Quadroplex Switch with multiple PoE Inputs

[wayneorack]

.... I ain't asking Chris for an economics lesson. ....

.

Smart move!

[sirhc]

The problem is it probably would not sell 500 units per year so the cost to design it and build it would be cost prohibitive.

I think we have built an AWESOME switch and I am still not selling 1,000 units per month which is where we need to be so that would be a tough nut to crack. Not saying we would not consider it if we had millions of dollars sitting around in a chinese bank account somewhere. *cough*

[coliinc]

We would definitely take a switch like this, if one was made. It would be ideal for our Fiber to the Curb business. Then from the Curb we run Ethernet cable to each customer and each customer provides power (POE) that powers the switch. One Customer unplugs the power, the switch stays up. The switch would also need one sfp port for the Fiber connection.

[sirhc]

We would definitely take a switch like this, if one was made. It would be ideal for our Fiber to the Curb business. Then from the Curb we run Ethernet cable to each customer and each customer provides power (POE) that powers the switch. One Customer unplugs the power, the switch stays up. The switch would also need one sfp port for the Fiber connection.

1) How many switches would commit to buy?

2) What is this switch valued at?

The reason I ask is we can build anything you can dream up it just comes down to R&D costs (NRE), how large of production runs which affect the cost per unit.

For example when we build prototypes it often costs THOUSANDS of dollars each (NOT INCLUDING NRE COSTS) just to build 5-10 test units.

[turtles2]

I dont normally go after those areas because of HOA but if I could go and say one dish per building I think I would do well. I think of needing 50 right of the bat if I can land an HOA agreement. But I am not sure I could land one I never have tried because of the issues I see from a technical side.

[turtles2]

On another note I would pay 200 to 400 depending on number of switch ports and if I could run OSPF for a ring I would pay 500 to 600. Must be outdoor and hard to tamper. When I say outdoor I need it to work sub zero. It is cold in MN.

[rebelwireless]

Frankly, that doesn't make any sense. You could run 2 power leads and a commercial power relay and plug an existing switch or router for far less.

I get this for cheap power, but when your talking 500-600 there are options available for less right now.

[turtles2]

I am not an electrician so I can't do that. Second I need to have an install with 1 small box. And UL listed.

[rebelwireless]

just for fun, I set this up on my bench.

2 ubiquiti injectors, a very simple schottky diode and a resistor per negative lead. positive leads are connected together.

The LED's and resistors are just a load. I had to stack some resistors up for power handling which is why you see 4 and I used 3 LEDs just to draw more power.

you can see that I can run with either supply, or both. When both are hooked up and I pull one, the LEDs dont blink or anything.
The resistors in front of the diodes are there to cause a voltage drop when the current draw increases, and that causes the supplies to balance out. These two supplies are .25v different so the higher output supply is used until the load reaches 250ma, then the 1 ohm resistor drops the voltage down -.25v and then both supplies run from their. There is some wasted power here, but it's pretty minimal.

IMG_20150513_191325.jpg (131.9 KiB) Viewed 19648 times

IMG_20150513_191303.jpg (128.62 KiB) Viewed 19648 times

IMG_20150513_191246.jpg (142.43 KiB) Viewed 19648 times

IMG_20150513_191238.jpg (144.54 KiB) Viewed 19648 times

[RavenWing71]

Lots of interesting ideas. rebelwireless has shown that the circuit for combining the POEs is quite simple. The real challenge in my proposal seems to be forcing the tenants to use the PoE injectors.

One of the reasons I think it makes sense to build it into the switch is the belief that you could rig a simple circuit that allows the CPU to detect power per port with it's GPIO pins. Then the OS can periodically scan the GPIO pins and choose to administratively down any port that doesn't have power.

So I don't expect that there is really much R&D needed. I had no idea of the prototyping costs, so I can see where the hesitation comes in. Perhaps Crowd Sourcing this product is a good idea?