Cause of unstable current when STEVAL-IDB007V1 does not process DIO0~DIO3
[Copy link]
During the development of the challenge, I encountered a problem. If the external IO is not processed when BlueNRG-1 enters sleep mode, the current will fluctuate unstably. When it is high, it can reach several milliamperes, and when it is low, it can be around 9uA. Under normal circumstances, the total current of BlueNRG-1 after sleep plus the standby current of the sensor should be below 10uA. This is normal. I have encountered similar problems in other microcontrollers before because the unused I/O of the microcontroller was not processed. This time I began to suspect that it was also a problem with the microcontroller I/O. Later, after analysis, I found that the problem occurred in LSM6DS3 instead of BlueNRG-1. ont=微软雅黑]The testing process was very painful and intermittent. It took me many days to find the cause. In order to find the problem with the STEVAL-IDB007V1 board, I even doubted several times whether I was too old and not suitable for this job. But fortunately, no matter how many detours I took and how much time I spent, I finally found this part of the problem. [align=l eft]During the test, many "hard injuries" were left on STEVAL-IDB007V1, including cutting and punching (because some of the traces on the 4-layer board are on the inner layer, it is necessary to punch to cut the traces) Below is the record I made when I was looking for the cause It took me many days to find the cause, and my brain was in a mess at the time. After sorting it out, some places are still very messy I started by using the following topology to measure the current of the development board When LSM6DS3 and LPS25HB are powered off and BlueNRG-1 is in sleep mode the current can reach several milliamps when it is high and about 9uA when it is low After analysis, it was determined that the four pins DIO0~DIO3 needed to be processed (after finding the problem, the wires that were cut on the board were restored and it was found that DIO1 did not need to be processed. DIO1 was at a high level when not processed). Pulling up or down can stabilize the current The following is the current change when other I/Os are suspended and the specified I/O is grounded In the schematic diagram, we can see that DIO0~DIO3 are mainly connected to LSM6DS3, corresponding to SPI's CS, SCLK, SDI,In addition to DIO0~DIO3, we can see that DIO12 of BlueNRG-1 is also directly connected to LSM6DS3. If the problem lies with LSM6DS3, why doesn't DIO12 need to be processed? Find the power supply route of the sensor by analyzing the Gerbre file Disconnect the sensor's power supply from VBLUE by cutting the trace on the PCB The current of the VBLUE network still changes, the highest is about 600uA, and the lowest is still 9uA Compared with the change of the sensor power supply below, it is better The total current of LPS25HB and LSM6DS3 in PowerDown mode should be about 6.5uA. After cutting off the power supply of the sensor, the idle current is still about 9uA. Why doesn't the current drop? Measure the sensor power supply (the network of the disconnected sensor division is named VBLUE_SENSOR), there will be a voltage of about 2.5V (this value is constantly changing), where does this power supply come from? Measure DIO12 as 0V, measure DIO4 and DIO5 (SCL and SDA of LPS25HB) as VBLUE_SENSOR voltage Measure DIO0, DIO2 and DIO3 as 0V, measure DIO1 as VBLUE_SENSOR voltage Check again and see that the disconnected VBLUE_SNEOSR is still connected to U10 In the schematic diagram, SPI is also connected to STG3692 through a 0 ohm resistor ont] After analysis, this should not be the reason, because there will be a 2.5V voltage when the VBLUE_SENSOR power supply is disconnected The main suspects are DIO1, DIO4 and DIO5. There is no place to provide power for these three I/Os Later, when checking the physical picture, the corresponding resistors were not welded It is certain that the problem has nothing to do with STG3692 Looking at the schematic again, I found a suspicious object PUSH2. I2C2_DATA is connected to VBLUE here through a pull-up resistor. I measured the voltage across R54, one end is 3.3V