I have a 500 W soldering plate instead of a proper preheater (like this one), and it is dirt cheap. To control the temperature I simply move the board closer or farther away from the plate and check with an IR thermometer until I get it to ~100-150 °C. Worked well so far for replacing SMD capacitors on graphics cards.
1 Like
I tested my measurements with 3D-printed mockups, and I’m finally confident to say they’re good to go. Here’s the DXF drawings and STEP models (bend radius accounted for) of the shield cans to keep everything in one place.
shield_can_top.step (252.4 KB) shield_can_bottom.step (473.6 KB) shield.can.bottom.dxf (11.0 KB) shield.can.top.dxf (13.5 KB)
Oh nice, waiting for you to do it with copper and see if it has any effect
1 Like
Cool - BTW why do you have made holes in the shields?
The eight holes in the top-side can are there to provide a semblance of airflow to the SMD components on the board beside the large ICs, which are going to be covered by thermal pads. It’s quite common in RF shields to make small holes like that, and they won’t hurt the shielding effectiveness. The holes on the bottom part are necessary to clear the tall SMDs on the main board, otherwise there just isn’t enough clearance between the PCBs.
Question about a shielding idea:
Has anyone ever tried to cover a thin metal plate with heat conducting pads on the inside and outside and use this as a shield?
Are such thin pads also available for a shielding between the BE1400 board towards the BPI-R4 board?
This may be a combined heat dispenser and shielding solution.
Hello, I am an electrical engineer. Thank you everyone for your time looking into the RX noise so far. I’m suspicious that the 3V3 regulator UP3 (MP8759) is one of the major sources. Possibly also U9, U14, U15 (MP2338). These regulators are all close to the BE14 when installed, albeit on the topside of the R4 PCB. Additionally, Betonmischer mentioning that populating the SFP cages improving the noise floor would make sense if the additional loading is stopping U9 from pulse skipping.
1 Like
But the problem with SFP cages is that the cage isn’t even connected to ground at all, not sure how it improves the signal
Since I don’t have a R4 on hand to test, can someone tell me if the two posts used for screwing down the BE14 are connected to the ground plane? I assume they are and was wondering if using tape and plastic screws so that the R4 ground plane and BE14 ground plane don’t connect through them makes any difference, better or worse.
I’d also be curious to know if a copper foil shield, taped to the bottom of the R4 PCB and soldered to the R4 ground plane specifically, makes any difference.
Hmm I would normally check but the R4 is being used as a main router for me
Yes exactly, I am saying that the SFP modules drawing power from the regulator is why it might be quieting the noise down. Nothing to do with the SFP cages being grounded or acting as shields themselves
Tbh tho I don’t think the SFP makes a difference, I got 2 filled slots and I still get -85dbm when 6ghz is at 80mhz width and -81dbm when at 160mhz probably even less at 320mhz
Yes I might be wrong, it’s only a theory and since I don’t have an R4 I can’t test it myself.
That said, -81dBm is a lot better than some other people seem to get. Example from openwrt forum:
Unknown if they have SFP modules or not, but a user in the same thread that is using at least one SFP is getting numbers similar to you
Finally guys some good news
1 Like
Yeah the channel selection matters too, have to find the channel with the least noise
Also, as mentioned in openwrt-forum, the value reported in LUCI/webgui might not be the correct noise value (all 3 bands always report the exact same noise level…), I now measure noise using the cli:
iw phy-2g survey dump | grep -B 1 noise
(replace phy-2g with the name of your device) - strangely (and probably the cause for the luci error), even if I only specify 2g-device, it reports noise for all wireless-channels:
root@OpenWrt:~# iw phy-2g survey dump | grep -B 1 noise
frequency: 2412 MHz [in use]
noise: -78 dBm
--
frequency: 5180 MHz [in use]
noise: -63 dBm
--
frequency: 5955 MHz
noise: -84 dBm
--
frequency: 5975 MHz
noise: -84 dBm
--
frequency: 5995 MHz
noise: -84 dBm
--
frequency: 6015 MHz
noise: -84 dBm
--
frequency: 6035 MHz
noise: -85 dBm
--
frequency: 6055 MHz
noise: -85 dBm
--
frequency: 6075 MHz
noise: -85 dBm
--
frequency: 6095 MHz
noise: -84 dBm
--
frequency: 6115 MHz
noise: -85 dBm
--
frequency: 6135 MHz
noise: -85 dBm
--
frequency: 6155 MHz
noise: -85 dBm
--
frequency: 6175 MHz
noise: -85 dBm
--
frequency: 6195 MHz
noise: -86 dBm
--
frequency: 6215 MHz
noise: -86 dBm
--
frequency: 6235 MHz
noise: -87 dBm
--
frequency: 6255 MHz
noise: -88 dBm
--
frequency: 6275 MHz
noise: -87 dBm
--
frequency: 6295 MHz
noise: -88 dBm
--
frequency: 6315 MHz
noise: -89 dBm
--
frequency: 6335 MHz
noise: -90 dBm
--
frequency: 6355 MHz [in use]
noise: -81 dBm
--
frequency: 6375 MHz
noise: -91 dBm
--
frequency: 6395 MHz
noise: -91 dBm
--
frequency: 6415 MHz
noise: -91 dBm
Yeah I saw that, I just report LUCI values so everyones on the same ball
1 Like
BPI-R4 seems to report the noise value for the last AP on the list. So, it’s important to turn off the 6 GHz band when, e.g., trying to find the best channel for 5 GHz.
Thank you for your input. Do you think it’s possible for the noise to leak through the 12V power circuit, too? The filtering on the BPI-R4 looks rather bare-bones to me. I could take a look at the ripple with an oscilloscope, but I’m not sure what PtP range would be considered normal for this application.
Would probably have to compare it to another 6ghz WiFi router to get some baseline