Starlight Trius Pro-834

[smr]

Can't find much about these Cameras, are they any good? 

https://www.sxccd.com/trius-pro-834

[ebdons]

It's an SX so that should be good news? no standard (black) version and thought they were going more cmos now but it's interesting as the QSI cam has the same sensor? it should be great in mono spec tho.

[newbie alert]

Atik use it in their 4120 camera, which must admit I'd never heard of until now.. 

[smr]

Does anyone know how it compares to the 2600MC Pro? It's on offer at FLO at the moment for almost the same price.

[Whirlwind]

2 hours ago, smr said:

Does anyone know how it compares to the 2600MC Pro? It's on offer at FLO at the moment for almost the same price.

Well the 834 is a mono camera to start and don't think there is a colour version so keep that in mind.  

CMOS is steadily replacing CCD in the market.  Not really because one tech is better than another but other sectors want the (much) faster readout and apart from science dedicated businesses CCD has a limited demand.

The 2600MC will give you a larger field of view, slightly more sensitive (especially in the blue) and faster capture rates with potentially tens of frames per second (which can help if you have a less well behaved mount and could use it for planetary).

The 834 should have more stable noise characteristics which will be easier to calibrate out (CMOS generally has walking floor noise).  CCDs also bin better than CMOS (for the former this reduces noise, but not really in the latter so much).  

The noise and full well depth when you read out the sensor are generally comparable at the same gain settings - however for CMOS you generally can alter these yourself so you get more flexibility if you want higher read noise, but larger well depths or vice versa.  A CCD is fixed in this regards.  *Calibrated* CCD images generally still win out from a noise perspective because of their stability

So it largely depends on what you want to do with it and the scope that it will be attached to.  If you want to do any science work (photometry, spectroscopy) the CCD still wins out (stable noise that is easier to calibrate).  If you just want photos of large nebulae the new CMOS give a lot of area for the same value.  On the other hand this helps mostly for mid to long focal length instruments as generally even a smaller chip can cover most objects with a short focal length telescopes. 

Also keep in mind what telescope you are going to use.  If you use mid/long focal lengths (or are planned for in the future) then larger pixels can be better otherwise you are sampling too finely. 

Just to note in terms of the 834 unless you are using very short focal lengths the 694 might still be the better choice.

 

[Adam J]

On 19/01/2021 at 18:58, Whirlwind said:

.*Calibrated* CCD images generally still win out from a noise perspective because of their stability

So it largely depends on what you want to do with it and the scope that it will be attached to.  If you want to do any science work (photometry, spectroscopy) the CCD still wins out (stable noise that is easier to calibrate). 

If it calibrates out its not noise though. Only fixed offsets can be calibrated noise is not a fixed offset (its noise) and hence cant be calibrated because its random.

Adam

[Whirlwind]

20 hours ago, Adam J said:

If it calibrates out its not noise though. Only fixed offsets can be calibrated noise is not a fixed offset (its noise) and hence cant be calibrated because its random.

Adam

So noise can arise in several forms.  Some are completely random and others are not.  Ultimately all you want remaining in an image is truly random noise as when you combine data these errors will balance out.  For example read noise is a fixed source, random noise is a natural consequence that you never measure something exactly.  So if your true value was 0 and you observed 5 images of the same pixel then you might get -2.-1,0,1,2, which averages out to a value of 0.  Calibration frames take out the things like dark noise (for example) but you are still relying on the measurement of any individual measurement of this value to be truly random - so you want your dark frame images to be 198,199,200,201,202.  These will average to a value of 200 and subtract away from your image.  On the other hand if the fluctuations are not truly random and has an underlying additive or subtractive element that is different to that when you take your images then you end up adding a fixed noise back into your image.  Because of the design of CMOS (each pixel has its own electronics) then you can end up residual fixed noise that is difficult to remove and in some cases isn't stable (for example biases are not generally taken with CMOS cameras because it can be unstable at short exposures).  There are different ways to approach this.  For CMOS it is common to aggressively dither to average out the fixed pattern noise that they can generate.  Even for CCDs it can be worthwhile to dither to remove some hot pixels that can not be managed through darks.  However these tend to be relatively isolated whereas for CMOS the effect occurs across the chip. 

