IR to LAN can it be that simple?

[neazoi]

Hi, recently I have found this schematic **broken link removed** but no description.
Do you think that it is too easy to connect two IR devices through standard LAN cards?

 

[betwixt]

That is basically how fibre-optic LAN links work. You still need the ethernet signals though and you may have difficulty extracting them at TTL levels from the network controller devices. I doubt it would work when connected to a standard RJ45 port.

I'm puzzled by the crystal frequencies which are different and also seem too high.

When I used this kind of circuit in the past it only worked at 10Mb/s so don't expect miracles! You will also need fast IR devices, the ones normally used in this application would be built into a socket designed to hold fibre optic cable plugs.

Brian.

 

[neazoi]

You still need the ethernet signals though and you may have difficulty extracting them at TTL levels from the network controller devices. I doubt it would work when connected to a standard RJ45 port.

Aren't the rj45 standard LAN port signals TTL level? If not, then a converter may be needed, but the author of the circuit has not described such a converter.

The author has described it as a 10MBps system, so I would not expect more indeed. I have no idea about the oscillator frequencies, but I am thinking of using an Si570 and program it at different frequencies to see how it goes, I may be able to achieve higher speeds by using a higher frequency oscillator and G logic instead of ALS, who knows.
However compared to RONJA this is a much more simple circuit, directly outputting to LAN.
I see possibilities if this works, like creating a small RF free LAN inside the house (for those considered on microwave RF safety).
I think , as a starting point I am going to "borrow" the TX/RX diodes from the RONJA project, although better may exist (less capacitance at reasonable price).

 

[FvM]

Aren't the rj45 standard LAN port signals TTL level?

No. They are balanced RZ signals, level lower than TTL.

You probably misunderstood the intention of the circuit linked in post #1. If it actually works in the shown dimensioning (there are various reasons to doubt this), it would transmit/receive through a fiber cable, but hardly through air. There's no realistic chance to transmit 10 or even 100 MBit via open optical channels without modulation. It might work with a laser or fast IR led with optics and a perfectly adjusted beam.

 

[neazoi]

No. They are balanced RZ signals, level lower than TTL.

You probably misunderstood the intention of the circuit linked in post #1. If it actually works in the shown dimensioning (there are various reasons to doubt this), it would transmit/receive through a fiber cable, but hardly through air. There's no realistic chance to transmit 10 or even 100 MBit via open optical channels without modulation. It might work with a laser or fast IR led with optics and a perfectly adjusted beam.

When you say without modulation, you mean without a high frequency carrier?
I thought that the 74as08 gate is modulating the UHF square wave oscillator, actually producing a square wave signal composed of zero and VCC voltage levels. You mean modulation of a HF sinewave carrier?

I can see that the rj45 levels are +/-2.5v https://en.wikipedia.org/wiki/Ethernet_physical_layer so that would be a problem, I do not see how 0 and 5v could be transformed to -2.5v and +2.5v at the speeds used by the circuit, perhaps there is a solution on this for compatibility to the RJ45 levels?

Would you expect reasonable performance on 1MBit i.e. 10 times less the mentioned speed when used on open air, with photodiodes like pbw35 and cheap IR LEDs or laser diode heads?

 

[betwixt]

The RJ45 signals are not square wave either so even if you translated the voltage levels you would have difficulty slicing them accurately. The circuit is designed to run directly from/to the TX/RX pins of a network controller IC without the physical layer present. Some early network cards had these signals brought out to 'D' connectors, from memory, they had 15 pins but I'm going a long way back.

The network controller IC will not work at 1Mb/s, the clock speed is extracted from the data stream so unless it is within the capture range of it's PLL it will not be able to function. The component values in the schematic would also have to be changed. They serve no purpose other than to shape the output of the light detector diode so the pulses look more like they are from a direct connection. At a different speed you would have to adjust the values to maintain the wave shape sent to the controller.

A further problem with more modern (> 1990) network adapters is they need DC across the common wires to let them know they are connected. You would either have to fake the DC (typically 9V) across the RJ45 pins or hardwire the sense pins on the controller to keep it enabled.

I'm not going to it's impossible to do but I doubt you would find it worth the effort.

Brian.

 

[neazoi]

The RJ45 signals are not square wave either so even if you translated the voltage levels you would have difficulty slicing them accurately. The circuit is designed to run directly from/to the TX/RX pins of a network controller IC without the physical layer present. Some early network cards had these signals brought out to 'D' connectors, from memory, they had 15 pins but I'm going a long way back.

The network controller IC will not work at 1Mb/s, the clock speed is extracted from the data stream so unless it is within the capture range of it's PLL it will not be able to function. The component values in the schematic would also have to be changed. They serve no purpose other than to shape the output of the light detector diode so the pulses look more like they are from a direct connection. At a different speed you would have to adjust the values to maintain the wave shape sent to the controller.

A further problem with more modern (> 1990) network adapters is they need DC across the common wires to let them know they are connected. You would either have to fake the DC (typically 9V) across the RJ45 pins or hardwire the sense pins on the controller to keep it enabled.

I'm not going to it's impossible to do but I doubt you would find it worth the effort.

Brian.

Thanks Brian, It saves me a lot of effort!

Have a happy new year all of you guys