There are a lot of things you can do with today’s powerful microcontrollers, but sometimes you really need a full embedded Linux setup. [Dylan Brophy] wanted to make it easier to add Linux to his own projects and designed the BeagleStamp.
Squeezed onto a 1″ square, the BeagleStamp puts the power of a PocketBeagle into an easy to solder module you can add to a project without all that tedious mucking about with individually soldering all the components of a tiny Linux computer every time. As a bonus, the 4 layer connections are constrained to the stamp as well, so you can use lower layer count boards in your project and have your Linux too.
The first run of boards was delivered with many of the pins unplated, but [Brophy] plans to work around it for the time being so he can spot any other bugs before the next board revision. Might we suggest a future version using RISC-V?
Ok, I have seen SoMs and SiPs, but I think it’s the first SiP on a SoM I’ve come across.
By the way, there’s a joined effort from multiple vendors to create an “open” standard for similar solderable modules in various sizes: Open Standard Module (https://sget.org/standards/osm/).
Ah, standards.
I’ll leave out the obligatory xkcd and instead drop https://en.wikipedia.org/wiki/Smart_Mobility_Architecture
;)
Yeah, it’s an odd choice. The Octavo OSD335x-SM SiP is designed so that you don’t have to create a whole board around TI’s Sitara AM3558. This seems to add very little to that.
“Might we suggest a future version using RISC-V?”
I’ve never understood the fascination with RISC-V. No, by all means no. Give me the Beagle and the ARM thank you oh so much. I have had wonderful success with the beagle in various forms, documentation is superb, ARM is well understood with fabulous documentation. The price is right. But somehow the word “open” has gotten attached to RISC-V. To my taste, what I just described is about as “open” as I might like.
The RISC-V thing involves what I call “PMS” (perceived moral superiority) that is somehow linked to RISC-V not involving licensing. I don’t see the benefit of this unless you are a silicon foundry — it certainly is of no benefit to the end user hobbyist.
Yeah people just like RISC-V because it is “open”. Yes the ISA is open but individual implementations don’t need to be, you still have companies licensing cores just like ARM does and all the best cores will tend to be closed source and licensed.
The RISC-V vs ARM debate doesn’t matter for hobbyists, it makes no difference for us. In fact ARM may be better as it is standardized when RISC-V can choose what extensions they want to add or add custom ones. You can still create open source boards and programs on closed source chips, no hobbyist is creating their own chips so for now it doesn’t matter if the hardware itself or ISA is open or not. Hobbyists typically aren’t making their own compilers either so the ISA being open doesn’t matter either.
Even for FPGA use you are limited to making your own processor or using an open source one which again isn’t the best cores out there, you still have to license them. RISC-V does make it easier to make your own cores and have it work with existing tools though.
As for the cost argument of RISC-V, that since it is open source it will be cheaper, it just isn’t true, the best cores still need licensed and RISC-V SBCs aren’t cheaper than ARM based ones, neither are microcontrollers really, that is perhaps the biggest fallacy of RISC-V, they are not cheaper.
Plenty of open source projects work on ARM based chips, from microcontrollers to SBCs. So what are you trying to say?
Also I never said all products with an ARM processor were perfect did I? One vendors not hobbyist friendly and very closed source chips do not represent every ARM based chip in existence. A lot of things you interact with in everyday life will have arm based chips in them, including a lot of systems that run android. There is nothing to prevent the exact same problem from occuring with RISC-V, is there? Nothing is stopping someone hooking up a proprietary GPU to a RISC-V processor, is there?
Your thoughts and feelings are pretty different from a lot of other people’s, you may not like them but many people do, they are popular for a reason, you might not like those reasons but they are still valid reasons.
If you have decent hardware and a way to use it, does it really matter what architecture or instruction set it uses?
You lumping all ARM processors together with a few SBCs that you don’t like just shows you don’t really know what you are talking about, ARM is popular for many different types of processor for a lot of reasons.
I fear Max is like so many people these days thinks ranting profanity makes him right. I think that have lots of options is great. I RISC v cores will probably make it into a lot of custom chips soon if it is not already happening. A lot of custom chips used the 8051 as core not because it was good but because it was free you could get dev tools for it. RISC V is also great because it will allow a lot of universities and researchers to try out new ideas in things like branch prediction and speculative execution. I can see lots of use cases for RISC-V custom chips.
I also think having the Beagle flavor of the ARM on this form factor is also great.
