This thing is some kind of anomaly I swear. Performance doesn't even make sense anymore lol

Im open to input and advice, but this is also a bit of a sanity check as I’m having a major “wtf” moment right now.

TLDR; How on earth is the buggy BE14 outperforming a MT7916?


I got tired of dealing with the BE14’s issues so I decided to order two MT7916 from AsiaRF, specifically the AW7916-NPD. After a couple weeks of messing with the AW7916-NPD modules, I got “useable” range, and unstable speeds.

  • I could pickup the 2.4GHz from about 40ft away, but only around 10-20Mb/s
  • The 6GHz is a whole other story I won’t bother with here.
  • 5GHz…The speeds would initially be between 800-900Mb/s and within about 1 second would drop to around 200Mb/s.

So that was pretty confusing considering the MT7916 is a pretty mature card that I’ve seen used in numerous other applications. I started wondering if maybe the R4 just can’t supply enough power to both cards on the 3.3v rail.

Sure enough two of them was definitely pushing it, especially with an NVME drive, an AW7915-BMD and both SFP cages populated. I removed the AW7915-BMD and one of the AW7916-NPD modules…but nothing changed at all.


I regularly check both the BPI and OpenWRT forums to see what kind of updates there have been. I noticed someone on the WRT forums posted a new MTK 4.1 build. Well, I have some time today so why not try it out.

I put the BE14 back in to try it out, I didn’t even bother putting it back in the case. Heck I didn’t even attach the antennas to the pigtails. I just stuck some small heatsinks to the BE14 and fired it up.

Right off the bat Im maintaining speeds over 800Mb/s and fluctuating up-to the 960Mb/s range…with no case, inadequate cooling and no antennas…

So Im throughly confused at this point…lol


I have done a few modifications over the last few months:

  • Replaced the solid aluminum top with a perforated one for air flow.
  • Custom CPU heatsink and fan.
  • Separate heatsinks on the eMMC, RAM, and SPF+ cages.
  • Upgraded thermal pads on the BE14 and the NVME drive.
  • Upgraded to pigtails with much thicker copper shielding.
  • Placed copper pads directly under the antenna connectors on the BE14 soldered to ground. (Made no difference, but I can’t remove them now.)
  • Added a 150x85x12mm heatsink to the bottom. (More mass more better right?)

“Buggy BE14” :face_with_thermometer:

  • So, the software is getting better and better. :+1:
  • That’s good news, right? :partying_face:

Very true. Though Im definitely happy to see the BE19 is supposed to be going into production, and they addressed the lack of FEMs on the BE14. Fingers crossed it turns out better.