Fix pwm period calc #3795

LMESTM · 2017-02-17T13:16:17Z

Description

This pull request solves an issue that was encountered in specific cases where pwm HW frequency is high and the pwm period was about 100ms (encountered on F746). This was a limitation due to hard coded pre-scaler values of 1 or 500 only. The solution is to have more granularity.

In order to apply the fix to all families and not only F7, the code is also made common between all families which makes maintenance easier.

Status

READY

TESTS RESULTS

pwm_test_results.txt

As done for other families, let's define the PWM channel in the PWM pins table definition rather than driver.

The pwmout driver is very similar for each STM32 family. The only family specific part is defined in pwmout_device.h file. It mainly contains few specific information: - The mapping of PWM/TIMERS to APB1 or APB2 so that we can get the clock - The clock calculation uses the right APB clock, which was sometimes not the case before and could have lead to errors in case dividers were enabled on APB clock settings. This case is now covered. - Inactivation of inverted support on feaw families

bcostm · 2017-02-17T18:33:37Z

targets/TARGET_STM/pwmout_api.c

+ TimHandle.Init.Period = (us - 1);
+
+ /* In case period or pre-scalers are out of range, loop-in to get valid values */
+ while ((TimHandle.Init.Period > 0xFFFF) || (TimHandle.Init.Period > 0xFFFF)) {


oops - needs to be corrected
will do when I'm back in office

Any update for this?

@LMESTM is off this week.

0xc0170 · 2017-02-20T14:06:40Z

targets/TARGET_STM/pwmout_api.c

+ /* In case period or pre-scalers are out of range, loop-in to get valid values */
+ while ((TimHandle.Init.Period > 0xFFFF) || (TimHandle.Init.Period > 0xFFFF)) {
+ obj->prescaler = obj->prescaler * 2;
+ if (APBxCLKDivider == RCC_HCLK_DIV1)


if you are touching this file, please fix formatting if possible

@0xc0170 I had a quick look but probably missed the point - what's wrong with current formatting ?

should be

if (condition) { do(); } else { do_else(); }

{ and } are missing there

0xc0170 · 2017-02-20T14:10:21Z

targets/TARGET_STM/TARGET_STM32L4/pwmout_device.h

+ PwmoutApb pwmoutApb;
+} pwm_apb_map_t;
+
+static const pwm_apb_map_t pwm_apb_map_table[] =


I believe there was earlier PR about 2 weeks ago that was defining some variables in the header file. I raised an issue there that this would be much better if it's done in a code file.
It works now as it is included in just one code file. however, this might become a silent bomb later if this HAL will include a new code file that would require some declarations from this file (it's possible and might be overlooked that this actually defines this map of pwm).

@0xc0170 yes you're right ! I will move the definition to a C file

Correct the while loop limit and add a safe guard to avoid infinite loop.

In order to avoid possible multiple definitions errors, move the table initialization to the C file instead of header file

Use the recommended style if (condition) { do(); } else { do_else(); }

LMESTM · 2017-02-27T17:03:45Z

@0xc0170 @bcostm branch updated as per your comments

LMESTM · 2017-03-06T09:50:23Z

@0xc0170 hey - I think the PR now takes into account your comments - let me know if anything missing

0xc0170 · 2017-03-06T16:15:23Z

/morph test

mbed-bot · 2017-03-06T19:34:40Z

Result: FAILURE

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 1641

Build failed!

0xc0170 · 2017-03-07T14:58:39Z

@LMESTM Please look at the failures

Here's the output (XDOT platform fails for all 3 toolchains)

13:29:01         [Error] PeripheralPins.c@88,0:  #54-D: too few arguments in invocation of macro "STM_PIN_DATA_EXT"
13:29:01         [Error] PeripheralPins.c@88,0:  #158: expression must be an lvalue or a function designator
13:29:01         [Error] PeripheralPins.c@88,0:  #158: expression must be an lvalue or a function designator
13:29:01         [Warning] PeripheralPins.c@88,0:  #1278-D: excess initializers are ignored

The link is above in the failure message to get this log

Typo with misplaced closing parenthesis leads to compilation error, which is fixed with this patch

LMESTM · 2017-03-07T16:23:29Z

@0xc0170 - thanks - there was a typo in this XDOT file. Fixed now.

0xc0170 · 2017-03-08T09:20:13Z

/morph test

mbed-bot · 2017-03-08T12:49:31Z

Result: SUCCESS

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 1654

All builds and test passed!

