Canon Digital Rebel 300D Astrophotography Mod

Amp Off Modification

Updated 9/10/2006

Home

Resources

Amp-Off Mod

As always, when embarking on any wild modification such as this, be sure you know what you're getting yourself into, and that you could easily damage your camera in the process. I've personally been pretty brazen with my CMOS sensor, and it's held out fine, but of course be sure to keep yourself continuously grounded to avoid that mysterious "it doesn't work and I don't remember doing anything bad" dilemma. And of course I will reiterate: I am not responsible for anything you do to your own system. All I know is that this mod worked for me, and it should "technically" work with any other canon camera using a 704F CMOS imaging sensor. There is even a strong possibility that it could even be extended to the newer 20D/350D or 30D.

In any case, here is where I'll discuss a little about the amp off modification I implemented on my 300D. It may not seem overwhelmingly necessary, as the amp noise isn't really all that terrible. However, as I said before, if the CMOS sensor is cooled to around 28 degrees F or lower (approximately), the amp glow becomes increasingly obtuse. I am not sure of the cause for this; all I knew at the time was if I wanted it to go away, I had to find another means of turning off the amp circuit. It turns out that this really isn't terribly difficult to do on this camera, although it did take a lot of time to nail down. Ultimately, it turns out that no external timing devices are required for this amp-off circuit (eg timers, microcontrollers, or other circuits that could synchronize the amp-off circuit with the current exposure). This is because the 300D already has a way of disabling the onboard amplifier during an exposure, which is one reason why it fairs much better than the comparable offerings from Nikon, which lack this feature (although I'm not sure about recent Nikon DSLR's). Hence, it is possible to tap into this already existing signal to trigger the tertiary amp-off circuit that I present here.

The circuit I have developed was done with components I had lying around. Therefore some of the components may not have been the most ideal for the task at hand. In any case, here is the amp-off circuit schematic.

The circuit above consists of seven resistors and three transistors, and can be divided into 3 stages. This first stage consists of the NTE2985, R1, R2, and R4. It is used for impedance matching purposes between the rest of the circuit and the 300D's amp off signal (located at Pin 27B). This first stage switches the biasing for the second stage, which uses a 123AP or 2N2222 NPN BJT in conjunction with resistors R3, R6, and R7. This second stage is used to bias the third stage, which uses a 2N3906 PNP BJT wired in parallel with a ~630 ohm resistance to provide a reduction to the CMOS sensor's voltage supply during an exposure. Basically you need to chose a resistance for R5 to give you approximately 3.7 V at the CMOS-side node during an exposure, and about 4.9 V to 5 V at readout. I'm not really going to go into much more detail than this, as the circuit is pretty self explanatory. It's very similar to Steve Chamber's original Toucam amp-off modification, except this one limits the source as opposed to the ground (ultimately the same thing). The only problem is there's about two billion ground connections on the 300D's CMOS sensor, and four connections to a +5V source, so it makes the latter easier to deal with. Additionally, I realize that there are more sophisticated amp-off mods that have been developed since then (using say, zener diodes), but this one is relatively straight forward and simple to implement on a sensor you have no datasheet for (or at least a datasheet you made yourself from meticulous probing).

In any case, I want to mention a bit about this MOSFET before going on with the rest of it. The NTE2985 is not necessarily the most ideal component to use in this circumstance, considering its size,

However, any similar N-channel enhancement mode digital (5V Vgss) MOSFET should work, but if you use a different transistor, I'd double check that a +3V signal at the gate will cause saturation of the device (e.g. if you place a 1k resistor between the drain and +5V, and connect the source to ground, then apply +3V through the 470K resistor at the gate, then you should read close to 0V across the transistor (Vds). If not, a different transistor is in order, or different biasing may be required). But I'd imagine that any digital MOSFET should be able to achieve this assuming the designers are aiming for a threshold voltage of around 1.5V.

Now for the pin locations. To make this circuit work you need to connect to the pins indicated in the following pictures, (click for a larger versions)

View of the backside, top region of the chip

View of the backside, bottom region of the chip

As can be seen, there are 4 connections to be made. Out of these, one pin must be lifted to allow one to connect a wire to the CMOS side and CAM side of the system. Furthermore, 3 additional pins must be disconnected. These are the additional voltage source pins (as said previously, there are 4 total). I tested fairly rigorously and did not notice any qualitative difference in image quality when I disconnected these 3 pins; they all appear to be connected to the same node, so I found no issues with it. However, it is completely possible to switch all 4 nodes simultaneously if one discovers a problem with it.

Ultimately, I only lifted one of the pins. They are very strong pins, and it was difficult to get even one of them lifted enough to fully disconnect it from the pad (just solder sucking will not disconnect the pin). So I decided to disconnect the other 3 voltage source pins by cutting into my flex cable with a razor blade. I simply removed a small portion around the lead that connected the voltage source to the camera. This way I have the option of disabling the mod by switching a cable, but I don't see myself doing that in the near future :)

If something is unclear, let me know and I will try to add more explanation.

Contents (c) 2006 Mike Kudenov unless otherwise noted and may not be used for commercial purposes for profit without concent.