Suggestions where/how to sell 125 Cmosis 12MP image sensors

[techydude]

My employer has stock of 125 Cmosis 12MP monochrome image sensors surplus to our needs.
They typically retail for us$1500+ ea in low qty, way higher from Mouser.

Majority are completely unused, in their original shipping trays.  Some are used, insofar as they've been installed into our cameras (ZIF-socket) and electronically tested (and passed), but that's all, never pressed into field service, never left the workshop bench or storeroom. (The 20 or so others have been in the field in active use we're keeping them)

Do you have any suggestions on how we might be able to find a new owner for them? A second-hand components sales channel who deal with this kind of product?

Specifically, they're the following models & qtys available:

Cmosis CMV12000-2E5M1PA (monochrome)
Qty available:  90

Cmosis CMV12000-2E12M1PA (monochrome, Near-Infrared sensitised).
Qty available:  35

Both models are 12 Mpixel, 4096 x 3072, APS-C (22.5mm x 16.9mm sensor area), 12bit 132fps / 10bit 300fps / 8bit 330fps, 5.5um pixel pitch, up to 64 LVDS outputs at up to 600 MHz DDR (+3 for clocking/control), sensor is controlled via SPI & a few GPIO inputs.

This sensor is also used in the Apertus Axiom Beta open-source cinema camera project, though they typically use colour (Bayer-filtered) sensors.

I can provide a datasheet on request, but I'm not expecting to sell any/many via EEVblog, I'm more asking if anyone knows of a second-hand sales channel that might be interested.

[wasyoungonce]

techydude

I might know just the place/person,  Astroccd.org..     A Ukrainian gent named Vakulenko has designed an built DIY astro cameras using various image sensors.  These are being built all over the world atm.   I reckon he'd be very interested in some mono sensors...depends upon the available data for the sensors...and availability.  He might not want sensors that can only be had in small lots or are too cost-prohibitive...not sure cannot speak on hid behalf. 

I'll post a like for him over ("grim") at this camera forum

[Asmyldof]

Had I had the budget, I'd easily find/think-of a Metrology project to put the NIR types in and make you an offer.

You might find a bigger-budgeted company in that area to just buy those 35 as development stock.

[dexters_lab]

My employer has stock of 125 Cmosis 12MP monochrome image sensors surplus to our needs.
They typically retail for us$1500+ ea in low qty, way higher from Mouser.

Majority are completely unused, in their original shipping trays.  Some are used, insofar as they've been installed into our cameras (ZIF-socket) and electronically tested (and passed), but that's all, never pressed into field service, never left the workshop bench or storeroom. (The 20 or so others have been in the field in active use we're keeping them)

Do you have any suggestions on how we might be able to find a new owner for them? A second-hand components sales channel who deal with this kind of product?

Specifically, they're the following models & qtys available:

Cmosis CMV12000-2E5M1PA (monochrome)
Qty available:  90

Cmosis CMV12000-2E12M1PA (monochrome, Near-Infrared sensitised).
Qty available:  35

Both models are 12 Mpixel, 4096 x 3072, APS-C (22.5mm x 16.9mm sensor area), 12bit 132fps / 10bit 300fps / 8bit 330fps, 5.5um pixel pitch, up to 64 LVDS outputs at up to 600 MHz DDR (+3 for clocking/control), sensor is controlled via SPI & a few GPIO inputs.

This sensor is also used in the Apertus Axiom Beta open-source cinema camera project, though they typically use colour (Bayer-filtered) sensors.

I can provide a datasheet on request, but I'm not expecting to sell any/many via EEVblog, I'm more asking if anyone knows of a second-hand sales channel that might be interested.

you could try contacting http://pccomponentscompany.com/

we have sold a huge amount of new-old-stock components though them

[mikeselectricstuff]

unless you're lucky, specialised stuff like this can be hard to shift, as anyone who would normally buy it wouldn't buy from a used source.
Perhaps contact the Apertus people, and has been mentioned maybe someone would be interested in building an astronomy camera around them.
I'd be surprised if you could get more than 10-20% of what you paid

[mikeselectricstuff]

If there are any small, specialis distributors, it may be worth contacting them - they may have customers who may be interested & not so bothered about the paperwork side of things, maybe  for development/prototyping.
Years ago I shifted a load of EL display panels that way - distributor had a customer who wanted them faster than the factory leadtime.
I would also imagine Apertus might be able to use a few for testing etc,. if sufficiently cheap.

