Detecting drops by measuring temperature of actuator chamber

US 2015/0321470 A1
Base, liquid discharge head, printing apparatus, and method for determining liquid discharge status
Canon Kabushiki Kaisha

With increased interest in high resolution page arrays, a major issue is maintaining print quality in the event of nozzle failures. Nozzles may fail for a variety of reasons — clogging due to debris or the ink viscosity increasing locally. Air bubbles may get trapped inside the actuator chamber, or the nozzle plate can become contaminated. Many of these failures can be overcome by performing some nozzle maintenance, while firing adjacent nozzles can compensate for more permanent failures.

While it is possible to regularly perform nozzle maintenance, this can be problematic with single pass page array printing. It is possible to spit drops onto the substrate either in between pages, or even in the images or background areas. but any more extensive maintenance, for instance cleaning the nozzle plate or purging ink from the nozzles, requires a halt to printing while the maintenance station and printhead are brought together.

Therefore it is becoming more desirable to know the status of all nozzles all of the time. With this knowledge nozzle maintenance can be carried out when required, rather than at regular intervals. In addition missing nozzle compensation algorithms can be used to maintain high print quality in the event of nozzle failures.

Popular techniques for determining missing nozzles are optical systems to either detect drops in flight or dots on the substrate after printing. Neither of these processes are easy to implement.


Here Canon explores building detection circuits under the heaters in each actuator chamber to detect partial or full drop generation failures. The detection circuits track temperature to determine if the drops are fired or not. In the figure heater 104 is generating a bubble within an actuator chamber. Underneath the heater is the temperature detection circuit 105 which records the temperature rise and decay. If no drop is generated then the ink remains in the chamber and the heater is completely uncovered by the bubble. If a drop is generated then some ink is left above the heater. These different conditions can be detected thermally.


On the left the temperature detection circuits are shown. These are underneath the heater, separated by an insulating film. There are two circuits, 105a under the centre of the heater and 105b under the periphery of the heater. Both circuits have serpentine-shaped tracks to increase the circuit length and therefore the resistance, making it easier to measure variations in voltage in the circuit. The connections to the circuits 110 are taken out by vias to tracks on a layer underneath the detection circuits, so they are less affected by the temperature changes.

The two circuits for each heater 105a and 105b are connected in a detection circuit. The outputs of differentiators 121 and 122 feed the comparator 123 resulting in Vout. If there is a drop discharge failure there is always a bubble on the surface of the actuator chamber above the heater, so the temperature gradually lowers. If a drop is generated then heat is removed from the chamber by the ink drop, so the temperature of the surface of the chamber decreases much more rapidly.


The graph shows the output voltage Vout from the comparator. If V2≥V1 then the output is high. If V1>V2 then the output is low. If no drop is generated then V1>V2 is always true so the output stays low. The time for the measurement Tk is chosen to be when the ink fragment left from the discharge should remain on the heater, and before the refill process.

The detection circuits are integrated with the silicon substrate of the printhead die. The detection process is very sensitive yet accurate. no complicated signal processing is required, so detection is very fast.

Low-cost moulded page array printhead from HP

WO 2015/116025 A1
Flexible carrier
Hewlett-Packard Development Company

This review is from a further batch of patent applications describing more details of HP’s moulded printheads. We discussed these in the September/October 2014 issue of Directions. HP introduced their low-cost page array printhead in the OfficeJet Pro X printer in 2013. The printhead used consists of a staggered array of four colour dies.

The new moulded printhead design has some important benefits. First of all it allows the use of much narrower silicon dies. Patent applications related to the first low-cost page array printhead design described the die size as 5 mm wide. The moulded printhead patent applications describe the dies as “slivers” 500 microns wide, so an order of magnitude narrower.

The use of moulded epoxy resin for the body of the printhead significantly reduces the cost of the substrate for the dies – considerably cheaper than others previously considered or used, such as silicon, glass, ceramic and so on. HP is also able to incorporate in-flight drop detection or image scanning into the moulding on the front face. In addition other patent applications describe ink property sensing and ink level detection.

The first of the patent applications we have selected for review describes the manufacturing processes for the moulded printhead, in particular the use of a flexible carrier sheet.