It is like the people that want to have big pickups hating on electric cars. Hey I like IC cars but I can also see why I would want an electric for commuting to work or as a van to go to home depot or just a local hauler.
I am pro choice in products and technology. The more options the better.
Moral sanctimony notwithstanding, eventually a newer implementation of RISC will overtake older ones. People probably don’t need to worry about that when making projects just yet but eventually somebody will have an edge because of it.
The benefit is lower prices without the “ARM tax”
The consumer doesn’t see savings yet. The ARM chips have the benefit of large scale production to keep the prices down. Eventually someone will probably start producing RISC-V in large numbers and then yes you may see a small difference in price.
What lower prices? Nothing with RISC-V yet is any cheaper than ARM.
RISC-V is an open ISA, the actual implementations can still be closed source nd all the best ones tend to be, then people have to license them if they want to use them, just like they already had to do with ARM.
RISC-V processors are not any cheaper yet, some may be one day but you will always have companies making the best RISC-V cores and licensing them.
The CH32V003 is a shockingly competent $0.10 MCU. The BL808 is an embedded Linux SoC available on the $6 Ox64
There are some shockingly good open-source cores. Some examples are Vroom! for cloud server class (though it’s still a WIP), pico-rv32 for MCU class, and SERV for extremely small (and slow) peripheral manager, among others.
Agreed for the most part, though. I know that SERV has been put in a fair number of ASICS, and I recall seeing some Vex-RiscV cores in some kind of open-ish silicon chip. The C106 and C910 on some of the popular boards have open-sourced versions, but the consensus is that the versions on silicon are a little different. Of course, with any of these, the actual floorplanning and implementation is going to be closed-source (unless they’re on the open-source skylake node or something).
https://hackaday.com/2023/03/02/a-ch32v003-toolchain-if-you-can-get-one-to-try-it-on/
When everyone is inescapably reliant on ARM architecture they absolutely will raise their price to license ARM cores. There’s a reason SoftBank bought and NVIDIA is trying to buy ARM and I promise you it isn’t to be a caretaker of open source technology. Venture capital always wants to gain massive market share and then exploit that market share. When that happens you’ll be glad other people have spent time maturing RISC-V.
RISC-V cores can be closed source and licensed, the best cores will tend to be so you are just replacing ARM with another company.
I will disagree here but in a kind way. I like choices. RISC-V is interesting because it will allow more innovation in the market. The open source ISA allows a manufacture to start with a core with a mature software base and move forward from their. That is good but it will take time to fully mature if it ever does. This stamp right now is cool and a good think. It will make it easier for people to build projects and prototypes with this device. The PRUs are really nice to have. I just wish they had made it a little bigger and put some boot eMMC on it. Would I like to see a RISC-V version? Sure why not? Choice is good. I wouldn’t mind an ATOM version, or another flavor of ARM version. Choice and diversity in the marketplace is a good thing.
Am I upset they did this with a Beagle system? No why should I be upset. This fills a need and is pretty cool.
So if anyone wants a RISC-V version of this… Make it and post it on here. I bet a lot of people will think it is also a really cool idea.
I was quite delighted to find that these chips (and some beagleboards) are supported by NetBSD. Makes my life easier having good long term support in place.
My first thought is that Dylan may not have checked the ‘Castellated Holes’ option from the board manufacturer.
In the picture it looks like some pins are and some aren’t. I would think if he hadn’t checked a box they would have all been unplated.
Maybe he tried to save some bucks by using “normal” through-plated holes which were ripped off while milling the edges.
For fabrication in bigger quantities, the tight courtyards (see tiny space between resistor arrays and SOC) seem quite challenging.
All in all it is a very cool project. I look forward to see this running.
If it’s OSHPark, they don’t technically support castellations, but they also don’t prevent you from trying to do it (and provide some guidelines).
It looks like in this case things were fairly aggressive in the attempt to keep the board to that small 1 square inch form factor. You’ve basically got a one-or-the-other situation there: you can either use a cheap process and give up space, or use an expensive process and keep it tiny.
Trying something the manufacturer doesn’t specifically support is generally not a good idea.
Trying to save the extra fee for castellated holes and trying to get them for cheap…could be a game of chance. If the routing bit is worn but “still good enough for milling”, the resulting PCBs would be headed straight for the bin.
Funny: I just checked with my preferred manufacturer, and checking “castellated holes” would add $0 for such a PCB.
“Might we suggest a future version using RISC-V?”
