looking for vendor agnostic embedded systems forum

Hey Yinz. (Yeah, from Pittsburgh)

Anyways, I've been focusing on embedded systems lately and have built up a collection of rants regarding the particular ecosystems I'm working under (STM32)...but that caused me to pose a larger question.

Are there any "very active" embedded system discussion forums that are agnositc from being sponsored by, or tied to, a particular companies products. Obviously my web search has been less than helpful, since I'm asking here. I'm looking for something less lurked by hobbiists and more by professionals, and where the sophistication level allows for discussions about hardware/software integration at the layer1-layer2 levels: hardware and basic ASIC programming: ie MCUs, RTOS, DSP, bare-metal programming...
 
I can recommend something from Arduino and Raspeberry Pi... Yeah, I know those are from particular companies, but hey, they are popular and common stuff that do get used by large companies that have formal engineering processes in place.

Yeah, they will end up being primarily hobbyist lurking places. Professionals are usually bound by NDA's, so it may be difficult to tease out good sophisticated discussions and determine if someone is a pro or not. Sometimes, a hobbyist sounds like a wannabe pro, and sometimes there's someone who's paid, but sounds like a n00b. And there's gonna be someone who's got things going so smooth, you have to wonder if they're paid to promote the place.
 
re - pi forums...ugh...as mentioned, mostly hobbiists...Thus is the nature of RPI: a glorified toy that entices idiot managers into thinking they can get something for nothing. As long as they stay in the educational domain I have no problem with RPI, but they are NOT professional systems.

Not too worried about NDAs. They are usually restricted to specific intellectual property, not industry standards or platform/ecosystem discussions.
ie: I've been fighting with the STM32 development ecosystem for months and it's a case of good hardware being being encumbered by really really lousy documentation and support tools. manuals obviously written by folks who don't speak english as a primary language, demos and eclipse based IDE that cannot have undergone any sort of quality control before being published, and a marketing centric culture at STmicro where user metrics and advertising are more important than providing stable tools...and God forbid that you complain on a company owned forum becuase that's just an excuse to get blacklisted.
 
Well, you start with a toy and see what you can make out of it before deciding to get rigorous and professional so that you can have quality stuff and can make promises that have money attached to them. Pi and Arduino are meant to be raw materials for prototypes for electronics.

It's kind of like playing with sand, only to realize that when mixed with water and a few other types of materials, you get concrete, which can be used on a professional construction site.

If you don't like the development ecosystem for the microcontrollers, this is your chance to come out with a proposal and a proof of concept of what it would take to make it better. Yeah, it will take work, time, possibly money, and you'll be developing documentation, repos, and standards, rather than microcontrollers.
 
a little more on this since I've done more research.
As discussed, most of the forums are hosted by a commerical entity so they are slanted toward the goals of that entity. I also noted a vendor-neutral forum that seemed interesting, but they block anonymous email accounts for signups: no yahoo or gmail...so that leave me out.

and I found an open embedded standards group with forums but they wont accept individuals, and seem to be more of a paying member organization consortium.
 
As a multipurpose operating system FreeBSD also includes embedded, but it ain't the overall main topic.
Though I (and others) like to help.

I doubt you'll find exactly, what you're looking for.
Two reasons:
1.) Embedded is not FreeBSD's main branchline, even if it's successfully used for it. If you browse this forums you'll find a lot of stuff about the topic, such as experiences about FreeBSD on certain boards etc.

2.)
I'm looking for something less lurked by hobbiists and more by professionals,
In my eyes there seems to be a crucial point.
Professional embedded systems are always part of commercial products. In contrast to 'hobbyists' companies seldom share their core competencies, especially not for free.

