31 replies to this topic

[bilgebay]

I am about to pull the trigger on a KAF-16200 based CCD camera.

 

At an extra 800 US,  I have the option to purchase a Class 1 sensor.

 

What would you do ? Is it worth it ?

 

I will be grateful for your input.

 

 

 

[JJK]

The 16200 spec sheet states:

Grade 1 (no column defects)

Grade 2 (up to 15 column defects)

[fetoma]

How many columns on the chip?

[kw6562]

How many columns on the chip?

4500.  I noticed in the data sheet that defects of 2 or more columns are not permitted - so up to 15 single columns on a Class 2 chip.  Also, sometimes those defects cool out at lower temperatures.  If it were my decision I would save the $800.  Just my 2 cents --Keith

[blueman]

I don't know, I have had class 2 chips and the columns got worse with age, not an FLI but none the less that was the case. A Class 1 chip meets a very stringent requirement.

Blueman

[josh smith]

I don't know, I have had class 2 chips and the columns got worse with age, not an FLI but none the less that was the case. A Class 1 chip meets a very stringent requirement.
Blueman

Does it matter how bad the columns are if using bad pixel mapping, pixel rejection, and dithering?

[blueman]

 

I don't know, I have had class 2 chips and the columns got worse with age, not an FLI but none the less that was the case. A Class 1 chip meets a very stringent requirement.
Blueman

Does it matter how bad the columns are if using bad pixel mapping, pixel rejection, and dithering?

 

Well, yes I think it does. Besides the hit on resale, sometimes the columns don't 100% get corrected or take a bit more work to complete an image.

Blueman 

[bilgebay]

Thanks everyone.

 

This is what Atik has to say about CCD grades

 

 

 

The Atik 11000 is fitted with a KAI 11002 class II sensor as standard. This allows for a small number of column defects. Alternatively, a class I sensor with no column defects can be specified. Does the CCD grade matter? Well, we feel the price premium placed on the class I sensor is not justified when the camera is going to be used for astro-imaging.

In short, this is because Kodak’s grading system uses a test in which the CCD is not cooled; exposures are for fractions of a second; and the sensor is read out several times a second. But conditions for the sensors are very different in astro-imaging because the sensor is cooled; exposures are typically for many tens of seconds; and readout takes several seconds. It is also expected that the images will be corrected with a dark frame. The upshot is that Kodak’s grading tests do not reflect astro-imaging performance in a useful way.

 

Looking at this, going for a Class 2 seems like the way to go but I'm still open to opinions.

 

Cheers

[WesC]

I have a column defect that developed in my QSI 683, its annoying and I have to do an extra step (PI cosmetic correction) to fully eliminate it, but in the end it does not affect my images.

 

If I had more than two though I would return the camera. 15??? No way!

 

I think if I end up going for the ML16200 I will spring for the class 1. I'm very particular though.

[gezak22]

Assuming 10+ exposures, wouldn't dithering and sigma rejection take care of column defects?

[pfile]

not necessarily since sometimes the columns are just "weak" rather than dead. so they don't "stick out" enough to be easily rejected like a hot or cold column/pixel.

 

if you are growing dead/weak columns quickly then you're not only going to be trying to match flats but you'll be trying to match darks/bias as well. with 1800s exposures this can be a PITA - aquiring new calibration frames all the time.

 

rob

[contedracula]

I'm agree with Atik Policy, the Grade I is justified in science not in amateur astronomy...with post processing correct all " cosmetic "defects

 

Thanks

[gezak22]

not necessarily since sometimes the columns are just "weak" rather than dead. so they don't "stick out" enough to be easily rejected like a hot or cold column/pixel.

If it's so weak that a proper sigma rejection does not remove it then it cannot possibly have a negative result on the final image.

 

if you are growing dead/weak columns quickly then you're not only going to be trying to match flats but you'll be trying to match darks/bias as well. with 1800s exposures this can be a PITA - aquiring new calibration frames all the time.

Yes, I can imagine that this would be a major PITA. If dead/weak columns grow, how likely is it that a class 1 CCD will develop similar defects over time?

[blueman]

 

not necessarily since sometimes the columns are just "weak" rather than dead. so they don't "stick out" enough to be easily rejected like a hot or cold column/pixel.

If it's so weak that a proper sigma rejection does not remove it then it cannot possibly have a negative result on the final image.

 

if you are growing dead/weak columns quickly then you're not only going to be trying to match flats but you'll be trying to match darks/bias as well. with 1800s exposures this can be a PITA - aquiring new calibration frames all the time.

