RT5350F-OLinuXino-EVB now have micro SD card


The new revision C of RT5350F-OLinuxino-EVB now has micro SD card which could be used for storage.

The newest images from our ftp server supports it.

OpenWRT summit 2015, Dublin recap


OpenWRT summit 2015 was in Dublin on 8 October right after ELCE so we took the chance to stay one more day and participate, this was right decision.

OpenWRT summit was organized by Prpl Foundation, the event is the first one to get together all OpenWRT developers and it become very interested event .

Few talks took my attention:

1. Adding new SoC to OpenWrt presented by Hauke Mehrtens
Hauke went through what you should consider when add new SoC to OpenWRT. It was brief without complex and boring stuff.

2.Project Turris – open router with OpenWrt presented by Bedrich Kosata
This was one of most interesting lecture at least for me as it talks for project Turris – build of completely open source hardware router for traffic monitoring and DDoS prevention.

3. Securing the Internet of (broken) Things present by Cesare Garlati
Now this was scary thing and show how the aim to connect everything to Internet without proper security measures could be dangerous. Showing how people can take over control on your car which is connected to Internet, or how they can change the volume of medicine which go in your veins by manipulating your internet connected medical pump – stuff which could kill people!

4. Introducing the sunxi target – Your next IoT gadget presented by Zoltan Herpai the Allwinner SoC maintainer in OpenWRT.

Let’s hope Prpl Foundation or OpenWRT people will set page and put the slides online, many people will be interested!

OpenWrt Summit 2015 will be in Dublin, Ireland on October 8th


This year LinuxCon Europe and Embedded Linux Conference Europe will be in Dublin, Ireland from October 5th to October 7th.

Olimex is bronze sponsor to ELCE for second year.


What we learn is that this year after LinuxCon/ELCE in the same place on October 8th will be FREE to attend OpenWrt Summit which is sponsored by Prpl Foundation.

You can learn more details here.

OLinuXino roadmap 2015 :)


Many people ask us about what we do with Open Source Hardware OLinuXino project. Indeed we didn’t blog about OLinuXino boards often, but this doesn’t mean we do not have fun and working on them.

So let’s make brief update:

RT5350F – the DIN rail version is already complete and wait for assembly, the issue is trivial the plastic box, it will take us most of the time to complete.

Other OpenWRT boards:

We completed and now prototype MT7620N @580Mhz with some more RAM – 64MB and two WIFI bands 2.4G + 5G and 5 10/100Mb Ethernet and prepare board with MT7620A @580Mhz and little bit more RAM 256MB, x2 Gigabit and x4 10/100Mb Ethernets.

When these will be released – we don’t know, RT5350F software took us about year to get NDA info about WiFi tunning, we don’t know what it will be with MT7620, so do not hold your breath 🙂


A33-OLinuXino (Quad Cortex-A7) now is complete and on prototype stage, still had no time to check if Linux-Sunxi community have Linux for A33 and at what stage it is.

Now we route H3-OLinuXino (with Quad Cortex-A7 and which have Ethernet and HDMI) and A64-OLinuXino (with Quad Cortex-A53, HDMI, Camera, MIPI/LVDS/RGB, Gigabit Ethernet).

Why we stopped to blog or announce so often the new boards we working on?

Because when we blog that we start working on new OLinuXino board many people start to make commercial plans with them and how they will obtain OSHW designs for free for their commercial idea/project. Sure this have nothing wrong, if some of them didn’t became too impatient, annoying and noisy if the release time do not match their expectations. They forget that we do this for fun and we do not have any firm release dates, so they try any forms of social engineering posting challenging or abusive messages thinking they would influence us somehow this way, well soon or later they realize they can’t but anyway this is annoying for us as we want to work in rest calm environment and enjoy what we do.

We learn while we do this stuff, AM3352 software support took us year and half, we couldn’t find vital info for RT5350F for more than year, I have no idea when MT7620, A33, H3 or A64 will have proper Linux support, so please do not hold your breath, these boards will be released when they are ready, if you are capable to help with Linux on these please let me know, we can send you preliminary prototypes, so we can work together to speed up the release.


ESP8266 has now lot of followers but they do not offer anything better


ESP8266 took the developers hearts for no time. To have WIFI stack and processor resources in tiny chip with only 4 external components necessary to operate was great success story by Espressif.

Needless to say this didn’t remain unspotted by others and now we are going to see boom of similar modules.

IMO they are doomed to not gain such success though. Why? It’s simple – nobody can make simplier device.

ESP8266 is already simple and well optimized. If you buy it in reel price for 5000 pcs is $1.50/each, if you can buy in volume I’ve heard you can buy it for less than $1.

Espressif sells about 1 million of these chips now per month. Their sales warned me that if I want to order more than 500K pcs I have to warn them 4 weeks in advance, small orders like 5-10K they ship same day from stock.

So whoever decide to beat ESP8266 should make something better which cost less $1 which is quite steep goal.

We saw now next competitor MXCHIP it’s nothing fancy STM32+MarvelWIFI chip and TCP-IP stack. it’s sold for about x2 end user price, but we are not impressed at all from it. Why? Because it’s in the same league like ESP8266.

We develop IoT Firmware for more than 4 months now and we work extensively with ESP8266 so we learned his good and weak sides.

Good side: incredible simplicity as hardware – you have everything in the chip. High integration, reliable work. Most of the problems are due to the not so good software implementation, although Espressif now rised their bug bounty award to $2000 for reported bug i.e. they are confident the bugs in their firmware are fixed.

