A10-OLinuXino-LIME EUR 30 Open Source Hardware Linux SBC first prototypes


In September we blogged about upcoming LIME. Now the first prototypes of A10-OLinuXino-LIME are assembled.


This is little compact board with dimensions only  84 x 60 mm a bit bigger than credit card size, but full of features:

  • A10, Cortex-A8, 1Ghz, NEON, VPU, GPU
  • DDR3 512MB
  • microSD card
  • optional 4GB NAND Flash
  • SATA + power supply
  • HDMI 1080p
  • USB Hosts x2
  • Ethernet 100Mbit
  • Lipo battery managment and connector
  • 200 GPIOs on 0.05″ connectors
  • PWR jack +5V only

When we designed it this time we selected proper enclosure from the start:


And here is the final result:



Once everything is tested with the software we will continue with mass production and if everything goes smoothly we will have LIME in stock in December.

Price will be EUR 30 for the base model without NAND Flash and enclosure. The Plastic box will be available for +EUR 5.00

RETRO COMPUTER MEMORIES: Acorn BBC Micro model B and Acorn BBC electron



BBC Micro and Acorn Electron – I like these two computers from my retro vintage computers collection as there is interesting story around them.

Everyone has heard for ARM processors, the fastest growing architecture, they are everywhere: in phones, tablets, handheld devices, GPS navigators, media players, set top boxes and in most of Android devices. Nova days we have even Cortex-M0 ARM processors for $0.30 which penetrates the holy territory of the 8-bit processors.

If you now ask yourself What ARM processors do have with the above retro computers, the answer is simple, because of these computers ARM processors were designed in first place.

How this happend:

Back in 1970s Chris Curry is Cambridge University professor with interest in the microprocessors, he launched together with his friend Herman Hauser two companies: first with research in the microcprocessors named Cambridge Processor Unit Ltd (CPU) and another named Acorn Microcomputer with aim to make mini computer based on the CPU processors.

According to Curry the Acorn name was choosen so it appear before Apple in the telephone directory, which show which company they were influenced at that time.

Some engineers joined to work in Acorn Computers and one of them Sophie Wilson designed few computer systems there. These were primary laboratory oriented kits with LCDs, keyboards, processors which allow laboratory exercises to be completed with students – looking same as Olimex’s development boards now 🙂

In 1979 Curry influenced by ZX81 computer launch decided also to move to home computer design and this is how Acorn Atom was build. Based on 6502 as any other home computer at that time.
After the Atom was released Curry and the team started thinking for more powerful processor, Hauser suggested compromise 6502 with enhanced peripheral co-processor which to unload 6502 from the part of the tasks so they designed the next version Proton.

BBC in 1981 selected Proton for their educational program and the computer was renamed to BBC micro. Acorn sold more than 1.5 million BBC microcomputers and this helped the company to make enough profit to start working on their own processor. With the launch of IBM PC in 1981 and the Apple Lisa Acorn saw that they need more powerful processor, first attempt was to contact Intel to licensee 268 and improve it, but Intel refused so they as University professors decided that its not so hard to make their own 🙂

At that time everyone has put efforts to make processors which machine language is close to higher level programming languages i.e. so named CISC (complex instruction set computing) processors. The idea was that programming language would be easy to translate and execute faster if the machine language instructions are close to it. This lead to very complex implementation of the processor, long signal paths, which didn’t allow high frequencies (most processors at that time run at 1-2Mhz) and it was taking about 100 man years to develop something like 6502. Acorn had no such resources so they went to different route – to make RISC (reduced instruction set computing) processor which was easier to develop with smaller budged and team. Sophie Wilson set about developing the instruction set, writing a simulation of the processor in BBC Basic that ran on a BBC Micro with a 6502 second processor. Another benefit from the RISC architecture came the smaller paths to implement and the higher frequencies and lower power supply the RISC processor required compared to CISC processor. First RISC processors were running at 4Mhz and the power consumption was less than CISC.  Smaller size also means less cost to manufacture.

