XiVO IOH first CPU prototype board fully working - part1

Posted on 16 December 2011 in Hardware by xcarcelle

to announce that our first version of our prototype for the CPU board (PCB prototyping the CPU and the different interfaces connected to the CPU) is now fully working and we have a full GNU/Linux system running on it to start stress-testing our platform (tests of currents and temperature in a full stress mode). We are now getting closer and closer to publish a fully-open design of an IPBX running GNU/Linux/Asterisk/XiVO in OpenHardware.

First we had our PCB pick-and-placed in our PCB factory (mainly on a Fuji machine) and you can see below a screenshot from the precision placement camera on the machine


The main concerns of a PCB design when one wants to first test the just "pick-and-placed" board is to follow different steps to prevent undetection of hardware errors of design. These steps are usually (at least, the ones we followed) the following :

  • Smoketest : of the powersupply block doing the different voltages level of the board (crucial for a SoC like the one we are using) by using different DC external powersuplies for the different ATX voltage levels (+12V, -12V, +5V, +5VSBY, +3V3) and limiting the current following the Intel EP80579 datasheet. This test was positive on each of the voltage level and confirming that our power-supply is functionnal in terms of voltage level and current limitations
  • Bringup : booting the board with a correct DDR2 DIMM (1Go) and flashing our version of coreboot build on our 2Mo SPI-flash using a dediprog SF100 SPI programer
  • Connecting the SATA HDD w/ a fully configured Linux kernel and trying to jump from the functionnal coreboot to grub then Linux decompression of the image

Please find below, a picture of the lab deployment of the board powered by external power-supplies and connected to a bus-pirate for the flashing of our bootloader through SPI connexions.


Once this smoke-test and bring-up test were fully positive and allowing us to move to next step with some stress-tests :

  • Temperature#1 : Running a memtest86+ on the DDR2 and testing the temperature of the CPU package w/ a temperature sensor
  • Temperateur#2 : Running an Intel utility for our SoC (pushing the CPU usage to 100% and memory to 50%) and measuring the temperature
  • Temperature#3 : Putting one of our board into a test metallic case and connecting the board w/ 3 parallel GbE interfaces fully loaded w/ iperf at 1Gbps and measuring the temperature in the case after several hours
  • Bitrate : Iperf'ing the 3xGbE interfaces and measuring the load of the CPU and the transfer bitrates

Please find below, a picture of our lab w/ 2 XIOH CPU boards running our test-program and 3xGbE interfaces connected together one-by-one


We will update our different tests on the CPU boards and keep you update on the follow-ups of our design steps.

Xavier Carcelle.