A long time ago, when I was still a rookie, I used the HiSilicon 3531 solution to port the company’s software (using the DVR solution to make an NVR).
I feel that the correct way to use BPI-R4 should be to use the closed-source OpenWRT to modify the system(be WiFi router, if not, frank-w’s image better).
Optimized web interface, optimized setting function.
Later, I felt that the problem should be as netizens said, the WiFi NIC hardware design may not be perfect, which may be the reason why BE1900 will be released.
So, regardless of the product introduction, functions, and utilities of BPI-R4, I will wait for BE1900 to see the situation.
If you don’t look at the WiFi function, compared with Raspberry Pi, BPI-R4 is a very good and very distinctive development board product.
Among the development board products currently on the market, there are almost no SFP+ networks, almost no 3 PCI-Es, and almost no two.
The only comparable ones are Intel N100/N150/N200 series products, but the price and scalability are different(and no GPIO 
).
Open source” and “developer board” are 2 different things. “Open source” means that the sources are openly available to everyone. This applies to both pre-release versions and finished products. “Developer Board” means that this board is to be seen as a basis for later own products. That said, the R4 and OpenWRT are two completely different products that have absolutely nothing to do with each other. The R4 is compatible with OpenWRT yes, but Banana Pi does not develop OpenWRT. OpenWRT has its own development cycle, independent of Banana Pi’s products. That’s why I don’t understand why it is always said that the software for the R4 is immature. When Banana Pi launches a new board on the market, OpenWRT doesn’t really care. Sure, the snapshots try to support the new boards, but that doesn’t affect the current major version. If there is a new major version at some point, the new boards will probably be supported. However, since OpenWRT is open source, it is up to the community to correct existing errors.
whereas “open source” with broadcom/MTK and the other china IP-stealers is more like “we give some blobs that either work or not” scnr 
I guess, it comes down on the definition of „development board“ - for me it is something like a RaspberryPi/Arduino board: a working board with all major hardware problems solved, available as board with no case and software in a more or less rough state (aka not suitable for end consumer like with nice GUI or so - so some/deep technical knowledge is required).
But, sorry to say, this is more of an internal prototype/beta board which major problems and nobody from SinoVoip does ever report back on any findings or helps finding solutions.
We are all guessing what the problems are for months (e.g. noise problem, some boards limit tx to a few dB), but obviously we are mostly software guys working on OpenWrt and don‘t have the hardware equipment (eg. spectrum analyzer to look for rf noise hotspots on the board) of a manufacturer like SinoVoip to find the root cause of the hardware problems and how to mitigate it with some hardware modifications…
So I would like to see SinoVoip to be a bit more active and at least report back if a found issue could be related to hardware or what their opinions are, so we don‘t have to freely guess what is wrong on the board…
i could imagine that this is a legal problem when they “find” something and report it. at least in the EU, when you “find” a bug that is on hardware side, i am very certain it must be replaced. at least when its within the warranty time. i would love to see the new pro router but i will wait this time too, before i buy it. if it also has those problems (heat, rf noise, m2 wifi compatibility etc.) guess i wont buy :>
I’m not looking for advanced features, bleeding-edge support, or to match someone else’s 1 km performance — just a reliable baseline firmware that actually works. Than I can add something on it.
Like many others, I kept the R4 because of its advertised expandability.
I own many routers and dev boards, — even lesser-known ones like the Radxa Orion O6, which has been an outstanding experience. Wi‑Fi 7 is no longer new or experimental — even the stock routers and free optical modems provided by ISPs now support it, I also bought a COMFAST USB dongle (983BE) back in August 2024.
Other routers based on the MT7988A/D series work just fine. As far as I guess, none of those vendors wrote their own drivers from scratch. They’re using MediaTek’s SDKs, just like SinoVoip.
This isn’t cutting-edge R&D — it’s basic integration work. So if everyone else can get it working…SinoVoip, c’mon
From recent posts, it’s clear that BPi still wants to turn the R4 into a more powerful platform — and they clearly have access to MediaTek’s proprietary SDKs, without reverse-engineering .
At the very least, they should be regularly updating the clean firmware images, even if they’re not open-sourced — just to replace the broken ones currently online. or just a reference to make sure our board is fine.