It convinced the Acorn engineers that they were on the right track. Before they could go any further, however, they would need more resources. It was time for Wilson to approach Hauser and explain what was afoot. Once the go-ahead had been given, a small team was put together to implement Wilson’s model in hardware.
The official Acorn RISC Machine (ARM) project started in October 1983, with Acorn spending GBP 5 million on it by 1987. VLSI Technology, Inc were chosen as silicon partner, since they already supplied Acorn with ROMs and some custom chips. VLSI produced the first ARM silicon on 26 April 1985 — it worked first time and came to be known as ARM1. Its first practical application was as a second processor to the BBC Micro, where it was used to develop the simulation software to finish work on the support chips (VIDC, IOC, MEMC) and to speed up the operation of the CAD software used in developing ARM2. The ARM evaluation system was promoted as a means for developers to try the system for themselves. This system was used with a BBC Micro and a PC compatible version was also planned. Advertising was aimed at those with technical expertise, rather than consumers and the education market, with a number of technical specifications listed in the main text of the adverts. Wilson subsequently coded BBC Basic in ARM assembly language, and the in-depth knowledge obtained from designing the instruction set allowed the code to be very dense, making ARM BBC Basic an extremely good test for any ARM emulator.

In 1982 Acorn released budged version with lower cost of BBC micro called BBC Electron, it was with co-processor IC ULA which had inside all logic and peripherals of BBC implemented.

Unfortunately Electron come to the market late as meantime Commodore 64 wiped all competitors in 8-bit scene, Atari and many others went in bancrupcy in 1985, Apple was about to sink also, same happen with Acorn. Olivetti take it over, but the new processor activity was kept in such secret that Olivetti even didn’t know that they buy it.

Good move from Acorn management was to present the new ARM processors to Apple and they decided to use it in their handheld Newton device as were impressed by the price, power and performance of this new chip, so Apple and Acorn began to collaborate of developing the ARM so they formed new company ARM Ltd in November 1990 where Acorn Group and Apple had each x 43% and the VLSI was both investor and first ARM licensee.

Now back to our retro computers:


Here is how BBC Micro looks on bot:


and back:


Note the many peripheral chips, this was one of most advanced computers at that time, it had, LAN, Graphics co-processor, Audio co-processor, needles to say some of most interesting games were on BBC micro 😉


The BBC Electron






as you can see all glue logic and peripheral were fit inside custom chip :


which allow BBC Electron to be sold under 200 GBP, sure this didn’t help to complete the mighty Commodore 64 which wiped all 8-bit market, but the research to make newer faster better processor for the Acorn computers lead to the design of one of most successful architecture at our time ARM

Retro Computer Memories: TRS-80 Model 100


TRS-80 Model 100 is one of my favorite computers from my vintage computers collection.


Although build in 1983 it’s in so cute and compact form that you immediately fall in love when you look at it.

This is the mother of all Notebooks and Laptops 🙂 with of only 1.4 KG with the batteries (4xAA 1.5V type) and size of only 300x215x50mm.


It was designed by Kyocera under the name Kyotronic 85, this model was so well designed for it’s time that Tandy, Olivetti and NEC decided to license the design instead to make their own.

More than 6 Million units were sold just by Tandy in US and Canada.

The processor inside TRS-80 Model 100 is 80C85 running at 2.4 Mhz. The RAM memory is humble 8K expandable up to 24K by slots inside and at the back of the computer.

The LCD display is 240×64 pixels 8 lines x 40 characters, there is build-in 300 bps modem, parallel printer port, RS232, bar-code reader, cassete input, RTC, external CRT connector.




The firmware is inside of 32KB ROM and contain Microsoft BASIC interpreter with good support for all hardware.


here is our Hello World program written in Basic:


What is remarkable with this firmware is that the Model 100 firmware was the last Microsoft product that Bill Gates developed personally, along with the Japanese hardware developers.
According to Gates, “part of my nostalgia about this machine is this was the last machine where I wrote a very high percentage of the code in the product”.

When introduced, the portability and simplicity of the Model 100 made it attractive to journalists,who could type about 11 pages of text and then transmit it using the built-in modem for electronic editing and production. The batteries allow up to 20 hours of work without connection to external power supply.

The Model 100 was also used in industrial applications as a programming terminal for configuration of control systems and instruments.

This is how main board looks inside:


and this is the LCD and keyboard board which is connected to the main board with ribbon cable:


This 30 years old machine is so well build, that works fine even today, all it need is 4 x AA batteries!

