117 replies to this topic

[orlyandico]

I have a really, really old Canon EOS 40D (un-modded).  This model is I think 7 years old.  However it works fine for terrestrial photography (I've had the shutter and rubber covers replaced) and is tolerable for AP.

 

Would a much newer camera be better?

 

I don't believe all this "ISO 12800" stories etc. because physics.  Also the newer cameras have much smaller pixels.  About the only advantage I can see to a newer camera is the tilt screen and Wi-Fi.

 

But... I can get say a used, modded 450D for a really low price.  I don't know if an older, modded camera would trump a new, un-modded one.

 

I'm not really looking for something that can do 5-minute exposures as I also have an ST8300M.  I want something for quick-and-dirty 2-minute or less exposures.

[GJJim]

I think live view and USB tethering are the most important features. It's possible to image without them, but assisted focusing is vastly easier than eyeballing it in the viewfinder. Another improvement is in power consumption. Despite the high performance of Canon's latest DIGIC chips, battery life seems to be much better compared to older models like the 5D.

[Jon Rista]

You might be missing some of the physics. Newer cameras don't simply give you "ISO 12800" as just a higher setting. Newer cameras actually have better sensors capable of gathering more light in each pixel. Newer cameras can often gather more light with SMALLER pixels than older cameras, even..

 

How? Higher quantum efficiency. The use of microlenses. The use of weaker color filter arrays. Lower noise electronics (more photons are "saved" from noise, by having lower read noise and lower dark current), etc. 

 

The 40D and 450D both have around 30% Q.E., neither use microlenses on their sensors, both have stronger color filter array filters. Both have higher read noise at ISO 800 (5-6e-). I don't know what the dark current is, but I suspect it is much higher than modern sensors.

 

In contrast, the 7D II for example has nearly 60% Q.E. The A7s has 65% Q.E. At those quantum efficiencies, combined with the lower noise, the OOC SNR per-unit-area per-frame is at a bare minimum twice what you get with the 40D or 450D.  The 7D II has lower read noise and higher FWC at ISO 800 than either the 40D or 450D. The A7s has MASSIVE dynamic range at ISO 800 with a FWC of over 20ke-, with RN less than 6e-. The A7s has more dynamic range, ~4e- RN and a 5400e- FWC at ISO 3200! That's better than the 7D II at ISO 800, in every respect. Combine all the technological improvements, and despite smaller pixels, you could have significantly more than twice the SNR (at midtone gray) than the 40D or 450D. 

 

The Samsung NX1 has the most advanced APS-C sensor on the market, using a BSI design for high sensitivity, with very low read noise, fairly low dark current (might not be quite as low as the 7D II, no concrete mesurements yet). The NX1 also has the most advanced DSP (sometimes called ISP, image signal processor) on the market, with a reconfigurable hardware matrix that allows hardware programmability. Samsung uses it for focal plane phase detect focusing logic...but astrophotographers could potentially use that matrix for any number of useful purposes. The NX1 sensor could be the beginning of a new wave of cameras to the market, as it's a mirrorless, however it has a standard DSLR ergonomic (albeit in a smaller, lighter body....good for astro.) 

 

So, yes. In my opinion, newer DSLRs are vastly superior to older DSLRs just on a raw sensor technology/OOC SNR basis. Newer cameras have plenty of other handy features for regular daytime photography. 

[mmalik]

Orly, bulk of the new camera research here... and here...; these discussions mostly boil it down for you. Ongoing talk here.... Regards

Edited by mmalik, 23 January 2015 - 11:23 PM.

[jgraham]

A modified 450D is a very capable camera. I have a Baader modified 450D and I frequently use it as a backup to my Baader modified 550D. The 450D is also my photometric camera. Low noise, 14 bit ADC, and relatively large pixels make for a very nice camera.

[SKYGZR]

I'm "sold" on the 500d (T1i)..relatively inexpensive, moderate pixel size, might be a bit "old school" for the APS-C sized sensor, yet "relatively" low noise, and capable ..(a bit of a step up from the earlier versions, not the greatest, yet "well rounded")...