[cdev]

How difficult would it be to mount one in a  box and turn it into a high resolution camera for computer vision?

[mikeselectricstuff]

My employer has stock of 125 Cmosis 12MP monochrome image sensors surplus to our needs.
They typically retail for us$1500+ ea in low qty, way higher from Mouser.

Majority are completely unused, in their original shipping trays.  Some are used, insofar as they've been installed into our cameras (ZIF-socket) and electronically tested (and passed), but that's all, never pressed into field service, never left the workshop bench or storeroom. (The 20 or so others have been in the field in active use we're keeping them)

Do you have any suggestions on how we might be able to find a new owner for them? A second-hand components sales channel who deal with this kind of product?

Specifically, they're the following models & qtys available:

Cmosis CMV12000-2E5M1PA (monochrome)
Qty available:  90

Cmosis CMV12000-2E12M1PA (monochrome, Near-Infrared sensitised).
Qty available:  35

Both models are 12 Mpixel, 4096 x 3072, APS-C (22.5mm x 16.9mm sensor area), 12bit 132fps / 10bit 300fps / 8bit 330fps, 5.5um pixel pitch, up to 64 LVDS outputs at up to 600 MHz DDR (+3 for clocking/control), sensor is controlled via SPI & a few GPIO inputs.

This sensor is also used in the Apertus Axiom Beta open-source cinema camera project, though they typically use colour (Bayer-filtered) sensors.

I can provide a datasheet on request, but I'm not expecting to sell any/many via EEVblog, I'm more asking if anyone knows of a second-hand sales channel that might be interested.

you could try contacting http://pccomponentscompany.com/

we have sold a huge amount of new-old-stock components though them

is there a typical "resell-to-original" price ratio?

[CatalinaWOW]

I can't suggest any channels but will point out that buying stuff from these guys you find prices that range from 50% to 150% of retail.  I am guessing that unless a maker project with this device takes off you would expect something more in the 50% range for the retail price.  A jobber is going to want to pay half or less of that.  So expect best offers in the $400 range, and many far below that.

[dexters_lab]

you could try contacting http://pccomponentscompany.com/

we have sold a huge amount of new-old-stock components though them

is there a typical "resell-to-original" price ratio?

hard to say as everything that we have sold has been part of office/workshop clearance so we never knew the original cost and the cost to us is mixed in with all the other stuff that was part of the clearance.

The website implies it's about 50% return though, the key point for us is items do sell and we don't have to do any of the sales or fulfilment.

[Asmyldof]

How difficult would it be to mount one in a  box and turn it into a high resolution camera for computer vision?

I dare say, if you're asking that, too difficult. They're high performance sensors, which require a well designed board to mount onto, featuring many high speed memory lanes it dumps data to and with a device you'll have to find or make (FPGA possibly) to convert sensor data lines into, presumably for you, USB or Firewire.

[PhillyGreg]

What were they looking to get for them?
Not sure if your employer is willing to hassle with selling one or two to an individual.
I've been looking to get a hold of one of these for a project for awhile but retail prices are out of my projects budget.

-Greg

[techydude]

Everyone, thanks for your comments and suggestions so far.

Perhaps we're going about this the wrong way.  Perhaps instead of trying to sell expensive 2nd-hand bare image sensors, we should instead be selling complete working camera modules (without optics & mounts) ?

Because that's in fact what we have 100+ of; our camera module is adapted from the Apertus Axiom Beta https://www.apertus.org/axiom-beta open-source cinema camera project.  That's based on the Avnet MicroZed (with the Xilinx Zynq 7020 SoC+FPGA), and a stack of Apertus-designed PCBs down to the Cmosis 12MP image sensor.

We forked that design, removed one of the (unnecessary-to-us) PCBs, and rotated the image-sensor 90º and moved it to the edge of the PCB stack, rather than centered like Apertus's design.  See pics attached.

