The main difference i found between 1.1 and 1.0 is the on board eeprom were configured (address strapping resistors) wrong and it was on address 0x50 instead of 0x57. This can easily fixed with some micro-soldering skills.
As far as I can tell the 1.0 boards sent out have their R73 removed, so the eeprom doesnāt cause any conflict. @frank-w can you please verify if your board has R73 or not?
Mine also came without an RTC module, a feature that I donāt mind not having, but the USB-C works on mine, it has the correct part installed, unlike @dangowrt 's board.
So the board you buy now should be a 1.1 and have all these issues corrected.
This is the WiFi7 NIC we plan to make in the near future,
but since the MTKās driver is not yet perfect, so it will not be released until 2024Q1 at the earliest.
Do you refer to mt76/master or net-next? Do you have a commit id where it is fixed? Strange that daniel has fixed this by using higher voltage on same socket,but maybe explains why the openwrt i tried was workingā¦but have not enabled all wifi devices
This new version of wifi nic has antenna aggregations (do not know correct name) to reduce the count of antennas?
We replaced the original MT7975N (2.4G, 4x4) and MT7977B (5G, 4x4) with MT7976C (2.4G 2x2 + 5G 3x3), and then used a diplexer to aggregate 2.4G and 5G together (same as R3mini),
although The overall speed is reduced a bit, but the number of antennas is reduced significantly, which we think is worth it.
As for the 14-antenna version, we will check it later and decide whether we need to continue doing it or not.
I might be asking a dumb question, but does the SFP port support a Direct Attach Copper (DAC) cable?
In my use case, my router is a PFSense PC with a x520-da2 card. My plan is using the BPI-R4 as a wifi7 access point running OpenWRT.
I have always used these x520-da1/2 cards for direct connection between servers / computers that are nearby each other. Less complications of using a module and should save energy/heat as well.
Yes, Iām mainly using the LAN SFP port in that way, and works great in 10GBase-R mode (with Mellanox, RealTek and Marvell link partners tested so far).
Same here, much less heat problems and less consumption, 10GBase-T 30m range copper PHYs eat about 2W more than a DAC cable on each end, so for one link thatās ~ 100Wh a day, so 35 kWh per year, and thatās real money, can easily buy you that DAC cableā¦
Another questions, on the PoE version, is the BPI-R4 being powered by PoE? Or is it actually powering another device?
I assume itās the one being powered, but just wanted to confirm.
Thatās correct. The PoE-version of the R4 (which isnāt produced yet and cannot be bought anywhere, as far as I know) can be powered over PoE on the 2500M/1000M/100M/10M RJ-45 port present instead of SFP2/LAN.
itās super easy to make however.
You need to remove the cage, solder in a LPJG0926HENL connector and an RT5400 PoE module and your done. You have the PoE model.
