I shoot darks, I might change it to 10.

Shooting 30, 60, 90, and 120 second darks at 0C and -10C. I included some shorter sets because when I image a broadband target (like M33), I am sure that more photons will pass through the IDAS LPS D-1 filter to the sensor.

Using a command in the imaging plan to set the gain and offset will insure that no matter what APT does while focusing or live view, I will get consistent gain/offset in the plan.

Thank you tons Steve and everyone else that gave me tons of information on this thread!

[Mac]

I learned a lot about Offset today. I am making a set of darks with an Offset of 50. The median ADU runs about 2000 compared to 500 with an Offset of 10. The default in ascom is 70, but it looks like I have a good ADU cushion at 50.
The next time that I shoot darks, I might change it to 10.

Shooting 30, 60, 90, and 120 second darks at 0C and -10C. I included some shorter sets because when I image a broadband target (like M33), I am sure that more photons will pass through the IDAS LPS D-1 filter to the sensor.

Using a command in the imaging plan to set the gain and offset will insure that no matter what APT does while focusing or live view, I will get consistent gain/offset in the plan.

Thank you tons Steve and everyone else that gave me tons of information on this thread!

I've ran my offset at 10 90% of the time and found 180 second exposures to be the minimum with the L-enhance.
It cuts down a lot of the light.

[Juno16]

I've ran my offset at 10 90% of the time and found 180 second exposures to be the minimum with the L-enhance.
It cuts down a lot of the light.

I went ahead and added two sets of 180 second darks. I will leave the offset at 50 for now since I made the library, but will probably drop it the next time. An offset of 50 is probably a bit of overkill, but should work fine.

Looking forward to getting out under a lesser moon. Hopefully, I will get a night next week. Thanks Steve.

[KenGS]

Hi Jim, If offset 50 works ok and does not result in pixels clipping to black then you can just leave it at that. It isn't so critical that it needs to be optimised. The effect it has on dynamic range is pretty small.

[Juno16]

Hi Jim, If offset 50 works ok and does not result in pixels clipping to black then you can just leave it at that. It isn't so critical that it needs to be optimised. The effect it has on dynamic range is pretty small.

Excellent!

I got a good collection of darks at an offset of 50 yesterday and feel better about the setting of 50 after your reply. I might bump it a notch lower next time, but good for now.

I’ve learned a lot from your responses and everyone's help on this thread. Much appreciated!

Thanks again all!