Argon ONE Pi 4

The Argon ONE was first released for the Raspberry Pi 3 following a successful Kickstarter. The designers Argon Forty have recently released the Raspberry Pi 4 version which has adapted to the changes the Pi 4 introduced.

This is the most beautiful enclosure I’ve tested in terms of both form and function. Last month I reviewed the Argon NEO which is the impressive little brother of the Argon ONE. They share some construction materials and a clear common design ethos.

The decision to invest in developing a board to allow all of the standard ports to be mounted at the rear makes this, in my mind, an ideal candidate for a Thin Client platform, a tiny form factor Linux PC or an elegant media player. Both in appearance and performance, it’s leagues ahead of the official Raspberry Pi 4 Desktop Kit, which uses the official Pi 4 case which I’ve proven time and time again isn’t fit for purpose.

Design

External

Like the FLIRC, Argon NEO the Argon ONE uses an aluminium enclosure to wrap around and protect the Raspberry Pi 4 and also act as a huge heatsink. The massive surface area provides impressive cooling, that even under maximum load for a sustained period didn’t even get to 60°C (more details below), well away from thermal throttling point of 80°C. The case has an integrated 30mm fan which can be controlled once some additional software has been installed. Ventilation slots are cut into the top case section, the rear removable top panel has fins and a channel to direct air into the case through a gap next to the external GPIO connector.

The rear top panel segment is attached with two magnets, once removed it reveals access to the full Raspberry Pi 4 GPIO header via a colour coded connector, with designation screen printed on the case. The connector has been effectively rotated to allow most HATs to still be connected with them overhanging the rear of the case. The pHAT I tried fitted almost flush to the rear of the main case. In reality, I’m doubtful that many people would use HATs with this case. The lack of access to the 4 PoE Pins means the POE HAT can’t be used. Whilst GPIO access is provided, it’s not possible to easily use the Pi Camera or Display ports.

Front of the case is shaped to allow the translucent lower section of the case to rise up slightly right, allowing the LEDs on the Pi to be seen. There also appears to be the option of installing some IR transmitters and receivers, though at least on the original version these weren’t supported, it remains unclear if they are now, or ever will be.

The rear of the lower section of the case has a small opening to provide access to the SD-Card.

The rear of the Argon ONE main case provides access to the Ethernet, 2 x USB 2.0 and 2 x USB 3.0 ports you’d expect. In addition, the Power, AV and twin micro HDMI connectors have been represented at the rear, creating a cleaner design and simplifying cable management.

The final component on the rear of the Argon ONE case is the power button. Depending on the state of the Raspberry Pi the button performs different functions (requires Argon ONE software to be installed)

Argon ONE – Power button functions chart

The software needed for the Argon ONE is installed via the command: curl https://download.argon40.com/argon1.sh | bash

Internals

The Argon ONE provides a small daughterboard to expose the 2 x HDMI and 1 x AV port on the rear of the case. This is inserted into the Pi 4 before being installed into the case. Note the USB-C power connector on the side of the Pi 4 is not used, instead, the USB-C input power is provided through a larger board mounted to the top of the case.

The larger board provides the USB-C power connector, power button, a fan header for the installed 30mm fan along with providing the means of presenting the Pi GPIO header through the top of the case. As mentioned earlier there are a pair of IR pad on the board, but they don’t appear to be supported at present.

The larger board has a cutout allowing a couple of large towers of aluminium through to reach the Pi 4’s processor and memory, once the supplied thermal pads are installed.

Once the daughterboard is installed into the Raspberry Pi 4, and thermal pads attached, the boards are installed into the main case by carefully aligning the GPIO pins into the 40pin connector on the topmost circuit board. These are then secured with 4 screws before the lower case is installed and secured with a further 4 screws. Finally, 4 rubber feet can be installed on the lower case section, SD card installed and powered up for the first time.