Ports for Upcoming Targets Fixes and Changes 3716: fix for issue #3715: correction in startup files for ARM and IAR, alignment of system_stm32f429xx.c files ARMmbed/mbed-os#3716 3741: STM32 remove warning in hal_tick_32b.c file ARMmbed/mbed-os#3741 3780: STM32L4 : Fix GPIO G port compatibility ARMmbed/mbed-os#3780 3831: NCS36510: SPISLAVE enabled (Conflict resolved) ARMmbed/mbed-os#3831 3836: Allow to redefine nRF's PSTORAGE_NUM_OF_PAGES outside of the mbed-os ARMmbed/mbed-os#3836 3840: STM32: gpio SPEED - always set High Speed by default ARMmbed/mbed-os#3840 3844: STM32 GPIO: Typo correction. Update comment (GPIO_IP_WITHOUT_BRR) ARMmbed/mbed-os#3844 3850: STM32: change spi error to debug warning ARMmbed/mbed-os#3850 3860: Define GPIO_IP_WITHOUT_BRR for xDot platform ARMmbed/mbed-os#3860 3880: DISCO_F469NI: allow the use of CAN2 instance when CAN1 is not activated ARMmbed/mbed-os#3880 3795: Fix pwm period calc ARMmbed/mbed-os#3795 3828: STM32 CAN API: correct format and type ARMmbed/mbed-os#3828 3842: TARGET_NRF: corrected spi_init() to properly handle re-initialization ARMmbed/mbed-os#3842 3843: STM32L476xG: set APB2 clock to 80MHz (instead of 40MHz) ARMmbed/mbed-os#3843 3879: NUCLEO_F446ZE: Add missing AnalogIn pins on PF_3, PF_5 and PF_10. ARMmbed/mbed-os#3879 3902: Fix heap and stack size for NUCLEO_F746ZG ARMmbed/mbed-os#3902 3829: can_write(): return error code when no tx mailboxes are available ARMmbed/mbed-os#3829

LMESTM added 2 commits February 17, 2017 14:11

STM32 L1: Define PWM Channels in PeripheralsPins.c

293b9fc

As done for other families, let's define the PWM channel in the PWM pins table definition rather than driver.

bcostm reviewed Feb 17, 2017

View reviewed changes

0xc0170 reviewed Feb 20, 2017

View reviewed changes

0xc0170 added the needs: work label Feb 20, 2017

LMESTM added 3 commits February 27, 2017 13:47

STM32: pwm period and prescaler calculation

8e5d2eb

Correct the while loop limit and add a safe guard to avoid infinite loop.

STM32: move pwmout device tables to C file

747e8e1

In order to avoid possible multiple definitions errors, move the table initialization to the C file instead of header file

STM32: fix formatting

3842f6e

Use the recommended style if (condition) { do(); } else { do_else(); }

0xc0170 added needs: CI and removed needs: work labels Mar 6, 2017

0xc0170 approved these changes Mar 6, 2017

View reviewed changes

0xc0170 added needs: work and removed needs: CI labels Mar 7, 2017

Fix XDOT compilation error

400b89e

Typo with misplaced closing parenthesis leads to compilation error, which is fixed with this patch

0xc0170 added needs: CI and removed needs: work labels Mar 8, 2017

0xc0170 added ready for merge release-version: 5.4.1 and removed needs: CI ready for merge labels Mar 8, 2017

0xc0170 merged commit fd6fdd5 into ARMmbed:master Mar 9, 2017

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix pwm period calc #3795

Fix pwm period calc #3795

LMESTM commented Feb 17, 2017

bcostm Feb 17, 2017

LMESTM Feb 17, 2017

0xc0170 Feb 20, 2017

bcostm Feb 21, 2017

0xc0170 Feb 20, 2017

LMESTM Feb 27, 2017

0xc0170 Feb 27, 2017

0xc0170 Feb 20, 2017

LMESTM Feb 27, 2017

LMESTM commented Feb 27, 2017

LMESTM commented Mar 6, 2017

0xc0170 commented Mar 6, 2017

mbed-bot commented Mar 6, 2017

0xc0170 commented Mar 7, 2017

LMESTM commented Mar 7, 2017

0xc0170 commented Mar 8, 2017

mbed-bot commented Mar 8, 2017

Fix pwm period calc #3795

Fix pwm period calc #3795

Conversation

LMESTM commented Feb 17, 2017

Description

Status

TESTS RESULTS

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

LMESTM commented Feb 27, 2017

LMESTM commented Mar 6, 2017

0xc0170 commented Mar 6, 2017

mbed-bot commented Mar 6, 2017

Output

0xc0170 commented Mar 7, 2017

LMESTM commented Mar 7, 2017

0xc0170 commented Mar 8, 2017

mbed-bot commented Mar 8, 2017

Output