Atmel SAMA5D4 Xplained Ultra

While talking about hobbyist I don't think there are people who really care about power consumption. Most of them would like to have some kind of board which will provided easy interfacing options with peripherals (Libraries). Again it depends if hobbyist just care about application or he care about learning a programming with ARM/ or any chips from scratch. Even now a days every board manufacturer provides example codes and libraries such a way that hobbyist just write a logic and make application. So to conclude when hobbyist decide to use any board he cares about how easy to start with any board. That could be main reason why this board also have given arduino shield support to attract arduino users to move to little powerful platform. That is why arduino platform for microcontrollers and raspberry pi for embedded linux platform are so popular even in school students.

Yes, you have to bitbake "once", then you have all the packages you get with the other distros.

bitbake generates the packages for you.

Once they are generated, you can create your own package server, and download packages

for test on a board.

It is not that hard.

BTW: Did You try http://www.linux4sam.org/

A lot is explained there?

It is a subtree of http://www.at91.com.

The Yocto "poky" project is a barebones project which is for people

that want to build a distribution. It is not intended to use as is.

You may want to download the Angstrom setup-scripts from github (Google for it)

Angstrom is Yocto compliant and created by the team behind Yocto (OpenEmbedded)

Best Case:

./oebb.sh config <atmel-machine>

# Edit conf/local.conf to add any extras.

. ./environment-<bla-bla-bla>.sh

bitbake systemd-gnome-image

I think will build linux, and a rootfs with python.

You will get a tarball which can be loaded onto an SD-card.

Piece of cake most of the time, but it can be broken.

Didnt test this for a few months and since this a user maintained

project things are not fixed automatically.

You need to subscribe to the appropriate mailing lists for help.

Check out www.openembedded.org

Hi Ulf,

Of course hobbyists aren't the target audience. But... If this is sold as a development kit, it IS very badly documented for those who are new to building Linux stuff. I've programmed a fair share of microcontrollers, and a few PC applications, but I'm new to embedded Linux. Boards like BBB have support to easily build stuff because of their package management. When I want to try something out on the Atmel board, I'll first have to 'bitbake', or build stuff on Yocto.... All of which is documented *somewhere* for *some* platform, but not for *this* platform. So if I want to see whether my application has a chance of running on this platform, I should in fact already have the skills of someone who's daily job is to build SBC applications.

My primary interest was to use the excellent Atmel peripherals combined with the benefits of  using a Linux kernel and Python, but that appears to be very, very tricky. Maybe gregoryfenton or Robert Peter Oakes can put something together, but I think Atmel should put something together for this goal. They also did this for their non-Linux platforms, using the ASF in Atmel Studio, so their microcontroller group knows what the importance is of providing good examples.....

You make the assumption that this is the target audience.

A semiconductor company does not target hobbyists.

They have low cost reduced functionality boards like the Xplained boards

and they have full-fledged development boards.

Both are intended for companies doing serious development.

The Xplained board can be used for limited production as well.

Yocto is the industry standard for this type of customers,

supported by companies like Atmel, Texas Instruments,

Freescale, Intel , Wind River, Monta Vista etc.

I.E: All the key players.

Ubuntu, Raspian, Debian, if they have support for a specific chip,

is based on the work done for Yocto.

When Linux is running, you don't have to learn about the different

peripherals on the chip. What you have to know, is how to

handle them using the Linux APIs.

That ís your job, and not the something a semiconductor Company

has any expertise in teaching you.

If there is a peripheral which is not supported by Linux, or has limited feature,

then of course you can ask the vendor to improve the support.

I agree there not here to teach us our job but...

