Pretty interesting to see that Arduino will be part of Qualcomm and that that new board they’re releasing is only $44. However, watching Jeff Geerling’s review, I didn’t realize the horsepower of the dragonwing chip is more akin to the RPi3 vs the RPi5…though the power envelope of this new Uno Q board is pretty low, they’re targeting 1W power envelope.
What say you?
It should be the last needed sign that arduino is not just for makers/tinkerers.
in the past qualcom chips have not been available like all the MCus on arduino boards were, while we see none pop up yet, there are already alleged hints that the non cutting edge processors might become available similar to how you can now buy normal mcus.
with fully open source reference designs (aka the UNO Q) this could open up a lot more possibilities for us all and push qualcomm sales quite a bit.
just think of all the newcomers today that might be decission makers in 10 years.
IMO edge computing needs to target as little power consumption as possible but the UNOQ seems to be a “sensible” choice to pull of as an initial board, a balance of risk to reward.
And here I was hoping that Qualcomm+Arduino would lead to a Cambrian explosion of LTE+ enabled microcontrollers. Without crazy, locked-in, vendor tooling. Let’s rant.
The parallels are staggering. In 2000s (now I’m dating myself) the only way to really do WiFi (802.11) as a hobbyist (and a fair amount of products) was to abuse an WRT54G (yes, the namesake for OpenWRT). Which meant rather complete Linux, overheads, separate DRAM and flash… bulk, power consumption, etc… ESP8266 (and later… way later, ESP32) cracked it open, and created the easy ramp to add WiFi to nearly anything.
We are in a similar situation with LTE; the only good way to build an embedded 4G+ device, is some form of a linux computer. Even if you’re literally just bridging UART to HTTPS, you’re taking on everything from PPPD to QMI and a full OS.
Yes, most /all?/ LTE modules are programmable. If you squint. And deal with NDA’d datasheets, vendor (ehm, model!!) lock-in, ancient toolchains etc. Not very different to how Atheros radios were “programmable”, before ESP8266.
Come on, Q. Get on with it. We have $1/mo SIMs, where is my $30-$50 Arduino platform compatible LTE modem with some usable GPIOs and UARTs and whatnot.
We are in a similar situation with LTE; the only good way to build an embedded 4G+ device, is some form of a linux computer
…Gonna be a no from me, dawg
I regularly build Zephyr based cellular devices. Guess it depends on your definition of embedded at the end of the day
Otherwise I was nodding my head along with the post
I find this rather worrying.
Qualcomm has a reputation for being extremely closed. This is not compatible with the Arduino culture.
We might end up with two fairly large ecosystems, both of which are considered open-source: Arduino and Raspberry PI, both being dependent on companies with very much non-open-source attitudes (Qualcomm and Broadcom). In the case of Raspberry PI, it has always been the case that some parts of the chip were NDA-only and not open-source. I wonder if we’ll start seeing the same with Arduino.
Getting my popcorn out to follow this nerd fight 
to see what I can learn from the shadows…
Are you able to build Zephyr for the actual baseband chip, or is it essentially a Zephyr “OS” with a modem peripheral (through, I guess, PPP, AT or QMI interfaces, all to a large extent proprietary)?
The “big deal” with ESP8266 was that it’s not a peripheral, but the main MCU. Which is why it eats the lunch of RPi Pico’s “peripheral WiFi” idea.
I’d say Raspberry Pi (in their own silicon, at least - RP2040, 2350s and RP1) is pretty much the antithesis of Broadcom. They are still (at least with RP1) heavily restricted by using the core IP under licenses, but they are moving in the right direction. The “secure boot” on RP2350 is the only one that’s actually openly documented, for example.
