Why does the newly soldered STM32 main control board run about 10 times slower than normal after programming?

bqgup

Why does the newly soldered STM32 main control board run about 10 times slower than normal after programming? [Copy link]

 

 

<i class="pstatus"> This post was last edited by bqgup on 2020-8-19 18:40</i> # Why does the newly welded STM32 main control board run about 10 times slower than the normal speed after burning the program## 1. Description of the problem of the newly welded STM32 main control board#### Recently, I designed a main control board using the STM32 RBT6 main control chip. After I completed the welding of the first piece, I burned the LED flashing program with a delay of 500ms into it. The LED light should be on for 500ms and off for 500ms in a cycle. The flashing frequency under normal circumstances is as follows: <ignore_js_op> <img id="aimg_495791" style="cursor:pointer;" aid="495791" src="static/image/common/none.gif" zoomfile="data/attachment/forum/202008/19/173803uat9uxcz9gm0j0mg.gif" file="data/attachment/forum/202008/19/173803uat9uxcz9gm0j0mg.gif" class="zoom" onclick="zoom(this, this.src, 0, 0, 0)" width="360" id="aimg_495791" inpost="1" onmouseover="showMenu({'ctrlid':this.id,'pos':'12'})" /> <div class="tip tip_4 aimg_tip" id="aimg_495791_menu" style="position: absolute; display: none" disautofocus="true"> <div class="xs0"> <p><strong>ddd.gif</strong> <em class="xg1">(3.36 MB, downloads: 0)</em></p> <p> <a href="javascript:;" class="loginf" >download attach</a> <a href="javascript:;" onclick="showWindow(this.id, this.getAttribute('url'), 'get', 0);" id="savephoto_495791" url="home.php?mod=spacecp&ac=album&op=saveforumphoto&aid=495791&handlekey=savephoto_495791">save to album</a> </p> <p class="xg1 y">2020-8-19 17:38 上传</p> </div> <div class="tip_horn"></div> </div> </ignore_js_op> #### However, in actual phenomena, the state of the LED light is indeed flashing according to a certain period, but the actual period seems very long, with about 10 delays of 500ms. The actual phenomenon is probably as follows: <ignore_js_op> <img id="aimg_495760" style="cursor:pointer;" aid="495760" src="static/image/common/none.gif" zoomfile="data/attachment/forum/202008/19/164821uw4jjespwf8j5eb5.gif" file="data/attachment/forum/202008/19/164821uw4jjespwf8j5eb5.gif" class="zoom" onclick="zoom(this, this.src, 0, 0, 0)" width="360" id="aimg_495760" inpost="1" onmouseover="showMenu({'ctrlid':this.id,'pos':'12'})" /> <div class="tip tip_4 aimg_tip" id="aimg_495760_menu" style="position: absolute; display: none" disautofocus="true"> <div class="xs0"> <p><strong>aaa.gif</strong> <em class="xg1">(7.55 MB, downloads: 0)</em></p> <p> <a href="javascript:;" class="loginf" >download attach</a> <a href="javascript:;" onclick="showWindow(this.id, this.getAttribute('url'), 'get', 0);" id="savephoto_495760" url="home.php?mod=spacecp&ac=album&op=saveforumphoto&aid=495760&handlekey=savephoto_495760">save to album</a> </p> <p class="xg1 y">2020-8-19 16:48 上传</p> </div> <div class="tip_horn"></div> </div> </ignore_js_op> #### In order to verify again whether there was a problem with the delay, I specially wrote a timer program. I set a 1s timer to control the flipping of the LED light. In fact, there was a problem with the timer. The program was about 10 times slower than the normal 1s. I immediately locked the problem to the crystal oscillator circuit. The circuit in the circuit diagram is the circuit I have been using. This problem has never occurred before. My crystal oscillator circuit is as follows: <ignore_js_op> <img