Metal and plastic RJ45 connectors, should you use STP of UTP cables?

Question

In model railway world. There exists an occupancy decoder (Piece of electronics to detect trains on tracks) which is named S88. It is nothing more than a bunch of shift registers, mostly CD4014.

You could chain up up to 16 S88 modules.

The earliest design used a flat cable (ribbon cable). But it proved to be extremely sensitive for noise at longer distances.

So the company which designed S88 made S88n. The VCC on the bus was raised from 5V to 12V and STP cables replaced the flatcables.

This is the pinout of the S88n bus

And one of the two CD4014

(The RESET line goes to a SR latch CD4044)

The S88n modules are wildly available and cheap. They are made by many manufacturers and there are several DIY projects of it.

One of the manufacturers has firmly claimed that the twisted pairs do add to the reliability. And that shielded cables have an actual bad influence. Mainly because the shield is not terminated.... in his own design. These connectors have plastic casing

The company which designed S88n in the first places, uses RJ45 with metal casing on their design. They predict and sell STP cables.

The bit timing is also relatively low (relative to the capabilities of the CD4014 chip). I found this website about bit timing. I know that the S88n is driven with a lower speed to avoid reflections and false readouts.

That leaves me with 2 questions: 1). If the RJ45 connector is not of metal and the shielding of the cable is left floating. Is it then true that using STP cables may reduce reliability? AFAAIK, you are would be adding an antenna.

2). If you do use metal RJ45 connectors and you connect the shielding to the GND of the PCB. Will STP cables increase the reliability?

Answer 1

Answer to question 1 us quie simple, if you dont have grounded connector on the board, you are not grounding the coat of STP, making the connection basically UTP. Sure there us some shield around, but if it is not grounded it acts more like "not there" then "there". I've seen this a lot in EMC issues people have with designing their devices. Metal shield is not really an effective shield if not grounded because the field doesnt couple into ground but to some random "potential" if you wish.

Answer to question 2. is more complicated. You cant say what will happen and what the environment will be. So no one knows if there will be a difference or not. But i'll but it this way, it's like putting tent in a tent. Chances of leakage will be smaller then with one tent. This analogy is good, because the twisted pair is already quite good at being robust to outside disturbance because fields coming from the outside induce in both cables equally and become common mode, meaning that receiver does not detect them at all (as it only look as difference between +s and -s, an this stays the same if disturbance is added to both). So you already have a tent. Do you need another one? Maybe, but it depends on your environment, will there likely be hail, etc and it depends on the use case, if you need the inside to stay super dry q100% it is vise to do it just because you can.

Same here, if the cables will be long, if you need very low BER, if you can't afford even occasional few miliseconds if issues kr if environment is know to be especially noisy with EMI (large currents, huge magnets, spark ...) it is a good idea