A10s-OLinuXino-MICRO progress


About the hardware:

We made two versions of the A10s prototypes – Rev.A with all components on top and Rev.B with decouplung capacitors on bottom near the BGA chip (both versions are on GitHub).

Then assembled both versions, none of Rev.A worked, so plasing the decouplung capcitors under the BGA is mandatory for the board to work correctly.

From Rev.B the boards work but not stabile, from time to time DDR memory crash and board freezes, this is because we decided that A10s is similar to A10 and the resistor matrixes on the DDR are not mandatory. What we see is that in Linux A10s have same processor ID as A13 which makes us to believe that A10s and A13 are same silicon die just with different ports wired to the different packages TQFP and BGA. The same unstability we see on A13 when try to remove these resistor matrixes, so in the new REV.C we have to keep them.

Another change we will do in Rev.C we are working on now is to add NAND flash on board, which will be optional i.e. we will offer boards with and without NAND assembled.

One bad feature of A10s with the BGA package is that it overheats significant when work at 1Ghz, same we see on the A10 MK802 dongle we have the processor IC overheats up to 60-70C so you cant touch it with finger even at ambient temperature.

A13 do not have such issues as it have big metal pad on bottom side which we connect to large GND ground plane so A13 never ever got temperature higher than the temperature it operates, this makes A13 much more suitable for industrial and automotive applications.

A10 and A10s overheat and we should think for proper heatsink which we to assembly on top of the BGA to dissipate this heat they generate.

As you can see from the picture we added heat sink to A10s as without it it freezes after some hours of work.

The new Rev.C will be ready by the end of February for more extensive testings, this will delay the planned mass production with few weeks.

About the software:


Android 4.0.4 image generated with A10S-SDK from Allwinner is working fine, but do not support Ethernet by default.



Angry Birds works well on HD screen:



You can see here in action Angry Birds on A10s  http://www.youtube.com/watch?v=fD3IImqEprI

The Android video player is pretty good at HD playback http://www.youtube.com/watch?v=rB-eQeKXlTQ

for some reason XBMC on Android works worse http://www.youtube.com/watch?v=_579b3by1JA

Linux support patches are submitted to Linux-Sunxi so A10s software support have no issues.

AM3352-OLinuXino preliminary schematic is complete


AM3352-OLinuXino preliminary schematic is complete and push to GitHub https://github.com/OLIMEX/OLINUXINO/tree/master/HARDWARE/AM3352-OLinuXino.

We still have to test this and that and in Monday will proceed with the PCB layout routing.

Both 2Gbit and 4Gbit DDR3 memory prototypes will be made, so board with 512MB and 1GB of RAM will be tested.

OLinuXino project is Open Hardware Linux Computer project initiated by OLIMEX. The first board from OLinuXino family design started in March 2012 and to December 2012  there are 7 boards in production based on iMX233 (454Mhz ARM9)  and A13 (1Ghz Cortex A8) processors:

A10-OLinuXino, A10S-OLinuXino and now AM3352-OLinuXino boards are on design stage, there were 9 busy months for us this year 🙂

A13-OLinuXino-MICRO first prototypes are ready


Here is the first look at A13-OLinuXino-MICRO assembled prototypes.

It took a lot of time as our assembly machines are overloaded and we run them on two shifts to meet the demand for our development boards, so the blank PCBs were made and sitting for two weeks waiting on tail for the assembly window.

What we have here is:

– A13 Cortex-A8 processor running at 1GHz
– 256MB of DDR3 RAM
– USB host
– VGA output 800×600 @ 60 Hz
– SD-card
– Audio Out for headphones
– LCD connector
– GPIO connector 1 and 2
– UEXT connector
– 100×85 mm

The goal is to release A13-OLinuXino-MICRO in production at EUR 35 retail down to EUR 28 for 50+ pcs OEM order.

For those who will say that 256MB is small amount of memory I would say that we run pretty amazing stuff now even on iMX233 with 454Mhz processor and 64MB RAM, so with x4-6 times faster processor and x4 times more memory lot of projects could be complete in Linux. Well, we will probably not run Andoriod 4.0 nor Open Office and Ubuntu with Chrome, but for this purpose it A13-OLinuXino-WIFi where you have double the RAM and 4GB NAND flash disk.