Before I spend another $100+ on the BE19, and potentially countless hours more on top of an already $200 platform, there’s only one thing I want to know:
Can it actually perform better or not?
Maybe sinovoip just give us some numbers — even simple test results, like throughput at various distances, stability under load, would be enough. Right now, we all agree the R4’s Wi‑Fi can’t even beat a $20 AX router like the Cudy TR3000 (2Gbps+ over 5Ghz) , which also can runs OpenWrt. 
i can only talk for another vendor, there is a clear NDA from MTK for them, which means: nothing of that SDK must be made public, nor anything related to it. The funny thing is tho, those use GPL software and then forbid to comply with the GPL. Anyway, thats another topic…
no one knows that until someone is willing to buy it and do those tests. if you want to know such stuff, its probably best to write them an email. but i doubt that they will release any information when the product is not even available to buy.
COMFAST USB dongle
Funny, i have the same, and it works like crap… it states it is connected as wifi7 to my AP but i lose all the time the connection and the speeds are terrible. Really terrible (20-30mbit/s). My mobile phone and the pcie wifi 7 card do the job very decent.
I found the sibling of Banana Pi’s BE19 on AsiaRF:
Very good. I surely believe there are many others than just me waiting to see, if this would complete “the best WLAN access point/router available”. Because others than BE14 range issues, I still believe this is.
I don’t see a shield for it, had the noise issues of be1400 been solved with the be1900
I’m absolutely certain this is yet another piece of e-waste. As much as I don’t want to believe it, there’s a high chance something’s wrong again — and it’s all because this is a “development board”!
The BPI-R4’s CPU frequency is way too low. At a similar price point, the ASUS RT-BE88U is powered by a powerful quad-core 2.6 GHz CPU!!!
Looks like provision for the shielding, though? the edge lines do look like they are present. Would there be the shielding on already (on such speculated lines) the photograph would have very little to show.
The BPI-R4’s CPU frequency is way too low. At a similar price point, the ASUS RT-BE88U is powered by a powerful quad-core 2.6 GHz CPU!!!
Are you sure that Cortex-B53@2.6ghz (custom version of A53) surpasses Cortex-A73@1.8ghz?
Yeah there’s markings for shielding which the be1400 also had but didn’t really improve the noise at all from other people testing
Not necessarily mere markings - should be the actual ground plane the shields attach to (via soldering).
Impossible to say for sure, of course, from just that picture.
If they are, this will, of course, yield better shielding than just having metal plates.
Yeah I just forgot they were called, but no one is sure if it’s actually connected to ground plane due to the lack schematics
I’m not sure whether the Cortex-B53 @ 2.6GHz (a custom version of A53) truly surpasses the Cortex-A73 @ 1.8GHz in performance. But what I do know is that I have an ASUS RT-BE88U, and both its wired interface speed and Wi-Fi signal strength are far superior to the BPI-R4 combined with the BE14. Honestly, I really regret buying the BPI-R4 + BE14 combo—it feels like I just bought electronic waste. Even the official firmware is outdated.
I understand your frustration!
But you’re comparing things that should not be compared in this way!
The BE14 was described as an iFEM and had no shielding soldered:
→ I did not understand the difference between an iFEM and eFEM because I’m a laiman
. This why I cannot criticize it in general …
I know there are some other problems with certain BE14 boards (e.g. EEPROM issues). And we’re not even sure where the noise came from!
The OpenWrt (open source) drivers for Wi-Fi 7 are not finished yet.
→ We should talk about this again in a year… let’s see how OpenWrt 25.xx has improved!
Don’t get me wrong:
If you want to make a general statement, I think it’s a good idea to include the following sentence beneath your individual comments 
:
No one should assume that setting up a Banana Pi development router board is as straightforward as using a fully developed ASUS router.
Congratulations:
You changed your opinion and redesigned the BE19.
eFEM
- Now there’s a really big eFEM on the board — it even looks larger than the ones from AsiaRF
.
Shielding
- Much more improved — and now separated!
External power supply
- Did I see it right — is external powering possible?
Power consumption
-
As always, I’d be very interested in the power consumption:
- Max
- Idle
- 2.4 GHz only
- 5 GHz only
- 5/6 GHz only
-
That info would be really helpful! For comparing BE14 and BE19.
like this:
