Messages

They come from the selected access mode template. stm32plus can drive an LCD through the 8 or 16 bit FSMC peripheral or using a number of optimised GPIO modes. You can find them all in this directory.

The 2.8V LDO is for VCI (2.60 - 3.00V). See data sheet "Table 1 R61523 Power Supply Specification"

IOVCC has a wider acceptable range but if you can power everything from 2.8V then you'll save yourself some components.

I know what's going on now. The build configuration for each example has resource 'excludes' set up so that the 'system' directory for irrelevant MCUs are excluded from the build. The excludes were not set up when I did the merge for the pull request for the f1md support so Eclipse is trying to compile the system directory for the f1md build configuration even when you've selected the f4.

You can either use git to sync to the commit before I merged that pull request or wait until the weekend when I'll create all the resource exclusions and push a new commit to github.

The 0.4mm connector is a hard one to solder but it sounds like you have the right equipment to do it. Careful not to melt the plastic with your heat gun! Good luck and feel free to post photos if you get it working

For the BL, I can just use a step up converter with max. current at 20mA, right? No PWM for 1st tests.

Yes definitely. The backlight is a separate circuit and can be driven for testing with nothing else connected. If I remember correctly I verified the suspected backlight pinout using a bench power supply, a resistor, a pair of very fine wires for probes, a microscope and my wife to tell me whether she could see a light come on around the panel when I touched the pins on the connector very carefully!

Here are more pics showing my Z800 MB fitted into a Thermaltake Armor+ VH6000 case with some fancy blue leds. The PSU is a TDPS-825AB from an xw serie desktop. The setup includes the temp sensor, firewire, usb connections to the front panel which silence many POST warnings about missing cables.  Now, only a back panel is missing, plus some fans providing input so to silence the POST warning about missing memory and rear fans. Job almost done :-)

Nice job, especially with the cables. The motherboard itself looks like a tight fit into the case. Did you drill your own holes in the motherboard tray to get it to line up correctly?

Hi,

You're in luck. The R61523 is one of the easiest panels to initialise. There's no complex oscillator or voltage multiplier settings. You just wake it up and set the orientation and colour depth.

Here's the post-reset wakeup sequence. Here's the sequence to set it to 24-bit colour depth. See similar files in the same directory for 16 and 18-bit modes. Here's the code for enabling landscape orientation.

Until the backlight is enabled you will see nothing at all so your first aim should be to get the backlight working. Here's the code to control the PWM output pin that can be used to drive the enable (EN) pin of a step-up LED driver. Please see my blog article for schematics and details of how I drive the backlight.

I think what was not clear to me was that I had to build STM32plus with float=hard and then in the projects I have to compile the library without the hard float.

That's strange. The library and the examples should match in terms of the floating point settings. You didn't say if you were using Eclipse or if you were building from the command line - which one are you using? I'm sort of guessing Eclipse but if you can confirm then it would help.

Wow, lots of questions I'll give you a full answer shortly when I've gone through the detail but basically you're looking at two issues.

• How fast you can write to the LCD bus and that's limited by the timings given in the controller datasheet combined with the frequencies you can get from the clock multipliers and dividers in the STM32. The extra grunt in the F4 doesn't necessarily help here because you'll be dividing down to keep the slower LCD bus in spec.


• How fast you can perform the graphics algorithms themselves - i.e. the work you're doing when you're not writing pixels to the LCD. The F4 clearly wins by a country mile over the F0 in this department.

Hi Marius,

The f407 discovery board is one of my main test targets, and it's also one of the automatic "Travis" builds that runs on Linux each time I push a commit to github so you definitely should have no issues. Please can you tell me the command line that you used to do the build? Basically to create a debug build on Linux from the directory that you cloned into you would do (from memory):

Code Select Expand


scons mode=debug mcu=f407 hse=8000000 float=hard install


If you are compiling from Eclipse then the key point is to make sure that you have selected the correct build configuration, first for the library and then for each example you want to compile. Build configurations can be accessed by right-clicking the project file in the project explorer.

Hi, can you let me know which compiler you are using? From the command line that gives you the errors, please type:

Code Select Expand


$ arm-none-eabi-g++ -v


You should get some output that includes the version information:

Code Select Expand


gcc version 4.9.3 20141119 (release) [ARM/embedded-4_9-branch revision 218278] (GNU Tools for ARM Embedded Processors)


Today I completed the migration of the entire website and forum from a shared host located in the American mid-west to a VPS located in NYC. I apologise for any glitches that you may have experienced today as the DNS changes trickled out across the internet.

If you have any issues logging in to these forums then please clear all your browser cookies associated with *andybrown.me.uk

I invested in an HP Z800 #02 which the seller indicated what works for XEON X56xx.

So I bought :

2x X5670 + Fans

+1 on what John said. Your seller is wrong. You need an 003 board to work reliably with dual X56xx series. Try taking CPU #2 out and booting up.

If you haven't already seen the updates in the blog article then please do take a look. I resolved all the issues I was having with oven insulation, fixed a few bugs and tuned my PID parameters. My oven now tracks the reflow profile very accurately indeed.

I needed to generate a high frequency 50/50 PWM signal out of PA2 on the F0. The device I was working with was fairly flexible on the exact frequency so I chose 12MHz because it's an exact division of the 48MHz core clock which means that a precise frequency will be generated. PA2 is the TIM2 channel 3 output. Here's how I did it:

Code Select Expand


Timer2<
        Timer2InternalClockFeature,
        TimerChannel3Feature<>,
        Timer2GpioFeature<
          TIMER_REMAP_NONE,
          TIM2_CH3_OUT
        >
      > clk;

  clk.setTimeBaseByFrequency(24000000,1);
  clk.initCompareForPwmOutput();
  clk.enablePeripheral();
  clk.setDutyCycle(50);


The timer is set to tick at 24MHz with a very low reload count of 1 so it will toggle every tick giving a frequency of 12MHz. Adjusting the frequency will get you different frequencies. For example, 48000000 will get you the maximum frequency of 24MHz and 12000000 gets you 6MHz.