Atik 314L

[Rocket Pooch]

Hi,

Gave it a go today just to get the readout noise, my HX916 Bias frames we're min 1,948, max 3,239 mean 2,921 the 314L is min 211, max 361 mean 273.  The number of hot pixels on the HX is in the 100's the 314L is in the 10's.

So the readout noise is 11 times less than  my hx916 GOLLY!

The camera is also 50% more sensitive see chart below:

314L with a ICX285, the HX916 with a ICX085

[RobertM]

Hi Chris,

That's really is low readout noise.  I just checked my SXV-H9 for comparison and it's 1110 mean with a min/max of 1035/1160 but there again it does take 7 seconds of so to download so I'm not really too surprised.  It also has about 10-20 hot pixels.

How much difference do you think the lower read noise will make for longish exposures ?

Cheers

[MarkS]

I'm not sure that what you are measuring there is actually the read noise. 

Instead, take 2 frames in succession and subtract one from the other.  The difference between the two will mainly result from read noise.  Divide by 2 then calculate the standard deviation of those pixels in a selected area (if your software doesn't like negative pixels then apply an offset to one of the frames before subtraction).  That will give RMS read noise in ADUs.  Even then, the gain on the two cameras may be different - you really need is the read noise in electrons.

Even the method above isn't entirely accurate but to do it properly is very laborious - I'll dig out the link some time.  In the meantime the bottom of this one describes a method: http://www.astrosurf.org/buil/20d/20dvs10d.htm

[Rocket Pooch]

Hi Mark,

I just subtracted a BIAS From a Bias for the 314L and it came out as 6.14 adv dark current.

For the readout noise I have subtracted one bias from another to get standard deviation of 28.4 ADU, I have not divided this by 2.

The figure is 63.3 on the HX916.  So maybe the 314L is twice as good then , still an improvement....

Chris


Robert lower readout noise, as far as I understands means that the photon's collected will register earlier in the image, so in theory if it was 0 then I would see all photons hitting the ccd if the QE is 100%.  If the QE is 50% every second photon registers.

[MarkS]

I should have written, divide by square root  of 2.

Your second set of figures make more sense.  A factor of 11 read noise reduction would be phenomenal (but maybe possible on the next generation of CCDs)

[Rocket Pooch]

ok, did the square root of two and got 4.34, the quoted value is 4, how good is this camera, bring it on!

[RobertM]

Just redone the H9 and it's not so bad,  Mean=7.5 so the read noise works out at 5.3 which is better than the quoted < 12 and typically 7.  That was calculated : bias frame - avg of 10 bias frames, stddev was high as you would expect given we're reading noise !

Not as good as yours Chris but not bad for a 4 year old camera.  Am I sure I've done this right - no way !

[MarkS]

ok, did the square root of two and got 4.34, the quoted value is 4, how good is this camera, bring it on!

Good to know that the equation actually works!   That's a very low noise level - nice one!   Can't wait to see some images ...