Within the constraint that all our sensors are monochrome (lacking the Bayer filter for RGB), would that be a much more saleable item?  What if we open-sourced our camera control software to go with it?
These are vague high-level ideas at this stage, I understand, and the specifics will matter to some more than others.  But it strikes me, as one of the reply-threads here already shows, that a more canned solution is potentially much more useful.

Anthony.

[mikeselectricstuff]

I'd imagine a camera is more saleable then a sensor,providing you could figure out who wants to buy a monochrome camera of that spec.
I don't know enough about astronomy to know how useful they'd be & how they'd compare with what;s out there ( or if they'd be useful without cooling)

[cdev]

I have a bunch of composite video cameras and the monochrome ones are really useful exactly because of their sharpness. But the capture card Id been using has some issues, so *if I could afford it* a higher resolution camera body that I could incorporate into a camera-like device sounds very appealing.

[PhillyGreg]

That's actually pretty interesting to hear, 100 might mean you have more of them than the Apertus folks have produced. Last I checked they had shipped like 25 of them.
What industry was this product targeted at?
In the photos it looks like you have the power and main boards. It looks like the PCI-E (Physical not electrical) connector may have been removed from the main board?
I'm assuming then the project would have been streaming the video over ethernet.

As for monochrome, in the photo/film world monochrome is rather desirable. Several companies have produced monochrome cameras.
Notably RED had a mono version of the EPIC camera, and it did sell rather well.

The bigger problem with the Axiom is so far they have been busy with color science, and there is no onboard recording method yet.
Folks that would pay more than just the cost of the parts, would probably be looking for a more complete system than what the Axiom has produced at this point.
Recently, there are more tech tinkerers in the industry, and more open source projects are becoming interesting to a point.

The major problem for a product would be something like the Blackmagic Design URSA, it uses the CMV12000, and is a complete ready to run camera produced by a company with a known history in the industry.
And it sells for 3k USD, the Axiom beta project dev kit was 3k EUR. On the industrial side there is the JAI SP-12000M-CXP4 that was $4500 or so last I checked out.

As pointed out in this thread, I think there are some folks with the interest and expertise to take a module like this and integrate it into a useful application or tinker with as a learning tool.
But for the most part I think they are not the people that could justify spending 3-5k on it.

That's essentially the spot I'm in on my project, I would have liked to work with the larger CMV12000 but the cost pushed me down to the CMV2000.
If I could have set up the CMV12000 and support parts minus the FPGA/CPU board for under 1K it would have been a no brainer, at 3k it gets to be an expensive experiment.
For now the CMV2000 has taught me quite a bit, eventually I'll work on moving up to the CMV12000

[Brumby]

While I'm not in the market - bare image sensors sound interesting, but working camera modules sound useful.

I think your market for a functioning module will be much larger than for the bare sensor.  Adding software can only make it more so.

[techydude]

That's actually pretty interesting to hear, 100 might mean you have more of them than the Apertus folks have produced. Last I checked they had shipped like 25 of them.
What industry was this product targeted at?
In the photos it looks like you have the power and main boards. It looks like the PCI-E (Physical not electrical) connector may have been removed from the main board?
I'm assuming then the project would have been streaming the video over ethernet.

Almost all correct
We had ~150 PCB sets made, have the wherewithal to make at least 100 camera modules, probably more.

My employer is in the "precision agriculture via aerial imagery" industry via multi-spectral analysis.  We developed a system that uses 3 such Axiom-Beta variants with different wavelength filters in each one (and a Nikon D800 DSLR for the human-friendly images), along with an Odroid UX4 to capture single frames from the 3 Beta-variant cameras over Ethernet, and store on USB3-connected Flash.  Along with a bunch of supporting software and hardware to glue it all together, and integrate into an existing airborne system currently based on DSLRs.  A combo of market and funding constraints mean we need to put further development on hold for a while :-(.  But either way, if we iterate on this design, it probably won't be based on 3 discrete cameras, and more likely on a higher resolution sensor too (i.e. CMV50k).