Noise

After booting and installing the Argon ONE software, the internal fan can be configured using the argonone-config command. By default the fan runs at 10% at 55°C, increasing to 55% at 60°C and 100% at 65°C, the utility allows custom thresholds and speeds to be set, or even for the fan to be set to always run at 100%.

I found the stock fan a little noisy, not terrible, but not great even at low speeds. Removing the fan revealed it as an RDKCooler RA03010HD5 30mm 5V DC fan which has a noise level of 33dBA, considerably higher than the Sunon MagLev MF30100V2 I’d tested in a previous case test.

Make/ModelCurrent
(mA)
Speed
(rpm)
Air Flow
(CFM)
Pressure
(mmH²O)
Noise
(dBA)
Price
(£)
RDKCooler
RA03010HD5
(30x30x10mm)
250110004.94.933?
Sunon (MagLev)
MF30100V2
(30x30x10mm)
8095004.74.3214.80

The Sunon fan is quieter, but still not close to silent, so I suspect much of the noise is caused by turbulence generated pushing the air through the case. The noise level increases with the top GPIO cover removed.

Price & Availability

As Raspberry Pi enclosures go, the Argon ONE is definitely on the more expensive side at $25, though really I’m not sure there are any cases that are comparable to what the Argon ONE delivers.

I sourced this from SeeedStudio.com as part of a larger order at a time when there weren’t any international stockists. Since then more international distributors are stocking the Argon ONE. At the time of writing I found it available at:

Performance

The results above show the impact on CPU temperature of running a 4 core stress on the Raspberry Pi 4 for 1 hour. Without fan running the maximum temperature reached was 59°C, well below any level of thermal throttling. Enabling the fan and its default profile (10%@55°C, 55%@60°C, 100%@65°C) shows that even under a sustained max CPU load, the Argon ONE only needed to run the fan at 10% speed.

WiFi Signal Strength

As noted in my previous case review post, metal enclosures tend to have a negative impact on WiFi signal strength.

WiFi Testing Results

Measurements of WiFi signal strength use dBm (0 to -100), closer to zero is better. A simple description of WiFi signal strength can be found here.

2.4 GHz Results

The first graph shows the averages from the tests for each configuration. The second graph shows all the measurements, to highlight where some configurations had patchy signals, not getting reading from an AP in all 5 test runs.

5GHz Results

The first graph shows the averages from the tests for each configuration. The second graph shows all the measurements, to highlight where some configurations had patchy signals, not getting reading from an AP in all 5 test runs.

WiFi Summary

The Argon ONE performs very slightly better than the NEO and FLIRC cases. But as expected the metal case still has a measurable impact on signal strength.

The 5GHz channel 108 access point wasn’t seen in any of the results today, this AP had the weakest signal in the last round of testing. Possibly down to atmospheric or other electrical interference that wasn’t present during the previous testing.

Conclusions

This case stands out from all the other cases I’ve reviewed, it performs well and looks great. The Argon ONE form factor provides some great advantages for those seeking a more sleek case. There are some limitations, such as the lack of PoE support, and lack of Camera and Display port connections, but I’m doubtful these are problems for the scenarios this case is designed to be used in.

In the future I might try and set this up as a Citrix thin client, potentially exploring PXE boot to avoid the need for local SD card storage. Though I’m not sure if this is possible.

Parts Tested

Note, Amazon UK links are affiliate links, meaning I may make a small amount of money in return from you following the link and buying the product.

22 thoughts on “Argon ONE Pi 4

  • 19th May 2020 at 3:00 am
    Permalink

    I too find the stock fan too loud. Especially considering how little the fan actually helps the CPU temperature. Thanks for the awesome graphs!

    Reply
  • 29th May 2020 at 2:23 pm
    Permalink

    Can you not swap the fan for another? If you could, which fan would work?

    Reply
    • 29th May 2020 at 2:53 pm
      Permalink

      Hi Taylor, if you take another look you’ll see I did test with another fan. “The Sunon fan is quieter, but still not close to silent, so I suspect much of the noise is caused by turbulence generated pushing the air through the case. The noise level increases with the top GPIO cover removed.”