style="cursor:pointer;" id="aimg_495768" aid="495768" src="static/image/common/none.gif" onclick="zoom(this, this.getAttribute('zoomfile'), 0, 0, '0')" zoomfile="data/attachment/forum/202008/19/165606n7eers451mjososh.png" file="data/attachment/forum/202008/19/165606n7eers451mjososh.png" inpost="1" onmouseover="showMenu({'ctrlid':this.id,'pos':'12'})" /> <div class="tip tip_4 aimg_tip" id="aimg_495768_menu" style="position: absolute; display: none" disautofocus="true"> <div class="xs0"> <p><strong>1.png</strong> <em class="xg1">(37.86 KB, downloads: 0)</em></p> <p> <a href="javascript:;" class="loginf" >download attach</a> <a href="javascript:;" onclick="showWindow(this.id, this.getAttribute('url'), 'get', 0);" id="savephoto_495768" url="home.php?mod=spacecp&ac=album&op=saveforumphoto&aid=495768&handlekey=savephoto_495768">save to album</a> </p> <p class="xg1 y">2020-8-19 16:56 上传</p> </div> <div class="tip_horn"></div> </div> </ignore_js_op> #### First, I suspected that there was a problem with the crystal oscillator. I immediately replaced the crystal oscillator, but the response was still slow (about ten times slower). I used an oscilloscope to measure the crystal oscillator pins and found that the crystal oscillator did not start oscillating. I checked the crystal oscillator circuit again and ruled out the possibility of short circuit between the two crystal oscillator pins. I also replaced the 22pF capacitor and the crystal oscillator several times, but the problem was still not solved. I followed the methods in major forums on the Internet and still did not work. The last modification was to replace a surface mount crystal oscillator (I used the pin-type one before). After testing, it still did not work, so I put the board aside and re-soldered a piece. After the second board was soldered, I quickly tested it with the program and found that the crystal oscillator started oscillating and the program delay returned to normal, which also verified the feasibility of the circuit. #### After two or three days, the second board has been running normally (the crystal oscillator is a surface mount crystal oscillator). I picked up the first board that I had soldered before. After powering it on, I suddenly found that it became normal. The flashing cycle of the LED light returned to normal. I guessed that it was because of the difference between the SMD crystal oscillator and the pin crystal oscillator. I immediately replaced the SMD crystal oscillator on the first board with the pin crystal oscillator. I found that the first board had the same old problem as before. The running rhythm was about 10 times slower than normal. I planned to replace the crystal oscillator with the SMD one, remove the pin crystal oscillator, and replace it with the SMD crystal oscillator. Take a look at the effect: <ignore_js_op> <img id="aimg_495785" style="cursor:pointer;" aid="495785" src="static/image/common/none.gif" zoomfile="data/attachment/forum/202008/19/172547o4el7fy7b49r85t5.gif" file="data/attachment/forum/202008/19/172547o4el7fy7b49r85t5.gif" class="zoom" onclick="zoom(this, this.src, 0, 0, 0)" width="640" id="aimg_495785" inpost="1" onmouseover="showMenu({'ctrlid':this.id,'pos':'12'})" /> <div class="tip tip_4 aimg_tip" id="aimg_495785_menu" style="position: absolute; display: none" disautofocus="true"> <div class="xs0"> <p><strong>ccc.gif</strong> <em class="xg1">(3.99 MB, downloads: 0)</em></p> <p> <a href="javascript:;" class="loginf" >download attach</a> <a href="javascript:;" onclick="showWindow(this.id, this.getAttribute('url'), 'get', 