[garret]

Would a much newer camera be better?

 

 

  

Edited by garret, 24 January 2015 - 03:02 AM.

[orlyandico]

Wow I didn't know the NX 1 had a back illuminated sensor. That's a biggie.

As for double the QE... If you have half the pixel area then you're right back where you started, at a given focal length. Of course the smaller pixel camera would give more detail (smaller pixels) so more information assuming the number of electrons per pixel is the same (which it won't be, given the smaller pixels).

Anyway I'll have a look at the NX 1. Sounds like a good idea.. What does it cost though. The CFO has been wanting a 6D for a long time now. But full frame for astro is a pain in the arse. I don't want to buy an FSQ 106 just to cover that big sensor.

[Jon Rista]

Orly: You are correct about per-pixel sensitivity. I tried to be clear that with double the Q.E., the per-unit-area sensitivity doubles. I used that term to differentiate from per-pixel sensitivity. For one, I don't think pixel sizes have quite halved. The 450D had, 5.1 micron pixels (? don't have mine handy), the 7D II has 4.1 micron pixels. That's an area factor of 0.65x. In terms of per-pixel sensitivity alone, the 7D II gains at 59% Q.E., and loses at 0.65x the area. That gives the 7D II a 27% advantage in overall light gathering capacity. That only factors in the increase in Q.E., though. There is also the reduction in read noise. Even better, there is the MASSIVE reduction in dark current (the 7D II has one of the lowest dark current rates of any digital camera on the market, apparently even 10x lower than the 6D, which was already pretty low.) The higher Q.E., reduced RN, reduced DC, all lead to a significantly higher SNR per-sub out of the camera.

 

Regarding the NX1, body only is $1500. Not a bad price, but certainly more than a used older Canon DSLR. I've got some data from an NX1. Personally, I think it's VASTLY superior to any Canon data, particularly at low ISO. At higher ISO the gap narrows, but I prefer the very neutral kind of data strait out of the camera. The 7D II still has that fairly heavy red shifted data in their RAWs, and it STILL has banding (even though it's lower). I saw no hint of banding in the NX1, and minimal hot pixels. 

 

I don't know if anyone has used one for astro yet. At the moment there is not an SDK, however Samsung seems to be fairly on the ball with firmware updates, and the SDK is supposed to be coming soon. It's supposed to be a very rich SDK, I gather that Samsung is taking direct aim at Canon's 7D line of cameras (and the specs of the NX1 seem to back that up...I'm quite impressed with it specs wise, and can't wait to rent one and give it a whirl for birds, wildlife and astro.) I have very high hopes for the NX1, however personally, I feel it's just a little too soon. I don't know that Guylain would be interested in investing time and resources in something like BackyardSamsung any time soon, not with less than a handful of people using any of those cameras for astro. I do think Samsung is a solid new player, and if they keep up their current momentum (which actually includes a 300mm f/2.8 lens that seems to rival Canon's own prime superteles, again at least specs wise), I think they could be a very, very solid contender for top Astro APS-C camera.

 

If you are really interested in it, I would check out LensRentals.com. I already verified that they have NX1's in stock for rent, and the cost is pretty decent. I plan to try one out myself, along with a Canon EF adapter and maybe one of their longer zoom lenses, once bird migration season starts again. It's the top camera on my list at the moment...it really, REALLY intrigues me.  

 

Regarding full frame...I think you could use more than just the FSQ106 with it. Any of the Edges have a flat 44mm field, and many of the AstroTech RC's do as well (particularly the larger ones). Personally I use my 600mm f/4 canon lens with my 5D III, although at that price your better off with the FSQ. I don't know of any of those other scopes are going to have as flat and unvignetted field as the FSQ106...it's base field is 88mm, so the 44mm center field should be pristine. I don't think I can even say that about my 600mm f/4....

 

Anyway, for APS-C cameras, personally I think the top dogs are the NX1 or the 7D II. The former is actually even cheaper than the latter, but it's not going to have much in the way of software compatibility. (Also, at the moment...I think the current firmware only supports 30-second bulb, but given how programmable the camera is, and given how fast Samsung already came out with firmware 1.2, it may not remain that way for long.)

[Tonk]

A modified 450D is a very capable camera. I have a Baader modified 450D and I frequently use it as a backup to my Baader modified 550D. The 450D is also my photometric camera. Low noise, 14 bit ADC, and relatively large pixels make for a very nice camera.

If you are not sold on one of the newest cameras then I can vouch for the 450D - its my current workhorse and is excellent

[orlyandico]

That settles it.. If a 10u user is happy with a 450D...

With my current ancient camera, sky fog overwhelms all other factors. Though I suspect the better read noise of a newer camera would be of benefit.

I don't see how the 7D II can have 10x lower read noise though. Electrons are quantized, and 10 e- read noise is quite high. I can't see 0 e- read noise...

[orlyandico]

..  I went through one of the other threads.  So the new Samsung sensor has 1.6 e- of read noise.

 

This table seems up to date -

 

http://www.astrosurf...lli/strum43.htm

 

the 60D is 1.99 e-  and is roughly the same age as the 7D.  So how can the 7D II have 10x lower read noise?

Edited by orlyandico, 24 January 2015 - 09:44 AM.

[josh smith]

While there certainly have been advances in newer cameras, what really matters is if you'll be happy with the results.  I can vouch that the 450D and or T3 were excellent astrophotography cameras for me for the price.  Search those cameras or setups similar to yours on Astrobin and see if there are any results you'd be happy with.  If there are, then you know it can be done and you would be pleased with those cameras.  You can see some of the examples on my astrobin page and Tonk sure has some nice work in his portfolio.

[jgraham]

Just for yucks, we get so few clear nights here that when it is clear I image like crazy and hopefully I get enough source images to keep my busy while I'm waiting for the next clear night. One of the setup that I use is a pair of nearly identical SN6s one two Orion Atlas mounts. I originally used a monochrome CCD on one and a color CCD on the other. I now use my 450D on one (right) and my 550D on the other (left). The images from the 450D are a tad quieter and go a tad deeper than the 550D, but the 550D images bin 2x2 better since it has a higher pixel count. I use the 450D for photometry and camera-assisted observing. The smaller image files transfer and process faster than the larger files from the 550D.

 

 

What a fun time to be an amateur astronomer!

 

[DonR]

I have a 350D (2005) and a T2i (2010), both unmodded, and I used both of them for a while when I first got the T2i.  Despite the newer sensor technology, the T2i does not produce noticeably better images than the T2i in my subjective opinion.  Yes, the T2i can operate at ISO 12800 with "ISO expansion", but the sweet spot for both cameras in my experience is ISO 400 to ISO 800.  The T2i may have a slight performance edge at ISO 800, but it's certainly not dramatic. 

 

The T2i has a higher dynamic range (14-bit ADU vs. 12-bit ADU), but in practice this does not make much difference for long exposure AP in my opinion.

 

The 350D exhibits "amp glow", a red glow in the lower right corner of long exposure images that is most noticeable on cold nights.  The T2i does not exhibit any amp glow, and I understand the same is true of the 40D.

 

On hot summer nights down here in the southern US, the sensor temp of my T2i can easily approach 100 degrees Fahrenheit - it stays around 10 degrees F above ambient during a sequence of long exposures.  I find it helpful on these occasions to reduce the ISO in order to keep the thermal noise under control.  The 350D does not report sensor temperature, but I suspect it stays cooler than the T2i, as I don't see such a dramatic difference when the ambient temperature is high.

 

Noise in calibrated light frames is a toss-up in my experience.  The 350D, with pixels more than twice as large as the T2i pixels, does a pretty good job of controlling noise.  Qualitatively I would not give either camera an advantage regarding noise in long exposures.  Of course the much larger pixels in the 350D means there are many less pixels on the sensor, but 8 million is enough for just about any amateur AP application.

