Where in the universe does one get Active (non EOL) MIPI CSI camera modules???

[product_dev_as_a_hobby]

Below I've colourfully expressed my use case, but I thought I'd just state what I'm looking for in terse language here.

I am looking for a MIPI CSI camera module using modern, non EOL image sensors the likes of which you might find in a smartphone, webcam, security camera, or laptop. I specifically am not looking for arducams, rpi cams or any other hobbyist aimed prebuilt modules (which typically are either not suited for mass manufacture, are too big, and or use EOL components). I want to go one step lower than that to just get the modules they make convenient.

This is for a number of reasons.

• I want to see what it takes to make the sausage; what I would need to do if I were designing for mass manufacturing where depending on a module using an EOL sensor from a company aimed at hobbyists would simply not work.
• The traditionally recommended modules either fall into the category mentioned above, or are just too big for the use cases I imagine using them in. I would even accept just a camera module that was smd mounted where Id have to use a flex pcb or similar. I just want the little black square, on the little bendy yellow rectangle.
• I want to be free: I want to learn to fish: I want to be able to just pick a camera module/sensor rather than relying on a company to make that easy for me with a hefty markup to boot (the sort that would make it impractical for usage in a mass manufactured product).

Picture this: You've done a raspberry pi/esp32/arduino cam project or two. 

Cool, but you want to advance. You want to figure out how to swim without the aid of big Arduino, or sour fruit sbc corp. 

You want to learn how to source and put together all the components for an idea you have, sourcing from actual vendors/distributors and designing in a manner that could eventually (big, ambitious, somewhat unrealistic wishes) be turned into something that could be manufactured)

You start seeing if all the types of parts you want to use are available.

You go to Octopart, Digikey, Mouser, Arrow, PCBWay, all the places. You think

Wow! I didn't know I could technically just make my own custom  SOC based pcb for wacky unique idea 1 3 or 5.

Wow, just about everything is available as an smd component and with careful reading of the datasheet and an understanding of communication protocols you might be able to make this work! Great!

But how in the universe do I get a camera module?

• Compact Camera Modules - Yea that oughta do it. I want a mipi csi flexicable with a camera on the other end, and not the typical oddly almost all EOL image sensors that all the typical PI cams and Arducams use. Nope. Nothing.
• MIPI CSI Camera? - That seems pretty clear right? Nothing.
• Camera/Image sensor/MIPI (in the usual distributor search bars)? - Once again nothing.

Its like I have a fundamental misunderstanding of the words camera, image sensor, lens, module or something, because I just can't seem to see where I would source this from if I were aiming to mass manufacture something. It's like I've finally stumbled upon the component that hobbyists, even the most devoted of hobbyists just aren't allowed to access because? Like, If I want a powerful modern SOC, I can just buy that right off of digikey, but I want a webcam camera module and suddenly everything seals up?? Utterly confusing.



Ok some might say, well just use a RPI cam like a regular person, and I totally get you from the point of view of practicality, but to me, the entire point, and I mean the *ENTIRE POINT* is that I want to figure out how I would do this if I was designing for mass manufacturing. So far a whole host of things I thought might be roadblocks turn out to just actually be available to just slap on a pcb with the appropriate nets, and careful reading of datasheets.

Cameras though? They just don't seem to exist.

To be very clear, cameras are absolutely everywhere. Cheap security cameras, laptop lids, phones, printers, the list goes on and on, so its just unbelievable to me that this would be the part that was unobtanium.


Conclusion time: Am I choosing the hard way? Yes. Is it weird that this seems waaaay more hidden than I would possibly expect? Also yes.

I truly hope for once for the sarcastic forum post linking the obvious answer that I somehow missed, because it feels like bizarro world not being able to find a source for a component this common with such a relatively simple/well known interface.

[Georgy.Moshkin]

Similar situation with mouse optical sensors. E.g., it is very difficult to source something like PMW3325 plus accompanying plastic lens. Mouser has some traces of obsolete stock.
It is much worse with image sensors.

[Siwastaja]

Personal experience: I had never really designed a manufacturable imaging device. Wanting to design a 3D time-of-flight imager, I googled around, found a relatively small and fresh (for a semiconductor manufacturer) imaging device manufacturer, who agreed to sell us a "devkit", basically a support license, for just a few thousand, and sell us individual chips even in just tens of units.