Yes, I can imagine that this would be a major PITA. If dead/weak columns grow, how likely is it that a class 1 CCD will develop similar defects over time?

 

Class 1 chips so far have not developed issues for me. Class 2 on the other hand did continue to degrade with increasing length and brightness of column defects.

However I will not argue that the Class 2 chips won't work, that of course is not true, of course they will. But when I sold the last one I had it did not sell for a good price, the columns were long, bright and had to be disclosed.

Blueman

Blueman

[gezak22]

Thanks for the input Floyd. In that case I agree that the $800 upgrade is well worth it.

[contedracula]

There are donuts that are born without a hole   despite, there're tons of wonderful image taken with Grade 2 Sensor

 

I love to see the Super Bowl for the match not for Advertising spot...buy a Grade 1 is wasting the money for using in amateur astronomy, as I consider useless to see advertising of Ikon XL-231 on amateur sites

 

Just my opinion

 

 

 

[bilgebay]

Thank you all... but.... you didn't make my job easier at all the fight within continues....

[gezak22]

Thank you all... but.... you didn't make my job easier at all the fight within continues....

The good news is that there is no bad decision. As others have said, Class 2 sensors produce nice images. But I feel that the cost of $800 (when compared to the cost of scope/mount/ccd/laptop/time) is small enough to justify the upgrade to a Class 1 sensor.

[JJK]

 

Thank you all... but.... you didn't make my job easier at all the fight within continues....

The good news is that there is no bad decision. As others have said, Class 2 sensors produce nice images. But I feel that the cost of $800 (when compared to the cost of scope/mount/ccd/laptop/time) is small enough to justify the upgrade to a Class 1 sensor.

 

I agree.  I'll be ordering a Class 1 sensor.

[pfile]

 

not necessarily since sometimes the columns are just "weak" rather than dead. so they don't "stick out" enough to be easily rejected like a hot or cold column/pixel.

If it's so weak that a proper sigma rejection does not remove it then it cannot possibly have a negative result on the final image.

 

 

not sure where you would get that idea. deviations smaller than whatever sigma you have chosen continue to participate in the integration and will affect the average of the stack of pixels. if you have to choose a really small sigma to get rid of the column deviation, you'll be rejecting pixels in other pixel stacks that you wanted to keep. and because these are long, linear structures they are pretty obvious to the eye even if there is just a small difference in the average.

 

also because these are columns, the Y dither has pretty much no effect on them and so statistically speaking you'll get a few frames where the bad columns line right up on each other, making them look like signal.

 

the "best" solution to this problem is to use cosmetic correction to fix up the bad columns. however if you have a sensor that's drifting fast, you may find yourself having to tweak the bad column list over and over again, which is a pain in the butt.

 

rob

[josh smith]

Some valuable information here based on real life experience. Thanks for the info guys.

[bilgebay]

After much deliberation, I decided to go for a Class 1 sensor. I hope it really is worth the extra 836 bucks

 

Thanks to all who contributed.

[pfile]

i think you will be happy with that choice. if i remember correctly, my STT-8300M was not advertised as class1 or class2, however, the sensor is exceptionally stable and has no bad columns even after 3 years. it's one fewer thing to worry about.

 

rob

[gezak22]

 

If it's so weak that a proper sigma rejection does not remove it then it cannot possibly have a negative result on the final image.

 

 

 

not sure where you would get that idea. deviations smaller than whatever sigma you have chosen continue to participate in the integration and will affect the average of the stack of pixels. if you have to choose a really small sigma to get rid of the column deviation, you'll be rejecting pixels in other pixel stacks that you wanted to keep. and because these are long, linear structures they are pretty obvious to the eye even if there is just a small difference in the average.

 

also because these are columns, the Y dither has pretty much no effect on them and so statistically speaking you'll get a few frames where the bad columns line right up on each other, making them look like signal.

 

the "best" solution to this problem is to use cosmetic correction to fix up the bad columns. however if you have a sensor that's drifting fast, you may find yourself having to tweak the bad column list over and over again, which is a pain in the butt.

 

rob

 

I did jump the gun with that statement. But with a proper dither pattern, sufficient frames, sigma rejection, and a mean combine, the effects should be minimal. Then again, I have no experience in dealing with column defects.

 

After much deliberation, I decided to go for a Class 1 sensor. I hope it really is worth the extra 836 bucks

 

Thanks to all who contributed.

Way to go!

[pfile]

yeah, the problem will manifest in areas of weak signal and the background... but if that's what you're trying to dig out, overlapping bad columns suck.

 

rob