      The case design is such that it is great for passive cooling, due to the thermal mass. Active cooling under low loads that rarely spin up the fan might be ok. However, if you are running an intensive workload the fan noise is likely going to be annoying.

      Reply
      • 29th May 2020 at 4:08 pm
        Permalink

        Thanks, Martin. I saw that. Let me rephrase my question. If I replace the fan with the quieter Sunon fan is it as easy as taking the stock one out and putting the new one in or is there more to it. Great read btw. Thanks 🙂

        Reply
        • 29th May 2020 at 4:34 pm
          Permalink

          Ok, the stock fan is connected via a 2 pin JST connector (IIRC) so to replace the fan you’d need to either fit the same style connector to the replacement or as I did, just cut the wires solder the new fan to the remaining wires to the connector and protect with heatshrink. Hope that helps.

  • 7th June 2020 at 9:53 am
    Permalink

    Hi, my question is do you use the 11000 rpm or the 9500 rpm fan ?
    MF30100V1 or MF30100V2 the db level is 23 for the 11000 and 21 db for 9500 stock fan is 33db at 11000rpm

    Reply
    • 7th June 2020 at 10:01 am
      Permalink

      Hi, I used the MF30100V2 I would have like to have tried the MF30100V3 which runs at just 6000rpm and claimed noise of just 10.2 dBA, however when I was doing some 30mm fan testing some while back they weren’t in stock anywhere here in the UK. Checking now and Mouser has them, but if only ordering a couple of fans the shipping costs are a bit steep. So won’t be testing those just yet.

      Reply
  • 8th June 2020 at 8:56 pm
    Permalink

    Hi, tnx for the info 🙂 last question do the script work normal if you use the 9500rpm fan ?

    Reply
    • 8th June 2020 at 10:09 pm
      Permalink

      Hi Stan, Yes script will still control the speed of the fan as before. The fan is only connected by two power wires back to the control board on the Argon one, the script varies the signal to control the speed of the fan, the fact different fans spin at different rates doesn’t matter.
      Martin

      Reply
  • 26th June 2020 at 3:02 am
    Permalink

    Hi, and I also agree that the ArgolnONE Pi4 is an exceptional case for the Raspberry Pi 4. I purchased two use in a development effort that uses a Relay HAT and Precision AD/DA HAT stacked on the GPIO connector. The problem I find is the RPi 4 will not start when the 2 HAT stack is connected to the GPIO. However if I start the RPi 4 first and then connect the 2 HAT stack everything works as you would expect, or if I power the RPi 4 directly it will startup with the 23 HAT stack connected. Have you experienced this problem in the past? I don’t like connecting the 2 HAT stack to the RPi 4 board when it’s powered up.

    Reply
    • 26th June 2020 at 8:27 am
      Permalink

      Hi, I’ve not tried > 1 HAT with the Argon. I’ll see if can do some experiments over the next few days.

      Reply
      • 26th June 2020 at 10:54 pm
        Permalink

        Yes with the PI out of the ArgonOne case I’ve been able to boot with 2 HATS and a TFT dispolayh on the GPIO.

        Reply
        • 27th June 2020 at 12:20 pm
          Permalink

          Harold,
          Are any of your HATs using GPIO 4 as that is used by the ArgonOne for the power button? Or the I2C address 0x1a which is used for controlling the fan speed?

          If no obvious conflicts, then I wonder if the ArgonOne has an eeprom on it’s internal hat like board? Per: https://forums.pimoroni.com/t/stacking-hats-causes-rpi-not-to-start/8995/4
          “In short, if your hats have no ID eeproms, you can stack as many as you want, assuming they don’t conflict with each other. And any “one” of them can have the ID eeprom. It’s only an issue if there are two ID eeproms. And even then, I think you can manually set them up to work. The issue is the PI can’t address the ID’s separately, it will read one ID that doesn’t match either HAT. Thats how I read it anyway.”

