My initial review of the Argon ONE M.2 Case for Raspberry Pi 4 was fairly positive, with the exception of issues I experienced with WiFi connectivity. In conclusion, I recommended that potential users of the case only use it if they plan to use it with a wired Ethernet connection. Following on from this I received feedback from a reader indicating that Argon40 had acknowledged the issue and an updated version might be available by contacting the Argon40 customer support team ([email protected]).
I first received some new M.2 cases from Argon40 at the end of April, these contained M.2 board versioned 1.6, my original was 1.3. However, after spending around 3 days testing various board version (the M.2. board, the small video board and the top Fan board) I determined that these boards were, if anything, slightly worse in terms of their WiFi reliability/performance than the boards I’d started off with, in my original M.2 case. Contacting Argon40 once again, they confirmed that the changes to the M.2 board to address WiFi issues required a board version 2.1 or later.
I now have the correct updated M.2 board, time to see how do things look.
The board I received is labelled 04012021V2.2, so version 2.2 was created on 1st April 2021. The matt black PCB solder mask makes it difficult to examine the PCB in great detail, but you can see that the top right section which was previously a large ground plane stitched with vias is now clear of copper. On the reverse of the board, the area is also clear, with the exception of a couple of traces that go up to near the top edge, connect through to a couple of pads for an unpopulated resistor and LED. Not sure why these weren’t removed completely.
The creation of a “keep out” area on both sides of this presumably double-sided PCB is what fixes the WiFi problem. With earlier versions, you had the Raspberry Pi 4 sandwiched between the top fan controller board and metal case and beneath the M.2 board with all unused areas left as copper ground planes. Surrounding the Raspberry Pi 4 with all this metal created a sort of Faraday cage blocking the RF signals used by WiFi.
The WiFi antenna on the Raspberry Pi 4 is etched into the PCB next to the Micro-SD slot, so creating the area ideally above and below the antenna on the Pi will provide the best WiFi signal. Certainly blocking it with copper ground pours is probably the worst thing you can do. I can only imagine the reason 5GHz WiFi worked a bit better is that the shorter wavelength somehow managed to find a path through all the metal.
Updated USB to SATA Controller
Whilst examining the M.2 board I noticed that the USB to SATA controller has changed on the latest V2.2 M.2 board. The original featured an ASM1153E and the latest one an ASM235CM. That’s a change from a USB 3.2 Gen 1 to a USB 3.2 Gen 2×1 controller, though the Raspberry Pi can’t take advantage of the maximum performance of either controller, so why the change. I asked Argon40 and the answer was simple, “a lack of supply of the older chip in the market. It is more expensive but we have no choice ” – Joseph Zapanta, COO, Argon40 Technologies.
Beyond the M.2 Board
The Argon team haven’t just removed the copper fill on the M.2 board. I’ve now got 3 versions of the upper Fan controller board. The original 200827V2.0 (not sure on the decoding the date code, maybe 27 Aug 2020), a later 02102021V2.8 (10 Feb 2021) and the most recent 04212021V3.2 (21 Apr 2021). Both the V2.8 and V3.2 fan boards have removed the copper fill from the small section of PCB between the GPIO header and the IR LED. In my testing, the changes to the top fan board alone didn’t help the WiFi signal at all.
WiFi Testing in a home environment is subject to many uncontrollable factors, with the Argon ONE M.2 I attempted to test with all the different PCB versions of each of the 3 board, the M.2 PCB, the Fan PCB and the Video PCB. I also tested with 2 types of M.2 SSD drive. The test was a simple test, measuring the time to download a 100MB file ten times, with the device under test just 3.7m away from the access point. I also tested different WiFi channels, 1, 11 and 100 (5GHz).
Changing the Video and Fan boards didn’t make a significant difference, so for simplicity I’ve excluded those variables from the graph.
As noted in my original review of the Argon ONE M.2 case, 5GHZ WiFi performance is good, with the faster SSD storage allowing it to outperform a bare Pi 4 with an SD card.
Using the 2.4GHz channel 1 the new V2.2 M.2 board completed the test with an average time a couple of seconds behind the bare Pi 4. Version 1.6 of the M.2 board wasn’t able to obtain or maintain a connection, the earlier V1.3 M.2 board managed to complete the test but it took nearly double the time of the unenclosed Pi 4.
Channel 11 provided better results for the new V2.2 M.2 board getting a slightly faster average download time than the bare Pi 4, but probably within the margin of error of the test. The older V1.6 and V1.3 board couldn’t obtain a signal whilst using channel 11.
Firstly, I must give credit to Argon40 for the way they handled the problems with WiFi performance. Ok, they haven’t issued an immediate recall of all stock from the channel, as not all customers will have issues (those using wired connections, in particular, will be fine). The lack of a recall of stock does mean if you purchase a case new today, you may get one that will suffer from these problems. This decision may also be in part due to the current and ongoing silicon shortage, which may mean the revised M.2 boards are lower than they’d like, so selling what they have makes some sense, if they accept the cost of needing to handle the replacement process, and perhaps some slightly disgruntled customers.
Argon40 have at least accepted their mistake, fixed it in a revised design and customers contacting customer support will get you a replacement. No debate about if it’s a problem or not, just owning it and doing the right thing. It’s what I’d want to see from any company.
Finally, I’m pleased to say the revised Argon ONE M.2 Case I tested addressed the biggest issue I had with the case, and I’m now happy to recommend it without reservation. (OK, I’d have preferred that the M.2 case supported NVMe drives, just to save on my limited testing budget.)