The Cameras

So far in my whole astrophotographic career, old and new, I have had just three (with a fourth entry in 2019 and a fifth in 2020) cameras.

One, the first Meade LPI (Lunar and Planetary Imager)

MeadeLPI

Image sensor: CMOS Hynix Semiconductor Inc. HV7131E1 (Type 1/3″)
Image size: Diagonal 6.5mm, 5,18mm x 3,90mm Effective pixels: 644(H) x 484(V) ~310K pixels
Total number of pixels: 648(H) x 488(V) ~320K pixels
Unit cell (pixel) size: 8.0m(H) x 8.0m(V)
Exposure: 1/1000s – 16s (1ms – 16000ms)
Max. Resolution: 640×480
Maximum usable frame rate @ 640×480 resolution: 3fps (not compressed)
Interface: USB

was not exactly considered a technological marvel even in its heyday (the early 2000s), when modified webcams like Vesta and Toucam Pro were generally considered to be superior.

But I was in love with Meade at the time and did not have the wish to fiddle with webcam modifications and so I bought a LPI for about 200 € in 2006.

At the time I took some decent image of the moon, Jupiter and Saturn with the Meade 2045D and the Meade ETX-80 and more recently (2017) a few more pics of the moon with the Skywatcher MAK-90, before buying the DSLR.

Always back in 2006, I bought my second camera, a Meade DSI Pro, first generation.

I paid 350€ on ebay for a second hand one and it was not a bad deal.

At the time the Meade DSI (both the color one and the monochromatic Pro) were the first “low-cost” dedicated cameras for DSO imaging and aroused a certain interest in the AP community, as one can read in this Cloudy Nights Report

DSI-Pro

CCD Sensor Sony ICX254AL
Pixels 510 x 492 pixels (250,000 pixels)
Pixel Size (in microns) 9.6 microns (W) x 7.5 microns (H)
A/D Conversion 16-bit 
Min. – Max. Exposure Time 1/10,000 of a second to one hour
Sensor Size 4.8 x 3.6 mm (6 mm diagonal)
Quantum Efficiency 60-65 % (estimated)
Size / Weight of product 3.25″ x 3.25″ x 1.25″/ 10 oz.

My estimation of the Q.E. of the first generation DSI Pro is based on some old CN thread .

DSI-ATIK-QE

The small sensor is a pain in the keister when you are trying to center an object, the images are ludicrously small by today standards, the big pixels tend to bloat stars, but when you compare its sensitivity with that of a recent albeit entry-level DSLR, well you realize the old camera has still a saying.

If nothing else, moreover, the small size of the images makes them extremely easy to process even in big quantity for today’s PCs.

With my return to astrophotography in 2017, I thought it was time to follow the trend and try a DSLR.

I chose a Canon because they are the ones most supported by AP software, the EOS 1300D (Rebel T6) in particular because it was then the newest entry level APS-C format DSLR in the Canon EOS family, so for less than 400 € I bought one with a 18-55 zoom lens kit.

Reflex-Canon-EOS-1300D-Objectif-18-55-mm-EF-S-IS-II

Specifications according to the APT site:

Schermata 2018-07-28 alle 23.38.02Schermata 2018-07-28 alle 23.38.22

The Q.E of an unmodified (like mine) 1300D is 37%: that explains why (as we may see comparing images in the gallery) the old Meade DSI Pro is so quicker at collecting light.

But the far bigger sensor of the DSLR makes object finding and centering quite easier, it also makes possible to take wide field pictures with camera lenses and it makes easier to shot in color.

Moreover its 4.3 µm pixels (compared to the 9.6×7.5 µm ones of the DSI Pro) make imaged DSO look twice bigger (and having four times the area) than in a image taken with the Meade camera.

The small pixels also allow the exploitation of  the full resolution power of the Skywatcher MAK-90 for lunar and planetary imaging, with the simple addition of a 1.4X Telextender (1750 mm focal length).

All in all I am quite happy with my Canon EOS 1300D and as of December 2018 I have taken more than 16000 shots with it.

2019 Update

The high sensitivity displayed by the old Meade DSI PRO camera enticed me into considering the purchase of a modern monochromatic camera.

After some studying I chose the ZWO ASI 178 MM.

ASI178MM

It is a CMOS camera (CCD ones are harder and harder to find, especially at the lower end of the price spectrum), but it sports a quite high peak sensitivity of 81% and remains sensitive a long way into the IR domain.

QE178MM

When compared to the EOS 1300D it is more than twice sensitive (81 % vs 37%) both in the green region of the visible spectrum and at the H-alpha wavelength (about 45% vs 20%).

On the other end, its sensor is quite smaller than the APS-C format one of the DSLR (7.4x5mm vs 22.3×14.9mm) and that means the FOV is also about one third per linear dimension (that is one ninth of the area) of that of the Canon 1300D.

So, for instance, the Tamron 135 with the ZWO ASI 178MM has essentially the same FOV displayed by the Startravel-80 when paired with the EOS 1300D.

On the other hand the monochromatic sensor has its fair share of pixels (6.4 MP in a 3096×2080 matrix) and they are very small, with sides of just 2.4 µm.