If I had to summarize the overall experiences I dare say:
If your primary target is to run FreeBSD as your embedded system, you need to find a suitable board. It's likely you need to compromise. Not on every board all features are supported. Or do the full professional way, and design your own board.
With Linux it's the opposite. You just pick a suitable board by its hardware features, price, and availability. Every board runs Linux.
Plus you get a larger community - including hobbyists.
Don't degrade this.
My experiences as a professional designing engineer teached me: Just because someone posts his results on his homepage neither means he or she's not a professonial, nor his or her stuff ain't.
You may exclude yourself from a goldmine if being too snooty.
Just because some one presents how he or she built some silly toy doesn't mean there is nothing within you cannot adapt to something else, even professional. Not seldom you find reliable, rock solid solutions by amateurs, 'cause they not only have the will but also take their time not to get into market as quickly and dirty as possible. 😁

Although RPis are marketed like toys, neither means they are. You'll find tons of professional, or at least professionally usable stuff, if you dig beside official sites.
And there are lots of other boards - even marketed more professionally.

I doubt there is some 'free and open, but professionals only embedded forum' - please, drop a line here if you found it.
Like everywhere else: It's either closed, or one has to deal with anybody: professionals, amateurs (both neither means to have more or less competence on a certain topic), noobs, and all other kinds.
 
re - FreeBSD in embedded vs larger picture....which is why I posted in off-topic
re - rPI .... fine learning tool but not something suitable for my venues: safety critical systems in aerospace, healthcare, govt/DOD, etc.
re - pro vs hobbiists ... absolutely nothing against the hobbiists and students. its more of a question of looking for folks who understand the "right way" vs the "expedient way"

anyway, not looking to start a debate. Just thought I'd report back on my findings...which seem to indicate that the field is a vendor sponsored arena.

FWIW, my embedded interests are AVR, STM32, and Xilinx Zynq/Ultrascale...I used to be quite enamored with PPC but it isn't used as much anymore.
 
^ THIS.

In some languages and cultures around this planet, it's even a well-known proverb.
...i don't get it.

but not something suitable for my venues: safety critical systems in aerospace, healthcare, govt/DOD, etc.
...you also could have named military, and nuclear power...

'I hate to give good people bad news.'
Sorry, Dude, not a chance.
By my experience on any kind of electronics/development in those fields (I did in nuclear, and medicine), you will not find anything open even remotely clearly linked to it.
You need to gather own experiences (and then you'll understand why they're not given away for free),
or (even better) join some com/org in such a field.
On such high sensitive topics (short for long elaboration about real, serious, actual QM, not paper tigers, fullfilling official regulatories, approvals, licensing... - taking responsibilities) companys are very sensitive.

But you're right in that point:
This by far is the opposite side of anything hobby-like. :cool:

The last pointer I can give you:
Don't focus on the technical side about how to do it right, too much (that's self-evident.)
Focus (way) more on (real hardcore) testing, and above all regulatory issues.

I wish you the best.
Peace out.
 
...i don't get it.
Well, if you are too picky, you may never find what you're looking for, and may miss out on everything else as a result.

One of the best-known examples is looking for a spouse/partner, and passing up on lots of imperfect opportunities and remaining single in the meantime as a result.

Another well-known example is going hungry because you're a picky eater. (Yeah, even with counter-examples of being allergic to stuff like peanuts, milk, wheat, etc). Or declaring that "At my house, we only eat white bread, and even then not every day", while thumbing their nose at other varieties of bread... Different cultures have fairy tales about how picky eaters end up missing out.
 
Just because some one presents how he or she built some silly toy...
Usually, if that silly toy is how to start calculator.exe on your machine, you should pay very close attention ;)

Sorry I can't upvote your text more than once. Especially that part about quick to market.
 
I think that OP's best bet is taking a few classes at a local college, preferably a big one. That's usually a good starting point for connecting with professionals.

As an example, joining ACM (acm.org, Association for Computing Machinery) is very easy in college, most professors would know how to join, and fees for students are fairly low compared to fees for those who want to join later. ACM is an example of such a club that is aimed at professionals rather than hobbyists.