 

The T2i does have some features I use and appreciate that are lacking on the 350D.  First and pretty important to me is the ability take long exposures with PC control over USB.  The 350D requires a remote shutter control cable in addition to a USB cable for PC control of exposures over 30 seconds.  Since it has a DIGIC III processor like the T2i, I believe the 40D will have USB control of long exposures too.

 

Live View mode on the T2i is very useful for focusing with a Bahtinov mask, as is the expanded 12800 ISO feature.  The 350D lacks live view and tops out at 1600 ISO.  Video mode on the T2i can be used for lunar and planetary imaging, with optional 5X zoom built in.  The 40D does provide Live View mode, but does not have a video capture mode like the newer EOS cameras.

 

To sum up my experience with these two cameras, finished image quality is a wash in my opinion, with the exception of amp glow, which usually requires cropping to completely remove from images acquired on cold nights with the 350D.  The newer camera, though, provides some very useful features that give it the edge in the final analysis, IMHO.

 

Don

Edited by DonR, 24 January 2015 - 10:07 AM.

[DonR]

P.S.  I want to add to the significant feature enhancements in newer EOS cameras the built in ultrasonic sensor cleaning feature.  This really works well.  I have manually cleaned the sensor on my 350D numerous times, each occasion being a tense and tedious procedure.  I have never found it necessary to clean the sensor of the T2i.  Flat field images show no "dust motes" - a very welcome improvement.  The 40D does include the ultrasonic sensor cleaning feature.

 

Don

[DuncanM]

The Nikon D7000/D5100 has the Sony 16mp sensor, and with the modded firmware available from Nikonhacker it becomes a very astro capable camera.  

[alan.dang]

http://www.clarkvisi...ion-canon-7dii/

The 7D Mark II isn't that great for read noise

 

(But it is great for astrophotography -- look at that M45 shot!)

 

-----

The NX1 does seem very good though.  I think those read-noise numbers came from my article

www.slrlounge.com/samsung-nx1-vs-full-frame-cmos/

 

I'm in a white zone, so it'll be interesting to see how it handles at a dark site.

[Jon Rista]

The 1.6e- RN on the NX1 is at what ISO setting? If it's at ISO 800 or 1600, that is REALLY freakin good. Better than I'd imagined. Better than just about anything else out there. 

[orlyandico]

Yes, but... 1.99e- for the 60D isn't too far off.

 

My 40D has USB control and the ultrasonic sensor cleaner and live view.

 

I saw this very interesting device on Yahoo Japan... external DSLR cooler for Canon cameras with a tiltable screen. Basically you tilt the screen out of the way so that the cooled finger can be right up against the chassis next to the sensor.  Not as good as a true cooler mod, but it's totally reversible.

 

 

That for me is an additional advantage... when I use ByEOS even with 1-minute exposures on the 40D, after an hour the sensor temp is around 37-38 Celsius (100 F). So any type of cooling would be useful.  I just don't know where to find this cooler other than Yahoo Japan auctions.

[SKYGZR]

L00Ks like quite a bit of weight hanging off the cam...not really desirable..

[Jon Rista]

Orly, where are you getting your read noise numbers? I'm not finding those values anywhere on the sites I usually visit. I have never heard of any Canon camera having less than 2e- RN at ISO 800 or 1600. According to sensorgen.info, it's over 3e- at 800 and just under 3e- at 1600 for the 60D.

As you get to higher and higher ISO settings, the difference between 2e- and 1.6e- is actually quite significant, at least from a dynamic range standpoint (it could mean a stop or two even.) It's probably irrelevant for subs of similar exposure length, though...

