Check your OpenWrt web interface for the noise level (lower is better) and see where it’s at before and after making changes.
Also take a look at the TX power settings available for your Wi-Fi card.
I’m still confused that the interfaces are called the same, and the sfp interfaces are called eth2, although I saw someone call them correctly.
My wifi card is be 14 with standard antennas.
in addition to the poor signal, i noticed that connected devices are reluctant to switch from 2.4 to 5 ghz even close to the router. 6g doesn’t start at all.
as a firmware I installed 24.10. 1 with OpenWrt Firmware Selector
and before that there was 24.10. 0 with patched bl2 for expansion of ddr memory
Woah -61 is real bad lol
The new SFP interface names have not been backported from snapshot yet. The noise level of -61 dBm is dire, that explains your issues. Things to do next:
- Try switching to a different channel. Find one that has the lowest noise floor.
- Put some devices (transceivers, DAC cables) into the SFP ports.
- If none of that works, try disconnecting the pigtails and check the noise level by logging in over Ethernet. If you see a drastic improvement, the pigtails are at fault.
Always make changes with only one band active (2 or 5 GHz). Reboot after disabling the other bands.
Band preference if you have the same SSID for different bands can be flaky without proper band steering (see usteer package). Lowering the 2.5 GHz Tx power (e.g., to 15 dBm) might help.
P.S. For the 6 GHz band, check what’s going on in logread when enabling it. Most likely, you have the wrong channel or bandwidth for the selected region.
Going back to my own experiments, I’ve had the EMI shield cans laser-cut to exact dimensions. Didn’t seem to change anything. Although, after I moved my BPI-R4 into a 3D-printed enclosure I designed with a proper grounding strip for SMA connectors, the noise level has been more consistent at -91 dBm on the 5 GHz band and 80 MHz bandwidth. The 6 GHz + 320 MHz station sits at -80 dBm. I still need to keep the SFP cages populated, but in the end I’m satisfied with this kind of performance. My MacBook has the noise level in the -90 dBm range, too, so it doesn’t get any better than that overall. As for the 6 GHz/320 MHz, that’s for close distance anyways, meaning noise is less of an issue, and we always have up to 10GbE over wire.
1 Like
- changing the channel gave me a couple of units, it became 80.
- connecting the SFP optical module did not improve the situation.
- then I disassembled the case and left only 1 antenna, it became 85-86
- after that I disconnected all the pigtails and it became -89-90
- I also tried to run the board in my hands, without the case, I got -91-92. Provided that only one 2.4 network was turned on. What should I do, change the pigtails and antennas? P.S. (from point 1-3 I went to check the range, there was no result)
You can keep the antennas, but if you’re using the kind of pigtails that are thin and black, it’s worth switching to RG178 cables (RP-SMA to U.FL connectors) instead. Also, if you only have one SFP module, try inserting info different cages. The metal case is a huge problem, though. If you’re willing to use a 3D-printed enclosure and provide some sort of cooling for the BE14 chips, there are several designs floating around.
Do this for the rg178 cables you’ll.want 20cm for.the cables
So it’s not worth having the shield can made?
I’m not sure I’d like to change the case, I just need the wifi to penetrate a couple of meters behind the wall, I don’t have a large area or many walls, the router is near the wall, it should penetrate it. and a couple of meters of free space behind it. Will I have enough pigtails for this purpose? I understand that I need 3 for 2.4-5 antennas, can I forget about the remaining 3 for 6 ghz? Well, in any case, I would consider replacing the case as a backup plan, I don’t know about cooling the chips, there are thermal pads there now.
I believe one of the reasons Synovoip decided not to install EMI shielding is the metal enclosure, which serves the same purpose of keeping outside interference out. But the metal case ended up doing more harm than good, so it is a good idea to add some sort of shielding after switching to a plastic case. My RF environment is relatively clean in the 5 GHz band thanks to concrete walls around the apartment, but it might be a different story for other use cases.
1 Like
Yes, I’d start with replacing the pigtails for RG178. If you’re planning to use 6 GHz, replace those, too. Also, buy the cheapest device you can (100M RJ45 transceiver, DAC cable, e.t.c.) to put into the second SFP cage so that both of them are populated. Find the Wi-Fi channels that work best for you. All of that combined should make a huge dent in the noise level.
Did I choose the right pigtail? I’m confused that the original ones are not completely round, but flattened on one side, they might not fit into the crop box without modifications. I’m also wondering, why do you need a second SFP?
Yes, it’s the right one for the standard antennas. The radio noise most likely comes from the power regulators on the BPI-R4 board. SFP modules will absorb some of the noise or shield the Wi-Fi card from it.
which pigtails are shipped from sinovoip? rg178? i am too lazy to open the case, remove the screws of the board, etc. :>
The thin 1.13mm ones.
thanks, what size should the cable be? somehow i am unable to find anything else than 1.13mm (europe/germany)
Look for RG178, RP-SMA to U.FL.
i did, all i found (on aliexpress at least) was 1.13mm with RG178. on the german shops i did not find any information about the cable at all 
Or is the RG178 always the correct cable size? Is this the standard? Because all cables i found with RG178 did not show any information about cable size on german shops
You only need to look for the correct cable type and connectors. The RG178 standard takes care of the dimensions.
1 Like