For images it is perfectly OK to have some residual noise if you can dither it out.  You just expose a bit longer (or more images) to beat the noise.  For photometry / spectroscopy where generally you want to aggressively avoid placing objects in different locations and you want each image to be close to perfect as possible, ideally you want pixels that respond the same all the time.  

Hence for a CCD after you have calibrated you generally have a floor value for each pixel that it will to a reasonable approximation randomly vary around.  For CMOS with for example aggressive dithering you tend to have both a naturally varying signal plus a background value that has been averaged out over many images (and hence the floor value is just that bit higher because average of the noise has been added to all your images). Hence CCDs tend to calibrate better from a noise perspective and at the moment are more useful work such as photometry and spectroscopy.  Although CMOS technology is advancing and it might not always be the case.

[Adam J]

7 hours ago, Whirlwind said:

So noise can arise in several forms.  Some are completely random and others are not. 

 

Ok I'll distil all that down to this. There is no such thing as none random noise or noise that is not totally random. If it's not totally random darks will remove it period so long as you have a sufficient number of dark frames in the stack. 

Bias instability in short exposures is not a problem for CMOS you just don't use bias. You use dark flats and it's all great. Once you do this there are no random offset s on the cameras I have used. 

As for spectroscopy etc that's a small corner of the hobby and so it not a factor to 99% of people in this forum. 

Average noise will not be added to your images when dithering. It will have been removed by darks. That's why data pedestals and offsets exist. These are then removed during the normalisation stage in stacking. They just don't get through to the final image. 

The only time any of the effects you are discussing would appear is in very old sensors or via user error in selecting the wrong camera settings or calibration flow. 

 

 

Edited January 22, 2021 by Adam J

[Whirlwind]

On 22/01/2021 at 02:30, Adam J said:

Ok I'll distil all that down to this. There is no such thing as none random noise or noise that is not totally random. If it's not totally random darks will remove it period so long as you have a sufficient number of dark frames in the stack. 

Bias instability in short exposures is not a problem for CMOS you just don't use bias. You use dark flats and it's all great. Once you do this there are no random offset s on the cameras I have used. 

At a simple level there are two types of noise, systematic and random.  It is correct to state that systematic noise can be removed (such as using dark frames).  However, this assumes that the systematic noise is completely predictable.  In some cases this is not the case and is effectively 'chaotic' (in that there are some many individual components that it is impossible to predict them all accurately - this does not make it random however.    Bias is a good example for CMOS - this isn't random noise but it could be considered systematic, chaotic noise.  Hence you don't calibrate out the bias because it is not predictable.  As noted there are other ways to proxy removing this noise so that you are only removing a stable signal.   

This is why for CMOS you can get more commonly fixed pattern (walking floor) noise and for CCD you can get bad pixels that don't calibrate out completely (and need other tools, e.g. dithering).  For CCDs fixed pattern noise is more limited to individual pixels so to an extent are easier to remove with just a few images and a bit of dithering and processing.  For CMOS as noted before every pixel has a source of systematic, but chaotic noise that generally requires more aggressive dithering (this is due to electronics being on each individual pixel) but as it is prevalent on every pixel (although improving in how much of it is there is).  In effect what you end up doing is 'smudging' this fixed systematic noise across the image.  As it is everywhere this effectively averages out this noise source across the image.  Although you then subtract off this average noise off it does result in a residual memory of that noise (as no one pixel is fully corrected for the noise because the correction is an average of the error).  For general astroimaging this is generally not a problem (as you just image for longer) and subsequently in the processing there is a floor which below is given a value of 0 (so you cut out this noise).  However where each exposure is important (rather than the combined image) then this type of systematic noise can lead to problems (especially for faint objects).

I agree that the spectroscopy/photometry are not important to most on the forum, I did not state otherwise, but the OP asked about the relative merits of getting a SX834 compared to a 2600MC so hence the intent was to provide a balanced consideration of the merits of going for the SX834 and its relative advantages.