Weak side: their RTOS is binary blobs which make hard to contribute, we see lot of missed events when you enable more timers etc, but we can’t see where these comes from as the software is closed. We do workarounds on higher level to make sure IoT Firmware works reliable though for end user who do not have to worry about what happens at low level. Another issues is lack of memory to run normal TCP-IP stack. https? forget it no memory, once one https connection is made there is no resources to handle other, so in practice is not usable. SSL? forget it, no memory for more than 512 bit encryption which is assumed compromised already, so basically with ESP we do not advice you to connect to anywhere else than your local WIFI net, as WIFI is encrypted, it provides you with good security, but if you want to make your “IoT” available to outside, you have to put between ESP and Internet something with real TCP-IP stack and good security, like RT5350-OLinuXino running real Linux, where you can install SSL with 2048 bit key, and hide well your devices which otherwise could be hacked from outside. We already work on such ‘broker’ which to allow access to ESP8266 IoT Firmware from outside.

So whatever comes next as ESP competitor will be in the same league unless have at least few megabytes of RAM to may handle properly https and ssl encryption, and I highly doubt one could make such for less than $1 to compete with ESP.

ESP8266 already deeply penetrated the low cost low security market. Good luck to the others trying to compete them!

The Long anticipated RT5350F-OLinuXino is finally in stock and we have some more good news


RT5350F-OLinuXino development started loooooong time ago, we made the hardware very quickly and hit the wall with the software 🙂

The Hardware had two revisions, we knew that this IC is overheating and need external DCDC power regulator to release it from the heat, but we first decided to try how bad is it with the chip steps down the 5V to 1.2V with internal LDO, well…. the chip was easily rising to 60C at room temperature 20C, so we made revision.B where we add external DCDC and the temperature of the chip cool down to 40C, so far so good.

The next obstracle was WIFI tuning, Ralink have bunch of registers which are not described in the ‘datasheet’ but you have to know what do with them if you want to tune your WIFI to work correctly.

OpenWRT wisely skips this area of memory as each module vendor is responsible to write the optimal info for the chip in these. It took us about year to figure out what to write in these registers, including obtaining NDA info about the chip features.

The result is good, the chip makes link from up to 20 meters when in house and up to 50-60 meters in open air which is about the maximum of this small PCB printed antenna.

RT5350F-OLinuXino is now in stock and cost EUR 15 in single quantity.

We didn’t stop here, we made also EVB with two relays, button, RT5350F-OLinuXino socket, UEXT connector, USB host, two 100MB Ethernet ports and EXT connector with all unused ports.



The RT5350F-OLinuXin-EVB is also in stock and cost EUR 24 including RT5350F-OLinuXino in it.

Then we decided to make new version with just one Ethernet and LED display in DIN enclosure:


you can see it here how it looks in plastic box:


RT5350F-OLinuXino-DIN is still not complete but will be in stock soon (hopefully end of June). The price will be about EUR 35. It will be nice solution for home automation, serving array of ESP8266 sensors and BLE modules as OpenWRT supports 6LoWPAN.


RT5350F-OLinuXino Update



We startd our OLinuXino with RT5350F back in July right before our Summer vacation.

As soon as I blogged about it we got tons of useful tips and advices on the blog and by e-mails, like info about RT5350F frequency tunning, Antenna design, Overheating internals, which was not obvious from the skinny info on the web.

We decided to build two prototypes one with integrated and one with discrete baluns and on the picture above you can see the version with balun on left and discrete components on right side. Here is how the boards look on back side:


Our first intention was to make this board with 64MB of RAM and 8MB of Flash, but after reading more carefully the datasheet we found that this will require two RAM memory banks and two chips, which would increase the size of the board, so we decided to keep the size and make only 32MB RAM, but to increase the Serial Flash to 16MB.

The prototypes start working from the very first time, which makes me to worry about what we did wrong 🙂

The schematics and CAD files are on GitHub. Note these are preliminary and we have to update them with the latest components exchanges like SPI memory now is W25Q128FVSSIG and RAM is W9825G6JH-6.

Although we got tip for the tendency RT5350F to overheat (it uses internal LDO for 1.2V which dissipates enormous amount of heat and can cause the chip to desolder from the high temperature, we are non-believers and wanted to try ourself… well the result is that RT5350F really overheats 🙂 as this 1.2V could be feed from outside too, this will be solved in the next revision and we will put small external DCDC to release RT5350F from this heat source inside the chip.

This is our test setup we made small carrier board with 5 Ethernet ports and USB host to may test the board functionality, do you see the big heatsink 🙂 :


the cable on the left side is JTAG for initial uboot programming to the SPI Flash, the small board on the right side is the USB-Serial for the console.

The OpenWRT is running fine, so after all tests are done next week we will experiment to lower the heating and make new prototypes.

What we still didn’t decided yet is what EVB to make for this board?

Initial though is:

  • x1 or 2 Ethernet ports
  • x1 USB host
  • x2 RELAYs
  • few GPIOs with optical isolation
  • UEXT with UART/SPI/I2C for connecting UEXT modules
  • small 4 digits 7-segment display
  • all this put in nice compact DIN enclosure

it will be easy to put on DIN rail and to measure things and connect to Internet:


Definitely with these problems of the overheating the board will have no problem to run on negative temperatures, just the problem will be if the ambient temperature rise above 40-50C 🙂 but let’s see what will happen with external DCDC in the next revision.

As always we are open to hear your suggestions 🙂