and the other end is 2.xV Because the official schematic cannot be retrieved by network label, I moved the BlueNRG-1, BlueNRG-2 development kits en.DM00298232.pdf document to retrieve the network label Unfortunately, Figure 29 in the manual:The STEVAL-IDB007V1 buttons and LEDs are provided in the form of pictures and cannot be retrieved by network label, so they are often missed. The next step is to disconnect R54 and VBLUE, and then measure the voltage. After disconnection, both ends of R54 are 2.xV. Doesn't the electricity come through R54? Press and hold the reset button on the development board and measure the voltages of DIO1, DIO4, DIO5 and VBLUE_SENSOR. They all become 0V. Release the button and power on and the voltages return to 2.xV Open the oscilloscope to measure the voltage of DIO4 and DIO5, only to see that the voltage decreases with the increase of current, but I still can't find the reason. Directly power VBLUE1 and pull RESETN to VDD with a current of 600 uA Connect VBLUE_SENSOR to GND, the idle current is about 47uA When DIO05 and DIO04 are short-circuited with GND respectively, the current is 126.7uA. When DIO5 and DIO4 are connected to GND at the same time, the current is 75uA When DIO1 is short-circuited with GND, the current is 249uA Current situation when VBLUE_SENSOR is connected to GND, VBLUE and idle When VBLUE_SENSOR is connected to GND, the current is about 42uA. After waking up and then sleeping, the current does not change. If the current of the sensor is basically obtained through the VBLUE_SENSOR network, the current when VBLUE_SENSOR is grounded is 42uA. When VBLUE_SENSOR is disconnected, the current that the sensor can obtain from the VBLUE_SENSOR network will not exceed 42uA. GPIO1 is connected to GND, the current is about 248uA, the current will increase after waking up and then sleeping When GPIO1 is connected to GND, the current is 248uA. If you do not disconnect GPIO1 from GND and connect BLUE_SENSOR to GND, the current will drop to 42uA When GPIO1 is connected to GND, the voltage of VBLUE_SENSOR is 1.5V. If you connect DIO4 or DIO5 to GND, the voltage of VBLUE_SENSOR is 0.5V Finally, when I used a multimeter to measure DIO11, I accidentally short-circuited it with pin5 of U10 and found that the current jumped to 600uA Current when DIO11 and pin5 of U10 are short-circuited Further testing found that if U10's pin4 (BlueNRG-1's DIO11) is connected to GND, the current is 42uA, and if pin5 is connected to GND, the current remains unchanged at 9uA. If pin4 and pin5 are connected to GND at the same time, the current is 42uA But as long as the connection to GND is disconnected and only pin4 and pin5 are shorted, the current will reach 600uA If VBLUE_SENSOR is connected to GND, the current is 42uA, and then short-circuiting pin4 and pin5 of U10 does not change the current Using a multimeter to measure the current from VBLUE_SENSOR to GND is 43uA, the overall current consumption of VBLUE is 45.75uA, and the current from U10 pin4 to GND is 48uA, the overall current consumption of VBLUE is 43.95 Use a multimeter to test the voltage of pin4 of U10, which is higher than the voltage of VBLUE_SENSOR Short-circuit VBLUE_SENSOR and GND at VBLUE_SENSOR Later I found the description of DIO11 in the BlueNRG-1 data sheet. DIO11 is in uplink mode in sleep mode. The pull-up resistor is 81KhOM corresponding to 40uA Looking at the data sheet of ST2378E, we can see that each of its I/Os is connected to VL through a 9K pull-up resistor YaHei]This explains how the voltage of VBLUE_SENSOR comes from. The current flows from VDD of BlueNRG-1 to DIO11 through the 81K pull-up resistor, and then flows to VBLUE_SENSOR through the 9K pull-up resistor of ST2378E. Although the pull-up resistor of IO11 can be disabled by code, the pull-up function will be restored after entering low power mode. /** Init Structure */ GPIO_StructInit(&GPIO_InitStructure); GPIO_InitStructure.GPIO_Pin = GPIO_Pin_11; GPIO_InitStructure.GPIO_Mode = GPIO_Input; GPIO_InitStructure.GPIO_Pull = DISABLE; GPIO_InitStructure.GPIO_HighPwr = DISABLE; GPIO_Init(&GPIO_InitStructure); After disconnecting U10's VDD, the current drops to normal, with an average current of 752nA, which should be the sleep current of BlueNRG-1 when no sensor is connected Measuring the current of VBLUE and VBLUE_SENSOR at the same time, we can see that the current is mainly consumed by the VBLUE_SENSOR power network