Well, if OP wants ASIC programming, bitcoin mining is not a bad place to start looking - ASIC programming is THE underpinning tech for bitcoin. There are places to discuss algorithm efficiencies in such contexts, and conversations do include pretty academic references... ;)
 
Thus is the nature of RPI: a glorified toy that entices idiot managers into thinking they can get something for nothing.
Here are 2 examples of products based on Raspberry Pi. Do you think they are toys? The prices are not for toys or "for nothing".
 
I should probably refrain stoking the fire, so to speak, but since you asked... Yes, I absolutely consider them to be toys, and wholely innapropriate for anything even close to safety critical or secure applications (which encompasses many industrial apps). Someone may be able to make me a liar, but as of fourish years ago the PI was based on a PROPRIETARY broadcomm ASIC, where an engineer was able to expose a sidechannel ARM processor and "some" of the chip features without running afoul of his corporate bosses. Furthermore, being that the chip is manufactured in a country that is often politically at-odds with US policy, they are a prime target for becoming a cyber-espionage tool...especially since they are only "partly open " and thus cannot be "fully audited"...and because the masses are so enamored with them.

Then lets talk about Raspin...or Debian desktop crammed on a uSD card for PI. It is in no way, shape, or form suitable as an embedded OS. Remember my comments about hobbiist vs pro? The pro would run as much in RAM as possible instead of simply cloning the disk subsystem to Raspian and allowign the uSD card to become the standard non-volatile storage mechanism. uSD controller circuits are quite fragile. It is way too easy to brick a uSD by simply pulling it at the wrong moment.

Anyway, I'm not gonna comment futher on this. Feel free to flame away. I'll be wearing my asbestos coveralls. ;^)
 
Yes, I absolutely consider them to be toys, and wholely innapropriate for anything even close to safety critical or secure applications (which encompasses many industrial apps)
Actually, a lot of industrial apps are really writtien for Windows (I'd know, I've seen plenty of examples on my most recent $JOB). I'd say that Windows are a far less appropriate choice for safety critical or secure applications than Raspbian - simply on the grounds that if you build adequate security into a Windows image, it will become pretty bloated with all that code. By comparison, Raspbian comes with just about all the necessary mechanisms already built-in (just write a few .conf files, and you're set), and it also supports GPIO (Phishfry is the resident expert on that kind of stuff around here).

Also, microSD cards are much more reliable these days than they were in the past - especially if you buy well-known brands like Samsung or SanDisk...
 
Yes, I absolutely consider them to be toys, and wholely innapropriate for anything even close to safety critical or secure applications (which encompasses many industrial apps). ...
Completely agree with all your points. But that doesn't imply that it is a piece of junk and completely doesn't work, if you use it smartly.

Clearly, the system is utterly not trustworthy. I would never put it into a place where the spies from its country of origin can possibly reach it. But my home network is very well isolated, and there is no way for an IP packet to make its way from China to any of my Pi machines: my FreeBSD firewall takes care of that.

The hardware is not trustworthy, but it works pretty well. The form factor is annoying too: every edge of the board has connectors on it, so it is hard to embed physically. For example, I'm just working on putting a Pi Zero into a small outdoor plastic case, using a custom-built "hat" (it's actually a base board that the pi sits on). Because of the form factor, replacing the SD card will require unbolting the pi, but fortunately that's only four screws and one cable connector.

The bigger problem is the software and system design. In spite of what everyone says, Linux is not an embedded system: it still has leakage issues, the WiFi stack isn't perfect enough. For now, I'm making do with automated weekly reboots: a one minute outage at a well-known time is easier to deal with than random crashes every few weeks. Another thing I've started doing: because the IP stack on Linux will sometimes freeze up completely (making the Pi unreachable), I have a pushbutton installed on the board, and a small service that monitors it: if you hold it down for 5 seconds, the pi will cleanly shut down. I've found that crashing the pi (for example by pulling power while code is running and writing to disk) will about 1 time in 20 or 50 cause the SD card to become trash.