Aug-15aThe first stage of the manufacturing process is to fix a flex circuit 66 with conductors 22 on to a flexible carrier sheet 68 with a thermal release tape 70. The flexible carrier sheet used can be cured epoxy sheet or a high temperature plastic. It is also possible to use semi-rigid and rigid materials such as metal, carbon fibre, composites etc. by adding grooves so that in the final stage of manufacturing the substrate can be peeled away.

The printhead die 12 is then positioned over the gap 72 in the flex circuit. The die is 500 microns wide, 100 microns thick and 26 mm long. Dry etching is used to form the ink flow port 56. This is possible and viable due to the thin structure.

On the front face of the die the ink channels 54 and actuator chamber 50 are formed in the usual way in a spin-on photo-imageable resin, and the nozzle layer formed on top. The die contains not just the heaters for drop actuation, but addressing and drive circuits too. Electrical contacts 24 connect to the flex circuit tracks 22.

Aug-15bAfter the die is in position on the substrate, the body of the printhead14 is formed by transfer moulding with tool 74. The transfer moulding process has been adapted by standard techniques used in the semiconductor industry for device packaging. Preferably no release agent is used in the moulding process, the concern being contamination of the surfaces that will come into contact with the ink.

The stiffness of the moulding can be adjusted, depending on whether the print bar will be used directly or whether it needs to conform to a separate support structure.

Aug-15cThe final step is to strip away the flexible carrier 68 and thermal release tape 70 to leave the finished print bar. Note that with this design there is a single wide ink manifold slot 16 stretching along the full length of the die, with separate ink feeds 56 into each actuator chamber 50. As shown there are probably two rows of nozzles, one each side of the ink feed slot, and therefore using a single colour of ink. This would be a similar configuration to the new HDNA (High Definition Nozzle Architecture) printheads that will be field upgrades for HP’s high-speed web presses. These have 1,200 nozzles/inch in each row, and therefore 2,400 nozzles/ inch for each ink feed slot.

Aug-15dHere a wafer-scale view is shown. There are 5 print bars being made on one wafer, with each print bar having 4 rows of silicon die “slivers”. 10 dies are shown in a staggered arrangement across the width of the print bars, giving a print width of 230 mm. The width of each print bar is around 16mm. With a potential saving on manufacturing costs and printhead size, these printheads are likely to be very competitive with Memjet’s low-cost page wide arrays.

The ability to make such narrow low-cost arrays of nozzles leads to the ability to offer extra rows for redundancy in case any nozzles become defective. However to be able to substitute for non-working nozzles you need to know which ones they are. In the second patent application two schemes for determining these are described, in both cases with all of the functionality within the printhead.

Konica Minolta’s silicon MEMS printhead

EP 2 716 461 A1
Inkjet head and inkjet drawing device provided with same
Konica Minolta, Inc.

This patent application seeks to overcome two problems. When designing a high density printhead with multiple rows of nozzles it is difficult to provide electrical connections to actuators as the wiring must pass between other chambers. In addition the electrical contacts and piezo actuators may be sensitive to moisture.

April-14aHere a section through the printhead is shown. At the bottom is a silicon nozzle plate 11 and a spacer layer 12 made from glass. Next is a silicon on insulator (SOI) substrate, with layer 13 forming the chambers and thin layer 14 forming the diaphragm. Thin film piezo actuators 31 are formed on the diaphragm layer. At the top of the printhead is the wiring substrate layer 21 which is also silicon. On the surface of the wiring substrate are the wiring

Connections between the external exible circuit board 25 with driver chips 24 and the actuators. The printhead assembly is connected to the wiring substrate layer 21 by a resin adhesive layer 30. This is patterned to provide cavities for the actuators and the ink channels. above the printhead is ink manifold 40.

April-14bOn the left the structure is shown in more detail. The wiring on the substrate 21 is more visible. Connections are made from the upper surface to the lower surface with vias. Both top and bottom surfaces of the wiring substrate are protected by films 26 and 28. As the wiring substrate only has ink passages through it, almost the whole surface is available for wiring tracks to the multiple actuator rows.

