EEVblog Electronics Community Forum
Products => Crowd Funded Projects => Topic started by: Oleg Artamonov on January 27, 2015, 07:05:07 pm
-
Hello everyone
Let me brag a bit about our project — we are developing coin-sized microcomputer which runs OpenWRT, with onboard Wi-Fi (including PCB antenna) and a lot of GPIOs, friendly for novices as well as emedded electronics professionals. It's open source, including hardware — we will post everything (schematics, PCB, BOM) in editable formats, not only useless PDFs.
We just hit Kickstarter: https://www.kickstarter.com/projects/1133560316/black-swift-tiny-wireless-computer (https://www.kickstarter.com/projects/1133560316/black-swift-tiny-wireless-computer)
(http://olegart.ru/wordpress/wp-content/uploads/2014/12/blackswift-sd-olegart-670x405.jpg)
Here is the official website: http://www.black-swift.com/ (http://www.black-swift.com/), but I'll describe the project below for everyone's convenience.
Short specs:
- Dimensions: 25×35×4 mm
- Chipset: Atheros AR9331
- Memory: 16 MB NOR flash, 64 MB DDR2 SDRAM
- Power: 5 V or 3.3 V or 3.6...6 V (onboard voltage regulator)
- Power consumption: 60 mA min (200 MHz, Wi-Fi disabled), 120-150 mA typ (400 MHz, Wi-Fi enabled), 300 mA max
- Wi-Fi: 802.11 bgn with PCB antenna (NB: there will be no U.fl antenna connector in the final design)
- I/O: microUSB (power), microUSB (USB 2.0), 2x10 pin 1.27 mm & 2?15 pin 1.27 mm (GPIO, power, 2?Ethernet, USB 2.0)
- Firmware: U-Boot bootloader, OpenWRT Barrier Breaker
Price: $25 basic version, $35 developer edition. Worldwide shipment by Hong Kong Post.
Developer edition has absolutely the same form-factor, but comes with onboard UART-USB adapter and ability to directly reflash NOR memory, so it should be great for people who likes to play with bootloader and/or firmware — unbricking never was easier. Also it comes with 1.27 > 2.54 mm adapter, but later adapter will be sold separetely as well.
Why we did it? We are a small company which specializes in professional electronics development, and we were not happy with existing options. For example:
- Cheap routers. Not suitable for any kind of professional design, and unconvenient for DIY or prototyping — not enough memory, not enough GPIOs.
- EL-M150, Carambola 2, similar boards — surface-mounted, so unfriendly for novices, unconvenient for fast propotyping, ineffective in terms of occupied main PCB space, some of them need external Wi-Fi antenna, others lack onboard voltage sources, etc.
- VoCore and other RT5xxx-based boards — slow due to SDR memory (DDR2 is not supported by Ralink's chipset)
- Raspberry Pi, BeagleBone Black etc. — great for enthusiasts, but way too big for embedded applications
- Rapsberry Pi Compute Module — needs SODIMM connector, unfriendly for enthusiasts and fast prototyping, lacks Wi-Fi
Also, the common problem with Chinese hardware: almost impossible to persuade manufacturer to change anything according to our specific needs. We tried and failed, they just don't want to do it (or don't know how to do it, most likely...).
Besides, most OpenWRT-based boards come with more or less stock OpenWRT — which means 100 Hz timer, no preemptive multitasking, no GPIO IRQ support on AR9331, space-efficient userspace software like busybox utilities and dropbear SSH... So finally you have to build your own firmware anyway.
What we tried to achieve here? Mainly, combine the easyness of use for novices and enthusiasts with convenience for professionals in embedded electronics.
- 2 standard microUSB ports — one for power, another for USB 2.0 interface, so you can connect USB peripherals with regular USB-OTG adapter and power the board with cellphone charger
- Onboard voltage regulators — 5 V to 3.3 V and 3.3 V to 2.75 V to power peripherals (although 3.3V-tolerant, it is recommended to use AR9331 GPIOs with 2.75 V)
- If you don't need USB power, you may power the board with 3.6...6 V or 3.3 V sharp.
