Raspberry pi camera (DIY)

[martinus]

(Please note that this thread was previously posted on IFAS and is reposted here at David's (bigeye) suggestion)

A friend very kindly let me experiment with his raspi camera (including removing the glued-in lens). Here's a quick overview of how it went:

1. Taking off the lens required scraping out the glue that locks the lens housing into the sensor housing. The IR filter is built into the lens housing.


2. I needed some way to mount the picam onto the telescope so I cut a square aperture into the lid of a film canister ensuring it was a snug press-fit. (I put the lens back in to keep dust off the sensor)



Here it is in the ultra high-spec 'housing'


Equally high-spec mounting was employed for the pi:



3. Took a couple of shots using raspistill (ISO 400, wb off):



Seeing was moderate with a fair amount of atmospheric disturbance.

Overall I'd say that the raspicam has potential as a cheap astrocam. However, the lens mount doesn't seem to be one of those standard M12 webcam mounts so it isn't compatible with the widely available webcam adapters and thus one cannot easily use a screw-on 1.25" IR filter.

[martinus]

I previously mentioned that the picam does not have an M12 mount (standard webcam lens) which forced me to modify a film canister lid. It works but is a friction fit and one cannot easily attach a filter to it, thus it is not ideal. In an attempt to improve the camera attachement I found an old webcam* and stripped it down. There was a detachable M12 which I tried to retrofit to the picam board. As you can see from the previous images the PCB has four mounting holes - two are in line with the sensor. Alas, it turned out that the hole spacing was not compatible with the M12, the picam board has a spacing of 21mm.

I then took a look on ebay and discovered that someone was manufacturing M12 mounts with a 21mm screw spacing! Not only that but they're only a few pounds (PM me for a link). The seller even advertises the M12 mount as being picam compatible. However, this is not exactly the case. You can see the new and scavenged M12 mounts in Figure 1. Note that there are openings on two of the sides of the scavenged M12 whereas the underside of the 'compatible' M12 are flat (Figure 2). Unfortunately this mount does not allow for the sensor attachment method used on the picam. Figure 3 shows a detached sensor and the picam board; the sensor is held on to the PCB with a small circle of double-sided tape, electrical contacts are made with a small, press-fit connector. Whilst the ribbon cable is thin the 'compatible' M12 mount does not have an opening on the base that accomodates the press-fit connector, this is evident in Figure 4.



Figure 1. A picam with a scavenged, incompatible webcam M12 mount (R) and a mount with the correct screw-hole spacing (L).


Figure 2. View of the bottom of the 'compatible' M12 mount.


Figure 3. Picam board with detached sensor.


Figure 4. M12 mount placed on the picam PCB. Note the angle between the mount and the board caused by the mount sitting on the sensor connector.


Turning the 'compatible' M12 into a compatible M12 was trivial - I cut a small section of the base away as illustrated in Figure 5. The resulting fit (Figure 6) was good and I was able to attach the M12 to the PCB with two small screws**. Whilst I was modifying the M12 I also cut a section out of the adjacent side of the base, I have a peltier cooler on order and am aiming to put a copper cold-finger on the underside of the sensor at a later date.



Figure 5. Base of the M12 mount showing the cut-away section.


Figure 6. Modified M12 mount attached to the picam PCB.


I attached the 1.25 inch adapter to the picam (Figure 7) and took it for a test-drive. The sky was a bit hazy and I'm not convinced I got good focus however I did take a few still shots of Mars (ISO 1600, auto white balance) just to ensure everything was working. An example of which can be seen in Figure 8.



Figure 7. Complete, focuser compatible picam.


Figure 8. Test photo of Mars.


As a bonus I was driving the mount using INDI (indi_eqmod_telescope driver) from the raspberry pi via a USB to serial cable I built. The raspberry pi was running as a server and Kstars on my laptop did all of the heavy lifting. Whilst kstars is not as feature filled as EQMod it is perfectly feasible to slew and move the mount and the tracking appears to work well.



* The old webcam had low sensitivity making it unsuitable for pretty much anything including astrophotography.
** The M12 mount I purchased was not supplied with screws, I used two spares I had lying around.

[keithm]

That's very interesting Martin and thanks for that.
I've been interested in the Pi for a long time but never really had a good use for one until now. I think I definitely have to give this a go.

[martinus]

Thanks, keithm.

It has a lot of potential for astronomy and is very cheap.

To date I have managed to control the mount successfully using INDI (http://indilib.org/), automate my Canon DSLR and obviously take photos and video with the pi camera.

As the board has a number of I/O (input/output) pins it would be straightforward to add sensors for cloud cover detection, servos to drive the focuser and even a magnetometer (compass) to aide in set-up if one were using a portable mount. OpenPHD has been successfully compiled for the raspi so autoguiding is also potentially feasible.

Just have to get enough time to experiment with it all.

[keithm]

I've read a lot about controlling telescopes with a Pi since you posted this and I'm seriously tempted to build a goto system for my EQ3-2.
I deliberately decided not to get it with the official goto system because I want to learn the night sky myself but this project sounds too interesting to pass up on!

I'll need the dual-axis motors, a Pi and SkySafari app that can connect to the Pi to control the motors. The tricky part is the interface between the Pi and the motors but I've seen a number of solutions for that. I just need to sit down to read, compare and select the best one.