Do you have access to a C200 that you can try this with? Just using a laptop and a plain Ethernet cable, maybe through a friend of an in-store demo.
Also, I do networking support for a living, and those instructions sound... odd to me. Should be much easier to just set a static IPs in the same subnet on the camera and computer, or even run a simple DHCP server server on your machine to hand out an address to the camera.
Agree - however you just dont know how Canon have coded the setup. It sounds like there is some sort of routine that the camera runs to hand out an IP - and it sounds convoluted at best. But this is why there is some sort of issue with the ethernet on this camera. Something odd about it for sure.
And no, I dont have a camera to try it on. I have yet to purchase the camera. As has been said many times here, work backwards. First find a camera that works for your purpose then get the housing/ports/extras setup and then worry about the actual camera itself last as thats the easy bit.
4 hours ago, ChrisRoss said:
Fitting a bulkhead should not be rocket science - they are sealed with an o-ring. Remove the plug, pass the internal cable through the hole then screw in the bulkhead and tighten. Pretty hard to stuff it up - use your normal o-ring cleanliness and lubrication and once it's in it stays there. It should be no different to fitting a flash bulkhead.
I have never fitted a bulkhead fitting to a housing so not sure what is involved or how precise it needs to be. So will let the dealer do it in his workshops to make sure it comes out to me right.
I am going to go ahead and get the ethernet setup from Dive and See.
* DNC-2084 Elbow style bulkhead connector is used for facilitating the transmission of LAN Ethernet data at speeds up to 1,000 Mbps (1 Gigabit per second) between underwater systems
* The new design allows installing the bulkhead connector in available - desirable direction on your housing. Elbow shape (90 degrees) of the connector keeps the external cable horizontally, instead of the standard vertical style of the connections
* Material: Marine Aluminum with Hard Anodizing coating
* Bulkhead connector has a water block, which is allowing to keep vacuum inside of your underwater housing
* HIGH SPEED CAT5E CABLE, Stranded T-568B Wiring
* 8 Contacts (Brass - Gold plated)
* Custom length and other connector types are available
* Depth - 240 PSI (160 meters)
* Compatible with LAN Ethernet Dive And See cables DNC-1101, DNC-1044, etc
300ft Underwater to Surface Cat6 Ethernet Network Cable, LAN
* Depth Rating 90 meters
* Connecting any CAT6 applications to the computer, router, hub, Network, distribute data or video *Z CAM E2Gigabit Ethernet for data, control & live preview *Transmit UltraHD (3840x2160) at 25/29.97 fps 4K/HD Camera such a AJA RovoCam *Gigabit Ethernet *Power over Ethernet(PoE) compatible *Supports IP Cameras * Cable compatible with Dive And See bulkhead connectors DNC-2025, DNC-2031, DNC-2071, DNC-2084, DNC-2082 * The bulkhead connector sold separately
44 minutes ago, John Doe II said:
Agree - however you just dont know how Canon have coded the setup. It sounds like there is some sort of routine that the camera runs to hand out an IP - and it sounds convoluted at best.
This isn't black magic; the procedure described sounds like the camera is set up to pull a DHCP lease, but with a direct connection between workstation and camera and no DHCP server involved, it times out and falls back to APIPA (automatic private IP address) - that's why it takes several minutes, for both the camera and the workstation to time out on DHCP discovery and assign themselves something in the 169.254.0.1 - 169.254.255.254 range, with a /16 subnet. Once this happens, you configure the random address as static, and the application autodiscovers the camera, probably by the means of network broadcast. It reads as a result of someone randomly poking around in the dark with no idea of what they're doing - you can arrive to the same result much faster by either configuring static IPs in the same subnet to begin with (and with a direct connection, you can go as small as a /30 subnet, instead of a /16), or by running a DHCP server on your machine.
7 hours ago, John Doe II said:
I have never fitted a bulkhead fitting to a housing so not sure what is involved or how precise it needs to be. So will let the dealer do it in his workshops to make sure it comes out to me right.
There is no precision involved, you just screw it in and tighten it, but if you are not confident I guess there is no harm in getting the dealer to do it. The only thing to be aware of is the same standard of o-ring cleanliness as you use for the main housing o-ring which you service yourself all the time. If the housing holds a vacuum after installation, you are all set.