- Onboard Wi-Fi antenna, so nothing external needed here
- All GPIOs, power, USB, etc. are on 1.27 mm connectors (standard PLLD type), so you can use it in a variety of ways — attach 2.54 mm adapter, solder wires directly to the board, solder it to the mainboard (if you need one) or make it detachable, align to the mainboard or place it at 90 degrees...
E.g. I have two projects with it already — one is a kind of weekend project with everything soldered to eurocard (Christmas tree lights controller (http://habrastorage.org/files/81f/643/d8f/81f643d8f3494072b14849f9eac83db1.jpg), and I'm not, repeat not, joking, I seriously needed that thing), and another is apartment's energy meter with web interface in case small enough to fit on standard DIN rail (even EL-M150 doesn't fit inside it).
And to make something like wireless mp3 player or print server or router (unexpectedly!) you need nothing to solder.
OpenWRT image will have some modifications:
- 1000 Hz kernel timer & preemptive multitasking enabled by default
- GPIO IRQ support on AR9331 (you know, there's a patch available)
- Ability to pass variables from userspace back to bootlader (i.e. to change CPU frequency on the next boot)
- Normal SSH client instead of dropbear (you can't use Eclipse to remotely debug your software with dropbear)
- et cetera
As I said, we will publish everything we did and changed in our wiki — http://www.black-swift.com/wiki?view=categories (http://www.black-swift.com/wiki?view=categories) (some things already published: basic board information, some OpenWRT tips & tricks). OpenWRT images, patches, as well as board schematics and PCB will be published around February, after we get final design and make sure everything is working the way it should. Also we'll prepare and make available such things as Linux VirtualBox images to develop OpenWRT programs (minimal with SDK and command line (http://www.black-swift.com/wiki/6?view=article), normal with X Window and Eclipse installed and set up), 32-bit OpenWRT SDK, et cetera, and some guides on how to use it (not everyone is a linux guru I guess).
As for now:
- We finalized design in January and expect samples in February, so mass production can be started in March-April. It will be third revision, but mostly it's final touches, bells and whistles — so I expect no problems with it.
- We have no investors and we are a small company, so we need funds to start mass production — as I said already, you can help us here (https://www.kickstarter.com/projects/1133560316/black-swift-tiny-wireless-computer)
-
Nice work!
I have a couple of questions...
1) Is the USB available on the pluggable connectors (not just the micro USB socket) ? (I have an application currently using a Carambola where the module plugs into a board with a USB hub IC on it and some other USB peripherals. If the USB is only available on the micro USB connector, it would render your module useless in my application).
2) Do you plan to CE/FCC certify it ?
-
Doh!, just noticed the pinout diagram on the kickstarter page, didn't spot it after a quick glance on your website. It does appear to include USB on the headers :-)
-
Why make a tiny module when it will need either a big battery or external power to be useful?
-
1) Is the USB available on the pluggable connectors (not just the micro USB socket) ?
Yes. Here is the pinout: http://www.black-swift.com/wiki/4?view=article (http://www.black-swift.com/wiki/4?view=article) (USB_DM & USB_DP on pins 1-2 of the 2×15-pin connector).
2) Do you plan to CE/FCC certify it ?
FCC — definitely, CE — probably as well.
-
Why make a tiny module when it will need either a big battery or external power to be useful?
Because you may have power already available — e.g. when using BSB on some kind of mainboard with some peripherals, and at the same time you may not have enough space to fit something bigger. Especially useful taking into account that Black Swift may be powered with 5 V or 3.3 V and in the former case can provide 3.3 V @ 700 mA for other devices.