The closed part of Broadcom are broadly (also) third party IP (VideoCore) - BCM itself is fairly accessible, as long as you don’t try to poke the host GPU…
Yep still an abstracted interface. I missed your point about writing code to go directly on the baseband chip. I don’t think that is ever a goal of mine, and I doubt that modem vendors will ever allow that, since the mobile operators won’t allow unproven firmware onto their towers (a la carrier certification)
So though it’s getting away from the original post, I do think it’s possible to make embedded LTE products today, but it will almost always be with many layers on top. The quectel BG95 is the mdm9205 under the hood (Qualcomm) but I have no expectations to be able to write firmware directly on the modem. Things like the nrf9160/9151 are more integrated and that’s normally what I build on top of, still with Zephyr RTOS.
I would expect any Arduino product of the future will have many more layers atop the already discussed parts. For instance, Arduino now runs sketches as an LLEXT (Linkable, Loadable EXTension) on top of…Zephyr RTOS
I read Dave’s post about this as well:
I think that the main value here is that Qualcomm will now have access to a vast customer list and flow of leads that far exceeds the monetary value of what the hardware revenue generates.
You can imagine that QC will leverage their newly acquired platform power to block competitors from easily entering the Arduino ecosystem in categories that they want to own.
It would be a nice thing (for QC not the industry at large) to the defacto standard tool for AI prototyping (like arduino was first for microcontroller prototyping)
There is the cynical view that you just have a big company who has the cash and can make poorly thought through decisions and it will be a failed experiment 3 years from now. All of the management that made this decision gets promoted up the ranks to other positions in the semiconductor space and arduino dies on the vine.
At the end of the day, most people are not buying Arduino brand hardware anymore, they are using clones and boards from Adafruit/Sparkfun instead. I think that Adafruit alone is a bigger contributor to the actual Arduino library infrastructure than Arduino themselves. That leaves just the IDE, so someone probably needs to just re-make that.
It is very reminiscent to when Intel got in bed with Arduino to make useless Galileo and Edison boards just that this time it is via acquisition.
Maybe you also remember the hot second when Broadcom made Wiced
Key signals where you can think that QC is actually doing a smart job will be when:
Any customer can buy a Qualcomm inside module/chip on Digikey
They start putting their radios into these Arduino boards
Key signs that things are going wrong
In the end, it will just be entertaining to watch.
Would you think differently if Nordic (for example) bought Arduino? Who would be a better owner (aside from it continuing to be independent)
Qualcomm has a reputation for being extremely closed. This is not compatible with the Arduino culture.
That was my exact first thought. I really don’t see what Qualcomm is looking to get from it. A fig leaf of open source? Seems implausible.
However, I find the Arduino product line to be somewhat irrelevant these days. Things like Teensy, Pico, Feather, and a whole host of similar “stamp” style processor boards are far better than the Arduino family of products. And far cheaper.
As to the Arduino IDE and library community, I’m not sure what Qualcomm will do to it but if it stayed static, it wouldn’t be a disaster. Since it is open source, it will always be available and there are a huge number of developers putting different processor cores up on it so it will remain relevant for a long time to come.
Phil
I think one possible reason — which could be completely wrong — is that Qualcomm is entering the microcontroller business with the QCC740, and this might be their way of penetrating a market where they’re practically unknown in that segment.
I agree wholeheartedly if you mean the designers present here. However, I suspect (i.e. have no proof) there’s a level of commercial enterprise buying Arduino branded boards to have a sense of security and the sales there could be many times greater than the alternatives the designers are buying.
I guess I would be surprised if many companies are basing products on “official” Arduino products. The Arduino company’s product line seems to be wandering around aimlessly. I cite the Reneasas based products as an example. They are up there with Qualcomm in terms of being closed - you still have to pay for their SDK. (Yes, I know Arduino picked it because of 5V I/O which seems to be a 20th century line of reasoning.) Meanwhile, the Uno R3 has been cloned all over the place. And you can buy ARM Cortex M7 based speed demons for $30-40 USD. Cortex M33 products for well less than $10 USD. And, ESP32 for about the same. I surely don’t see what value the Arduino company offerings bring to the table.
Let’s hope that the Uno Q fares better than the Yun, the support of which was chaotic (to say the least)!