Monday, March 20, 2017

Eric Fossum Founds Gigajot Startup

Eric Fossum updates his LinkedIn page, saying that he has founded Gigajot Technology LLC devoted to "R&D for commercialization of the Quanta Image Sensor."

GoodCall site also reports that "In June, after [Jiaju] Ma completes his Ph.D., he will also make the move from inventor to entrepreneur. Ma will launch a startup, Gigajot, with Fossum to commercialize the QIS... Ma came to Dartmouth in 2012, and since then, he’s worked with Fossum on the Quanta Image Sensor (QIS), which can greatly enhance low-light sensitivity."

Thanks to AJ for the pointer!


  1. Always unnerving to find yourself in the headlines first thing in the morning. Anyway, the head of the startup is (Dr.) Saleh Masoodian, also my student, who recently defended his PhD dissertation thesis. JJ will join in a few months, and I am just a co-founder.

    We will see how it goes. An experimental 1Mjot QIS implemented in a stacked BSI CIS process seems to work fairly well and papers were submitted to IISW. We appreciate the support of Rambus, DARPA and a world-class foundry in Taiwan in the implementation and characterization of that experimental device.

  2. Albert Theuwissen - Harvest ImagingMarch 20, 2017 at 1:25 PM

    Eric, are you co-founder or co-sponsor, or maybe both ?
    Does this mean that CMOS is joining the club of the dinosaurs (like you claimed in 1995 to be the case for CCDs) ?
    Anyhow, wishing you success and do not forget to have some fun as well.

    1. My day job will remain with Dartmouth for the foreseeable future. And yes, I am also a very early investor too. And no, I don't see a demise of CIS technology anytime soon although I believe there will be some nice niche markets for photon-counting image sensors like QIS. We have achieved 0.17e- rms read noise (best device) which is definitely fine for photon-counting applications.

  3. Best of luck to the company. I am personally very skeptical of commercial organizations with co-founders who retain tenured academic positions. Nothing screams "low confidence" or "figurehead" like a co-founder who is not 100% behind their creation. SAB member - sure. Board member - by all means. But a co-founder implies serious skin in the game.

    1. My guess is that you don't completely understand the term co-founder or the various roles co-founders perform in a start-up.

  4. Impressive background, not considered by anonymous poster. Eric's "skin in the game" is the few bucks he chips in and his name. Anonymous has no skin in his comment. It would be great if Vladimir didn't approve anon comments that are not rumors about products.

    Eric while I don't understand much of the QIS I like the Delta-LogH; digs deep into the dark while crushing the Blooming.

    Those multiple planes, are they like monochrome Foveon but temporal?

    Is there going be some Hardware Compression like Run Length Coding or even EBCOT to reduce the size of the Data?

    What is the Spectrum, as far as Gamma and Microwave or one category less on each end?

    Thanks for answering if you wish to.

    PS (Anon): This is where I'm "a Fan" - T2SL:

    1. i remember i watched some vedio demos shot by the sensor from this group during a conference back to 2011! the impressions that i got during that time are: very interesting idea (biomimetic), impressive light sensitivity & DR, but still a lot to improve the noise, specially pixel&column FPN....i hope they manage to get the best out of their idea & technologies nowadays.

    2. @Rob Not sure I understand your question about Foveon. But the bit-planes are temporal fields of binary data, combined later to form a frame of image pixels. (and then there is the multi-bit QIS with 3 or 4b per field per jot)
      Digital data compression - many options here and mostly application dependent. Probably best done on a 3rd layer in a stacked structure.
      Spectrum is "silicon" so mostly visible. The concept may work in other materials too but IR is fraught with ITAR issues. No idea how Northwestern does their work legally.
      @Anon - if you were referring to our group by "this group" then it wasn't our video you saw. First working QIS devices (1Kjot) were made in 2015, and the first 1Mjot arrays (stacked BSI) only demonstrated in 2017.

    3. @Eric, I meant the Northwestern group (H. Mohseni from that group gave the presentation with demo, if i remember correctly). just to make it clear for others.

      btw, i am highly interested to watch a demo from QIS, although haven't got a chance yet...

  5. Thanks for your reply Eric. Yes it's odd, some of the Projects that Universities and Companies do and, how they think they can protect Bleeding Edge research from export.

    Pay Tuition, maybe hold a Doctorate, and work on sensitive projects.


    BTW: The least noisy and most sensitive Image Sensors I've seen lately are: , Excelon: or from a few years ago On Semi's KAE-02150: .

    NW's T2SL Video with Demo: .

    Coolest Camera - 10 trillion FPS captures the temporal propagation of light: .

  6. Is there any information available on the actual pixel like publications or patent applications?


All comments are moderated to avoid spam and personal attacks.