[Samir Kharusi]

I have never quite grasped why people have the suspicion that camera makers, like Canon, would go to the bother and expense of marketing a new model if they do not think that they have made tangible improvements... Anyway, as regards dark current from 3 generations of Canon DSLRs, regular histograms from 20 minute darks, ISO 1600 at top, vertically expanded below them:

 

Frankly, IMHO just about any Canon DSLR from the 20D onwards, when modded it will deliver superior astro images overall compared to an unmodded newer model. Modding is by far the best astro enhancement one can make to any of the consumer cameras. Since the modding involves a significant commitment of time and or $, I would always be tempted to jump in with the latest Canon. Other makers can come in after the modding aspects have been sorted out, and there does seem to be significant recent efforts at branching out from Canon. Good. Frankly, with the latest models, dark current is no longer the big discriminant, IMHO. For me, Read Noise and pixel size to match my optics are more critical. I yearn for < 4 microns for Hyperstar use, but refractor owners are better off with > 4 microns. Why sacrifice pixel capture area if you can still sample at 2 pixels/FWHM with larger pixels?

[liors]

Just get a Canon 6D, this camera has a newer chip/sensor, lower sensitivity to noise, HDR, high resolution of 20.2mp and pixel size of 6.55um.

Wonderful full frame camera and the best for AP today that I know, of course it`s better to get it modified for AP.
Camera in not cheap but worth every Dollar.

[DonR]

If you like to consider statistical data, tables or plots when evaluating a camera for AP, don't be fooled into thinking that less dark current is necessarily better.  Dark current is signal, not noise.  The bulk of the dark current is an offset intentionally added by the manufacturer to assure that the signal never falls to zero.  The offset signal is easily removed by bias subtraction. 

 

Like all digitized signals, the dark current signal contains noise - generally, the higher the dark current signal, the more noise it contains, but higher dark current signal means that the signal-to-noise ratio of the dark current is higher as well, assuming thermal noise and read noise are similar.  If you plot dark current as magnitude vs. number of data points at each magnitude (in other words, a histogram), and if you zoom in far enough to see it, the noise will appear like "grass" riding on top of the signal.  Samir's bottom row of histogram plots above shows the dark current noise, which is evident in all three histograms on the bottom row, but appears to be slightly higher relative to the signal level in the 500D histogram than it is in the other two.

 

If your image exposure times are long enough to achieve good results, by far the most abundant type of noise in the images is shot noise, caused by inherent variations in the rate of photons reaching each pixel of your sensor over time.  This noise increases with increasing exposure time, but the signal increases faster, so longer exposure times means higher signal-to-noise ratio.  Besides increasing exposure time, another way to increase the signal-to-noise ratio is to make the pixels larger, so that more photons contribute to each pixel of the image.  Larger pixels in the same sensor size means fewer pixels.  An 8 megapixel camera will produce an image with higher signal-to-noise ratio than an 18 megapixel camera, assuming the same quantum efficiency, the same sensor size, the same optics and the same subject and imaging conditions.  As pointed out above, the quantum efficiency of newer cameras is higher than that of older cameras, but that just means that more amplification is needed to produce the same digital signal level in the older camera - ISO is a standard, and new and old cameras produce the same digital signal level with the same ISO setting and the same exposure time of the same subject through the same optics.  Increasing the amplification produces more noise, but in my experience with the Canon 350D and the Canon 550D, it's pretty much a wash after calibration.

 

As Samir points out above, matching the pixel size to the optics is another consideration.  My old 350D achieves 1.3 arc-seconds per pixel at 1000 mm focal length.  With an 8 inch aperture, the airy disk size is 1.3 arc-seconds.  So under perfect seeing and tracking conditions, the image of a faint star would fit closely onto a single pixel in the 350D.  Unfortunately, I have never imaged in perfect conditions.  Until I have access to a much larger imaging telescope and an observatory at the top of the stratosphere, 8 megapixels is OK.  That said, the feature improvements that manufactures like Canon have made over the years make a strong case for their newer DSLR models.  I rarely use the 350D anymore.  I would like to see a modern DSLR with an approximately APS-C sized sensor and 8 megapixels or so.  Unfortunately I think that is unlikely, as consumers react favorably to more and more pixels in DSLRs.  Dedicated astronomical CCD imagers are the only way to get close to that currently.

 

Don