21 hours ago, Barmaglot said:
This isn't black magic; the procedure described sounds like the camera is set up to pull a DHCP lease, but with a direct connection between workstation and camera and no DHCP server involved, it times out and falls back to APIPA (automatic private IP address) - that's why it takes several minutes, for both the camera and the workstation to time out on DHCP discovery and assign themselves something in the 169.254.0.1 - 169.254.255.254 range, with a /16 subnet. Once this happens, you configure the random address as static, and the application autodiscovers the camera, probably by the means of network broadcast. It reads as a result of someone randomly poking around in the dark with no idea of what they're doing - you can arrive to the same result much faster by either configuring static IPs in the same subnet to begin with (and with a direct connection, you can go as small as a /30 subnet, instead of a /16), or by running a DHCP server on your machine.
Yep - understand all that. I was also in IT - not quite as heavy as network support but i was involved in installing WAN links and setting UP Public IP's and Private IP's on various subnet levels. I am not quite sure what the 169.x.x.x range is all about - generally when you get a 169.x.x.x the connection is not getting through.
Overall it does seem as the ethernet link can be made to work with enough perseverance. Its surely a better way to go then the alternatives and gives up to 300ft of usable cable length.
14 hours ago, ChrisRoss said:
There is no precision involved, you just screw it in and tighten it, but if you are not confident I guess there is no harm in getting the dealer to do it. The only thing to be aware of is the same standard of o-ring cleanliness as you use for the main housing o-ring which you service yourself all the time. If the housing holds a vacuum after installation, you are all set.
OK well that is good to know. There is of course a vacuum system fitted to the C200 housing by default so that all helps. I have been in touch with Saga re the WACP on a Subal but will take that over to the right forum section. Not going to bother with WACP on the Seacam housing since I have the Nikonos seacam conversion lens for that housing. It has taken a long time to get all this gear together but almost there now.
21 hours ago, Barmaglot said:
This isn't black magic; the procedure described sounds like the camera is set up to pull a DHCP lease, but with a direct connection between workstation and camera and no DHCP server involved, it times out and falls back to APIPA (automatic private IP address) - that's why it takes several minutes, for both the camera and the workstation to time out on DHCP discovery and assign themselves something in the 169.254.0.1 - 169.254.255.254 range, with a /16 subnet. Once this happens, you configure the random address as static, and the application autodiscovers the camera, probably by the means of network broadcast. It reads as a result of someone randomly poking around in the dark with no idea of what they're doing - you can arrive to the same result much faster by either configuring static IPs in the same subnet to begin with (and with a direct connection, you can go as small as a /30 subnet, instead of a /16), or by running a DHCP server on your machine.
Do you think it would help at all to use a class A subnet IP instead of a class C ? Get right away from the 192.168.x.x ranges ?? something simple like 10.10.0.1 /29 ? So as you say, static IP's on each end with a small number of useable IP's like a /29.
That might do away with all the hunting for the IP's that the camera seems to be doing. And if that is indeed useful then as you say a /30 is even better - be faster for the camera's IP tables to scan (it will most likely be a linux variant of OS onboard the camera ...possibly BSD .... and if that is the case it would be using some form of IP Tables). It does seem to be some IP address issue though.
40 minutes ago, John Doe II said:
Do you think it would help at all to use a class A subnet IP instead of a class C ? Get right away from the 192.168.x.x ranges ?? something simple like 10.10.0.1 /29 ? So as you say, static IP's on each end with a small number of useable IP's like a /29.
Classful routing hasn't been a thing since sometime in the 80s. That said, unless something is hardcoded in that camera in some weird fashion, it shouldn't matter which subnet you use for what is effectively a point to point link.
47 minutes ago, John Doe II said:
I am not quite sure what the 169.x.x.x range is all about - generally when you get a 169.x.x.x the connection is not getting through.
It's not 169.x.x.x, it's 169.254.0.0/16, which is reserved for link-local addresses/APIPA.
Classful routing hasn't been a thing since sometime in the 80s. That said, unless something is hardcoded in that camera in some weird fashion, it shouldn't matter which subnet you use for what is effectively a point to point link.
we use 10 and 172 ranges all the time.
3 minutes ago, John Doe II said:
we use 10 and 172 ranges all the time.
Yes, but they're not class A/B/C/whatever; those have been retired long, long ago. People use 'classes' as shorthand for /8, /16 and /24 subnet masks, but that is incorrect.
1 minute ago, Barmaglot said:
Yes, but they're not class A/B/C/whatever; those have been retired long, long ago. People use 'classes' as shorthand for /8, /16 and /24 subnet masks, but that is incorrect.
Techs at the teleport always referred to them as class A/B/C. we just followed suit.
