8-port 100M switch RTL8309N

This is an eight-port 100M switch, which has been verified and tested 7*24h (no minor problems have been found for the time being).
 
I made a bill of materials and downloaded it in the attachment. (It contains LiChuang coding and TB shopping cart one-click copy homework) ! ! !
2022.9.16 I am making an acrylic shell! ! ! (It will take about a week)
At the same time, I will make the shells of five-port 100M and eight-port 100M switches! ! ! (It will take about a week)
 
The following are my debugging experience and insights! ! ! The chip used is from Realtek. Model RTL8309NB-VB-CG (new) or RTL8309NB-CG (old) In fact, the two chips are the same, but the difference between the new version and the old version is that the new framework has lower power consumption. The reference datasheet is marked with the RTL8309NB-CG chip model.
 
This 8-port 100M switch has an internal LDO input of 3.3V and output of 1.0V (but there are only a few words about LDO in the datasheet). I guess the usage and instructions are the same as those of my other open source project 5-port 100M switch. After all, they are chips from the same company (RTL8305NB-CG datasheet has detailed instructions on the internal LDO).
 
There is also a polarity correction function. At first, I thought it was just a matter of swapping P and N. When I read the datasheet again, I found that polarity correction is only used in 10M mode, and polarity is irrelevant in 100M/1000M.
So in order to make the routing beautiful, I swapped the polarity of the winding difference, making the entire layout more beautiful without so many windings. (I suspect that the RTL8305NB-CG chip also has this function, but the datasheet does not say, and I have not tried to verify it.)
The schematic diagram is not parallel to indicate those that have undergone polarity swap.
 
The material selection is very casual, using whatever is available (in fact, I choose whatever materials are available in the school), and you need to change the type or electrolytic capacitor yourself. There was not much heat when it was running at the beginning, but it got a little hot after it was fully connected. I need to open source a heat dissipation copper block.
 
As for the shell, I use whatever I have. I just use eight copper pillars to support it (I plan to customize acrylic for the shell, and I am currently learning how to make acrylic).
 
As for the cost, the switch chip is 8 pieces, 3.3vLDO is less than 2 pieces, and each 100M Ethernet interface is less than 2 pieces. The total cost should not exceed 20. Compared with the 8-port 100M switches on the market, it seems to have some advantages. Study it and take it away! ! !
 
 
If it is useful to you, please like and collect it to support my research and development.
If it is useful to you, please like and collect it to support my research and development.
If it is useful to you, please like and collect it to support my research and development.
(Important things should be said three times!!!)