0);" id="savephoto_495785" url="home.php?mod=spacecp&ac=album&op=saveforumphoto&aid=495785&handlekey=savephoto_495785">save to album</a> </p> <p class="xg1 y">2020-8-19 17:25 上传</p> </div> <div class="tip_horn"></div> </div> </ignore_js_op> #### It's so disappointing! It's still slower than normal. I plan to leave it for another day or two to see if it can return to normal. ## 2. Find the key to solving the problem#### I took the soldered board and carefully checked it under the light. Suddenly, I found that there was a little bit of solder hanging on the No. 1 and No. 2 pins of the main control chip. The No. 1 and No. 2 pins were short-circuited!!! I quickly used a welding gun to remove it, and then used alcohol to remove the impurities attached after welding. The board returned to normal working condition. <ignore_js_op> <img id="aimg_495794" style="cursor:pointer;" aid="495794" src="static/image/common/none.gif" zoomfile="data/attachment/forum/202008/19/182618ir5ij3ojggj9djg5.gif" file="data/attachment/forum/202008/19/182618ir5ij3ojggj9djg5.gif" class="zoom" onclick="zoom(this, this.src, 0, 0, 0)" width="640" id="aimg_495794" inpost="1" onmouseover="showMenu({'ctrlid':this.id,'pos':'12'})" /> <div class="tip tip_4 aimg_tip" id="aimg_495794_menu" style="position: absolute; display: none" disautofocus="true"> <div class="xs0"> <p><strong>eee.gif</strong> <em class="xg1">(3.18 MB, downloads: 0)</em></p> <p> <a href="javascript:;" class="loginf" >download attach</a> <a href="javascript:;" onclick="showWindow(this.id, this.getAttribute('url'), 'get', 0);" id="savephoto_495794" url="home.php?mod=spacecp&ac=album&op=saveforumphoto&aid=495794&handlekey=savephoto_495794">save to album</a> </p> <p class="xg1 y">2020-8-19 18:26 上传</p> </div> <div class="tip_horn"></div> </div> </ignore_js_op> #### According to the precautions given by the forum master on using an oscilloscope to test the crystal oscillator, I used the oscilloscope probe X10 to test the sixth pin. The crystal oscillator started to oscillate normally. I succeeded! ! ! <ignore_js_op> <img id="aimg_495795" style="cursor:pointer;" aid="495795" src="static/image/common/none.gif" zoomfile="data/attachment/forum/202008/19/183429w9q7ki7702150l0r.gif" file="data/attachment/forum/202008/19/183429w9q7ki7702150l0r.gif" class="zoom" onclick="zoom(this, this.src, 0, 0, 0)" width="640" id="aimg_495795" inpost="1" onmouseover="showMenu({'ctrlid':this.id,'pos':'12'})" /> <div class="tip tip_4 aimg_tip" id="aimg_495795_menu" style="position: absolute; display: none" disautofocus="true"> <div class="xs0"> <p><strong>rrr.gif</strong> <em class="xg1">(12.45 MB, downloads: 0)</em></p> <p> <a href="javascript:;" class="loginf" >download attach</a> <a href="javascript:;" onclick="showWindow(this.id, this.getAttribute('url'), 'get', 0);" id="savephoto_495795" url="home.php?mod=spacecp&ac=album&op=saveforumphoto&aid=495795&handlekey=savephoto_495795">save to album</a> </p> <p class="xg1 y">2020-8-19 18:34 上传</p> </div> <div class="tip_horn"></div> </div> </ignore_js_op> ## Summary#### 1. The functions of the chip crystal oscillator and the pin crystal oscillator are the same. You don't have to worry too much about this when you encounter a problem; #### 2. It suddenly got better after being placed for a period of time, indicating that the conductive materials such as miscellaneous tin did not short-circuit the pins by accident; #### 3. After rational analysis, the clock in the program was initialized according to 72M, and the external 8M crystal oscillator did not work. The delay became slower, 72/8=9 times slower than normal.

