LeafLabs Garden

A place to share, learn, and grow...

1. 

   bubulindo
   Member

   I think you forgot to include a new rcc.c

   There are some functions to activate the clocks missing when I try to compile. I've added that, but I ran into another problem while trying to wrap up everything in a .cpp file.
   Turns out that the problem was the rtc_dev *RTC that shared the class name. LOL

   I also had problems including time.h in the .cpp file, but I had more errors at the time.

   In the end, I managed to compile everything as a library, but the test program hangs the processor. I'll start from the .c files tomorrow to see if I started the clocks correctly.

   Posted 4 years ago #
2. 

   Rod
   Member

   The functions to activate clocks are in a modified rcc.h (as opposed to rtc.h) as 'static inline'.

   Did you swap out your entire libmaple for my version, or just install the rtc.c and rtc.h files?

   Posted 4 years ago #
3. 

   bubulindo
   Member

   just the rtc.c and .h :\

   I looked up your rcc.h but didn't find these functions. I'll have another look.

   Posted 4 years ago #
4. 

   bubulindo
   Member

   Ok... I give up.

   I copied all your files, put them in the right place.
   Compiled the test program with no errors, downloaded the program and it hangs.

   I tried all clocks with no luck, different prescalers, etc...

   Tomorrow, I'll try to see where it's hanging. The only place I can see is when setting up the clocks we have a cycle running without a timeout. Maybe that is where it sticks?

   Have you ran this successfully?

   Posted 4 years ago #
5. 

   Rod
   Member

   The inline functions to start and stop clocks are at the end of rcc.h.

   Seriously, rename your libmaple directory to libmaple.0.0.12, and install mine as libmaple. There should be no new issues, just new functionality. The library should just build cleanly for you. Less trouble for you and I'm interested in anything that doesn't just work in this scenario. Better I should fix any problems than require LeafLabs to.

   I did put a test program in the examples directory with the other test programs. I now see there is no direct connection between examples in that directory and the IDE. My test program just compiles on my Mac (while cleanly building libmaple.a from scratch) and runs on a Maple Mini for me.

   Posted 4 years ago #
6. 

   Rod
   Member

   Didn't see your second post there for a bit...

   I had four problem areas while I was trying to get this to run. The first was getting the clock to tick at all. The second was with the synchronization between the RTC domain registers and the APB1 registers. The third was getting the global interrupt to be generated and the fourth was with the EXTI interrupt.

   Yes, it works fine for me on a Maple Mini using the LSI clock.

   If nothing works, my guess is that the synchronization is not working. You need clocks running on both sides of the interface for things to work properly. See section 2 of app note AN2821 (google 'stm32 AN2821') for the full list, but the brief summary is APB1 clocks for PWR and BKP domains as well as the RTC clock. If it come out of the rtc_init() routine, all this should have been done. It will hang on rtc_wait_sync() and or rtc_wait_finished() if the clocks are not there.

   Since the RTC runs when the MPU is completely shut down, there is a fair bit of complexity in the intertface between the two.

   To debug, stick with the LSI clock. It's got to be there. Look at the CRL register for the RTOFF and RSF flags to come up. Watch for the seconds count register to increment. If you get that operational you are in good shape.

   You said you have an Olimexino board (nice looking board, BTW. Like the SD card). Hopefully you are not having a $(BOARD) problem. You have other LeafLabs example programs running, right?

   Posted 4 years ago #
7. 

   bubulindo
   Member

   Yes, I had other programs leaflabs running before, I've already got the CRC working and already included the SdFAT library to work with the board.
   It is pin compatible and has the same chip and firmware as Maple. Don't think the problem is there.

   I also have that application note here with me and everything from the code seemed to match it. Maybe it has something to do with the calibration registers?

   I do notice that the USB gets irresponsive too. Maybe there's some linkage problem somewhere.

   I'll troubleshoot further today.

   Posted 4 years ago #
8. 

   Rod
   Member

   I don't think Maple and Maple Mini are pin compatible...

   But, I agree, I think you have that issue covered.

   I've done the SD thing before on a PIC and its a pain.