- There are two sensors on the power network of VBLUE_SENSOR, namely LSM6DS3 and LPS25HB. Judging from the configuration of the pins, the current should be consumed by LSM6DS3, because LPS25HB is only connected to the I2C pin In order to verify this judgment, it is necessary to disconnect the power supplies of LSM6DS3 and LPS25HB. From the GERBER file, it can be seen that the power supplies of LSM6DS3 and LPS25HB are connected through the middle GND (green) and PWR (cyan) layers If you want to disconnect, you need to punch a hole at the specified position to destroy the inner layer connection The following is the PCB board after punching The current of VBLUE and VBLUE_SENSOR after removing LSM6DS3. You can see that the current of the two power supply networks has dropped. The problem of unstable current of STEVAL-IDB007V1 when BlueNRG-1 is in sleep state has been found. The problem lies in LSM6DS3. Why does LSM6DS3 have this problem? Because pin2 and pin3 of LSM6DS3 on STEVAL-IDB007V1 are directly connected to ground. I am going to buy a LSM6DS3 development board online for testing, hoping to find the reason. Another question is that when VBLUE_SENSOR is powered by DIO11, the maximum current it can get from DIO11 is about 40uA, but why can the current consumption from VBLUE reach 600uA? There are two sensors on the power network of VBLUE_SENSOR, namely LSM6DS3 and LPS25HB. Judging from the configuration of the pins, the current should be consumed by LSM6DS3, because LPS25HB is only connected to the I2C pin In order to verify this judgment, it is necessary to disconnect the power supplies of LSM6DS3 and LPS25HB. From the GERBER file, it can be seen that the power supplies of LSM6DS3 and LPS25HB are connected through the middle GND (green) and PWR (cyan) layers If you want to disconnect, you need to punch a hole at the specified position to destroy the inner layer connection The following is the PCB board after punching The current of VBLUE and VBLUE_SENSOR after removing LSM6DS3. You can see that the current of the two power supply networks has dropped. The problem of unstable current of STEVAL-IDB007V1 when BlueNRG-1 is in sleep state has been found. The problem lies in LSM6DS3. Why does LSM6DS3 have this problem? Because pin2 and pin3 of LSM6DS3 on STEVAL-IDB007V1 are directly connected to ground. I am going to buy a LSM6DS3 development board online for testing, hoping to find the reason. Another question is that when VBLUE_SENSOR is powered by DIO11, the maximum current it can get from DIO11 is about 40uA, but why can the current consumption from VBLUE reach 600uA? There are two sensors on the power network of VBLUE_SENSOR, namely LSM6DS3 and LPS25HB. Judging from the configuration of the pins, the current should be consumed by LSM6DS3, because LPS25HB is only connected to the I2C pin In order to verify this judgment, it is necessary to disconnect the power supplies of LSM6DS3 and LPS25HB. From the GERBER file, it can be seen that the power supplies of LSM6DS3 and LPS25HB are connected through the middle GND (green) and PWR (cyan) layers If you want to disconnect, you need to punch a hole at the specified position to destroy the inner layer connection The following is the PCB board after punching The current of VBLUE and VBLUE_SENSOR after removing LSM6DS3. You can see that the current of the two power supply networks has dropped. The problem of unstable current of STEVAL-IDB007V1 when BlueNRG-1 is in sleep state has been found. The problem lies in LSM6DS3. Why does LSM6DS3 have this problem? Because pin2 and pin3 of LSM6DS3 on STEVAL-IDB007V1 are directly connected to ground. I am going to buy a LSM6DS3 development board online for testing, hoping to find the reason. Another question is that when VBLUE_SENSOR is powered by DIO11, the maximum current it can get from DIO11 is about 40uA, but why can the current consumption from VBLUE reach 600uA?
| 
提升卡
变色卡
千斤顶
.
京公网安备 11010802033920号
