Expand Quote
I am thinking about "building" my own external recorder. I have an extra firewire card on hand and i think about buying an raspberry pi 4 compute module and io board which has an pcie socket. Unfortunately nobody has tested an IEEE 1394 card on the pi. At least the card works without any problems on linux. The initial costs to test it would be:
Compute Module 4GB with WIFI ~50 Eur (maybe 2GB would be enough)
IO Board ~38 Eur
(PCI-E Firewire Card ~15-20 Eur)
If that works one could think about designing a case for 3D printing but one would need to get the card in a good position by using a pci-e riser ribbon cable (9 Eur) and figure out how to mount it properly. Maybe one could simply remove the slot bracket for the pc case and hot glue it somewhere or use the mounting holes for the slot bracket to mount it to a self designed case. The whole case and mounting situation is what I am the most unsure about.
In case one figures out all of the above a sufficient power bank(12V, 2.1A is needed for the board when using the pci slot) would add 60 - 80 Eur, at least that was the first one I discovered. I think my pcie firewire card is pcie x1 so it should be limited to 12V and 0.5 A. PCI-E x4 is limited 2.1A and 12V so that might not work.
IMO the steps would be:
1.) Try to get the card working on the pi
2.) Setup a hot spot on the pi or connect it to a hot spot from mobile phone
3.) a) Setup web server which emits dvgrab commands
b) Setup video preview in web server
c) Optional: Simple hardware buttons via GPIO to start/stop recording and a simple led
4.) Design case.
If 1 works, then 2 and 3 would be really straight forward. 4. would be a pain in the ass. At least for me as I am completely unfamiliar with 3D printing and designing for it. And I don't know the cost for such a print (and highly likely failures). There are some models for printing cases for the pi + io module on which one could base a design.
In the end my guess is that the price would be approximately the same like a lot of these recorders (250-300), at least in material cost for developement, but one would have some advantages:
1.) Possible to directly record to usb stick
2.) Preview on mobile via web server
3.) No mechanical hdd like in some of those recorders
4.) Maintainable. One should be able to swap all of the components.
Disadvantages:
Basically the biggest problem would be to get it compact and robust. Figuring out the whole mounting of the card, the power bank and the pi and then mount it to the camera. And I have absolutely no idea about modelling for 3D printing. Even the cooling of the pi might be a problem. The power bank is also comparable heavy and large (20.9 x 19.6 x 5.2 cm; 450 Grams). For reference the fire wire card is 1cm x 6.5 cm x 9.5 (hxwxl) without the slot bracket.
It is very likely that I think way too much ahead, after all nobody has even tried to get this kind of card working on a pi let alone with success. One might need to compile the kernel to have the necessary modules for firewire. Unfortunately I won't have time for any of this before mid October as well and I am still not sure if this is even a good idea. I just saw the pi 4 compute module has a pcie port and this immediately came to my mind. This post somehow completely lost structure..
Edit: If it does not work on the pi an alternative one could use an Odroid H2+ (that is x86) or something similar with a m2 slot. But the power supply would be a lot more difficult. One would need an extra source of power for the m2 to pcie adapter. And I don't know anything about that stuff. But apparently people even got GPUs working on it (with computer power adapters).
I have actually had this exact project idea and purchased the cm4 module and io board along with 2 different pcie firewire cards utilizing different chipsets and had no luck getting drivers to work with arm linux firewire. I am by no means a linux expert, I'm 100% linux newbie, but I did recompile the kernel a few times including all possible firewire related items I could identify through my research and the best I got was the card was recognized and listed with the linux lspci command but dvgrab could not utilize the card at all. I then moved on to the other option you mentioned and searched around for affordable x86 based single board computer options like the Odroid. I can't remember exactly why I wrote that option off, maybe because of price or something the manufacturer responded to me via email about it's m.2 capabilites, who knows, you mentioned people got video cards working so I imagine any card should work. I ended up settling on the Zimaboard which is $99 and uses passive cooling which would be great for this project, as well as a full size pcie slot which you could attach an extension to then plug the firewire card in and build it all into some nice neat 3d printable box of some sort. Then all you need is to add a battery and setup linux to boot and start up dvgrab I guess. A small cheap screen of some sort would be awesome as well assuming you could keep the costs down. Zimaboard was a kickstarter project that I wasn't early enough to get in on but they did just start allowing preorders from their website and mentioned shipping shouldn't be too far off.
I am injured right now and found my raspberry pi 4 cm + io board and pci firewire card and just googled it again.
Turns out someone got the generic firewire_ohci driver running on that setup (
https://github.com/geerlingguy/raspberry-pi-pcie-devices/issues/443). Apparently i was missing the 'dtoverlay=pcie-32bit-dma' in the boot/config.txt at the time i tried 2-3 years ago. I still have to figure out also the actual necessary steps I did, e.g. I don't know if certain udev rules where actually necessary. And i had a few different problems.
Capturing from tape seems to work fine with my TRV-900 with dvgrab, but I can only "trick" it capturing live video by setting the TRV into playback mode starting dvgrab and switching into standby, then it seems to record mostly fine. Otherwise it only records 3-4 frames no matter what. I guess some control signals are not transmitted correctly or I am missing some settings on the TRV. But I am afraid of frying the port on the camera. Right now I can't tell where the actual problem is it could be either the Camera/Card/Cable/Driver or a combination of all. I just ordered a similar native PCI-E card mentioned in the Github issues on AliExpress + Cables and see where it goes. If it should work well I might look into finding a PCI-E riser + a smaller battery then my powerbank and figure out how to design a case but that would be in some far distant future.
While the current setup is nowhere close to being robust enough for using it in action it might be still cool for capturing old stuff without having to deal with getting a firewire driver running on modern windows, where I had several driver issues a few years ago already. If anybody is interested I might write more detailed instruction.
Would be cool if anybody has a pi5 + firewirecard and could test it there but i guess the breakout boards for the pcie slot are not (widely) publicly available (yet).
Edit: To be clear. This is by far no means cost effective or a replacement for the external recorders in the foreseeable future.
I guess you ruled out the odroid because of powersupply issues of its m2 slot. You would need to have an adapter with an additional powersupply. Never heard of the zimaboard, will look into it.