Thanks for the links! The 'Making of Beaglebone' video was great to watch - I'd never seen how the selective wave soldering was done before. The 'Flashing and burn-in" at 3:15 with all the blue lights on dozens of boards looks a little scary :-)

Regarding the Adafruit tutorial, I'm not sure if the 'opkg update' is advisable on that web page, I think that step should be avoided currently.

The GSoC link is interesting. I hadn't realized there was any interest in reviving Minix though. The work on the JTAG is interesting, especially if there will be some drivers for programming popular devices (e.g. Atmel, Xilinx), to save having to buy so many separate tools, although costs have gone down nowadays. I wonder if 'LCD debug/patchwork' includes the ability to easily customize the frequency and resolution without people having to compile a driver, so that any generic TFT can be driven.

Thanks for the links! The 'Making of Beaglebone' video was great to watch - I'd never seen how the selective wave soldering was done before. The 'Flashing and burn-in" at 3:15 with all the blue lights on dozens of boards looks a little scary :-)

Regarding the Adafruit tutorial, I'm not sure if the 'opkg update' is advisable on that web page, I think that step should be avoided currently.

The GSoC link is interesting. I hadn't realized there was any interest in reviving Minix though. The work on the JTAG is interesting, especially if there will be some drivers for programming popular devices (e.g. Atmel, Xilinx), to save having to buy so many separate tools, although costs have gone down nowadays. I wonder if 'LCD debug/patchwork' includes the ability to easily customize the frequency and resolution without people having to compile a driver, so that any generic TFT can be driven.

There are many ways to solder the bottom side, that's just one approach.  Others that have been used are to put the board into a heat resistant carrier that's machined out to mask off certain areas, it then goes over a normal solder wave.  Another would be that the bottom side isn't reflow soldered, the components are glued on and then the whole lower side is soldered by going over a wave.

The problem is always going to be what sorts of components are on the bottom, the microSD card and micro hdmi were always going to make traditional wave soldering hard.

The sequence of events in the video has been somewhat creatively edited, note that you see them enter the wash with no selectively solderd parts, but seemingly come out a few seconds later with all the selective soldered parts attached...  In the old days, the network socket with integrated magnetics might have been a no-wash part as it's un-sealed and retain water which could cause damage. So there's all sorts of design choices that can cause the production process to be changed to suit.

Would have been nice to have seen a bit more detail with some wider shots to get a sense of the scale of the machines used as most people will never get to see what's involved in making their everyday gadgets.

and I wonder what comes first, the 'flashing & burn-in' or the functional test ?   Note the connection being made to TP4 to allow programming the onboard eeprom during the functional test - would a normal sd-card flash image work properly (or even boot) if the eeprom is blank ?

glue dots to seal the boxes seems to be new too, none of mine have had one of those.

Derek's video does look interesting - looking forward to watching it.

I'd noticed that connection on TP4 too! and gone straight to the schematic ;-)

shabaz wrote:

 

The 'Flashing and burn-in" at 3:15 with all the blue lights on dozens of boards looks a little scary :-)

I was in stitches when I saw that, immediately picturing the screams of pain and horror issuing from some quarters.

I hadn't realized there was any interest in reviving Minix though.

Minix could turn into something very interesting for embedded work.  I looked at several different microkernels a few months back, and they all seem to have different pros and cons.  It's a very active research area to find a good microkernel model that still provides an effective POSIX-compliant subsystem and is able to harness the vast mountain of Unix-type software that realistically cannot be ignored.  Minix isn't likely to become a big-time microkernel for everything, but it's still worth exploring I think.  Expanding one's mental horizons is always good.

Linus doesn't want to admit it but microkernels are definitely the future, and I fully expect Linux to move slowly in that general direction without anyone saying so openly and honestly, at least not until Linus retires from the driving seat.  Eventually he'll fudge some kind of political excuse like "We're restructuring the kernel to provide more runtime modularity, but avoided all the pitfalls preached by the microkernel advocates" ...  Well, whatever excuse works for you Linus, lol.

The point is, monolithic systems have no future, and that should be so obvious to any software engineer that it's barely worth justifying.

Morgaine.

Morgaine Dinova wrote:

 

BBB resources are going from strength to strength.  This is all highly encouraging.

 

Morgaine.

As we suspected months ago, the real-time possibilities with the BBB are really helpful. I think it will tempt people to use the BBB in places they would not have considered in the past.

For example, I've been working (slowly!) on a reflow oven for about 6 months now (I don't get much time to work on it) and I was going to use a low cost AVR, but I'm now considering the PRU for control because it could continue running even if there was some fault in the Linux application software (I have a hardware cutout if all fails), and the Linux application side is nice, because I would like to have a capture of data that I can easily retrieve and graph. With an AVR, I would have had to send serial data and post process; with the BBB I can in theory just dump the entire file to flash and SFTP it any time.

Morgaine Dinova wrote:
 
Linus doesn't want to admit it but microkernels are definitely the future

I need to read more about it, but I'm seeing a similar thing in industry, where it's very convenient to develop "features" for products once, as long as the underlying architecture allows it to be scalable to consume resources on a single CPU shared with other features, or ported across to a separate CPU as a single or multiple instance of the feature, or to separate physical boxes. Time for me to read some of Tanenbaum's book, it's been on the bookshelf for ages. I don't really know much about Minix.  Although it must have changed an awful lot since he wrote the book.

Maybe we should create a separate post or blog to follow the GSoC work, but basically Vlad's project (full name Vlad Victor Ungureanu, from Romania) is explained here, and he's started coding. It's very interesting - if I understand it correctly, it should allow an android device to deploy and run software onto the BBB, and also direct the display from the BBB  to the android device if desired. I suppose in theory, BBB apps could then appear on an app store, and you could plug your mobile into the BBB to transfer it.

I've not had a chance to look into the other projects yet.

Hello! My first GSoC proposal was to boot a kernel+initrd over usb from an Android device and afterwards communicate via ADK with the device to use it as touchscreen. This proposal was made on the assumption that U-Boot SPL was already on the board and could initiate transfers. In the end, after discussed with the mentors, we decided to take the rough way ROM booting the board(board does not have anything on eMMC, no MMC inserted). ROM, U-Boot SPL and U-Boot implement same way of thinking RNDIS->Ethernet2->IP->BOOTP + TFTP. In a couple of days I will have a working Linux example if anybody is intrested( https://github.com/ungureanuvladvictor/BBBabb ) *beware code at this point is really sloppy, don`t get scared*. The progress of the project can be read at http://beagleboot.vdev.ro/ . Now the blog is mainly empty but after i get it working on Linux I will explain the flow of operations and how to use it. If a new thread is created it`s fine by me, I can post from time to time there to keep people up to date with the progress.