Here is the obligatory info: I take no responsibility for any damage one may do to their cameras by using the information contained on this site. It is important of course to realize that you will void the warranty of your camera by partaking in any of these mods. All information on this site is copyright (c) Mike Kudenov 2006 unless otherwise noted, and may not be directly utilized for profit without permission. Now, on to the mods...
8/31/09: This weekend I was fortunate enough to find some time to continue work on the 350D mod. I added a peltier cooler with a cold finger assembly to the camera, similar (conceptually) to a commercial unit I saw a while ago - though it does not have many (if any) dew/condensation prevention features yet. Ultimately, not the most efficient way to transfer the energy - but it works. In any case, here are some darkframes from the system as it stands,
I anticipate that the sensor is close to 40F. Note that before, the refrigerator was at 40F, so the previous darkframes probably had a sensor temperature higher than that. These frames were processed from Raw in Iris, then saved to a bmp before compression as a jpg in photoshop. I'll post some raw data once I get the amp-off circuit soldered to a prefboard. Currently, some extra noise, in the form of horizontal lines, are present within the amp-off image. This is likely due to the proximity of the power supply/peltier to the breadboard/amp-off circuit. I am anticipating it will vanish when everything is soldered up.
Here are some photos of the cooling. It's definitely a quick and dirty job, since I've only spent about a day or two on it.
6/23/09: It has been a while since I last updated anything here, but recently I have attempted to modify the Canon 350D Digital Rebel with a similar amp-off modification. Currently, the circuit is on a breadboard connected to the sensor via. leads, as depicted below. Placing the whole setup in my refrigerator (~ 40 F or 4.4 C) yielded the following darkframes:
These are preliminary results. I won't have final results until I can cool the sensor independent of my refrigerator. Also, I want to mention that I have other photos posted that I have taken with the 300D on my main site. More to come.
11/6/07: Some more pics I took last weekend and a few weeks ago with the camera and the R200SS: Orion nebula, Andromeda, Pleiades, Comet Holmes, Pelican. The pelican nebula could have come out better, but there were temperamental clouds that night which limited the exposure time to an hour.
5/31/07: Here are some shots of the lagoon nebula and the trifid nebula taken from Las Cinnegas (about 40 miles south east of Tucson) a few weeks ago. 10 minute exposures on a Vixen R200SS 8'' F/4 at ISO800.
1/15/07: Here are two more photos taken this weekend; more h-alpha shots taken with the modified camera on the ED80. First is the horsehead. It's a stack of about 4 30 minute ISO800 exposures with some averaging done between them: Horsehead. The next is the Rosette Nebula and is composed of two 30 minute and one 45 minute ISO800 shot: Rosette.
1/9/07: I just recently received word from Filip Lolic of his implementation of the amp-off circuit using SMD devices; very compact. Additionally, I have another image I took just this weekend I wanted to share. It's Orion again, but this time with a Baader 7 nm h-alpha filter (from alpine astro) using the modified 300D on an ED80. The outer region is ISO1600 at 45 minutes, then the exposure decreases as one approaches the core. It can be seen here. This image is fairly raw. I only used curves and never got around to doing any darkframe subtractions on it.
10/24/06: Over the last month, I've been able to get some more data
(e.g. real results). I've used both my MK66 and my Orion ED80 on a variety of
objects. Below are some un-processed images taken in
light polluted skies with no LPS filter. These are full sized PNG's. The first two have tracking error
because my guidescope was way misaligned before I realized it, but the rest are fine.
9/26/06: Here are the measurements I obtained while probing the CMOS sensor. Some have asked so here it is. Just simple voltage measurements taken with a DMM. This simple datasheet for the 704F can be downloaded here. Please be aware that all the data in the "notes" column is just my best guess and may be completely incorrect. Additionally none of my notes on the rough transient response are included, but I may add those at a later time. Ultimately it'd be best to get o-scope readings, but I never had to go that far to get where I wanted with this mod.
Back in March I set out something I've wanted to do for quite some time: modify a digital SLR in a way that'd make it a dedicated camera for astrophotography. I'm not going to go through a lot of disassembly details here (other sites go through disassembly, of particular note is Gary Honis' site), but I do plan to post as many results as possible.