For example, electricity meter in residual-current circuit breaker's case (http://www.black-swift.com/images/wiki/projects/energymeter/energymeter.jpg), 5VDC AC/DC converter is under the boards and BSB provides 3.3VDC for ADE7757 metering IC. A bit tense example, I know, but it gives an idea :)
-
Nice idea but please add pin pitches & finished hole sizes to you dimensional drawing !
-
Nice and simple. Perfect. Backed, but will you be selling these afterwards as well? If it works well, I can think of a boatload of uses.
-
Nice idea but please add pin pitches & finished hole sizes to you dimensional drawing !
Will do. Thanks for suggestion!
Backed, but will you be selling these afterwards as well? If it works well, I can think of a boatload of uses.
Of course we will. We'll open regular online store on black-swift.com, with the same worldwide shipping from Hong Kong.
-
Why make a tiny module when it will need either a big battery or external power to be useful?
Useful module to retrofit to existing equipment. Perfect price point at $19 shipped as well, seems like incredible value. If only it had RS232/485 onboard!
-
:-+ Looks like a nice IOT type device. I'm thinking of backing a few.
Useful module to retrofit to existing equipment. Perfect price point at $19 shipped as well, seems like incredible value. If only it had RS232/485 onboard!
The logic level UART RxD and TxD is there, as well as CTS and RTS. True RS232/485 requires different voltages, line drivers and transient protections, as well as screw terminals or DB-XX connectors to connect to the real world. All that is much better done on a plugin daughter board like the developer board.
The I/O pins also bring out the I2S ports, so you can sew it into a teddy bear with a battery pack and DAC/audio amplifier board, and put a talking teddy bear on the Net :) Or just use a USB audio device.
You can also put it into an old wired front door Intercom box you might find at a garage sale, put a battery pack, USB webcam, USB audio w/microphone, pushbutton on the GPIO port, and you have a WiFi intercom from your front door. Pop up a notice on your desktop PC when you're home, or on the SmartPhone when you are away, and you can see and speak to your visitor at the door.
It's small enough to fit into both those examples, and many more.
It's very usable because it's open source.
I think I might use it to turn a teddy bear into a speaking password manager... squeeze its nose in a secret manner to unlock it.
Oh shoot, now you know where my passwords are. I'll use the giant panda instead :)
EDIT: I'm in...backed the pro board. They we're going fast :)
-
Are you considering heat dissipation options?
Under worst case conditions is likely to need some help (e.g. sealed outdoor applications)
-
Are you considering heat dissipation options?
Under worst case conditions is likely to need some help (e.g. sealed outdoor applications)
I'm not sure if you're addressing that to me or the developers... however, I know this SOC is used in alot of routers without a heatsink, and it's probably 60C or 70C inside the case of a router when you stack it with other devices.
The datasheet says the package can run at 110C (Junction 125C). Obviously these are maximums and you want some headroom there. But I think that if it gets too hot a simple glued on heatsink would suffice, like they put on memory chips or southbridge chips
-
The I/O pins also bring out the I2S ports
There's a problem with I2S — no stable driver available for AR9331. We're working on this, but I can't promise you any solution at the moment.
Anyway, if you need audio I/O — regular USB sound cards work great with the board.
-
NB: we just added bundle rewards - 2 or 3 boards with $5 to $15 NZD discount compared to regular price.
So if you want to order more than a single board or tried to order multiple boards by paying 2 or 3 times more that specific reward required (some people did it that way) - you may want to reconsider your choice to get a small discount.