maychang

"Using an oscilloscope to measure the crystal pins, I found that the crystal did not oscillate."

Which pin does the oscilloscope probe touch? Pin 5 or Pin 6 of the microcontroller? If the probe touches Pin 5, it may cause the quartz crystal to stop oscillating. The probe should be in X10 gear and touch Pin 6 of the microcontroller.

maychang

In addition, the oscilloscope shows the quartz crystal oscillating, is the frequency correct?

bqgup

maychang posted on 2020-8-19 17:51 "Using an oscilloscope to measure the crystal oscillator pin, it was found that the crystal oscillator did not oscillate" Which pin does the oscilloscope probe touch? Pin 5 or Pin 6 of the microcontroller...

I used the oscilloscope at X10 to directly measure the crystal pin

bqgup

maychang posted on 2020-8-19 17:53 In addition, the oscilloscope shows the quartz crystal oscillating, is the frequency correct?

The crystal oscillator does not start oscillating.

maychang

bqgup posted on 2020-8-19 17:57 I used an oscilloscope at X10 to directly measure the crystal pins

The crystal has two pins, and the probe should touch the pin that is connected to pin 6 of the microcontroller.

maychang

bqgup posted on 2020-8-19 17:58 The crystal oscillator did not start oscillating when it ran half a beat slower

If the crystal does not oscillate, the microcontroller cannot work at all and no instruction will be executed. So your "running speed is about 10 times slower than normal" is very strange.

maychang

bqgup posted on 2020-8-19 17:58 The crystal oscillator did not start oscillating when it ran half a beat slower

It is best to use an oscilloscope to check whether the LED is oscillating when it is flashing normally, and observe whether the LED flashing frequency changes.

Use an oscilloscope to observe whether the LED is still flashing when the crystal does not oscillate, regardless of the flashing frequency.

bqgup

maychang posted on 2020-8-19 19:26 If the crystal does not oscillate, the microcontroller will not work at all and no instruction will be executed. So your "running speed is slower than normal...

Not surprising, there is also an internal crystal oscillator

chunyang

It is good that the OP makes a careful summary when encountering problems. This is one of the keys to improving technical ability.

A few additional words about the author's summary:

1. The package has nothing to do with the function, but it should be noted that different crystals and crystal oscillators (note that crystals and crystal oscillators are different components) may have differences in other parameters besides frequency, but they are also unrelated to the package.

2. It is not that it will be fixed by leaving it for a period of time. Instead, the contact resistance of the poor welding residues between the pins will be high or low with the movement of the board and temperature changes. When it is low to a certain level, it will affect the operation of the local oscillator circuit.

3. If the problem is determined to be the crystal oscillator circuit, all components in the circuit, PCB routing, welding, etc. should be checked. The crystal oscillator circuit is sensitive to distributed parameters, especially the high-frequency overtone crystal oscillator circuit, which requires special attention.

After summarizing, the OP will probably not go around in circles next time he encounters a similar local oscillator problem.

bqgup

chunyang posted on 2020-8-20 15:46 It is very good for the OP to make a serious summary when encountering problems. This is one of the keys to improving technical ability. Regarding the OP's summary, I would like to add a few words: 1. Enclose...

Thank you for your help, the problem has been solved

joeymm

maychang posted on 2020-8-19 19:26 If the crystal does not oscillate, the microcontroller will not work at all and no instruction will be executed. So your "running speed is slower than normal...

For STM32, if the external crystal oscillator cannot vibrate, it will switch to the internal RC oscillator source. Since the frequency division coefficients behind the two oscillator sources are different, the delay time is different even if they have the same delay.

maychang

joeymm posted on 2020-8-23 17:12 For stm32, if the external crystal oscillator cannot vibrate, it will switch to the internal rc oscillator source. Since the frequency division coefficients behind the two oscillator sources are different, although...

External and internal, does it switch automatically?

LeoMe

You don't even need to measure. When measuring, you have to consider the effect of your probe capacitance on the crystal oscillator. Maybe the crystal oscillator is stabilizing and it stops when you measure. Doesn't the HSERDY flag in RCC look good?

LeoMe

LeoMe posted on 2020-8-24 08:50 You don't even need to measure. When measuring, you have to consider the effect of your probe capacitance on the crystal oscillator. Maybe the crystal oscillator is in a state of stagnation, and it stops when you measure. Look at the HSERDY flag in RCC...

The crystal vibration stops when it does not move. 1. You can turn on the interrupt of the clock safety function. If the clock stops, enter the interrupt and do the corresponding operation. 2. After the clock is initialized after power-on, check whether the HSE and LSE are vibrating. If not, turn on a light and simply and roughly card a while1 to feed the dog, waiting for after-sales service to solve it.