   Mind sharing where you started on the sdFAT thing, or maybe your code via github? I just bought a copy of your board on-line.
   Not sure it will work for me with the application I have in mind. Will likely need more memory and so a Maple native, but it'll get me started on SD, let us talk apples to apples and I can use it with a GSM shield I have on yet a third project.

   Posted 4 years ago #
9. 

   bubulindo
   Member

   I got the library from here:

   http://code.google.com/p/maple-sdfat/downloads/detail?name=mapleSDfat.zip

   It's then a matter of putting it in the right place, setting the HardwareSPI to 2, I think, and use the functions. This one comes with a warning over a unsigned signed comparison and there was another error, but it was easily fixed. Can't remember what it was now.

   It was a breeze.

   I was going through the CRC calculator files I coded last week and I noticed one thing that happened to me then that is similar to what we have here.

   When you define:

   rtc_dev *RTC = &rtc;

   I was doing the same thing on the CRC calculator and it would not move, I stopped using the pointer (since there is only one CRC calculator) and it started working fine.

   One other thing I notice is that by adding these files, delay(), micros() link gets broken. Same with the time.h library. So I think the problem might be somewhere in the includes throughout the files maybe.

   Posted 4 years ago #
10. 

    bubulindo
    Member

    Ok... I installed everything in my windows machine and managed to run your example with the LSI enabled. HSE and LSE are still not working.

    Strange how it didn't work with any of them on my other computer. :\

    Posted 4 years ago #
11. 

    Rod
    Member

    Cool. That is great.

    My Maple Mini doesn't have a 32 kHz crystal, so I haven't tested LSE.

    Posted 4 years ago #
12. 

    bubulindo
    Member

    Found one issue...

    This:

    static inline void rcc_start_lse(void) {
    *bb_perip(&RCC_BASE->BDCR, RCC_BDCR_LSEON) = 1;
    while (*bb_perip(&RCC_BASE->BDCR, RCC_CSR_LSERDY_BIT) == 0); //error...
    }

    Should be:

    static inline void rcc_start_lse(void) {
    *bb_perip(&RCC_BASE->BDCR, RCC_BDCR_LSEON) = 1;
    while (*bb_perip(&RCC_BASE->BDCR, RCC_BDCR_LSERDY_BIT) == 0); //error...
    }

    Same with:

    static inline void rcc_start_hse(void) {
    *bb_perip(&RCC_BASE->CR, RCC_CR_HSEON_BIT) = 1;
    while (*bb_perip(&RCC_BASE->BDCR, RCC_CR_HSERDY_BIT) == 0);
    }

    Should be:

    static inline void rcc_start_hse(void) {
    *bb_perip(&RCC_BASE->CR, RCC_CR_HSEON_BIT) = 1;
    while (*bb_perip(&RCC_BASE->CR, RCC_CR_HSERDY_BIT) == 0);
    }

    Still not working.

    Posted 4 years ago #
13. 

    Rod
    Member

    I agree with those two errors and have added the fixes to the pull request. New commit is 706941b.

    Posted 4 years ago #
14. 

    bubulindo
    Member

    The program hangs when I poll for the clock ready bit.

    I had a look around google and some people said that they had problems with the 32kHz crystal with this chip. So I'm leaving that part behind until I find an oscilloscope to see what's going on with that crystal.

    Could it be that the crystal pins need to be configured in some way that the Maple initial configuration?? I didn't find any documentation regarding that... so it's a wild guess.

    However, the HSE is working otherwise the board wouldn't run. And I still have the same hangup problem there. I notice that you turn on HSEON bit, but shouldn't that be already on?? Same with the RDY bit. Or are these registers only for the RTC configuration?

    Thanks for the help.

    Posted 4 years ago #
15. 

    bubulindo
    Member

    Ok... 2 down, 1 to go.

    I commented the first line in the rtc_start_hse() and just verified that the clock was stable. That seems to work properly. After all, HSE was already running otherwise the board wouldn't "move".

    Any ideas regarding the LSE?

    Posted 4 years ago #

RSS feed for this topic

Reply »

You must log in to post.