So what you see in those 2 photos is:

- MicroZed (7020) with 16GB µSDcard

- standard Apertus Power PCB (v0.25)

- standard Apertus SensorIF PCB (v0.18 r1.3)

- inbetween the 2 Apertus PCBs is our adaptation of the Apertus 'Main' PCB. Basically I ditched the 'west' CPLD which was all about streaming video out, we didn't need it.  Kept the 'east' CPLD, but rearranged things a little. At that stage Apertus had no HDL for that CPLD (other than to wire-thru a SPI bus from MicroZed PL through to Cmosis Sensor), so we developed our own CPLD code that met that MicroZed SPI bus with a SPI-slave, and then i/f that via Wishbone to a SPI-master (to the Sensor), and 2 or 3 I2C-Masters to i/f to the EEPROM & TempSensor chips.  Not a pretty design, and if I had my 'druthers, it would've ditched both CPLDs all together, but at the time we wanted to maintain as much compatibility with Apertus's codebase as possible.  That was a year ago, and we've since re-written a lot of that stuff, as well as our own application-level code.

'Reminds me - we had intended to hand over that CPLD code to Apertus.  Not sure if they've already coded their own, which they'd intended to be a pair of LVDS lines to/from the Zynq with an embedded clock.  We stuck with the single 4-line SPIbus implementation, because we didn't need much speed just to control the sensor & read from a couple of I2C peripherals.  I'll make a note to check where they're up to there and offer it to them.

So as a single-frame camera (not video), at least a 1 frame / second rate is easily viable.  We currently have 3 Betas sending 1 frame as low as every 2.0 seconds to the Odroid XU4 via NFS over Ethernet without going to too much trouble, and I think the bottleneck there is the Flash write-speed anyway, so even faster is quite likely possible.  We put quite a lot of time into optimising the frame-buffer read-out for either raw or pgm image format, down to about 200ms (i think it used to be over 2s), and with flexible region-of-interest options too.

If this 'complete camera modules' idea has legs, it might behoove us to have a bunch of original Apertus Main PCBs made, to restore all that functionality & s/w compatibility.  It'll depend on the potential user's application...

The bigger problem with the Axiom is so far they have been busy with color science, and there is no onboard recording method yet.

Don't hold your breath for the latter, I don't even think it's on their roadmap - Beta is intended purely for an external HDMI/DP/SDI-connected recorder, at least as far as motion video is concerned.

As for the former, well, you can't rush perfection.

Folks that would pay more than just the cost of the parts, would probably be looking for a more complete system than what the Axiom has produced at this point.
Recently, there are more tech tinkerers in the industry, and more open source projects are becoming interesting to a point.
The major problem for a product would be something like the Blackmagic Design URSA, it uses the CMV12000, and is a complete ready to run camera produced by a company with a known history in the industry. And it sells for 3k USD, the Axiom beta project dev kit was 3k EUR. On the industrial side there is the JAI SP-12000M-CXP4 that was $4500 or so last I checked out.
As pointed out in this thread, I think there are some folks with the interest and expertise to take a module like this and integrate it into a useful application or tinker with as a learning tool.
But for the most part I think they are not the people that could justify spending 3-5k on it.
That's essentially the spot I'm in on my project, I would have liked to work with the larger CMV12000 but the cost pushed me down to the CMV2000.
If I could have set up the CMV12000 and support parts minus the FPGA/CPU board for under 1K it would have been a no brainer, at 3k it gets to be an expensive experiment.
For now the CMV2000 has taught me quite a bit, eventually I'll work on moving up to the CMV12000

Don't look now, but Axiom have just clarified their "path from dev-kit to production camera" in a blog post last week, and the dev-kit (i.e. no case, just the PCB stack & sensor) is now EUR$3990 for new sign-ups (only EUR$2100 if you backed the October 2014 Indiegogo campaign), and the full-pretty-case product is to be EUR$5990.

Obviously there's a certain market being aimed at with Apertus' dev-kit offering, as there would be for our hypothetical camera module.  And the 'Beta II' in its pretty case is something else entirely...

I'm not quite ready to put a price on our camera module PCB stack with CMV12k mono sensor, but I suspect we'd be under half of Apertus' new asking price.

Anthony.