Problem stacking Ha images - Zero or insignificant signal detected

[dnewbury]

Been using PI with DSLR for a while but only just now starting with CCD (Atik 460EX mono). I'm having trouble stacking Ha subs. I am not using darks. I have taken flats and bias. Have generated a superbias, calibrated and integrated the flats, and then calibrated/cosmetic corrected the Ha subs (all following LightVortexAstronomy tutorial). However, when I look at the calibrated Ha subs in SubFrameSelector, all the FWHM values are zero. Eccentricity values also all zeros. When I kick off the ImageIntegration for the registered Ha lights, I get an error: "Zero or insignificant signal detected (empty image?)".

The Ha subs are 2 minutes @ 1x1 using Baader 7nm filter.

Attached is

• single Ha sub
  superbias
  master flat
  single cal and cc Ha sub

I don't have any experience or reference to go by (with CCD camera, processing Ha, managing calibration frames for CCD etc). Wondering if it could be:

• slightly out of focus?
  just not enough signal?
  bad flats/bias/
  PI operator error? - this is always an option :=)

[pfile]

problems like this are not uncommon with narrowband exposures - many times there is almost no signal in the background, which can lead to issues during bias/dark subtraction.

first off - 2 minutes is kind of short for an Ha exposure. i regularly use 1800s @ f/5.

check the statistics of an uncalibrated sub's background vs. the statistics of the superbias frame. you can do this by defining a preview on the light frame on the background and then using the Statistics process to see the mean and median values. if the superbias has higher values, then that's what causes the "no signal" problem. then the real problem is figuring out why it happened. you can also tell this was the case by looking at the statistics of the calibrated frame - if there are a whole bunch of 0 pixels it also means the bias levels were higher than the light.

there is some chance that the camera gain was set differently between the light frames and the bias frames. that's something else to check.

if the bias is proper then you may need to add a pedestal to the lights to avoid the underflow/clamp to 0 during bias subtraction.

rob

[dnewbury]

Hi Rob,
Thanks for the insight. I've got some work to do before I get to 1800sec subs    I don't think the camera gain would have changed because that's not something I've been messing with (unless SGP somehow changed it?).

You are right - the mean/median of my uncalibrated sub background (approx 5.2x10-3) is less than my superbias (approx 5.8x10-3).

I'm not familiar with the PI Pedestal function (although I think I understand the concept). Some quick searching still has me scratching my head. I tried going back to the Ha lights calibration stage and adding 100 as a Pedestal for "Literal value". But this doesn't seem to make a difference. Can you make a suggestion on most appropriate method of applying a pedestal in my circumstance? How do I determine the appropriate value?

Any additional insight appreciated.

Dave

[pfile]

there's kind of 2 halves to the pedestal during ImageCalibration.

first is the addition of the pedestal. that's controlled by the "Output Pedestal" section of IC (which is in the Output Files section). then there's the removal of the pedestal. that's in the next section. the idea is that you add the pedestal to your calibration frames using the Output Pedestal (say for instance for some reason the bias signal was greater than the dark signal; when you calibrate the darks they would come out with a bunch of 0 pixels, much like your lights have). now you've got dark subs which contain a pedestal of 100ADU and are tagged with the PEDESTAL fits keyword to indicate this. when you make a master dark, the PEDESTAL keyword is carried over to the integrated dark. then when you calibrate your lights with the master dark, IC will then subtract the pedestal value from the calibrated light, if it finds that keyword (that's what it does by default - the literal value is to subtract some other value, say for instance if there is no PEDESTAL keyword, but you know the calibration frames have a pedestal.)

the idea here is that the background value of a light is probably going to be much greater than the pedestal (due to light pollution or actual signal) and so removing the pedestal is OK. but in your case where there's pretty much 0 background, that would not work. so you want to put the pedestal in your lights, and leave it there. it will essentially be removed in processing when you manipulate the histogram.

in your case i think if you set the Pedestal to a literal value of 100, IC probably just subtracted another 100ADU which adds insult to injury

anyway, i think if you set Output Pedestal to, say 80, that might take care of the problem. since you're only running IC once, the pedestal won't be removed.

however, i guess the root question is why the bias frames have lower values than the background of the lights. i understand that your camera uses a sony sensor which has vanishingly low dark current, so that might have something to do with it.

maybe someone that uses one of those sensors can chime in - i don't know if they forgo bias/dark calibration completely - i've heard that people don't use darks with those sensors but i don't know if that also means they don't bother with bias either... i guess you can also just skip bias calibration of your frames and see how they turnout.

rob

[dnewbury]

Woo hoo!