Ordered bare imager chips, just Chip Scale Package with 0.5mm pitch solder balls. Read the datasheet very carefully, then designed the PCB, just your usual 4-layer stuff from the usual cheap suspects, and another MCU board, with flat flex connectors on both PCBs.

Lenses, I tried to look at Western suppliers and it was a dead end, so went to Alibaba and ask Chinese lens producers. All were happy to sell samples at a few $ each. Ordered a dozen from different manufacturers, looked at the drawings to find all the mounting hole configurations, and made the PCB footprint accept all those different lenses.

Ordered PCBs and stencils, soldered using a hot air gun, wondering if I have any chances to get the imager chip working, wrote some code, and the thing worked. First prototype in a few months of calendar time, and much less development time. Surprisingly easy.

The next step was getting a lens for the illumination LEDs, because a lot of light was wasted from the 90degree field of view infrared emitters. Western suppliers failed to supply again, so this time I went to Alibaba with the question of custom design and manufacturing of a 1D acrylic lens. Most suppliers replied and one offered optical sumilation (sic) free-of-charge with prototypes injection molded for under a $1000 for the mold + lenses. The lenses did what they promised, and we later ordered a larger batch from the same supplier.

The fewer you have middlemen, the easier it gets because the less you have to worry about availability, weird issues, delays (i.e., weird EOL parts used because the module was designed a decade ago), etc. You can go to the lowest level. Really, from reading your interesting post, I don't think the module is a necessity for you. It makes some things easier, such as not having to look for a lens or solder small components, but if you hit a dead-end looking for a module, don't waste more time on it but by all means go to the deep end. I did, without much of issues, even being kinda rookie at that sort of stuff, and I will do it this way again if need arises.

[mikeselectricstuff]

The problem with anything like this that's used in very high volume products like phones, is the manufacturers just aren't interested in lower-volume users - it's just not worth their time.
RasPi cam is the closest you'll get to something in current production

[coppercone2]

i think thats a mis guided idea because if smaller people use their products it becomes a staple not some BS some big corporation used once and managed to go extinct because of some strategic level executive management fuckup that has nothing to do with the product

the little people are not as dumb as the big corpos

[product_dev_as_a_hobby]

Personal experience: I had never really designed a manufacturable imaging device. Wanting to design a 3D time-of-flight imager, I googled around, found a relatively small and fresh (for a semiconductor manufacturer) imaging device manufacturer, who agreed to sell us a "devkit", basically a support license, for just a few thousand, and sell us individual chips even in just tens of units.

Ordered bare imager chips, just Chip Scale Package with 0.5mm pitch solder balls. Read the datasheet very carefully, then designed the PCB, just your usual 4-layer stuff from the usual cheap suspects, and another MCU board, with flat flex connectors on both PCBs.

Lenses, I tried to look at Western suppliers and it was a dead end, so went to Alibaba and ask Chinese lens producers. All were happy to sell samples at a few $ each. Ordered a dozen from different manufacturers, looked at the drawings to find all the mounting hole configurations, and made the PCB footprint accept all those different lenses.

Ordered PCBs and stencils, soldered using a hot air gun, wondering if I have any chances to get the imager chip working, wrote some code, and the thing worked. First prototype in a few months of calendar time, and much less development time. Surprisingly easy.

The next step was getting a lens for the illumination LEDs, because a lot of light was wasted from the 90degree field of view infrared emitters. Western suppliers failed to supply again, so this time I went to Alibaba with the question of custom design and manufacturing of a 1D acrylic lens. Most suppliers replied and one offered optical sumilation (sic) free-of-charge with prototypes injection molded for under a $1000 for the mold + lenses. The lenses did what they promised, and we later ordered a larger batch from the same supplier.

The fewer you have middlemen, the easier it gets because the less you have to worry about availability, weird issues, delays (i.e., weird EOL parts used because the module was designed a decade ago), etc. You can go to the lowest level. Really, from reading your interesting post, I don't think the module is a necessity for you. It makes some things easier, such as not having to look for a lens or solder small components, but if you hit a dead-end looking for a module, don't waste more time on it but by all means go to the deep end. I did, without much of issues, even being kinda rookie at that sort of stuff, and I will do it this way again if need arises.