-
Made a board with AVR (ATTiny24A) to use it as peripheral controller - h/w PWM, ADC, realtime operations, etc.:
(https://fbcdn-sphotos-h-a.akamaihd.net/hphotos-ak-xfp1/v/t1.0-9/10351611_878396428877115_3494243782986957570_n.jpg?oh=aad02f55cca7465fd410e3177a050d90&oe=55689F87&__gda__=1432861966_7224fe79d36de30fe2f079ba2085170c)
Patched avrdude 6.1 to compile for OpenWRT 14.07, so no need in external AVR programmer:
(https://fbcdn-sphotos-d-a.akamaihd.net/hphotos-ak-xpf1/v/t1.0-9/10891529_791457520949185_4753707072722999261_n.png?oh=b6ecce0c5d3f8e32b841b0c4e29935bc&oe=5559C9C8&__gda__=1433214626_25c493d3e192c2e618124724382541fa)
AVR flashing is done through SPI, so the same SPI can be used to communicate with AVR's firmware.
P.S. Should I call it "Arduino compatible!" and get a huge bunch of media attention?.. :D
-
Nice design, backed! I've always been a bit curious about OpenWRT, but had doubts about the cheap routers, I wasn't sure if i would be able to use it without more hacking around than I wanted to do. And, not too much GPIO. So it's great that you have made an option that is actually intended for embedded Linux use. Not as easy to solder to 0.05" holes but I like the super-small form factor, it can go inside more enclosures.
-
What does this board offer over other the many other similar boards like;
http://www.aliexpress.com/item/AR9331-Open-Source-OpenWrt-Linux-WiFi-IoT-Module-Mesh-Version/2011607350.html (http://www.aliexpress.com/item/AR9331-Open-Source-OpenWrt-Linux-WiFi-IoT-Module-Mesh-Version/2011607350.html)
http://www.aliexpress.com/item/AR9331-single-chip-radio-frequency-wireless-WiFi-module-can-brush-openwrt-TC-AR30SX-AR9331-stamps-mouth/1890996958.html (http://www.aliexpress.com/item/AR9331-single-chip-radio-frequency-wireless-WiFi-module-can-brush-openwrt-TC-AR30SX-AR9331-stamps-mouth/1890996958.html)
http://www.aliexpress.com/item/Easy-Link-M-mini-ar9331-V2-0-openwrt-ultra-low-power-wireless-module-ultra-small-size/2002371818.html (http://www.aliexpress.com/item/Easy-Link-M-mini-ar9331-V2-0-openwrt-ultra-low-power-wireless-module-ultra-small-size/2002371818.html)
-
Speaking just for myself, the difference is likelihood of at least some minimal support, user forum with critical mass of english speaking users etc. I have little confidence i could get some unsupported china sourced design working on my own. Also, do those products have FCC cert? Could you trust them if they claimed to ? Maybe they they are solid products, but how do you know?
-
Speaking just for myself, the difference is likelihood of at least some minimal support, user forum with critical mass of english speaking users etc. I have little confidence i could get some unsupported china sourced design working on my own. Also, do those products have FCC cert? Could you trust them if they claimed to ? Maybe they they are solid products, but how do you know?
The chipset is well documented and used in a lot of other designs - all of which are very close to the reference design. Thus, any forum which is for any device based off the AR9331 will be able to provide a significant amount of support. I have little confidence you could get the NZ sourced design working if you couldn't get the chinese one working.
These products don't provide a FCC cert, but currently the Black Swift is not certified either - they will work on getting it certified once the kickstarter is complete, so it is currently in the same boat as the chinese modules. Maybe this Black Swift is a solid product, but how do you know?
-
As you suggest, I don't know any of the developers personally and I don't know how good any the products are. It is a gamble. But based on my experience with a variety of embedded Linux platforms (also a few wifi modules), I prefer to place my bet where I think the odds are best. In this case, the Kickstarter and forum threads like this, suggest to me the "Black Swift" designers are making some continuing effort beyond just making the PCB. That's what made me feel I'd have a better chance with it. Even with the same chipset, there are a lot of ways to mess up a PCB (esp. with a 2.4 GHz transceiver). In my experience, some China-sourced designs are put on the market with minimal testing and no support / bugfix/ followup of any sort. Maybe others have a different experience but based on what I've seen, I'm ready to take a chance on this one.
