ESP32-S3 USB COM Port disappearing

[Chevelle]

Yes, I know. There have been several posts about this in the past. I believe that I have read them all but the problem still persists.

I have built about 12 PCBs with the ESP32-S3-WROOM device. They are rather simple. They have a USB connector, mostly for programming over GPIO19 and GPIO20s . It uses GPIO18 and GPIO8 for serial communication between it and another board. It does not use the traditional TXD0 and RXD0 pins for UART communication. It also talks to a TFT display via SPI. There are momentary buttons for EN and BOOT, each pulled up to 3.3V through 10k resistors. They also have 1uf capacitors to ground to keep them quiet.

I built about 8 of these boards with no issues. Lately though, I have been having trouble with each one I build. It has nothing to do with the firmware because I never get that far. When the boards are first powered up, I get the expected cycling of the USB COM port with the beeping from Windows. Then it stops, permanently. Reseting the device (EN) does nothing to help and so holding down the BOOT button doesn't have any effect.

One thing I did check is that when both the EN and BOOT buttons are not pressed, the voltage at the EN pin is 3.3v as expected. It is just 2.6v at the BOOT pin. It should be pulled up to 3.3v. It seems that the ESP32-S3 is keeping it at 2.6v. If I measure the voltage at the BOOT pin while pressing just the EN button, the voltage goes to 3.3v. (These are clean DC voltages. I used a scope to see if there was a lot of noise or some signal on the BOOT pin.)

I have tried different port, different cables and different computers with no change.

This is maddening.

Any help is creatly appreciated.

[ahsrabrifat]

Replace the 10kΩ pull-up on BOOT with a 4.7kΩ or 2.2kΩ resistor and see if it helps.

[Chevelle]

This is getting even stranger. I let the latest board just sit. Maybe an hour. I then held the EN and BOOT buttons down while applying the USB connection. The board came up in boot mode! I quickly programmed the board and I thought all was well. I then saw the post about changing the pullups and so I did that to 2.2k. Again, when I powered the board up, no problem. After about 5 minutes, the COM port dropped and I am right back where I started.

FYI, this latest board is populated with only the few components that are required to power up the ESP32, nothing else.

[Chevelle]

This problem is getting more and more serious.

As I said before, there are two boards in the system. They talk to each other over a serial connection. Both boards use the USB CDC method to be programmed (GPIO19, 20). I have been trying to build out more of these boards but to no avail. I have been building them in small sections, not just populating the whole board.

First, I add the USB connector, the associated resistors and capacitors to allow for port detection, the ESD protection, the 3.3v regulator and its capacitors. I then connect a USB cable to check that the 5v in is being regulated down to 3.3v. This is always been OK.

Next, I add the ESP32-S3 along with the EN and BOOT buttons with their pullups and caps.

Without anything else added to the board, I connect it to the PC. As expected, the USB goes into its booting cycle. I use the buttons to put it into boot mode. The COM port IS properly detected. I upload the firmware and all it well. After about 5 minutes, the port is dropped and won't come back.

From then on, when the EN button is pressed, either nothing happens (no beeps from Windows) or Windows beeps but does not identify the port.

I think there are two possible explanations. One is that I have a batch of bad ESP32s. The other is that there is something in Windows that stops trying to detect ports.

This project is virtually stopped until I can get this resolved.

[Sprite]

Any chance you can post a schematic, preferably with the components you didn't place crossed out or something?

[Chevelle]

Here is a schematic. The bits in the yellow are not populated. As you can see, there is hardly anything there.

I built up another one this morning. I came up just fine. I programmed it with a test routine to write to the LittleFS area. As I was about to load the actual program, Windows dropped the COM port and I have not been able to get it back again. FRUSTRATING!

https://www.dropbox.com/scl/fi/vctoauou ... zw0wu&dl=0

[Sprite]

If anything, R7,R8 and C6,C7 are superfluous, but I don't think they should entirely kill your usb connection. You could try removing them and replacing R7 and R8 with jumpers, though. Aside from that, I don't see any glaring issues with your schematic, I'm afraid.

[Chevelle]

Those components have shown up in the literature for sort of best practices. I agree though, probably not necessary. I have tried it without them and the same thing happens.

As I have been doing this more and more, I have been able to resurrect the other boards in the set. When I first power it up, it comes up as it should. Sometimes I am able to get the test program loaded. Either way, after a few minutes, the COM port drops and the EN button does nothing. (Or it seems to do nothing since I can communicate with the ESP32.)

I disconnect the USB cable and wait. When I power it back up, it connects again and if I am able to upload the final code, it works. So far, none of those boards have dropped their COM ports. Very strange!!!

[Franco]

In a similar case, the issue was the power management features of windows that were disabling the USB port? Did you try to reprint the scenario on another PC?