Thanks for this. That admittedly sounds quite daunting, especially because I kinda have this idea in my head that I can make this process a lot simpler buy just designing with parts that PCBWay or other pcb + assembly services can put together negating the fiddly efforts on my part, so having to source and assemble things individually for something that is really just a hobbyist product made for self (and maybe others if it turns out to be good), sounds like quite an added challenge. At the same time, the thing I'm trying to make (a drone with very different requirements and setup to traditional FPV race style drones (custom firmware required, awkward control scheme, flight time over all else, wifi controlled, sub 250g)), feels like having the camera is a necessity to the project, so I have to figure this out one way or another.

I also don't want to order a sample giving them the impression Im a potential big customer and have them mistakenly dedicate effort to me if I never materialize my hobby project into anything of a larger scale directly.

For me, I do have fewer requirements, I basically just want the image sensor (and any necessary co-processor) and the lens; the basics needed to film 1080p video back to a MIPI CSI interface. I can treat LEDs if I find the need for them as separate components, and part of the focus is not adding extra weight where I can avoid it while minimizing space usage.

I guess the useful follow up question is are you then saying that basically my process for trying to accomplish this would necessitate basically finding an image sensor of decent value from a typical supplier but then sourcing the lens housing for it separately and marrying the 2 (The plastic lens assembly with the image sensor)?

[Siwastaja]

I guess the useful follow up question is are you then saying that basically my process for trying to accomplish this would necessitate basically finding an image sensor of decent value from a typical supplier but then sourcing the lens housing for it separately and marrying the 2 (The plastic lens assembly with the image sensor)?

Yes, I'm saying that's one viable option. Although if you are looking for the most miniaturized sensors such as ones used in laptops then I guess it's a whole package, a module as you say, assembling something so tiny with proper alignment becomes a problem. But for larger sensors, say in the half centimeter - centimeter, the availability of PCB mount lenses is good and small alignment errors less critical.

[product_dev_as_a_hobby]

I guess the useful follow up question is are you then saying that basically my process for trying to accomplish this would necessitate basically finding an image sensor of decent value from a typical supplier but then sourcing the lens housing for it separately and marrying the 2 (The plastic lens assembly with the image sensor)?

Yes, I'm saying that's one viable option. Although if you are looking for the most miniaturized sensors such as ones used in laptops then I guess it's a whole package, a module as you say, assembling something so tiny with proper alignment becomes a problem. But for larger sensors, say in the half centimeter - centimeter, the availability of PCB mount lenses is good and small alignment errors less critical.

It doesn't really have to be all that small. I was just hoping for the sort of thing that could fit in sort of a 10mm by 10mm space, record 1080p 60 with decent enough quality, and communicate back to a main board over MIPI CSI.

About these lens arrangements, they are soldered? Can you give me some leads on what I would be searching for there? Like lets say I use (just as an example) a OV02746-H34A-Z, what component would I be looking for to complete the module (in terms of the plastic module with the lens)?

[amyk]

Go to Ali and look for replacement phone parts.

[Georgy.Moshkin]

Go to Ali and look for replacement phone parts.

Yes. Also buy a smartphone, prepare setup for sniffing MIPI / I2C initialization sequences:
1) Turn on the smartphone, open camera with best settings and record everything thats going on
2) Replay this initialization sequence using your custom MIPI board to connected image sensor module
Idea with this approach is that you don't need any documentation. Just replay existing sequence. Image format can be relatively easily obtained by investigating raw image stream. But this option will probably be gone soon, after they implement encryption with individual s/n for each image sensor. In other words, you will need to respond to image sensor with a right code based on it's s/n or something like that.

[Siwastaja]

About these lens arrangements, they are soldered? Can you give me some leads on what I would be searching for there? Like lets say I use (just as an example) a OV02746-H34A-Z, what component would I be looking for to complete the module (in terms of the plastic module with the lens)?

The keyword for me was, IIRC, "CCTV lens". Stuff like this: https://www.alibaba.com/product-detail/large-fov-cmos-ttl-8mm-ov5640_1600719872225.html?spm=a2700.details.0.0.4c583971vIXsLl . You have to test a few to see if the optical quality is sufficient for your purposes. Mounting is usually 2-4 holes on the PCB. Some larger ones can be screwed in place with M2 screws or similar, some have plastic pins only and need to be glued. I think I tested nearly 10 different ones, all below $3 or so, and was pretty happy with about half of them. My use case was low resolution and insensitive to chromatic errors thanks to narrow band monochrome imaging, but otherwise it was quite demanding, low flare / as little internal reflections as possible was the key point, and differences between the best and worst manufacturers were large.