And indeed, the SD card it the weakest link. I just got another 5-pack of Sandisk cards, because it seems that each Pi destroys one card per year. When I say "destroy", about half of them are total wipeout of ext4 (where even fsck doesn't fix everything, and you need to wipe the disk and reinstall), or actual SD card failure (where after a crash the card is actually not fully readable, and attempts to read it with dd throw IO errors).

The beautiful thing about the Pi ecosystem, for a hobbyist, is that software for nearly everything already exists. Try to do Dallas 1wire interface using GPIO pins? That's in the kernel, and for temperature sensors even works perfect. Decide that the wave forms and hardware characteristics of GPIO pins are unsuitable for 10m long wires, and instead put in a DS2484 via I2C? Well, that's also supported in the kernel. Need a second serial port, solder an SC16IS740 UART into it? Took me 5 minutes to find the "dtoverlay=..." line, and support is perfect in the kernel. The board got wet because my case had a leak; decide that a humidity sensor needs to be added. Solder a BME280 onto the board, that was hard, but software is trivial: code exists. Decide that the SHT41 is easier to solder: It also has software support. I'm supposed to be a professional data scientist and software engineer (that's what my business card says); when playing with Raspberry Pi at home, I find that 90% of my work goes into PC board layout and soldering, since the software side is so effortless.
 
Too bad plan9 never became popular enough. It is pretty simple (kernel took a minute to compile from scratch on the original pi). Had even wifi working a long time ago on pi3 & pi4. If only more people helped out with drivers for various gpio/I2C/SPi based devices.... I have long felt it would make a great embedded platform.
 
being that the chip is manufactured in a country that is often politically at-odds with US policy
If you want a MCU PCB that is not manufactured in China, you have to make your own.
Based on a MCU produced in the US only (I don't know currently if there is any at the moment; STM, TI, NXP, Microchip... produce (almost) everything(?) in China (or Taiwan); not every lot always in the same fab (well, with app. ~>5M pcs you may deal with them some circumstances)).... which also is supported by FreeBSD. :-/ Tough call.
 
Vendor agnostic? Have you looked at the Beagle Board? They share the layout, open source. TI wants to sell the chips, they want you to adapt the layout, roll your own.
 
Based on a MCU produced in the US only
In a country other than China and Russia? That's easy. For example, Infineon produces MCUs at Global Foundries, which is in Singapore and the US. Global Foundries is the fab of choice for the Pentagon, and probably through that connection to some of the more interesting non-existing agencies.

STM, TI, NXP, Microchip... produce (almost) everything(?) in China (or Taiwan)
There is a huge difference, in terms of both politics and trustworthiness against espionage, between China and Taiwan.

STM uses its own fabs (some are in Europe), as does TI (I think all of theirs are in the US). If you talk to chip makers, or read their advertising materials, they are VERY aware of the geopolitical risk of using manufacturing sites in countries that are run by unreliable governments (such as China), or at risk of war (such as Taiwan).
 
Then lets talk about Raspin...or Debian desktop crammed on a uSD card for PI. It is in no way, shape, or form suitable as an embedded OS. Remember my comments about hobbiist vs pro? The pro would run as much in RAM as possible instead of simply cloning the disk subsystem to Raspian and allowign the uSD card to become the standard non-volatile storage mechanism. uSD controller circuits are quite fragile. It is way too easy to brick a uSD by simply pulling it at the wrong moment.
1. You can install normal Debian (arm64 version), not only Raspbian.
2. Pay attention to Compute Module version of Pi.
3. It is not necessary to use SD card - you can boot from normal SSD.
4. Try to order 100 or 200 pcs of other type of compact computer other than standard Intel/AMD x64 board. And find the same model after 2 years.
5. Between toys and "safety critical or secure applications" there are other classes of devices that need computer and the world will not finish if the computer stops working after 7 years (for example).

And BTW, are you developing a toy? Or it is safety critical application but you have bad documentation and expect to find more information in forum.
 
Back
Top