How is ink level determined?

[The Hat]

the printer can still remember when the cart/chip was installed and will cause an out of ink warning by memory, while ignoring the inactivity of prism.

the tests done with blocking the prism sensor and what happened to the ink level warnings -- the ink counting still occurred.

I taught that’s what I just said, and the chip can’t count, the printer does all the Maths...

 

[stratman]

I taught that’s what I just said

OK. My The Hat to English translator wasn't turned on.

 

[Drjim]

From an engineering viewpoint only the transition from opaque to reflective can be used to start down-counting. The printer cannot use the current state of the sensor or it would get nowhere. At least that's how I would code it. I woukd think it reads the sensor for a cart and writes it's state back to the chip. It then looks for a change to reservoir empty.

If Canon were devious printer would detect a change from empty reservoir to not-empty and refuse to print since that means you refilled it. Apparently they are not mean.

 

[stratman]

From an engineering viewpoint only the transition from opaque to reflective can be used to start down-counting.

If you are saying that the prism must be functioning to have a count down to Empty then that would be incorrect. In past experiments on the forum, the action of the prism has been blocked with tape in a "new" cartridge and the ink level monitoring marks the cartridge as Empty seemingly appropriately.

f Canon were devious printer would detect a change from empty reservoir to not-empty and refuse to print since that means you refilled it.

How would that work with a reset EEPROM chip?

 

[mikling]

Drjim is close to the money. The prism is activated within a certain window where the normal operations "expects" it should be around. The prism when activated allows the firmware to look for the point where the reservoir can be positively identified as empty. When this point is reached, the firmware then writes back to the chip and the real tighter estimated countdown then begins. At the point where the reservoir is empty, may other variables are removed or has less weight on the countdown estimation thus providing more accuracy.
The prism is not active throughout the chip cycle only at a certain point. The window appears to be around 10-20% of the ink level when it expects it to be low. That this window is wide enough to accomodate all kinds of issues makes its operation pretty much transparent. However, forced experimentation can determine exactly where that is. So within 10-20% of when the reservoir should be empty, the prism reading is being sought in order to write back to the chip.

 

[Drjim]

From observed data it would seem that the printer recognizes a full count and counts down from there. If the prism detector later shows a change if state that confirms the count. But is not really necessary as observed from early tests. I would think that the printer uses a strategy to protect the head from no ink.

 

[mikling]

From observed data it would seem that the printer recognizes a full count and counts down from there. If the prism detector later shows a change if state that confirms the count. But is not really necessary as observed from early tests. I would think that the printer uses a strategy to protect the head from no ink.

Exactly why it needs a tight countdown routine close to the end. The estimated path to empty from start is worse than hurricane predictions as we can see this weekend. Once it got close to the keys the path was better determined. Similarly, when it sees the prism reading it can then estimate much better when the end is actually near.

Indeed after the prism point, the printer further down the count proceeds to actually prime the head more frequently to ensure that ink is indeed present at the head. This would then put a higher ink output into the waste ink pad and another reason where when one can, to operate the printer before the low point to minimize ink into the pad.

These routines have also been revised throughout the generations and only Canon firmware engineers would know the details. All we can do is observe and surmise.

 

[stratman]

Yet Canon BCI cartridges had no chip, just the prism, and print heads did not burn out en masse. That's progress!

 

[mikling]

That was before class action lawsuits against printer mfrs for leaving ink inside the cartridges when the printer declared them empty.
The Epson lawsuit initally that sided with the consumer ended up with chips as the only reasonably cost effective way to allow the printer to better estimate the ink level. Wait.....the lawsuit covered chips that left up to 20% inside the cartridge even when declared empty.
Then Epson had to engineer a more cost effective way to really determine empty.
Canon OTOH with their sponge, had a way out to allow the customer to squeeeeeeeeze as much ink out as they could by overriding and then the user did push the envelope by using the cart to the very end and ended up burning out nozzles. When an Epson printer declares empty; It ain't going nowhere. NO MORE PRINTING. PERIOD. That is totally opposite to a Canon with a sponge.

 

[PalaDolphin]

If Canon were devious printer would detect a change from empty reservoir to not-empty and refuse to print since that means you refilled it. Apparently they are not mean.

Yeah, they don't have VW engineers.