          From a brief bit of reading I’m not sure how to check the presence of an eeprom on a hat, since it appears the this happens on an i2c bus which isn’t available after booting.

  • 28th June 2020 at 6:56 pm
    Permalink

    Martin

    Thanks for your response and suggestions. I’ve been troubleshooting the assembly in an attempt to determine what is causing the ‘No Boot’ situation. I have determined the problem is on the daughter board that connects the fan and has the input power circuits and GPIO reroute circuits. I checked that the HATs I’m using do not use CPIO 4 or I2C address 0x1a.

    Interestingly it appears that CPIO 4 is always at ‘ground potential’, a condition I would not expect if it’s being used as a control point.

    I’ll persist with my efforts and I’ll post any major successes I have.

    Reply
    • 28th June 2020 at 7:01 pm
      Permalink

      Harold, GPIO 4 is configured in pulldown mode, so will be at zero volts until you press the power button. I believe if any of your HATs are using classed as Official HATs then a) They should have an eeprom for identification, there may be a risk of conflict here. b) The official HAT standard doesn’t support stacking HATs.

      Reply
  • 28th June 2020 at 9:57 pm
    Permalink

    When I separate the stack I get the same results when I connect one specific HAT (WaveShare High Precision AD/DA Board). My other HAT (DockerPi 4 Channel Relay) does not exhibit the pro blem. The WaveShare Specs indicate GPIO 4 is not utilized. Does the official standard specifically exclude stacking or just does not mention it?

    I’ve been using the RPi 4 but I’m now reverting back to the RPi 3 to see if it makes a difference. I don’t expect it will but then again you never know…

    Thanks again for your comments and suggestions.

    Reply
    • 28th June 2020 at 10:54 pm
      Permalink

      The spec doesn’t cover stacking HATs, as such it’s not officially supported, however as you know it can be done and mostly works. This Adafuit page shows how each HAT needs to have its own address if using I2C, which is connected on the WaveShare.

      My ArgonOne shows that the Fan controller is using address 0x1a, via the command: `i2cdetect -y 1` I’d suggest you check what address each of your HATs are using, specifically the WaveShare HAT to determine if there is a conflict. Though I’d I’m not sure an address conflict should cause the Pi to fail to boot, but worth a quick check.

      Reply
  • 2nd July 2020 at 5:15 am
    Permalink

    Is there a PI 4B version of this shell?

    Reply
  • 2nd July 2020 at 6:37 pm
    Permalink

    The Argon One Raspberry Pi 4 case is a great concept but it does fall short when HATs are used with it. You can see from my previous replies the nature of my problem and what I’ve determined previously. Martin in followup to your earlier suggestions I am happy to report that I do not have any address conflicts between the HATs I’m using. My I2C addresses are 0x10 and 0x1a for my relay card and Argon One case respectively. My other card, the High-Precision AD/DA will be addressed via SPI if ever I can get the assembly to Boot when it is installed.

    I’ve been told by Waveshare Sales Team that the product (Argon One case) has current protection, and installing the relay module causes the instantaneous current to be to high and is causing the problem.

    I find this to not be the case in my testing as the relay cards I’m using do not cause a surge of current from the power supply on startup. In any event I’ve ordered some different AD cards that will be addressed with I2C and I’ll post future results here as I can.

    I’d hate to see the ArgonOne case be less than a fully functional product because in all other aspects it’s a great idea, design and product in general. Then too it might just be that after 60 years of Electrical Engineering I just hate to give up…

    Reply
    • 2nd July 2020 at 6:43 pm
      Permalink

      Thanks for the continued updates. If you like the Argon ONE but don’t want their board/fan controller getting in the way, you could look at the Argon NEO, but that doesn’t’ provide all the ports at the rear like the ONE does.

      Reply

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.