Tuesday, August 04, 2020

Sony Reports Drop of Image Sensor Sales, Reduces FY Forecast

Sony reports a drop of image sensor sales in its last fiscal quarter started on April 1, 2020:


  1. Well, they deserve it for the way they want to do business. The SONY distribution (I can speak only for Europe) is totally corrupted, one quote X, the other quote 3X price for the same part. They demand MOQ which is not acceptable, so we cant make make a prototype - and no product will happen. No engineering samples, except hugely overpriced ones, whereas others do provide early samples for the initial engineering or even mechanics. No preliminary documentation for parts on the roadmap, or no roadmap at all. Some parts have features disabled and kept secret from the customer, and other parts are not for sale - and they keep doing inside sales with them to produce crippled cameras. So yes.. better to look at other vendors, who treat their customers with respect.

    1. I cannot share this view. Sony does not sell directly but via distributors in Europe (except large customers I think?), documentation and support is provided by distributors. I work with 2 distributors of Sony in Europe and we integrated various sensors (pregius type of products, so maybe a different story since the sensors prices are quite high). We got/get good support, documentation and also roadmap infos and preliminary datasheets. There are evalboards etc.

    2. Albert Theuwissen - Harvest ImagingAugust 5, 2020 at 9:17 AM

      For those of you who attended my talk at IS London 2019 know how difficult it is to get even simple data sheets from the big vendors ... And BTW, not only from Sony.

    3. I cannot confirm this, too. In fact it is the opposite, we get excellent support from our (European) Sony distributor. The only negative thing that I can say is that Sony EOLed a sensor that we use in a surprise move (which causes us a lot of headache now).

    4. Hi Albert, I did not attend your speech in London, so I dont know what you mean ;-) one thing I miss in datasheets often are comparable units/figures for performance. Every vendor gives different figures, sometimes even within the same vendor the units are different, sometimes in candela per square hectoliter per 1/50 seconds or whatever strange units. It is hard to compare the performance of sensors based on non standard, I'd like to see something like the EMVA1288 figures in each datasheet. But datasheets in term of documentation of how to use a sensor is always easily available. I have integrated sensors from various vendors, large and small, I've seen a lot of datasheets but I never had the feeling it is difficult to get them. Sensor vendors usually want to have their products designed in and supply relevant documentation. I never had problems to integrate a sensor the last 20 years due to missing documentation. If there were problems the reason were my own errors in 99% of the cases ;-)

    5. If they would use the same standard metrics in each datasheet and made them easily available, would you not just start comparing them and take the best sensor you could find? If you have the best sensor out there, that probably is fine, if some competitor has a better sensor, you would be out of luck.

    6. Well no. Its not just the pure pixel performance that defines the attractivity of a sensor. In the end what happens is that you ignore this figures because you cannot compare them anyhow.
      I've only a view on a limited segment of the market ("industrial", global shutter sensors). As I have written in some comments before, the pixel core performance is in most cases good enough, so I dont care too much about this when selecting a sensor. of course, it is nice to know some core figures (e.g. like full well - which is the one figure I miss most in sony datasheets ;-), but decision to take a sensor is taken based on other properties in my environment.
      One feature is how a sensor fits a optical environment, e.g. how well it fits Cmount if you want to use Cmount optics, microlens arrangement/CRA, mono/rgb, how large the pixels are, if its output matches the speed of the following elements in the processing chain, like cable or FPGA. The size of the package, thermal power. Also the higher level features like multi ROI or advanced exposure (e.g. multiple exposure into one frame), if the type of interface matches the FPGA environment. If there are known features, e.g. "brothers" or "cousins" of sensors are already known or integrated. If the sensor is long time available etc, how well it matches a predecessor sensor that is discontinued etc. In industrial applications you dont just jump to another sensor if its core performance figures are a bit better.
      My feeling is that many image sensor manufacturers are too much biased towards core pixel performance (which is very good or at least good enough in most cases anyhow) and tend to underestimate the importance of "integrateability".

    7. In my presentation at IS London 2019, I showed the difficulties I had to get a data sheet of the IMX174, an older Sony device. Actually I was not able to get a data sheet with performance numbers, so at the end of the road I had to measure the performance myself (which is also fun to do but is costing quite some time).
      Moreover, sometimes the vendors provide you with performance numbers expressed in mV (for a sensor that has a digital output !) without any indication of conversion factor/gain. These numbers are completely useless !
      As mentioned already more in this blog, there is an attempt to come to a standardization in measurements and data sheets, being EMVA1288 (based on the PTC technique of Jim Janesick). But unfortunately the big imager suppliers in the Far East and USA do not subscribe to this standard. Maybe an advice to the chair of EMVA1288 : before expanding EMVA into more technical measurements, try to expand EMVA into the offices of the big CIS vendors.

    8. Hi Albert,
      you're right. I had a look at the performance figures of a new Sony sensor and tried to compare it to a older one (also from Sony) we already use and there were different units used and I gave up to get a feeling how they compare...
      The IMX174 is first gen pregius, and frequently integrated. Of course you know that industrial camera manufacturers like FLIR/PointGrey do publish EMVA1288 measurements - like this: https://www.flir.de/support-center/iis/machine-vision/whitepaper/sony-pregius-global-shutter-cmos-imaging-performance/, or Allied:
      https://www.alliedvision.com/de/produkte/kameras/kameradetails/Manta/G-235.html, or matrixVision: https://www.matrix-vision.com/industries-lecteur/cmos-sensor-imx174-von-sony-in-der-verkehrsueberwachung.html, or Basler: https://www.baslerweb.com/de/produkte/kameras/flaechenkameras/ace/aca1920-155um/ and others too...

      Of course this includes a bit more than the pure sensor, I learned that you can do quite much in firmware (you can ruin the sensor performance a bit ;-), but I think Flir/PointGrey/Alied/MatrixVision/Basler and all the others do a good job on this. The values given in the datasheets also are quite similar.

  2. Buy Chinese products, they provide free samples.

  3. Sony sensor customers in Europe should consider themselves lucky. Sony dropped its only distributor in the US over 10 years ago and the Sony sales office in Los Angeles won't provide any support whatsoever for potential sensor customers. I'm pretty sure the Canadian customers for Sony get their sensors through the Canadian office of one of the European distributors.


All comments are moderated to avoid spam and personal attacks.