Open Bosbessenbasje opened 8 years ago
Wow, thanks for report :) I've stopped to use Spark for neopixels, because i was not able to run SD libraries with DMA to read data from SD and pass it to LEDs. Which of STM's are you using now? I have one private program for massive stm 407, which are on the Discovery board. It listen for commands from UART, then can push data from SD to leds, up to 10kf rames per second :p With APA102 leds :P
2016-08-10 3:32 GMT+05:00 Bosbessenbasje notifications@github.com:
Hi,
I just wanted to tell you (so not really an issue, well, sort of), that I have succesfully adapted tour NeoPixel DMA library for the STM. My thanks are great.
Just one (actually two) issues I wanted to point out:
- in line 143 and 157, you disable the wrong timer.
- there is a problem if other threads are running which results in blinking or flickering of the string. After a long search and playing around, I found that this was fixed by swapping line 137 and 138. So first start the DMA, then the timer. That solved my problem.
Regards from the Netherlands and thanks again for all the work in this class.
Marcel Boom
PS. You might want to change some comments, as they are clearly from other implementations/tests...
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ekbduffy/Spark-Neopixel-DMA/issues/1, or mute the thread https://github.com/notifications/unsubscribe-auth/AGWLaN91nvhfDDYmUa4xKkYQa6ENBcj3ks5qeQAAgaJpZM4Jglxn .
Best regards, Davletshin Ivan
Wow, impressive. 10k frames per second is quite fast. I am using the stm32f103c8. I was using an lpc1343 before that, using the spi trick as the lpc can generate the exact needed baud rate. I have been building a more or less fixed pattern to light up stairs depending from which side the stairs are approached. But it als triggers the alarm (using an nrf24l01 on spi). In the latter was the problem. The lpc has only 1 spi bus which means that communication with the Nrf messed up the led string and because of the needed timing the led string messed up the rf communication a well.
I started using the loc because it is very cheap (€1,75 on aliexpress) and has two spi ports. However, the stm cannot generate the required spi baud rate accurate enough and at that time I asked google what alternatives there where. Ending up here.
After fixing the mentioned issues, the led string is no longer messed up with the rf interrupts, which made me happy. The fix is not needed/required if you push a lot of data all the time to the LEDs without interrupts, but in my case this is different.
Now I am only rewriting your code (to comply with our coding standards and house rules) and then for some final extensive testing.
Loc = lpc (auto correct on the iPad is not used to technical stuff ...)
Loc = stm.... :-)
Great :) commited change. But i'm totally switched to APA102, it's working right on SPI, so it can be driven from Raspberry PI and other linux machines, which haven't accurate timings for SPI interface. Also i've already tried to drive them with esp8266, which is pretty small :) So i'm advising to try them :)
Hi,
I just wanted to tell you (so not really an issue, well, sort of), that I have succesfully adapted tour NeoPixel DMA library for the STM. My thanks are great.
Just one (actually two) issues I wanted to point out:
Regards from the Netherlands and thanks again for all the work in this class.
Marcel Boom
PS. You might want to change some comments, as they are clearly from other implementations/tests...