Sunday, December 13, 2020

First Commercial Curved CMOS Sensor from Startup CURVE

CURVE-ONE startup (France) announces its first commercial curved sensor (English version starts on p. 2).

"The first public studies in the field of curved sensors are back in the 2000’s. If most of the largest world class sensors companies developed their home-made process, none of them made it a commercial product. The go-to-market strategy applied by the start-up CURVE, supported by the European Commission through the European Research Council programs, made it become an available product. Fruit of years of research and development for astrophysical instrumentation, the commercialization of curved and freeform sensors is now a reality.

The imaging of extended scenes (wide field), has always been a challenge as optical systems naturally curve the focal surface. The bending of photographic plates has been used routinely in many fields, even astronomy with the use of Schmidt telescopes which naturally have a convex focal surface.

The advent of flat electronic sensors has been a revolution, forcing optical designers to introduce additional optics to fit the flatness of these sensors surface. The classical problem of the planisphere appeared then: a huge distortion is created by the imaging system on the edge of the field. Additional complexity came along with the distortion: field flatteners increase the volume and mass of systems, and chromatic aberrations appear. Also, the imaging response is not uniform across the field.

The development of curved sensors is a bio-inspired approach. Mimicking the eye retina, this new technology impacts every future imaging system. By directly correcting the field curvature in the focal plane of imagers, the use of curved sensors suppresses the field flatteners. Less optics means less misalignments and instrumental errors, increases the stability and image homogeneity, and reduces the dependence to environmental condition. Thereby it improves image acquisition quality and then reduces image post-processing costs.

CURVE is now targeting the mass production of its curved sensor, with the support of the European Commission as well as the support of the European Space Agency."

13 comments:

  1. Well, after reading the benefits of the curved sensor I started to think why we need even optics in front of it if it is such a good shit :-)

    ReplyDelete
  2. Nice achievement ! By the way, can it work with variable optics such as zoom lens please ?

    ReplyDelete
    Replies
    1. Hi,
      we made an attempt with a x2 zoom system and a curved focal surface. It still works and helps saving one third of the optics. There is no need for the sensor to have an adaptable curvature in most cases

      Delete
  3. They dont state this on their website, or I did not see it in the first moment - but the sensor itself is a standard sensor, right? The parameters and foto look very similar to CMV12000, right? So they dont develop a full sensor or adapt a sensor design to something they can then make spherical - but they take a available commercial sensor and somehow change it to a spherical sensor?

    ReplyDelete
    Replies
    1. Hi, you are right, we take available sensors and just change their shape before packaging. It makes it a plug-and play product, with no need to have a specific PCB development.
      The Curve-One team

      Delete
    2. Hi Manu,
      thanks! do you have a list of sensors you have already available? Or would you develop a curved variant on demand for a certain sensor? Or do you have collaboration with certain sensor manufacturers to get unpackaged sensors? If yes it would be interesting to know what options are there.

      Delete
    3. Hello Manu,

      did you test sensors from Sony Pregius series or Gpixel GMAX series (popular in applications you might target)?

      Delete
  4. Some time back i tried to contact them but did not get a reply. The contact form on their website today says 'error please try again later'. Lets see when they are really ready to sell a product. I dont know... but with such topics around "breakthrough technology by research institutes" i sometimes get the impression that the announcements and real availability are a bit 2 different worlds. I thought of this again with the imec <2u/pix swir thin film sensor announcement a few days back. Again slides with single digit dollar cost (that gets announced for years now). But we are still in the x1000$ world and will be there for some years in my opinion - also CQD sensors that are available today are expensive, large etc.
    I doubt we will actually be able to solder a working curved sensor to a PCB soon in the context of a camera that gets then deployed into a 24/7 robotic application in 100s or 1000s / year - but will be happy if curv-one proves me wrong.

    ReplyDelete
  5. I remember on trade shows coming across early prototypes of this tech, where they would mainly demo this sensor with a more industrial application example. A comparison with an off-the-shelf camera and lens combination showed the advantage of having much smaller and simpler optics with a curved sensor.

    In this announcement, it seems they have pivoted more towards space and observatory applications? I wonder if this is due to the smaller volume and less cost sensitive requirements of this market. I hope cost and scalability issues can be overcome if there are any, as this is really cool technology.

    ReplyDelete
  6. In the web it shows a convex design, but all the times i have read about curved ensor was speaking about concave. the image looks concave also (but this could be an optical effect. any info about this?

    ReplyDelete
    Replies
    1. Dear,
      We can shape the sensors either concave or convex, spherical, toroidal or cylindrical.
      On the picture, the sensor is spherical, concave, as you noticed.
      Most of the optical designs use concave shpaes, but sometimes the convex shape is required, for Schmidt telescopes for instance.

      Delete
  7. It looks from the pictures like it is curved in 1 dimension but not two (i.e. a cylindrical curvature, not spherical).

    Is that true? It is certainly easier (Gaussian curvature blah blah blah).

    If so, doesn't that eliminate most of the advantage? It might make more sense for a widescreen panoramic sensor.

    ReplyDelete
    Replies
    1. if you look at the image at https://www.spiedigitallibrary.org/conference-proceedings-of-spie/11180/111802Y/Curved-sensors--experimental-performance-of-CMOS-prototypes-and-wide/10.1117/12.2536025.full?SSO=1, it seems there is curvature in x and y (especially visible in the right image of figure 1), isnt it?

      Delete

All comments are moderated to avoid spam and personal attacks.