ESP32-S3 troubleshooting -

[pinchy]

Hi guys!

I'm using the ESP32-S3FH4R2 in a new project. This is the variant which has 4MB internal flash and 2MB internal RAM.

Received the new prototype boards and the ESP32 appears "dead" so I've obviously connected something wrong somewhere. Nothing comes up on UART0 (literally nothing, no trace on TXD0 either) and I'm not getting any oscillator on XTAL_P and XTAL_N. All connected VDDs are measured 3.3v. PU is pulled high. Schematic attached. I have a couple of trouble shooting questions:

• With the S3FH4R2 variant, is VDD_SPI required to be connected to an external 3.3V voltage source? Docs are a little unclear here and I've currently got this pin floating
• can SPICLK_N and SPICLK_P be used as GPIO?

Any suggestions on what to do next to debug would be appreciated. It is of course possible that I've defined the chip wrong.

Screenshot 2024-08-13 at 5.17.04 PM.png (276.83 KiB) Viewed 785 times

[ESP_Sprite]

Schematic looks OK. I think VDD_SPI normally has a capacitor (1uF or so) connected to it for decoupling; if you do the next revision, you should include that. It does not need external 3.3V and not having it should not stop your ESP from booting (but you may have issues e.g. with deep sleep later.)

Specifically here, starting the crystal osc is one of the first thing an ESP does. If it does not, it's one of three things: the crystal circuitry is broken (e.g. I've seen people have 10nF load capacitors instead of the 10pF they needed), something is wrong with the power supply (e.g. VDD3P3_RTC not connected, or EPAD is not grounded) or CHIP_PU is not high. I'd suggest to look in those directions.

[pinchy]

wow thanks for the fast response!
I've deployed ESP chips in 1,000's of units now and this is the first time I'm reaching out on the forum, so thank you! (Also my first time with the S3)

Noted on decoupling VDD_SPI, I'll add that to the new revision notes.

• Good call on the crystal decoupling caps - that's the kind of mistake I could see myself making! In our case, I've used 18pF (0402ZA180JAT2A) coupled with a 40MHz crystal (XRCGB40M000F1S1DR0) which has a load capacitance of 12pF. I've assumed a stray capacitance of 3pF. Which should at least trigger something. There's another oscillator on the board with the same characteristics and it works, so I don't think it's that.
• I can confirm CHIP_PU is high on boot. (Although even if it weren't, I'd expect to see an output on UART0 and there's nothing there)
• VDD3P3_RTC (pin 20) is directly connected to 3V3, with a 100n decoupling cap right next to it
• EPAD is grounded HOWEVER it looks like the stencil layer is missing from my component file ,- which means no solder paste would have been applied during manufacture.

I'll lift the chip when I get back into the office tomorrow and check - thank you for pointing me in that direction. Fingers crossed.

[ESP_Sprite]

Yeah, a missing EPAD connection would do it. One correction:

I can confirm CHIP_PU is high on boot. (Although even if it weren't, I'd expect to see an output on UART0 and there's nothing there)

You would not. CHIP_PU is the enable for the entire chip; if it's low the chip will be dead as a doornail: the clock osc won't start, the CPU is kept in reset and no signal would come out.