-
I would suggest you at least keep the footprint for the antenna connector somehow. I wouldn't trust a PCB antenna for some applications :-+
-
Maybe not too elegant, but since the whole module is small, could be mounted at feed point of cantenna, dish reflector etc.
-
Speaking just for myself, the difference is likelihood of at least some minimal support, user forum with critical mass of english speaking users etc. I have little confidence i could get some unsupported china sourced design working on my own.
One of the reasons we made Black Swift is that it was impossible to get any support from Chinese manufacturers, let alone persuade them to modify their designs to suit our needs. It seems most of them just mechanically produce something they have documentation for, without any understanding of how it works and how one can change it.
P.S. From my previous experience - once we ordered Dual Link DVI cables from Chinese factory, around 1-2k pcs. They just soldered DL-DVI connectors on SL-DVI cable (it has fewer wires) - and it seemed to me they just can't understand what's wrong here. Had to change the factory for our next order - found another guys who had their own test facility and employees with engineering education. Another story, pretty recent - one manufacturer asked to change person who controlled the process from the client's side with the reason "he's too angry at us, we want the previous guy back, he was kind".
-
Based on lack of response from the developers, this is looking like another suspect Kickstarter project. Deliveries were supposed to begin this month but there have been no updates since April an no word from the developers despite numerous pleas from supporters.
Oleg, are you there?
-
Well, an hour or so after I posted the above, they sent out an update. The first one in over 2 months! I guess a prod on the EEVblog gets results! ^-^
Apparently the've been waiting for FCC approval....
-
Do you need FCC approval for a "kit" module like this? I thought you only needed it for final products.
-
It is not required, but if they were to obtain modular approval, it would allow people to integrate it into the final product without re-certification provided certain conditions are met (antenna with the same pattern and gain is used and that module is the only source of intentional radiation).
But what does it mean "waiting"? FCC takes about 2 weeks to prepare the paperwork after the lab test.
-
Good points. Maybe the FCC thing is a red herring.
From their update (https://www.kickstarter.com/projects/1133560316/black-swift-tiny-wireless-computer/posts/1272027): "FCC certification took a bit longer than we expected — but now we are expecting to receive final verdict (positive, of course!) within a few days. At the same time, we started mass production of Black Swift boards — PCBs are ready, we are waiting in line for the final assembly."
-
but now we are expecting to receive final verdict (positive, of course!) within a few days.
The final verdict is known at the day of the test in the lab. They usually generate a report right on the spot. All FCC does is enter all of the information into a database and assign an ID.
Or they are doing something with some sketchy lab.
On the other hand, the picture from the first post seems to be real. And they are not hugely over due.
In any case, if passing FCC takes you more than a month, you are doing it wrong. And should probably ship units as is, and certify later for mass orders.
-
But what does it mean "waiting"? FCC takes about 2 weeks to prepare the paperwork after the lab test.
It's common to just consider the third party lab test a part of the certification (technically incorrect maybe, but common). Thus a failed lab test can mean certification takes longer, for certain definitions of certification.
Now they could just be bullshitting any way, but saying what they said is suspect is reaching a bit.
-
Apparently, they got the FCC Part 15C cert on June 25. They show a document from MiCOM Labs in Pleasanton CA, addressed to Alexey Andreev at Tradezone HK Limited
"Let us inform you that Black Swift's FCC ID is 2AETB-AR933" https://fccid.io/2AETB-AR933 (https://fccid.io/2AETB-AR933)
https://www.kickstarter.com/projects/1133560316/black-swift-tiny-wireless-computer/posts/1278390 (https://www.kickstarter.com/projects/1133560316/black-swift-tiny-wireless-computer/posts/1278390)
-
Unless I'm mistaken (might well be, not my area of expertise) for modular approval (i.e., for an approval that carries over to another product that incorporates the module) the module needs to be shielded:
"1. The modular transmitter must have its own RF shielding. This is intended to ensure
that the module does not have to rely upon the shielding provided by the device into which
it is installed in order for all modular transmitter emissions to comply with Part 15 limits.