I somehow missed the pedestal option in Output files. Yes - makes sense now. I have managed to successfully complete registration and integration of the Ha subs.

Thanks so much for your responses.

Yes - I've heard that many with this sensor bypass the darks (which I'm also hoping to do). I think that bias is still beneficial. Looks like I now have a way to deal with low signal Ha and bias calibration. Of course will also try achieving longer exposures.


Dave

[pfile]

cool - glad to hear it!

rob

[Don Walters]

I'm having the same problem and I'm just beginning my journey into narrowband, so you'll have to forgive my ignorance, and I hope I'm doing this correctly:

I opened one of my H-alpha subs and defined a preview. Statistics module (for the preview) says my Mean=982 and Median=980.  When I open the master bias (I am not using superbias) Mean and Median are both 1012.  So that means I have the same issue?  If I understand correctly, the H-alpha filter has caused my background to be less than the bias.

So what do i need to do?  I think you're saying I need to add some value to the output pedestal... when I calibrate the light frames???  Sorry, I am a bit lost.  I read your post a few times, but it was hard for me to follow.

The workflow I am using is:
ImageIntegration: create the master dark and master bias
ImageCalibration: Calibrate the flats
ImageIntegration: Create master flat
ImageCalibration: Calibrate lights
StarAlignment: register the lights
ImageIntegration: Final integration of lights

This is also my first time using this workflow.  Before now, I have been using BatchPreprocessing script with my OSC camera, so doing it "long hand" is new to me, but I needed to use it because I was getting stuck on this error, and otherwise had no way to adjust the pedestal value.

Thanks in advance for helping me through this!

[pfile]

well, to be quite honest, i'm not sure why the background in the master bias is > the background in the Ha light. after all, one would expect a dark (or a light for that matter) to have higher signal than a bias, since in theory the bias should have 0 exposure time. both the dark and the light should at least have thermal signal which exceeds the bias signal. so it's not necessarily the Ha filter that's causing this. there's some chance it's due to the camera firmware trying to suppress dark signal in the non-bias exposures. you didn't mention what kind of camera you are using so it's hard to make an educated guess.

so anyway, yes, try setting the output pedestal to 100 in step 4 - ImageCalibration: Calibrate Lights.

rob

[dnewbury]

Hi Rob,
Thanks for the insight - would like to learn more about why it happened. In my case, the camera was an Atik 460EX mono, which is new to me. I took O3 and lum subs as well, and these did not have any issues with the stacking. I wonder if I'm doing something wrong with my preparation of master bias/superbias.

Dave

[Don Walters]

Thanks for your reply!  I am using a ZWO ASI178MM-Cool which is a CMOS camera.  I am also new to mono imaging with filters. Before now, it was only OSC for me. 

When I took the bias, I used Sequence Gen Pro and set the exposure for 0 seconds and fired off about 250 and stacked them using the ImageIntegration process.

I have not changed my gain/offset, so I can't imagine why this would be happening.

I was getting the error using both Ha and OIII subs.

I will try the pedestal when I get a chance and let you know how it went.

Edit:  Oh, and my Ha subs are 600" exposures.

[pfile]

SGP can set the frame type to Bias - maybe try that? there could be a difference between a 0s exposure and a true bias exposure depending on the driver.

rob

[Don Walters]

I just wanted to follow up to this post.  So I went on Cloudy nights trying to get help removing the Amp glow from my subs, and Jon Rista pointed out that my darks had a very different median than my lights and bias.  He wrote:

"There is a significant mismatch in offset between your darks, and your biases and lights. It looks like your biases are dead-on according to what you told me about your settings, with a 253 ADU median level. The lights are also very close with a 257 median level.  The darks, however, have a median level around 980-990, with one being over 1000 ADU. This is a 4x greater offset than your lights, which is the primary reason you needed the pedestal. You should not need the pedestal with proper darks."

So I created a new set of darks and I no longer need to include the pedestal.  In hindsight, I guess I should have uploaded some of my data to this thread.  Hope this helps somebody!

http://www.cloudynights.com/topic/546843-how-to-remove-the-amp-glow-on-zwo-178mm-cooled-pics