This means that the same object at the same focal length will appear 4.3/2.4 = 1.8 times bigger on each side (and will hence cover an 3,2 times bigger area) when imaged with the ZWO ASI 178MM than when recorded with the Canon EOS 1300D.

The effect is almost the same of the addition of a 2X Barlow lens in the optical train and it is of course quite helpful in Moon and planetary imaging.

But it may help also with DSOs, particularly by allowing the photography of smaller objects (particularly galaxies, globular clusters and planetary nebulas) with the same telescope or camera lens.

It is true that the same object is spread on a wider area, but the higher QE of the sensor partially compensates that. More specifically, given a 3.2 times bigger area and a twice sensitive chip, you need “just” about 1.6 times the exposure time to obtain the same S/N ratio on an object that visually will appear almost twice bigger per side.

Moreover, the absence of a Bayer mask in front of the sensor should further improve the resolution and possibly the S/N with respect to the calculation above.

Some more immediate and less “mathematic” comparison can be made looking at the pictures of the Rosette Nebula in the gallery (reproduced below for your convenience) made with the Tamron 135 and the ASI 178MM on one side and the Bresser AR 102-XS and the EOS 1300D on the other.

Always keep in mind that the color pic enjoys a more than 1.5 times longer exposure time (75 vs 45 minutes) and a lens of more than twice the aperture (102 vs 48 mm) than that of the monochromatic picture.

NGC2238-20190224-45x60s-Ha
NGC2238 – Rosette Nebula – 45x60s with Tamron 135 F2.8 and ZWO ASI 178 MM and UHC-E filter.
NGC2238-20190301-151x30s-UHC-AR102
NGC2238 – Rosette Nebula – 151x30s – Bresser AR 102-XS and EOS 1300D with Astronomik UHC filter.

A note about binning

Traditionally CCD cameras allow a physical (“analog”) binning of the pixels, by reading them cumulatively (2×2, 3×3) through the electronics itself of the sensor.

CMOS cameras do not allow that type of binning, but a software (“digital”) one can be performed at acquisition time, if the acquisition software allows (for instance Nebulosity 4 does), or in post-processing (StarTools has a “bin” feature).

If we analyze the S/N improvement obtained with this technique (I refer to binning during acquisition, but the post-processing one does the same thing), we can see that moving from bin 1×1 to bin 2×2 acquisition mode, the CMOS camera noise is multiplied by 2 because we quadratically add the signal of 4 adjacents pixels, so RMS = SQR (4) = 2 (i.e. digital binning).

With the CCD, the noise would be the same in bin 1×1 and bin 2×2 (“analog” or “on chip” binning gives this advantage).

At the same time, the signal is multiplied by 4 in both situations.

The gain in SNR passing from bin 1×1 to bin 2×2 is therefore:

For CMOS: 2X
For CCD: 4X

CMOS is hence less efficient than CCD as regards binning, but there is anyway a benefit in improving S/N ratio through binning also with CMOS.

2020 Update

In the midst of the lockdown due to the COVID-19 pandemic, I needed some morale boost and so bought a new camera.

To be honest I had planned before to buy a dedicated OSC with more sensitivity than a DSLR and also without any built-in filter of sort.

I chose the ZWO ASI 385MC, which externally is almost identical to the 178MM, because I was favorably impressed by the ZWO camera I had previously purchased, because it has quite good specification, particularly red and infrared sensitivity, and because the size of its sensor is similar to that of the 178MM, which allows me to use the same focal reducer I profitably used with the 178MM.

385MC-QE

The camera looks very promising in the H-alpha region and it’s absolute Q.E. is in the 80% region.

And indeed it was up to expectation, providing images, for a given amount of exposure time, really not far away from those produced by the monochrome 178MM, as far as sensitivity is concerned, while resolution wise, also because of pixel size, the 178MM still reigns.

NGC3628-20200523-91x30s-2045D-F4
NGC3628 – 91x30s – Meade 2045D at F4
M51-20200620-14-160x20s-2045D-F4
M51 – 160x20s – Meade 2045D at F4
M101-20200621-122x20s-2045D-F4
M101 – 122x20s – Meade 2045D at F4
NGC7000-20200912-70x60s-85mm-Ha-lpc-cbg-csc

North America nebula, 70x60s – Neewer 85 mm – H-alpha

Towards the end of 2020, together with the new Celestron C6, I wanted also to buy a camera with a sensor big enough to cover enough field to work, say, with the C6 at F6.3 or more and at the same time more sensitivity than the EOS 1300D.

For a sum of reasons, not all of them entirely rational, I opted for buying an already modded Canon EOS 2000D from Teleskop Service.

This modded version is the one without the red filter and without substitute for it, with full sensitivity from about 410 nm to about 690 nm.

Incidentally, the removal of one filter allows the camera to focus at infinity with the Kelda 135 mm and the UHC-S clip filter, something I could not achieve with the EOS 1300D.

IC1805-20201210-121x30s-135mm-UHC-lpc-cbg-csc
Heart and Soul Nebula – 121x30s – Kelda 135mm with EOS2000D and UHC-S clip filter

Besides the extended (tripled according to TS) sensitivity in the H-alpha region, the 2000D also sports a Q.E. of 51%, against 37% for the 1300D, and also significantly reduced read and conversion noise.

Back to Contents

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s