More sensor analysis questions

[galljj]

Robin

Almost all the time when I attempt to run the sensor analysis I get twin peaked histograms. Which I understand is not good. Once I got it to run without producing the twin peaks.

Is there a problem with my set up that creates this, or a "feature" of the camera, which is a rising tech 224. I run it s an ASCOM camera.

Does it matter if I use HCG or LCG setting ?

Where should analog gain be set in the 0 to 100 range , and does the settng for black level matter ?

The one time it ran to closure, I got the below, which seems odd compared to other plots I've seen.

john

test output.png (353.94 KiB) Viewed 3471 times

[oopfan]

Hi John,

I ran sensor analysis for the first time a couple days ago. Here was my experience:

The first attempt aborted with a message saying that SharpCap could not adjust the exposure any lower. I understood this to mean that the brightness level of the illumination was too high. For the second attempt I placed a tee-shirt over the objective and then it ran to completion. See attachment for a screenshot.

I see from your screenshot that your frame rate is only 0.5 fps. That is what my rate was on the first attempt. Like you I got some strange "hiccups" in the shape of the curve. Furthermore the analysis seemed to take forever at that frame rate. If I remember correctly it ran for 30 minutes before it finally aborted.

On the second run I reduced the "capture area" from 1920x1080 to 480x270. That reduced the byte transfer from the camera to the PC by a factor of 16 which enabled a frame rate of 15 frames per second. (I didn't think about it at the time but I could have experimented with adjusting the USB transfer rate.) Anyhow, at a rate of 15 fps the sensor analysis completed very quickly -- 5 minutes if I recall.

Just my experience. I hope it helps.

Brian

[oopfan]

John,

When you finally get a good sensor analysis would you do me a favor and upload a screenshot? I can't find a datasheet for your camera but I am looking at the one for the ZWO ASI224. It seems to me that your Read Noise is crazy high.

Thanks,
Brian

[galljj]

Brian:

Thanks for the tips. Here's what I did:

I moved from doing the analysis on a low end laptop (not even useful as a doorstop) to my desktop machine. Initially that made no difference - runs were unsuccessful and took upwards of 20 minutes.

I then reduced the capture area to 320x240 as you suggested, running in HCG mode and with frame rate set to high and analog gain set to 100. The system completed the analysis in just over 1 minute. The results make more sense (or, at least they look "better" in terms of sensor performance). It is not clear to me how this translates to in-the-field performance, or how this relates to performance for other cameras. Here's the screen capture of the output:

camera test 2.png (184.26 KiB) Viewed 3455 times

[oopfan]

That's great news, John!

Now I see that your curves look similar to what is published in the ASI224 manual.

[oopfan]

John,

I would try running the analysis twice, once at Low Conversion Gain (LCG), and once at High Conversion Gain (HCG). Compare and contrast. From what I read on CloudyNights, LCG limits the gain to 30dB, and HCG to 60dB.

Regarding your question about the gain setting, my understanding is that the sensor analysis routine is looping over all combinations of gain and exposure in order to generate the results, so it does not matter what gain you set before starting the analysis.

Brian

[oopfan]

John,

Is your camera color or monochrome? Your screenshot suggests that it is color, looking at the checkerboard pattern. Try running the analysis with a "color space" of "MONO".

My old ZWO ASI120MC color camera had these modes: RGB24, RAW8, RAW16, and MONO8.
My new Altair GP290M monochrome camera only has MONO8 and MONO12.

If your camera has a MONO mode then try it. I am sure that I don't fully understand the impact on sensor analysis, if any. Looking forward to Robin's expertise.

Brian

[galljj]

Brian

It is a color camera. There is no mono mode available.

I ran in LCG and got the following results, whihc are essentially te same as the HCG mode. In one way this makes sense, because the sensitivity of the sensor should be independent of that mode setting.

camera test lcg.png (170.83 KiB) Viewed 3440 times

[oopfan]

Thanks. Let's let Robin sort this out. The only thing that I find peculiar about your result is the erratic change in Read Noise at Gain values of 100 and 162. Gains above 162 act in a more predictable manner.

[admin]

Hi folks,

I can definitely explain the fact that you get two peaks in the histogram - you are using a colour camera via the ASCOM driver, but the ASCOM driver may be reporting the camera as mono or getting the bayer pattern wrong. When SharpCap knows the correct bayer pattern for a colour camera it selects only the green pixels to sample, but if it has the wrong pattern or thinks the camera is colour then it may be selecting all pixels or the red/blue pixels to sample. Try adjusting the debayer preview option until you get the correct colour output and then see if the two peaks go away. I must admit that I haven't tested that specific case, but I'd expect that to work.

Selecting a small sensor area (100x100 to 200x200) will help stop the analysis from being too slow. Another useful thing is to pre-test the illumination level. Set the gain to max and try to get illumination that means the exposure is in the 5-10ms range. Then put the gain to minimum and check what exposure that needs - hopefully less than 1s or the process may be slow!

I am not sure about the low read noise at very low gains - I haven't seen anything like that before. I do have an Altair 224 here, so will try running the analysis on that when I get a chance and see what comes out (maybe via the ASCOM driver as I've done it before using the native support and it was fine).

cheers,

Robin