An embedded engineer friend made one, and it wasn't too hard. The TSOP chips really cut down the complexity. Doing it with a 100 kHz frequency is really going to be hard. The chip takes care of automatic gain control, the modulation does most of the ambiant light filtering and CFL and florescent lamp interference, The receiver has a built in filter. Choosing one that supports NEC5 coding would probably be a good choice. That protocol specifically sends a burst just for the AGC even though you would not be using the protocol.
Reading has suggested using both a narrow and wide emitter for remote controls.
In this case, there is the audio problem that has to be dealt with too. Hopefully, some mode which can play audio based on some trigger.
It's not even clear we are trying to build the "same" project in this thread, but I could see a single gang box for the emitter and a dualgang box for the detector side. The detector could contain a speaker. I'd consider braille markings on the IR plates.. You could put say a 1/4" thick, 1" wide strip of wood from floor to ceiling with a cutout for the box. That would allow a blind person to find the plate and "read" the braille markings. Primary use would be direction of travel. Make one side 1" wide. and the other 2" wide. You could code N, S, E and W by board sizes, above and below the detector.
e.g Count in binary N (00), S(01), E(10), W(11), S; So, N would be two "wide boards", south would be a "wide" at the bottom and a narrow board at the top.
the Mneumonics are you need to remember N, S, E and W order, the LSB (least significant bit) is toward the floor
Wide represents a "0" and narrow represents a "1", Note that a 0 is "wider" than a "1".
How does one determine the direction of travel?
You could get weird and put the speaker in the ceiling. It's not necessarily weird because ceiling speakers are available. and it saves the expense of cutting a grille in an outlet cover plate and having a small speaker. IC's that support 12V power are widely available and you may find many ready made amplifier PCB's. All likely stereo.
You might have to use PTFE jacketed fire alarm cable and you have to buy that stuff in large rolls. The conductors are usually solid..
Sparky ran the cables and mounted the big stuff. I only had about 10 cables and a few locations, so the cables were labeled like 1A2 and 2A1 which is cable "A: from location 1 to location 2 and from location 2 to location 1.
So, when you have to have a bunch of cables from say the "power supply", the best way would be just run the cables into an enclosure with DIN rail terminals. The use a piece of conduit to the power supply. If you have to replace the power supply, all you have to do is replace the patch cables.
It ALWAYS makes sense to terminate the wiring to DIN blocks and use patch wires within the enclosure. I can;;t stress that enough.
Anyway, the cables ran in conduit on the wall to above the ceiling and the conduit had a large junction box mounted to the top. The junction box had individual cord grips for each wire, so it was basically a pull port. Fire is unhappy because it can;t travel or emit nasty smells. Class II wiring doesn't need conduit.
Sparky also mounted flush strobes and mushroom panic buttons. I did not pay attention to how that was done.
During a move and expansion, I did have to penetrate a firewall with LOTS of cables. They wanted to insert a PVC pipe, then bundle the cables and fill with firestop - Yuk. 6 or so cables that went to hydrogen sensors had to be handled separately and in their own conduit.
Sparky was nice and came up with a large box for running cables that helped connect everything low voltage together.
A wiring trough or wireway e.g.
https://www.zoro.com/wiegmann-wirin...x-36ind-rscg040436/i/G0749384/feature-product my panel, the power supply, the battery box, the hydrogen alarm (wall mounted). Six pieces of tubing to sample the gasses. One cable to the fire alarm panel. There were at least 3 other cables that had to exit and penetrate a firewall..
These are just ideas when "brainstorming".