CASELAW-EPO - reviews of EPO Boards of Appeal decisions

T 391/23 – What does an umbrella have in common with a fingerprint sensor?

chat_bubble 0 comments access_time 5 minutes

EP 3 701 420 B1 relates to a compact optical sensor, such as a fingerprint detector. The sensor is placed under a display panel for detecting and imaging light returned from a fingerprint on top of the display panel.

Existing solutions use absorbing channels as collimators to lead the light to the pixels below the display have the downside of also absorbing desired light.

The patent proposes to use an array of microlenses, to transmit the light from the fingerprint to the sensor. The configuration of the microlenses is such that light arriving at an angle of incidence lower than a certain predefined value is “focused to the sensor array”, whereas other light “is not detected”.

Brief outline of the case

The patent was revoked by the OD for lack of sufficiency.

The proprietor appealed.

The board confirmed the revocation.

The proprietor’s point of view

The proprietors argued before the OD that crosstalk should not be taken into account, but, in their grounds of appeal, and during the OP before the board, affirmed that the invention was devised to reduce crosstalk.

The proprietors submitted that a significant reduction was sufficient. In a real system there were always sources of noise, for instance background light. Light incident at a larger angle was also a source of noise.

The skilled person would consider crosstalk light “not detected” if it could not be distinguished from the other sources of noise. This was common general knowledge.

The skilled person would design the optical system using e.g. an off-the-shelf CMOS detector. The system to be designed was a fingerprint sensor and this purpose had to be taken into account. In practice the skilled person would be satisfied that such light was not detected if the signal to noise ratio was good enough for a fingerprint sensor.

It functioned much as an umbrella – some raindrops would still reach the person holding it, e.g. due to wind or ground splashes. But the invention nonetheless worked. It reduced significantly large angle light to obtain an adequate signal to noise ratio for fingerprint sensing, which was the problem addressed by the claimed invention.

The skilled person understood that in order to obtain a good image of a fingerprint, the sensor should only receive light from low incidence angles. An angle of view adequate for an efficient fingerprint system would be less than 10°, even less than 5°.

The skilled person would use simulation as an aid in that configuration. The proprietors reported on three simulations carried out with the Zemax simulation tool, two based on the configurations allegedly disclosed in the patent, and one based on an “optimized example“.

The two simulations showed a reduction of the power of light received at larger incident angles in comparison with the case where the reverse side of the microlens is completely transparent, irrespective of the sensitivity of the sensor.

The opponent’s point of view

The opponents argued that the claim interpretation of the proprietors had no basis in the patent specification, which contained no discussion of optical noise whatsoever. The patent only mentioned noise as electrical noise. Background light was mentioned, but in a different context, namely a discussion of the illumination to be used, not of types of noise.

The opponents also stressed that the proprietors’ statement that the skilled person would understand the notion of (non)detection as making an implicit reference to other noise sources was a mere allegation not backed up by any evidence.

The term “not detected” used in the claim implied, by itself, that no light incident at larger angles reached the detector.

The skilled person skilled would understand from the patent specification that the invention concerned constructional arrangements ensuring that light from larger angles did not reach the detector, disregarding stray light.

The board’s decision

There was disagreement as to whether the feature according to which “light returned from the fingerprint with an incident angle of more than said predefined value is not detected” is already satisfied if light received from angles above that predetermined incidence angle has reduced power, and if so, by how much. A subsidiary aspect of this question is whether the considered light should include crosstalk light.

The board agreed that crosstalk must be considered as the claim refers to the light incident on the microlens structure, not on a single lens, and crosstalk is specifically addressed by the patent specification.

The specification mentioned in several places that light did “not hit” the sensor and was therefore “not detected“. But this was only by way of example and did not imply the inverse, i.e. that light was not detected only if it did not arrive at the sensor.

The specification is focused on geometrical optics, namely on various constructional details aiming to guide light so that, assuming ideal conditions, it reaches, or not, the detector following its path, depending on its angle of incidence.

Thus, even if the person skilled in the art knows that detection, or rather identification of a signal , requires a certain power level in comparison with that of a noise signal, the context of the specification leads to the understanding that when the patent refers to light at certain incident angles as not being detected, it means that in ideal conditions such light does not reach the detector at all.

Returning to the umbrella analogy: ground water splashes aside, the example umbrellas still allow some rainwater coming sideways to reach the bearer, whereas according to the claim this does not happen.

The board agreed that the skilled person understands how an optical system works, understands the principle of the invention, and can use simulation tools to configure an optical system as taught and verify its properties.

However, in order to arrive at a working configuration, the skilled person would have to experiment with quite a few parameters, including angle of view, microlens size, pitch, microlens thickness and material. No working examples are disclosed for the skilled person to start from and no guidance is given as to which combinations of parameter values might be made to work.

The optimized example can only show that the claimed invention can be carried out for a certain predefined angle of view, but not that the patent puts the skilled person in a position to obtain that configuration without undue burden.


It is never good for a party to vary its interpretation of a claim between the first instance and before a board.

Simulations can be used, but they should refer to what is disclosed and not merely to an “optimised” example.

Light being “not detected” can only mean that light is not reaching the sensor and not merely attenuated.  

Last but not least, it does not seem that in any written submission of the parties, and especially not from the proprietor, the analogy with an umbrella has ever been mentioned. It seems thus to have come from the proprietor or from the board during the OP. The minutes of the OP before the board are silent on this point.

The result is not pleasant for the proprietor as with revocation of the patent, an infringement action against the opponents before a German court has come to an end. The appeal procedure was accelerated for this reason.

T 391/23


Art 83

Share this post


Leave a comment

Leave a Reply

Your email address will not be published. Required fields are marked *