First, I wanted to remove the IR blocking filter from the camera, as many others have already done. This increases significantly the effective responsivity of the camera to the H-alpha wavelength. Once I removed this filter, I measured its spectral response on a spectrometer I have access to at work, which yielded the following results,
For the above data I was forced to utilize an incandescent lightbulb as a reference. Unfortunately, these bulbs have lower output in the blues / UV than other sources, hence the reason for the lower SNR near these wavelengths. However, it at least provides a reasonably accurate quantitative description. As can be seen, the filter offers about 25% transmission near the H-alpha wavelength (~656 nm) and cuts off very smoothly starting from 550 NM out to 675 NM
The next issue I wanted to address is the thermal noise associated with long exposures. This requires a cooling system, and in hindsight it appears I went a bit overboard on mine, since I was coming from the grounds of CPU processor cooling with peltiers. To see pictures of the original design with its enormously oversized heatsink, check: side-view, back-view, front-cmos. In this design, the cmos sensor is mounted out infront inside a cold chamber that contains desiccant. Otherwise, it is at the same pressure as the outside atmosphere. As long as one doesn't change elevation too severely, this configuration works very nicely to keep the moisture out.
From this, while it was obvious that the heatsink was doing its job (as can be seen, ice is present on the coldplate in the images) it was way overbuilt and extremely heavy for what was necessary. Furthermore, I wanted to get rid of the small flex connectors that can be seen inbetween the two cables, which meant making the heatsink a little shorter. Doing this yields the current design (9/10/06), which can be seen below,
Now, as can be seen above, the cables are contiguous until they arrive inside the cmos' chamber (using 8 inch long flex cables, Digikey part #WM10245-ND). There is one male to male flex adapter (as is discussed in the resources section) present inside the chamber in the front. Shielding on this cable uses a copper tape purchased from Digikey (part # 3M1181B-ND) along with aluminum foil, insulated with electrical tape. Shielding these cables is very crucial and their connection to ground must be firm and their physical stability must be high (they can't be allowed to move significantly when prodded), or noise may be present on the final image. So far, from the testing I have done, this configuration proves more effective than my previous one (where I had two flex cables of differing length connected by an additional male-to-male flex adapter). While the previous configuration allowed a better selection of lengths, the additional flex adapter caused problems that were non-periodic and hard to pin down.
The last issue I wanted to address with this mod was that of the amplifier. It turns out that when one cools the CMOS chip, the amplifier noise becomes much worse. This is likely due to biasing inside the sensor changing with temperature, making the internal amp-off circuitry behave less effectively. This is just my best guess from all of my experimentation with the chip, and it may be incorrect. In any case, the bottom line is with the amp off, I can extend exposures out to one hour without significant issues.
For instance, here is a 15 minute ISO1600 default 300D uncooled darkframe (from the cmos sensor mounted in the chamber, but the peltier turned off) and a 15 minute ISO1600 default 300D cooled darkframe. The cooled darkframe has less thermal noise, but more amplifier noise, than the uncooled one. However, with the amp off mod enabled the difference is quite noticeable. Here is a 15 minute ISO1600 amp-off mod 300D uncooled darkframe and finally a 15 minute ISO1600 amp-off mod 300D cooled darkframe. With the amp off mod, the amplifier noise is difficult to spot without inverting the colors in the frame.
Ultimately, even with my previous setup with the additional flex adapter, I was able to obtain these images off my telescope before the monsoon season started. These are unprocessed pics. They are unprocessed to demonstrate the (what I consider low) noise levels more clearly.
M27, 20 minutes, ISO1600, below freezing, amp off. IDAS LPs filter, MK66 at F/12, unprocessed (some tacking error but not horrendous).
M101, 30 minutes, ISO1600, below freezing, amp off. IDAS LPs filter, MK66 at F/12 unprocessed (tracking is poor on this one, leading to poor contrast in the object. There's also an internal reflection in the lower right, but it shows that 30 min exposures and probably longer are easily possible).
The air temperature at the time was 80 degrees F. The CMOS sensor was around 20 - 25 degrees F.
Finally, here is a set of new 15 minute cooled darkframes taken with an air temp of 78 degrees, cmos temp at around 28 degrees F. These were taken using the single pin amp off mod per the circuit in the amp-off mod section; hence, it appears that the previously aforementioned column defect is not present despite the single pin switching. Rather, as was mentioned before, the error was being caused by a cable grounding issue that has since been resolved by utilizing a single contiguous cable and different grounding techniques.
300D, 28F, ISO1600, 15 MIN_1 Dark (5.8 MB)
300D, 28F, ISO1600, 15 MIN_2 Dark (5.8 MB)
300D, 28F, ISO1600, 2 SEC Dark (5.6 MB)
10D, Room Temp, ISO1600, 1 SEC Dark (5.6 MB)
Finally, here's a picture of the camera mounted on my MK66. Reducing its size from previously by making the heatsink smaller has vastly improved the handling performance on the mount. In the next few days (hopefully by sept 15th) the monsoons will be over, and I'll finally be able to get some solid clear skies for which to test this system out more thoroughly.
Contents (c) 2006 Mike Kudenov unless otherwise noted and may not be used for commercial purposes for profit without consent.