On top of the piezo actuator is a gold bump 16. At the corresponding point on the wiring substrate is a solder bump 29. Heating the printhead to above the solder melting point of 139C makes the connection. the printhead and wiring substrates are fixed together by the resin adhesive layer 30 in the presence of dry nitrogen. This results in the actuators being encapsulated in a dry inert environment, preventing both the electrical connections and actuators from deterioration in the presence of moisture.

Identifying missing nozzles

US 2014/0085369 A1
Method of identifying defective nozzles in an inkjet printhead 
Zamtec Ltd

Patents previously led by silverbrook Research are now led and held by Zamtec Ltd, also known as Memjet IP holdings Ltd. The Memjet page array printhead has over 70,000 nozzles and it is important to be able to detect any nozzle failures so that any necessary corrections can be made.

Mar-14a The conventional way to carry out the detection of missing nozzles is to print a line of drops 101 from the nozzles. if a nozzle isn’t working then a blank space 104 is left. The spacing between nozzles 103 that are simultaneously printing is determined by the resolution of the scanning system that will evaluate the image. In addition spaces 102 are left between the ends of one array of nozzles and the next to discriminate between the line segments printed by different groups of nozzles.

There are several problems with this method. The pattern is very low density, with most of the printed test area empty, and is therefore inefficient. in addition the nozzles are being driven in an unrealistic way which may result in some artefacts not being recorded. For instance poor nozzle refill rates may only be apparent when adjacent nozzles are also generating drops, which will never occur with this test pattern. Also misdirected jets may lead to misinterpretation as to which nozzle’s behaviour is defective.

The solution is to use a more complex test pattern where the pixel pattern generated is encoded and then later decoded. A Hadamard matrix is used for this purpose. This is a square matrix with entries only either 1 and -1, and where the sum of each row or column (except for the first ones) is zero.

Above an example pattern is shown. The nozzles across the array are divided into a number of cells. In addition a secondary scheme known as a Maximal length sequence or M-sequence is used. The values of these are shown above the pixels; 4 rows of pixels are shown. Note that every pixel sequence for a nozzle prints two drops, so the duty cycle is 50%.

Here a section of a sequence is shown. After printing the image is scanned and analysed. If a nozzle isn’t working then no pixels are printed, as with the 4th nozzle from the left. If a nozzle is firing intermittently, such as nozzle 8 from the left, then this will also be detected.

Large drop volumes from Xaar printhead

WO 2014/023981 A1
Droplet deposition apparatus and method for depositing droplets of fluid
Xaar Technology Limited

The Xaar face shooter architecture with through flow ink circulation has been very successfully commercialised in the 1001 printhead. This patent application discusses some improvements, in particular using a different shaped nozzle to allow larger drops to be generated on the same nozzle pitch.

Feb-14In the Xaar design the piezo walls 3 move to reduce the channel volume 2 to create pressure waves to generate drops. The ink is continuously circulated through the channels, as shown by the arrows, at a rate around an order of magnitude greater than the flow through the nozzles at maximum drop frequency. Attempts to elongate the nozzle in the flow direction were found to allow more efficient removal of debris from the vicinity of the orifice. In addition the effciency of drop ejection increased due to the acoustic wave being present at the orifice for longer.

However, the directional accuracy of drops deteriorated in the direction of the elongation when the nozzle was made elliptical. Experiments showed that the benefits of elongation could be retained just by elongating the inlet side. If the outlet (orifice) was circular, or up to an aspect ratio of 1.2, then the directionality of drops was similar in all directions.

Feb-14bBy increasing the nozzle area through elongation then the drop volume ejected increased as shown in the table on the left. Larger drop volumes are desirable for some applications, such as printing of ceramic tiles, one of the main applications for the 1001 printhead.

Lexmark’s recirculating thermal ink jet printhead design

US 2013/0182022
On-chip fluid recirculation pump for micro-fluid applications
Lexmark International, Inc.

There have been a lot of developments of recirculation architectures within printheads since the launch of the Xaar system.  Already we have reviewed Canon and Hewlett-Packard’s proposals for thermal ink jet recirculation, and here is Lexmark’s.

Recirculation systems enable fresh ink to be brought to the actuator chambers and nozzles.  This is particularly important for fixed array printheads.  With scanning heads, ink can be ejected from each nozzle during the period when the head is outside of the substrate width.  This is not possible for fixed arrays.