It is also intended to prevent coupling between the RF circuitry of the module and any
wires or circuits in the device into which the module is installed. Such coupling may
result in non-compliant operation."
https://transition.fcc.gov/Bureaus/Engineering_Technology/Public_Notices/2000/da001407.txt
Which might explain the delays... The FCC ID they got is specifically not for modular approval though: "Modular Type: Does not apply", so perhaps they just realised they couldn't get modular approval for the module as designed?
So let's hope nobody tries to integrate the module into another product...
-
How the hell did they get a limited modular certification?
In such a case, an operating condition on
the grant of equipment authorization for the module would state that the module is only approved for
use when installed in devices produced by a specific manufacturer, typically the Grantee. If LMA is
sought, the application for equipment authorization must make this fact clear. It must also specifically
state how control of the end product, into which the module will be installed, will be maintained, such
that full compliance of the end product is always ensured.
PS. on the other hand the grant makes no mention of being limited to a specific manufacturer while DA 00-1407 says it should ... so maybe there are some addendums I'm not aware of.
PPS. FCC 07 - 56 does seem a bit looser in it's language, no mention of specific manufacturer.
-
I think it is just a paperwork mistake/omission. I'm pretty sure the intent here is to allow them use it as a module in their own products, but not for general public. Which renders this certification pretty much useless.
-
As I said, FCC 07-56 is a little less strict in it's language ... might be able to get away with just following their instructions for shielding during the design of a product. Of course they'd actually need to share the design of the shielding they used for the certification.
-
Oleg,
It's been 1 1/2 months since the last update on the project page and now 2 1/2 months since initial delivery estimate. No responses to requests for an update on the project page.
What's the deal?
-
And no Oleg.
We need an icon for runners.
-
Well, apparently this is not dead yet.
New update today (https://www.kickstarter.com/projects/1133560316/black-swift-tiny-wireless-computer/posts/1359126)
We are not dead.
Last week Dmitry Zherebkov, original project leader, left the project.
For the past 2 months, the project was in deep financial crisis.
Now we have investors interested in bringing the project back to life.
We are reorganizing our structure now, and I believe we’ll start actual production in under 2 weeks. It means you’ll get you rewards in 6 weeks.
Our new name is Unwired Devices LLC, and our website is www.unwireddevices.com (http://www.unwireddevices.com). Black Swift was renamed to Unwired One as well.
www.black-swift.com (http://www.black-swift.com) and www.black-swift.ru (http://www.black-swift.ru) websites and corresponding email addresses are not connected with this project anymore
Good on them for at least keeping backers informed. Though, the question is what happened to the $78,000? We'll see if they deliver..
-
Hmm, "Dmitry has left the project", "we ran out of money" and "the two events are unrelated". ::)
I think this project had quite a few "red flags" (excuse the pun), so not really surprised to see it come unstuck.
"we are hoping to get some investors on board to finish the project"
My cynical mind interprets that as "we will launch some other scampaigns and use the money to fund this one". :)
To be fair, I don't think it was a deliberate scam. Just hopelessly incompetent and inexperienced.
-
Just received my Pro board!
I had lost hope, but it looks like they came through. Not cleaned very well (lots of flux on the headers to connect the module to the breakout), but that's alright. I'll hopefully get some time to bring it up and play around with it this weekend.
-
Just received my Pro board!
I had lost hope, but it looks like they came through. Not cleaned very well (lots of flux on the headers to connect the module to the breakout), but that's alright. I'll hopefully get some time to bring it up and play around with it this weekend.
Yeah - I finally got an email with tracking no for mine last Friday. I was losing hope as well. Kudos to Alexander for following through although his communication lately has been poor. I'm looking forward to see what this little board can do.