Cannon Pixma MG2960 image sensor reverse engineering - CIS datasheet request.

[Majorstrain]

Hi all,

I've been reverse engineering a Cannon CIS from their Pixma MG2960 multi function inkjet range.
Previously I've used a Rhom CIS but I'm modifying my project to use the Cannon CIS.

The Rhom was a 300 DPI single analog output, and used a simple clock and start pulse.

The Cannon is a dual 300/600 DPI with the resolution selected by the number of clock cycles the start pulse is high for. 4clk's -300DPI / 6clk's 600DPI
The pin out's, reference voltage (0.9V) and timing I have determined from probing the printer.

But I'm missing something.
The output stays at 1.8V (equivalent to full light sense) during the active part of the cycle.
It doesn't change with dark or light objects placed in contact with the sensor and with the LED's off and ambient light, shading the sensor gives no change in results either.
The preceding started pulse and black level outputs seem to be correct.

I was wondering if anyone had a datasheet for a Cannon 2 or 3 analog output CIS that might give me some ideas.

The other option is a datasheet for the driving chip so I can probe that effectively.

Number on the side of the sensor - R  QK1-8992 08 06 517  (***New Edit***)

Sensor PCB numbers are - 7H6-0748 and CHT BAAO
The chip numbers are A4102 , TI 721 , AJLH , G4 with and under score


Thanks,
Phil

[Majorstrain]

Some progress has been made on nutting out the signals and clock rate that I have to feed the contact image sensor.

The issue with not getting any change on the output is effected by the sensor scan rate.

By scan rate, I mean how long the interval is between full scans. The sensitivity is grater with a faster scan rate.  Rhom was unaffected but the rate. You could scan it once a day if you wished.

If I scan once a second the output seams to saturate, but if I scan at 1000 times a second the output operates as it should. The clock speed is the same for each variation.
I'm still getting unreliable output though.
Obviously haven't got the number of clocks and other signals quite right.

The sensor clocks out the light level of each pixel to the analog output pins each clock cycle.
There is always dummy clocks before and after the active part of the sequence though.

Rohm operates slightly differently in the number of dummy clocks and the start trigger signal.

Still hoping someone has some leads or information.

Cheers,
Phil

[ZenCode]

Hi Phil,

I came across your post while working on exactly the same project – reverse engineering the Canon CIS from a Pixma MG2960. I've run into the same wall you have.

Like you, I've been searching high and low for datasheets on this sensor. The PCB numbers are 7H6-0520 – exactly the similar as yours. I also looked for documentation on the A4102 chip), but turned up nothing useful – it seems these are proprietary/confidential parts that Canon doesn't release publicly.

Since I couldn't find any datasheets, I took a different approach: I photographed the PCB from multiple angles and traced out the connections between the sensor flex cable and the main board. I've mapped out which pins go where, but without the internal pinout of the sensor itself, I'm still missing the critical piece of the puzzle.

I noticed you mentioned you've already determined the pinouts, reference voltage (0.9V), and timing from probing the printer – that's further than I've gotten. I also saw your note about the scan rate affecting the output sensitivity – that the output saturates at slower scan rates but works properly around 1000 scans per second. That's really interesting and suggests there's some integration time or reset mechanism we need to account for.

Would you be willing to share the pin definitions you've measured? I'd be happy to cross-reference with my connection tracing and see if we can fill in each other's gaps. If you're interested, I can also share my photos and connection map – maybe between the two of us we can fully reverse this thing.

Let me know if you'd like to collaborate on this.

Cheers