Custom ESP32-C3 not detecting COM port (crystal problem?)

[ollieelliot]

I have a custom ESP32-C3 board derived from the reference manual, it has all necessary decoupling capacitors and inductors for Vdd. I've measured 3.3V on Vdd going to all power pins on the ESP32. My USB is connected to the ESP32 as well (I use the internal USB/JTAG converter) but my PC won't detect a USB device when I plug it in to my PC. I've started to suspect something is wrong with the crystal so I measured it with a oscilloscope and it is a clipped signal, not a sinus. The frequency is at about 10Khz, but the crystal should be at 40Mhz (10ppm, CL = 8pF). I originally used 12pF capacitors for the crystal and have also tested 10pF and 8.2 pF. D+ and D- are connected to GPIO19 and GPIO18, respectively. I tried changing the crystal but still the same result.

I'm assuming the strapping pins won't matter to just get a COM port detected by my PC.

I have attached my schematic for the ESP32 and oscillioscope measurements for the crystal signal when using 12pF and 10pF capacitors.

Right now I'm clueless as to what the problem could be and what other methods I should try to find the problem and fix it.

All tips would be much appreaciated!

[ESP_Sprite]

C18 is not connected in your schematic (on purpose or otherwise). You do need that as it forms part of the RC network that properly resets the C3. Also, Vdd_spi generally needs a 1uF capacitor to ground. Both things should not stop the crystal from functioning, though; I'm afraid I have no idea what's up with that.

[ollieelliot]

C18 is not connected in your schematic (on purpose or otherwise). You do need that as it forms part of the RC network that properly resets the C3. Also, Vdd_spi generally needs an 100nF capacitor to ground. Both things should not stop the crystal from functioning, though; I'm afraid I have no idea what's up with that.

C18 was excluded due to lack of space, thought I could get away with skipping it. For VDD_spi I coulnd't find any requirements. I did however notice that I have accidentally swapped D+ and D- going into my ESD protection IC in my schematic, that would obviously contribute to not being able to establish a connection with my PC.

[ollieelliot]

Update: I've swapped the USB data lines so that they are correct now. However, the Crystal problem still remains. I get the same signal on the data lines as on the Crystal (shown on the attachments).

[ESP_Sprite]

Are you sure the crystal footprint is correct and the crystal is soldered properly?

[ollieelliot]

Yes I'm sure. However, I haven't tried pulling down GPIO9 to enter download mode after correcting the data lines from USB, I'm guessing that the crystal won't start to oscillate properly if the MCU isn't booting properly? Currently the MCU is trying to boot from memory but there is nothing on my memory since it is new.

[ESP_Sprite]

The thing is that on all current ESP chips, the crystal needs to be running before the CPU will try to boot from anything, so regardless of bootstrap pins or flash contents you should still see activity on the crystal.

[ollieelliot]

The thing is that on all current ESP chips, the crystal needs to be running before the CPU will try to boot from anything, so regardless of bootstrap pins or flash contents you should still see activity on the crystal.

Alright, I think I understand. I've tried choosing the right booting mode now. Also tried adding a 1uF decoupling for the enable pull-up.

I'm not very good at how the MCU and crystal works together, how does the MCU start the Crystal in order for it to start oscillating? I'm starting to wonder if the C3 is damaged, since I've used a heating gun at 400°C to solder the chip. Could it possibly be broken due to too much heat? Maybe I should replace the chip? And solder by hand to minimize damage

[ollieelliot]

I've tried removing excess flux and still nothing. There is about 0.5 V DC on the Crystal, how do I know what DC I should expect? Couldn't find anything in the datasheet. I'm trying to find if there is some way to check if the C3 is working properly together with the crystal.

[ollieelliot]

The thing is that on all current ESP chips, the crystal needs to be running before the CPU will try to boot from anything, so regardless of bootstrap pins or flash contents you should still see activity on the crystal.

Can you see anything from my schematic that would cause problems? The series component for the crystal is excluded in my case, will that have an impact on this?