Weird power supply interference problem

wuwai

Weird power supply interference problem [Copy link]

 

I recently drew a board that uses stm32 to control an H-bridge circuit similar to a motor driver to drive a 125Khz antenna (the solution is copied from an official AMS board). When controlling the 4 IO ports to oscillate at 125Khz, I found that it would seriously affect the power supply voltage. Sometimes it would pull up 3.3V and directly burn out the stm32 and LDO. Sometimes it would pull down 3.3V to reset the microcontroller, and at the same time interfere with the 9V DC input. The following is a comparison between the circuit schematic I drew and the official schematic (you only need to look at the drive circuit in the lower left corner of the picture and the power supply at the top). Compared with the official one, there are the following differences: 1. The official schematic shows two power supply methods, external DC and USB power supply. My circuit does not have USB power supply 2. The official external DC has an anti-reverse diode, but mine does not 3. The actual official input power supply voltage is also 9V, but it is marked as 12V in the schematic 4. The official input DC has two 220uf tantalum capacitors in parallel, but my picture only has one. However, in the actual test, two were connected in parallel, and the C4 100nf in the picture was also connected in parallel. 5. The official microcontroller used is PIC, and I use stm32f103, but the simulated waveform is the same as the official one Figure 1 Schematic diagram drawn by myself Figure 2 Official schematic diagram Schematic diagram explanation (use the official diagram as a reference): 1. U6 and U7 are TC4426ACOA 1.5A gate drive circuits, U4 and U5 are IRF7389 P-channel and N-channel two-in-one MOS tube chips, R7, R8, R9, R10 are 18 ohm 2W resistors, and A1 is a 125Khz antenna [backcolor=rgb(239, 245, 2. Through the 4 IO ports of the single-chip microcomputer, the two gate drive circuits are controlled to control the opening and closing of the MOS tube to realize forward conduction and reverse conduction, so that the middle antenna can oscillate forward and reverse to emit electromagnetic waves. The flip frequency of the IO port is 125K, so it can emit electromagnetic waves with a frequency of 125K. For example: Control PMOS1 and NMOS1 to high, then the N-MOS of U5 is turned on; at the same time, control PMOS2 and NMOS2 to low, then the P-MOS of U4 is turned on, and the antenna emits a forward electromagnetic wave. 249)] Control PMOS1 and NMOS1 to low, then the P-MOS of U5 is turned on; at the same time, control PMOS2 and NMOS2 to high, then the N-MOS of U4 is turned on, and the antenna emits a reverse electromagnetic wave Repeating this process can continuously emit electromagnetic waves The actual control of the IO port of the microcontroller is as follows: Figure 3 IO port control 249)]The voltage of the official board and my own board tested by oscilloscope are as follows: (yellow is 9V, blue is 3.3V. The voltage changes in the figure are caused by controlling the emission. The voltage is normal when not emitting) Figure 4 Official board voltage Figure 5 My own board voltage Why is the 3.3V of my own board disturbed? In this picture of my board, the 3.3V voltage is pulled down to below the normal working voltage of stm32, which will cause stm32 to reset continuously. Sometimes I will measure that it is pulled down but not below the normal working voltage, and sometimes it is pulled up directly and burns stm32 and LDO power chip. The 9V of the official board will also be affected when transmitting, but it will not interfere with the 3.3V voltage

maychang

This kind of interference (or mutual influence) cannot be seen from the electrical schematic diagram. It is related to the circuit board layout, wiring, mechanical structure, shielding, etc.

huaiqiao

"To drive a 125Khz antenna" Is this reflected in your schematic diagram above? Which one is it? "When controlling the 4 IO ports to oscillate at 125Khz, it is found that the power supply voltage will be seriously affected." --> How does this IO oscillate?

gmchen

There are two reasons for this: The first is that the H-bridge driver must have a certain dead time, otherwise it is easy to cause the power supply to directly feed through the two field effect transistors (short circuit). The second is the factors such as the printed circuit board layout mentioned above.

LIQINGYONG

Add an EDS diode at both ends of your motor and see if it's about 9V.

wuwai

huaiqiao posted on 2018-3-20 16:55 "To drive a 125Khz antenna" Is this reflected in your schematic diagram above? Which one is it specifically? "When controlling 4 IO ports to vibrate at 125Khz...

The 125Khz antenna is A1 in the circuit diagram, which is the device that replaces the motor position. The vibration method is shown in Figure 3 in the article, which is to control the four IO ports to realize the positive and negative power supply to the antenna. The example above Figure 3 in the article is more detailed.

huaiqiao

wuwai posted on 2018-3-20 17:32 The 125Khz antenna is A1 in the circuit diagram, which is the device that replaces the motor position. The oscillation method is shown in Figure 3 in the text, which is to control four IO ports to implement...

I usually use LDO for voltage regulation instead of direct power supply. The power supply needs to be protected, such as adding a transient suppression diode like SMDJ36CA. A fuse needs to be added to the incoming power supply. I didn't see what device your U2 is, but it is best to add an ESD device to this device. Then if the speed is high, you can consider using optocoupler isolation. In addition, I noticed a more important point in your circuit. The 3.3V VCC, including the VDD pin, has no decoupling capacitor. I also want to ask, your CPU burned, is it the power supply short circuit to gnd or U2 2pin and 4pin short circuit to ground? Or is it something else? H bridge, although I am not familiar with it, I think it can also be properly protected and preventive measures. In short, to put it bluntly, your circuit construction is a bit like a development board, because the development board only considers the function, not the protection and EMI issues. Don't scold me, the original poster.

wuwai

Let's learn from each other. Maybe you didn't understand my circuit diagram. Figure 1 is the board I drew myself, and Figure 2 is the official evaluation board. The official board is good, but my board will affect 3.3V. The U2 you mentioned should be the U2 in Figure 2. This is the LDO model AS1360. Which 3.3V VCC and VDD don't have decoupling capacitors?

wsdymg

      The PMOS and NMOS signals have the same phase and cycle. The actual switch tube will have a momentary short circuit state when switching between states, which will directly affect the 9V power supply. It can be seen from the waveform that the two interferences are actually synchronized, and your 9V waveform interference is more serious than the official board waveform interference. In addition, when designing circuit decoupling capacitors, decoupling capacitors are usually designed for each 3V3 of the chip. It is recommended to try dead zone processing first.

wuwai

wsdymg posted on 2018-3-20 20:49 The PMOS and NMOS signals have the same phase and cycle. The actual switch tube will have a momentary short circuit state when switching between states, which will directly affect the 9V power supply. ...

I understand what you mean. I will modify the program first and add dead time processing. The waveforms of the PMOS and NMOS output by the official board are indeed slightly out of sync. It may also be to solve this dead zone problem. I will verify it first and come back later to post the four control pin diagrams of the official board first

wuwai

Conclusion: The problem is found, it is the dead zone of the H bridge. Since I control the MOS tubes on each side of the H bridge to flip at the same time, the two MOS tubes on the same chip may be turned on for a moment, causing 9V to be grounded instantly, so 9V will be greatly disturbed. The dynamic suppression ratio parameter of the LDO I use is not good, so it will cause 3.3V power supply fluctuations. Now add dead zone delay when controlling the pin, or use the PWM dead zone control of the stm32 advanced timer.

一株草

It's a layout and wiring problem. Has the ground been processed?

jh2311750663

Suggestion: produce PCB drawing!!!

jh2311750663

The driving part is exactly the same as wireless charging!

博智慧达

jasonxie123

Professional import and export agent

沃菲特电子

Learned