Proposed Parallax CMUcam4 Design

Kye

Hey all,

I've finished the first draft for the Parallax CMUcam4. Attached are the EAGLE files for it and a PDF of the schematic.

I'm looking for any suggestions for extra features that I should add to this board and any suggestions about what to do with the large empty space on the board.

I am not interested in creating a separate copy of the CMUcam4 firmware for the Parallax board. So, all pins must function in a certain way by default and any modifications cannot conflict with the default functionality.

I'm thinking about putting a small through hole prototyping area in the large empty space on the board. This would allow you to attach whatnot to the camera board, like a big LED or something.

Anyway, please give me some feedback!

Thanks,

(Please look at the Eagle files before asking me to route something impossible on the board).

Thanks,

MacTuxLin

The mini breadboard like the one on PropBOE immediately crossed my mind but after looking at your board, the huge space of about 600 x 600 mil? ... thru'hole for proto-typing probably.

Kye

Bump,

(Will take any suggestions until Thursday... Then I'm sending this off to Parallax).

Thanks,

Tubular

Kye,

Firstly congratulations on what will be a really nice and useful product. There is much to like about this one. I had a day off yesterday so gave your request some thought in between painting window sills...

Regarding what to put in the spare area, a small proto area is probably the most flexible, but
1) pads for some 5mm ultra bright white or IR leds for illumination (depending on focal range and sensitivity of the camera) - would help operation in dark environs
2) a 3 pin laser pointer as per Joe Grand's board - would enable distance measurement.
3) a DIP-8 footprint using the same SPI pins as the uSD card - to accept a fast flash IC as an alternative to uSD, or:-
4) a DIP-8 footprint using the local I2C (PROP_I2C_DAT etc), for IMU sensors such as Rayman's 3AD or Vanmunch's gyro. (or Parallax's IMU boards)

This last option might give you the option to control the pan/tilt servos to level the camera, eg on a near space balloon mission, or all terrain robot

Regarding the board layout - not much to suggest other than moving the pan/tilt servo connectors together would offer more flexibility for connecting other devices that might need two data pins (eg keyboard, mouse)

cheers
Lachlan

Roy Eltham

Can you post an image or two of these, since I don't have eagle (or anything similar here yet) to view them.

Tubular

Roy, here's a pdf of the board. There is a pdf of the schematic in Kye's original post.

This is generated from the original post - it may have changed by now

Tubular

Actually regarding the laser, it would be better to have a line generating laser, then you could do basic 3d scanning (to complement 3d printing).

I think Phil suggested a model he has used with his TSL1401 line scanner, but can't find the thread.

Edit: This was the link Phil provided...
http://www.dealextreme.com/p/2-5mw-line-laser-module-3v-46388?item=2

They have them in 5v versions too

Kye

I think designing the product to support a large prototyping space would be the best. This will allow you to choose what you want on the board.

I should be able to route the SPI lines out to the prototyping area. It will be a little hard but should be possible. The lines will be long however. I can also route I2C out to the prototyping area. Similarly, I can route the PAN and TILT connections out to the prototyping area.

I should be able to give you enough space to attach an 8-pin DIP to the board for code storage or what not too.

I'll try and expose as many pins that I can as sensibility as possible.

Parallax can choose to cary extra devices like SPI flash and lasers if they want to endorse you using them with the CMUcam4.

Kye

I'll release some frame work code to device the camera at 30 FPS in either YUV or RGB mode. This should allow people to do some fun things on their own.

Because the camera bus is exposed you could in theory connect multiple propeller chips to the camera to do multiprocessing.

The camera has the ability to double or triple the bus current to adjust for extra loading. So, I can also expose that feature too.

If you did attach a laser, then using a 4 core solution on the CMUcam4 you could get 15 FPS distance measurements. One frame the laser would be on, then the next frame the laser would be off. 4 propeller cores have plenty of time to preform background subtraction and centroid calculations on 640x30 pixels.

If you wanted more resolution then you could put the camera in SXGA mode. The 4 cores could give you a 7.5 FPS frame rate, but with a 1280x15 image instead. You could get really good distance measurements then.

You'll need an extra core for doing the floating point calculations, and one core to do the buisness logic. But, that should be it.

Thanks,

Tubular

Hi Kye,

Yes I tend to agree a proto area, while small, offers the most flexibility, and lasers can easily be mounted separately.

You could bring the pins out to be compatible with the L3G4200D ( 8 pins in single line SIL). Parallax have a series of modules which use the same Vin, Gnd, SCL and SDA pin locations (Altimeter, Compass etc).

The exposed camera bus thing is interesting. Each board in the stack could be tasked with different image analysis, extracting a different feature etc. Wow!

Think I found a good way to spend my cog air voucher.

vanmunch

Looking forward to getting one of these