I’ve never understood the fascination with RISC-V. No, by all means no. Give me the Beagle and the ARM thank you oh so much. I have had wonderful success with the beagle in various forms, documentation is superb, ARM is well understood with fabulous documentation. The price is right. But somehow the word “open” has gotten attached to RISC-V. To my taste, what I just described is about as “open” as I might like.
The RISC-V thing involves what I call “PMS” (perceived moral superiority) that is somehow linked to RISC-V not involving licensing. I don’t see the benefit of this unless you are a silicon foundry — it certainly is of no benefit to the end user hobbyist.
This seems like an admirable project, why drag the RISC-V groupthink into this?
Dang does this place have a heck of an echo. Let’s see if I can spot the same drivel a few more times…
The ARM A8 has been a great embedded Linux platform at 1GHz and the TI and Samsung offerings run without heat sinks in most environments.
The problem is continued support for 32 bit kernels and the speed of USB options and LCD interfaces. There are much cheaper more capable boards of 40x40mm or less with Allwinner H3 or Rockchip 3328 or Amlogic parts but they all need significant cooling compared to the A8 core.
An all-in-one RISCV would probably be popular but isn’t most of the IP you need for the peripherals licensed?
i don’t get it, so your project still needs a separate microsd?
Either a micro SD or a separate eMMC chip. There is no flash in the Octavo chip.
I too don’t get why it doenst bring everything you need to run linux on this chip. it gets bigger, but then it would really be a “drop in” to any project!
It is nice to see this. I have have great success and have enjoyed working with the TI 335x series of parts. A big part of that is the excellent 4000+ page technical manual. Anyone who has worked with ARM based SBC from other makers (Allwinner, etc.) knows that good documentation is worth its weight in gold. When I have chosen one of the beagles for a project, it has almost always been because of the two PRU processors which are quite unique.
Is there really a viable RISC-V based alternative? I certainly don’t expect to see anything like the dual 200 Mhz PRU units with it, but is a RISC-V core that runs at 1 Ghz with 512M of ram and a good set of peripherals available in a decently priced SoC?
Of all the features I care about in a 1″ Linux Stamp, RISC-V as opposed to ARM is not on the list.
Agreed. I tried twice to make some comments about this, but the Hackaday censors don’t care to allow discussion along these lines and my posts were rejected.
Dont take it personally. It happens to all of us.
Looks like they were just delayed?
That happened a few times to me too. I assume there is some automated script that get triggered by keywords (using a few URL’s would do it) and then the script holds it back for manual review, which does take some time.
With all those spam bots (and even people being paid to create spam) it’s an understandable compromise.
They are showing now, just delayed probably.
Anyway, how dare you speak out against RISC-V, you should be so ashamed of yourself. /s
I do feel bad. I have been cranky the past few days and speaking my mind. I guess I can’t blame the Hackaday censors, my cranky posts made it finally after a delay of several hours. I guess I am short on patience also.
Hey at least your post was not a psychotic profane rant like one of the replies was.
I sort of disagree with you. I would like to see a RISC-V in “addition” to this ARM version. Notice the in addition part is not instead of. I like options.
My dream SBC right now would have a 2.5 Gb network connection, HDMI out and two M.2 M slots running PCIe Gen4. For under $100. But as they say it is good to dream. I would take a 1 GB and two Gen 3 M.2 slots for under $100 and be very happy.
“Don’t attribute to censorship what could be explained by technology issues.”
Or something like that…
While a cool project, I don’t see these chips available at the usual retailers. The PocketBeagle itself seems to be in very short supply and above the retail price.
As I’ve only just begun my electronics journey, perhaps there is some secret sauce for getting hard to find chips besides Mouser/Digikey/Element14?
Mouser has it for shipment on May 8:
https://eu.mouser.com/ProductDetail/Octavo-Systems/OSD3358-512M-ICB?qs=lc2O%252BfHJPVaz0crE%252Bcp5FQ%3D%3D
85EUR per chip when you can get an Allwinner A13 for 1EUR on Aliexpress. Oh yeah.
For now if you can trust the source, I’ve been there for the A31…
Kaili, there are companies dedicated to this and the obsolete. Be prepared to be “scalped” on price, and as above risk paying for something that’s not what it seems.
Personal recommendation avoid ST parts at all costs.
If “easier soldering” is desired. Why not considering V3s? QFP package and will work on 2 or 4 layer boards.
Why even link to that old BeagleV article? It’s vaporware they never produced.