When a board is released like this as an Evaluation Board, with deliberate features like LCD, HDMI, Some Arduino like compatibility, it is specifically targeting a type of hobbyist or developer that is looking to evolve their skills to the next level or want to get more juice and capability out of the platform and IMO that is not the one who has fun re-compiling the OS and working right down at the chip level to get the basics working like ADC, PWM, DAC, GPIO, LCD, HDMI etc. This is most defiantly not classified as doing our job for us, I would classify this as enabling us to get on with our job and not be slowed down with the nuances of the physical hardware which really is a very specific skill set. I am talking generically here, I for one am ok delving into the hardware on some platforms but many like windows and Linux have evolved to the point where is is not easy to do and therefore I prefer to get the libraries that abstract the hardware and build My Aps (My JOB) on top of that

To that end, these SAMA5Dx Xplained "Evaluation kits" so far have let the team down. I am now in communications with Atmel support via E14 specifically to try and resolve this issue and give all the road tester a sporting chance to actually get more than, the "Yup is runs Linux but what's this Yocty thing, nothing like Raspbian, Debian, Ubuntu that I'm familiar with and the documentation as to how to use the hardware (Other than the actual ARM CHIP PDF) to do something is none existent.

I am hoping to get together a set of examples to read / write to the various peripherals on the board and if im really lucky a set of images that will actually support the HDMI out of the box or the LCD without having to learn a whole slew of things todo with modifying boot loaders, compiling kernels etc etc

Wish me luck, and if you don't believe me or think I'm alone in my view, just go look at the reviews for the SAMA5D3 from mid last year. It was not a successful road test by a long shot

I hope some of you folks protesting against what I have just stated have the luck to be selected and allow you "Experts" to give the rest of us the boost we need so we can "Do Our Job"

This post does not aim to insult anyone or put down any one persons skill set but to simply highlight the fact that there are so many varieties of needs out there and so may options available for the hard core chip level developer that these boards are trying to bridge a gap in the market, I speak for my self when i say this, some of us have no interest in developing the drivers and lowest level integration code for hardware, We just see the potential in the board and simply want to tap into it with our applications, others have different needs and that's ok too. But please, give the rest of us a break already.

Rant over

The www.yoctoproject.org website has a lot of documentation on how to add

applications. Semiconductor vendors are not here to teach you your job.

There are other boards available if you don't like this kind of boards.

In my IoT application, I need Linux, Console port, USB Host and Ethernet. Thats it.

This type of board then fullfills the requirement.

Using Beaglebone right now but the TI Linux support sucks, with hotplugging problems

on both USB and Ethernet. The TI/Beaglebone boot process sucks as well.

Thanks Simon,

I've been looking for a datalogger solution for use with a MPU-9250 motion chip, and there may be some libraries for this Atmel series available.

Scott

Hi Scott,

Atmel do deliver a free of charge SW package called “Softpack”.

All information can be found with below website link:

http://www.atmel.com/devices/ATSAMA5D44.aspx?tab=documents

All SW information relative to Linuxcan be found at Linux4SAM website: http://www.at91.com/linux4sam/bin/view/Linux4SAM/

The Android link is : http://www.at91.com/android4sam/bin/view/Android4SAM/

Cheers

Simon

Razvan Cojocariu wrote:

 

There are plenty of Cortex-M chips that have ethernet peripherals and it's not that hard or complex to make use of them, actually the process is pretty much the same. Yes, it's easier to slap an OS on the Cortex-A family and use it like that, but that's stupid/bad engineering if all you need is ethernet connectivity, not to mention wasteful.

According to me, as the system complexity increases, the developer will struggle a lot with a bare metal approach compared to an operating system based one.

Even we can do internet related stuff using cortex M, it still depends on how much we can do. I believe it will be more efficient, flexible and user friendly with a linux kernel and that requires an MMU which a cortex M series cannot provide. Now talking about A5, the cortex-A5 is having the lowest power consumption comparing all the application MPU series. Also if application need to play media files (video/audio) the best option will be a low power application processor like cortex A5.

I should agree with you for some reasons like people using (or insulting) raspberry pi for doing very simple tasks which even an 8 bit MCU can do, but that is a different story. For a hobbyist an MCU is easy because it is a single chip solution and easy to get started. Now a days most of the MCUs (atleast ARM) are coming with SMD packages and no DIP. So mostly nowadays hobbyists prefer to buy a development boards or evaluation boards for MCUs (say stm32f4 discovery, msp430 launchpad, stellaris launchpad, TIVA C etc etc), so why can't they think about boards with application processors also?