MICRO have no Ethernet by default but supports MOD-RTL8188CU WiFi adapter and the AX88772B USB to wired LAN adapters which will be for sale next month.

We try the SD card with the Debian image and as expected the bootloader starts then the kernel stops as there is no enough memory. So the goal now is to disable MALI400 and the GPU which alone take 200MB 🙂 and to make kernel which fits in smaller footprint.

We assembled few of these and the first who report that disabled the Mali and GPU and managed to ran Linux-Sunxi in 256MB will get one of the first assembled MICROs 🙂

Sneak preview of A13-OLinuXino-MICRO


A13-OLinuXino-MICRO specifications:

– A13 Cortex-A8 processor 1Ghz (we found we can overclock up to 1.2-1.5Ghz)
– power managment AXP209 removed – fixed core voltages
– 256MB of DDR3 RAM
– GPIO extensions same as normal A13-OLinuXino
– LCD extension same as normal A13-OLinuXino
– micro SD-card
– VGA output, Audio output
– USB host
– UEXT connector
– JTAG connector
– console UART connector
– power supply +5V DC only
– reset button
– bootloader button
– status LED
– operating temperature -25+85C

With this board we decided to challenge how down we can cut the features to reach rock bottom pricing.
This board is still on design stage (actually we already completely routed it) and we are going to prototype it next week, so please do not ask when you can buy it. In matter of fact we are still not sure if it will work at all, as we removed the AXP209 power managment IC and decided to try if A13 will work just with DCDC fixed power supplies.
If we succeed the price of this A13-OLinuXino-MICRO will be around EUR 35 retail – and this for board with 1GHz Cortex-A8 processor and 256MB RAM.

When we discussed this with A10 developers they warned us that current Linux distribution for A10 is memory hungry and only Mali400 and VGPU when enabled need 100MB of RAM.
Well I’m pretty sure there are many applications where fancy 3D graphics are not necessary, or the Mali400/VGPU apetite may be lowered with some tweaking?
Considering what amount of software we already managed to run on iMX233 ARM9 with 64MB, I think A13-MICRO have his niches where will perform just fine!
Well Andorid 4.0 on 256MB may not work nice, but Linux should be OK.

So crossing fingers to see what will happend with the prototypes 🙂

For these who miss Ethernet we did some experimenting with USB-Ethernet adapters (starting from $3 on ebay) and A13 Linux Kernel supports them both USB 1.1 (based on QF9700) and USB 2.0 (based on AX88772) versions the later is with several times faster transfer, so we probably offer these as USB accessories to A13-OLinuXino. And of course we will offer the WIFI alternative USB-RTL8188CU 150MBit as well.  The LCD7″module, Bluetooth USB dongles, 3G modem dongles and all accessoaries we preparing for A13-OLinuXino-WIFI will also work on A13-OLinuXino-MICRO.

A13-OLinuXino first prototypes are ready for testing



Today we assembled the first A13-OLinuXino OSHW Linux computer prototypes and now prepare to test them with Android and Linux image.

The board on the picture above is the BASE model, there will be also A13-OLinuXino-WIFI-4GB which will have on-board RTL8188CU WIFI 802.11n 150Mbit module + 4GB NAND flash on board.

If everything is OK with the tests we will send few prototypes to selected developers and prepare to run production in September.

Our target is to release A13-OLinuXino for EUR 45 and A13-OLinuXino-WIFI-4GB for EUR 55.

A13-OLinuXino routing complete!


Today we completed the A13-OLinuXino routing and I push the CAD files to GitHub. Note that these are not the final, we have to make some final touch and to add text, logos etc, but I know how many of you waiting for the CAD files with no patience so I push these work CAD files for your review.

If everything goes well we may have prototypes by end of next week 🙂

A13-OLinuXino board routing progress


We decided to re-format the PCB in standard Nano-ITX form factor 120×120 mm.
The high speed DDR3 memory part is completely routed now, the board will be 4 layers, so generally most of the work is done.
Now the peripherials have to be routed and next week we may be ready for prototype PCB run.

Previous Older Entries Next Newer Entries