With small nozzle sizes and relatively volatile aqueous inks, evaporation of the ink solvent from the nozzle may cause an increase in local ink viscosity and hence a deterioration in jetting properties within a second or two of becoming idle.  To maintain nozzle performance this means ensuring each nozzle fires drops within that period to bring fresh ink to the nozzle.

Nozzle failures are often more noticeable for fixed arrays as it is not possible to cover up defects by multiple passes.  Therefore it is necessary to be even more careful in keeping nozzles working with a fixed array.  To keep nozzles working, drops can be fired onto web substrates between pages.  The amount of ink consumed can be considerable.  Lexmark estimates heavy users, printing large multi-page jobs, may lose 15% of the ink to nozzle maintenance.  However light users printing only short jobs infrequently may consume 80% of the ink on maintenance.  As well as excess ink consumption the printhead life is shortened, as only perhaps 20% of the actuator cycles are used for printing.  The use of recirculation within the printhead can therefore eliminate a lot of external nozzle maintenance, ink wastage and increase the working life of the printhead.


Various schemes are proposed, but all involve the passage of ink through the actuator chamber from one side to the other, driven by a pump.  Here the ink flows in from the manifold via inlets 60, through channels 65 to heater chambers 22.  Heaters 12 generate bubbles to fire drops from nozzles 24.  The ink circulation path continues via channels 75 and through outlet 60, driven by pump 50.  The pump is another heater, similar to the drop generators, but of course with no corresponding nozzle.

When the pump heater is actuated, ink flows preferentially through the large area outlet rather than back to the heater chambers.  When the drop generator heaters are actuated, ink preferentially flows from the nozzle rather than back to the inlet or to the recirculation outlet.

Note that the flow doesn’t have to be continuous.  The printer can determine which nozzles have not been fired for the past one or two seconds, and then drive the pumps corresponding to just those nozzles.

Fixed array printhead for higher viscosity inks

US 2012/0069104 A1
EP 2 436 520 A1
Liquid discharge head and recording device using same
Kyocera Corporation

Here Kyocera wishes to improve image quality when using higher viscosity inks, such as UV curable. These inks have a higher viscosity than aqueous inks – 8 cP is mentioned, and therefore the actuators must be driven to generate higher pressures to form drops at the same velocity. Unfortunately, at the higher pressures satellite drops are formed and image quality deteriorates.

Mar12a  Mar12b

The solution is to introduce a restriction in the descending passage leading to the nozzle. Here a section through the printhead is shown. Ink flows from manifold 5 into actuator chamber 10 via connecting passageway 6. The actuator consists of two piezo layers 21a and 21b, both around 20 microns thick. In between the piezo layers is a common electrode 34, and on the top the individual electrode 35. A stack of Fe-Ni plates 4, each separately etched, defines the ink chamber and channels. The plates are fixed together using a thermosetting resin with a curing temperature of 100-150C. The reason for the relatively low temperature curing is to reduce the chances of stresses being set up between the different materials on cooling.

<a href=””><img class=”size-full wp-image-75 alignright” title=”Mar12b” src=”” alt=”” width=”297″ height=”218″ /></a>Drops are fired using a bi-polar waveform. To begin with the piezo actuator rises, increasing the chamber volume. The drive waveform is then reversed driving the actuator down and reducing the chamber volume. The pulse width between these two events is set to the acoustic length between the manifold 5 and the nozzle 8. This allows the positive reflected pressure wave from the expansion phase to be added to the drive phase so that a stronger overall pressure wave is generated. This is the well-known “fill before fire” technique. Drops generated are 5-7 pl.

Returning to the objective of these patent applications, that is to reduce the satellite drops being generated, the solution is to introduce a restriction in the descender. This increases the damping and hence pressure oscillation at the nozzle. The feature is defined by plate 28, where the descender diameter is reduced from Sd1 to Sd3 with a height Ld3. The distance Ld2 is also important. Ratios of Ld3/Ld0 of 10-15% and Ld2/Ld0 of 20-40% were effective at eliminating satellite drops during tests.

Mike Willis