QHY8 Settings ???

[jcm]

My QHY8 arrived this week. Over the last couple of weeks I have been researching what settings to use. The QHY8 is the first CCD camera I have owned that has a gain and offset setting.

Many people suggest starting with the gain at 0 or 1 and the Offset at around 114. Other users said they had the gain about 25 to 30 and the offset at 30 to 60. TJ ( on this forum ) uses setting of about 31 for gain and 30 for Offset.

I found with my camera that an Offset of about 55 gave me a minimum pixel value of about 850 which according to the Nebulosity manual this is fine. I checked with Photoshop and that gave readings of about 900.How come with the same camera other people are using values > 100 ?

The gain setting I found by actual imaging the Rosette neb. I know what a raw 4min image of this looks like so I adjusted the gain till I got something similar , not scientific , but my gain value is now approaching TJ's.

Bern suggested in another thread that one set of settings ( 1 gain , 114 Offset ) may be for user of Maxim DL and the other set for users of Nebulosity.

There must be some reason for the differences. :?

John

[Ben Ritchie]

I was confused by this, and ended up just using the setting suggested by Bern (which I forget..!). They seemed fine.

I think one issue is that different software uses different scales for the gain and/or offset, in particular gain is sometimes quoted as a percentage and sometimes not. So i'd be careful that the settings you're reading about match the software you're using.

[mcrossley]

I use 10% and 110 for the gain and offset in Maxim. The numbers vary according to the software and the driver, some use a percentage for the gain, others a number, and the number range also varies! You need to know how they map to the underlying values used by the camera.

The higher gain values will give you a 'brighter' image at the cost of some dynamic range. Using a low gain means you have to stretch the image a bit more in post processing but you can capture a greater range from dim to bright. Even a minimum gain will not lose signal, I have a couple of analyses that people of done on the effects of different settings somewhere...

Once you select the gain, you then need to adjust the offset so that no pixels have a value of zero. I'd suggest a minimum of at least 500. The camera can only really capture about 12-13 bits of data, so there is lots of 'spare' in the range to fit the offset into.

Mark

[jcm]

If I use 110 for offset using Nebulosity the minimum pixel value for a bias frame comes out as 7000.

But I suppose as long as you test the camera with the software you are using then you should have the correct setting.

John

[narrowbandpaul]

the gain is the conversion from e- into DN which you see in your image.

To maintain the dynamic range (full well/read noise) the gain should be set so that full well is reached at about 65000Dn assuming a 16bit ADC

The scientific method would be take a series of flat field exposures around the few thousand DN. Crop the image to around 40x40 pixels. Work out the mean and standard deviation for the box. The use

Gain = average/ (std dev)^2

reapeat for different exposures around a few thousand DN

the ideal is full well/65000 (find full well from internet)

adjust until gain is near this value.

or get a small source, and set the gain so that the overexposed value is about 55000DN. The fact that the pixel saturates below ADC saturation means that the pixel is full. Now adjust the gain so that the overexposed value goes to around 65000DN. This will ensure tha full well is reached near ADC saturation, preserving linearity and dynamic range.

I dont know how the gain is shown to you with the QHY, but i think its a meaningless number, like a ratio or something (as 0 gain would produce a completely black image). I dont know why they dont just state x e/DN...

the offset shouldnt matter much, 500DN is fine. Too low and negatives will appear through read noise, which will get clipped by the ADC

7000 is too high john.....

good luck

Paul

[jcm]

Thanks for the info , Paul.

Using Nebulosity.

I have ended up with Offset at 55 , this setting keeps the lowest pixel value of a bias frame between 550 and 950 , using a wide range of gain settings.

A gain of 30 ( 49%) is what I shall use for very faint DSO's. With objects like M42 I can substancially lower the gain and still keep the exposure times reasonable.

John

[narrowbandpaul]

Your welcome John

Mr QuiHY said on the QHY forum that gain has an impact on SNR....which I was very confused about.

If you take say a 1 minute exposure it will contain some electrons from the signal, some dark electrons...if you take exactly the same exposure with a lower gain (ie higher DN/e-) then you still collect the same number of electrons. The only difference is the image is brighter because each electron is given a higher DN.

I asked a guy I know in San Fran, who is in the semiconductor/image sensor business whether I was correct. He said that gain has only a very small effect on SNR.

Some theory states that you should set the value of gain (e/DN) to the numerical value of your read noise. This limits the effect of ADC quantizing noise, which is fixed in terms of DN to 1/sqrt(12)

But that appears the only advantage that gain has over SNR.

I think the best value of gain, is the one that optimises dynamic range.

And I believe gain is somewhat analogous to ISO for a DSLR. The image is brighter, but there is no actual SNR improvement

I could be wrong however...it has happened.

Dont know how much that helped?

Cheers, and hopefully you will find the best gain setting.

Paul

[mcrossley]

Found that data that one guy determined for his QHY8:

Driver     Actual
Gain      Gain
   0        0.40e-/ADU
   10      0.35e-/ADU
   20      0.30e-/ADU
   30      0.25e-/ADU
   40      0.20e-/ADU
   50      0.15e-/ADU
   63      0.09e-/ADU

The camera accepts gain settings in the range 0-63, what the application displays will be scaled to this range. Even gain 0 gives an ADU of 2.5 per electron. He measured the read noise as 7e- and the dynamic range at gain 0 as 71.5 dB.

Mark

[narrowbandpaul]

very helpful table mark....

seems like the driver gain is a meaningless number, or is it expressed in a silly unit like dB.

[mcrossley]

It's just a number as far as I know. Given the well depth of the CCD is 25,000 then gain 0 = 2.5*25,000 = 62,500 which just fits in the 16bit range, with 3,000 left for an offset.

At gain 10 = 2.86 * 25,000 = 71,500 which has overflowed the 16bit ADC range = lost dynamic range.

Not quite that simple of course as there is anti-blooming non-linearity at the top end of the range to take into account. But from these figures it would seem you are best using the minimum gain available.

[narrowbandpaul]

from the numbers you have specified, you do indeed seem correct that the gain '0' is the best choice.

You could perhaps use gain 5 with a lower offset....

i wasnt sure of the well depth of the sony chip, but I knew it wasnt particularly high.

The non linearity experienced when near full well is just due to antiblooming, rather than the output of the ADC becoming non linear...or is the ADC non linear near saturation?

perhaps both.

I think both!

agreed though, a gain about 0 would seem the best choice...

of course the exact value will fluctuate from camera to camera!

best doing a PTC for your individual sensor

Cheers

Paul