User Manual

Page 84
DISCUSSION OF ULTRASONIC INK LEVEL SENSING
Q: What is MSI’s concept for detecting low ink level?
A: Right now, ink level in an ink-jet print cartridge can be measured using resistance (if the ink is partially
conductive), or can be estimated by simply counting the number of droplets ejected. There are situations
where neither approach will work – when the ink does not contain carbon, or where the cartridge is
replaceable separately from the print head. MSI has based their approach on other work involving high-
frequency ultrasound, and propose a kind of ultrasonic switch for each chamber of the cartridge. This “switch”
would be a small patch of piezoelectric polymer, stuck onto the outside of the cartridge, which can send and
receive an ultrasonic pulse into the wall. If the ink level is above the point where the patch is located, then
most of the ultrasonic signal will travel on into the ink, and only a weak echo returns to the sensor. If the ink
level is below the patch, then most of the signal returns, and the echo is strong. A simple voltage threshold is
used to detect which condition exists.
Q: Sounds easy. What’s the catch?
A: The basic principle is pretty straightforward, and has been demonstrated in the lab using electronics
based on readily available discrete components. We know it has to be a low-cost solution, both for the sensor
and the associated electronics. The integration of the electronics into an ASIC should be quite practical. The
operating frequency of the device is high (about 20 MHZ) and we need some gain (+40 dB seems likely) – but
the development of mobile phones and high capacity hard disk drives has made this requirement seem quite
realistic. Our biggest challenge is, quite simply, fixing the sensor onto the wall. If we don’t achieve good
consistency in this area, the sensor would not be reliable. It is unlikely that MSI will be doing this part of the
assembly, so we need to work together to make sure that the process works.
Q: What about multiple chambers in a single cartridge? I’m interested in monitoring color
cartridges.
A: Obviously, we could arrange separate sensors for each chamber. With our piezo film technology, this is
easy since we can form independent patterns on a single piece of film. So an arrangement with three active
signal electrodes, and a common ground, would work well – but this would require four contacts, and some
mutiplexing on the receiver amplifier input.
At this point, we began to think of ways to combine three sensors (for example) into a single, extended one,
to simplify the interconnection and the associated electronics.
The obvious possibility is to treat the three separate walls as if they were one, allowing the three echoes to
“add up” on arrival. If any one out of the three echoes were to increase in amplitude (as the ink level fell
below the sensing point), we could detect this and flag the condition. This sounds fine until we consider the
influence of tolerances on the echo amplitude. The basic piezoelectric coefficients of our material don’t vary
much along the length of a roll of film, but we would need to consider roll to roll variation, temperature
influence on sensitivity, adhesive bonding variations, adhesive property temperature variations, wall
tolerances on thickness and parallelism, and the variation of these with temperature.
Q: It’s beginning to sound “risky”. What is the basic signal/noise ratio?
A: Typically around +10 dB amplitude change, from ink to air on the inside of the wall. The precise value
depends on the wall material, and slightly on the ink composition. But if we allow ± 3 dB on the starting level
to cover all tolerance ranges, then add up three return echoes, we don’t think we would have a very good
“switch”!
That’s why we exploited another concept we originally developed for a different kind of liquid level sensor –
creating different path lengths for the ultrasonic signal for each chamber. This would simply be done by
arranging fractional differences in the wall thickness of each chamber. This separates the three returning
echoes so they arrive one after another, with approximately equal amplitude. We still have a tolerance
associated with the amplitudes, but they don’t add, and a single threshold should suffice to detect any one out
of the three going “dry”. The same principle applies for any number of chambers (within reason!).
Q: How much thickness change is required for each chamber?