There are plenty of Cortex-M chips that have ethernet peripherals and it's not that hard or complex to make use of them, actually the process is pretty much the same. Yes, it's easier to slap an OS on the Cortex-A family and use it like that, but that's stupid/bad engineering if all you need is ethernet connectivity, not to mention wasteful.

I want to agree with you @shabaz. Most choices of processors come down to resource questions. The more libraries (ie interfaces needing drivers), the more ram/flash is needed to contain them. Cortex-M  have made progress for sure, but the Cortex-A start with better allocations in this regard. As a side note, more code lends to needing faster processor as well.

Just my two cents worth,

Clem

The board looks fantastic.

 

I disagree, the only ARM chips that have something to offer for the hobbyist market are the Cortex-M ones. The Coretex-A family has always had features and constraints that made them unappealing even to the hardcore hobbyists.

 

Most of the engineers (I'm including hobbyists) that I know have used (do use) Cortex-A. This is because it is realistically quite difficult to create Internet-enabled products with Cortex-M alone. Both series have their uses, and both are used by hobbyists.

I disagree, the only ARM chips that have something to offer for the hobbyist market are the Cortex-M ones. The Coretex-A family has always had features and constraints that made them unappealing even to the hardcore hobbyists.

Trying to send the message that "Hey, this is easy, anyone can do it!" is both misleading (even considering the ASF) and disrespectful towards the people that take pride in being able to actually understand, program and use these chips.

I have one of these boards and the TM7000 LCD, so far the support is exactly the same as the SAMA5D3 Xplained which is not alot.

It still comes with POKY / YOCTO as its default OS and the same lack of step by step instructions, hopefully I may be able to find out enough along with gregoryfenton to help the road testers out with their reviews

A down side of the LCD is that I have to de-solder about 20 SMD resistors from one set of pads to another to allow it to work with the D4 vs the D3 (Yes the interface for the LCD must be different ??, go figure), it is nice that the LCD allows for this but a set of jumpers would have been nicer. or even better, the Xplained board maintaining the same LCD interface.

Now a caveat to this is I have so far only performed a high level overview and examination of the software and board but I have yet to discover any images with native support for the HDMI or LCD which was a common theme with the D3 version but as I said, early days. I would have expected this to be an upfront and obvious set of documentation provided with the board but its not so we will have to deal with that. I am reaching out to various communities to try to pull together enough information to create a guide but so far no joy. Ill keep you all posted

The shop datasheet has the wrong link.

element14 Technical Page points to Digilent PmodRTCC Peripheral Module

I think it may be SAMA5D4 Xplained Ultra

In my opinion the chip is aimed at anyone with the need to use this level of functionality,

This board is aimed more at the folks with a bit of Arduino etc in their blood and want to evolve beyond that, there are other versions of the same chip / eval kit without the Arduino like headers (SAMA5D44 EK for instance) aimed at the more professional engineer types you refer to. this board is the same as a launchpad with a CC3200 or TivaC ARM chip on it or a RIOT board as far as the target audience

The headers are aligned for use with shields but in no way limit you to just that use, there still direct connect to the chip and there are also other headers to complete the set

Give me a chance, I am backtracing my steps so I can document everything

I have had mine running for days running dump1090 and feeding passing plane information to fr24.com

Videos look really good on the lcd display too

Let's hope the board support package is better than the ATSAMA5D3... That one had one Yocto image, but no step-by-step to tell how to build your own applications.

Arduino headers? Shields? What a bad joke. I bet the engineers that made this wanted to stab themselves in the face for being forced to make such an abomination.

1) This chip is meant for product engineers.

2) They don't use ardunio so they don't have shields.

3) This board has probably sacrificed the functionality you expect from a dev board (proper pin breakout), in order to support shields.

Result: frustrating board.

I'm sure the chip is great, since Atmel has the best peripherals around and their other ARM chips are a joy to work with, but this board is silly and they should come up with a proper one.

My only hope is that whatever is available via the Arduino headers is also available via the other headers.