ESP32 Dev Kit WROOM32 doit.am boards don't want to boot without pressing EN button

[valentinAlkan]

I'm having a very weird problem that I know is not related to the code I'm running. I have a bunch of ESP32 Dev kits I'm using for a project, they're made by doit.am, and about half of them refuse to boot into run mode unless I press the EN/FLASH button. They do this regardless of the code, I've checked with several different example sketches and they all exhibit the same issue. I need a fix for this as these will be deployed to a customer very soon. I never noticed this issue because my test modules weren't doing it, but I don't have enough time to get different modules. Furthermore the doit.am boards are a different pinout than pretty much every other ESP32 module, so I'd have to redesign my carrier boards and that's not possible. So if anyone can help me I would greatly appreciate it.

I'll run through what I have seen so far.

• Plug in usb cable
• Put ESP32 into programming mode by pressing BOOT and EN at the same time
• Load any sketch or example code
• ESP32 sits there and does nothing until you press EN/RESET.
• Once reset manually, ESP32 works normally.
• When next powered up, boards sit there and do nothing until you press EN/RESET
• This mode is not programming mode as I still have to press BOOT and EN at the same time to load sketches - otherwise esptool.py times out.

I have noticed a great deal of variation in how the boards act. All were bought at the same time and are presumably the same 'batch'. I have 25 of them, and about half of them have one form or another of this issue. Some of the boards are able to be reset by the USB serial line, others are not and must be put into programming mode manually. Others still work normally as one would expect.

I had them running on a benchtop power supply with an amperage reading, so I could tell something was up when they weren't pulling enough milliamps. When I press EN, the current consumption goes up to normal levels and the chips come alive and do their thing flawlessly. Also, rapidly power cycling them (and I mean very rapidly - less than 100 ms) will break them of their spell when in this low-power state and cause them to boot normally.

On the boards that will automatically load the program via the USB (that is to say, without having to manually put the boards in program mode) , this 'mode' almost appears to be programming mode, but on some of the other boards that require to be manually put into programming mode, I noticed that even when in this post-boot mode, they wouldn't accept a sketch unless put into programming mode manually again.

So I'm stumped. I technically have a workaround - to rapidly powercycle all boards, but that is a very last resort to have to tell the customer to do this.

I suspected the strapping pins were at play here, but I double and triple checked them and tied GPIO0 and GPIO2 to 3.3V and it has no effect whatsoever on this issue.

If anyone has any ideas, I would appreciate them as I'm kind of at my wits end.

[kurtow]

I have the same problem,

I bought three 'ESP32 Dev Kit WROOM32' boards from AZ-delivery.
Two boards are running normally. The last board which I have bought three weeks ago
shows the same behaviour.

[valentinAlkan]

I think the pullup on GPIO0 is the culprit. It looks like GPIO0 is floating when it should be pulled high unless the EN button is pressed. @kurtow, can you try hooking GPIO2 (blue LED pin) to 3.3V and see if you are still getting this issue? That should prevent it from entering boot mode even if GPIO0 is floating. I'm running a test where I hook GPIO2 to 3.3V and cycling the power on the ESP32 and wait for a response from some code that it's running. If after 30 seconds I get no response, I start the test again. I'm automating it, but if you don't mind just doing it a few times manually and seeing if you get any failures, I would appreciate it. I'm seeing about a 50% failure rate if you don't strap GPIO2 to 3.3V, but if you do strap it to 3.3V, it seems to work perfectly.

[valentinAlkan]

Also strap GPIO12 to ground. I still got the occasional failure but after strapping GPIO2 high and GPIO12 low, that seems to have taken care of it. I definitely think its weak pullup/down resistors.

[valentinAlkan]

Been looking into it some more, turns out it may have been related to a power supply that wasn't providing a good 5V. Will do some more testing and then close if that's what it was.

[helmut64]

I have a similar problem, no reset on power-on. Battery and extern power is fine, I reviewed this on the scope. Interesting is when I hold the reset key during power-on and release it, it will not reset. If I press reset after power-on it will do a proper reset.

On my own designed board I have no pullup in IO0 I because the docs says there is an internal pullup. I will try an extern pullup.

[valentinAlkan]

I made an automated test setup whereby I can cut the 5V to the ESP32 by hooking the power through a relay. I have a raspberry pi controlling the relay, and a python script running on the pi that turns the relay on (powering the ESP32), and waits for 60 seconds for a ping from the ESP32. If the pi gets the ping from the ESP32, it logs it as a successful boot, and then cuts the power for 2 seconds, starts it up again and repeats. If it gets no ping in 60 seconds, it counts it as a failed boot.

All the ESP32 does is connect to a wifi network and send a packet to a specific IP (the pi).

I've tested with a couple different power supply setups. On my original 5v supply (https://www.cui.com/product/resource/p78a-0500.pdf), I got about a 40% boot rate with GPIO2 floating. With GPIO2 strapped down, I consistently got a 98% boot rate. When I bypassed by 5V regulator and used a benchtop power supply, I got 100% successful boots over a few thousand cycles, with GPIO2 strapped low and also with it left floating.

I then started running the ESP32 off the raspberry pi's regulated 5v supply and also got a 100% boot rate over 3600 cycles and counting, with GPIO2 floating.

I am sufficiently satisfied that the issue was caused by a noisy power supply, however it seems that these particular dev boards are quite sensitive to power fluctuations.

[dtboy1995]

[esp32 devkitC V4] What do i need to do if i want the power to start automatically, instead of pressing EN

[dtboy1995]

The usb power went off and the program did not start.
But sometimes the program will run after the EN button is pressed
Who knows why, ask for help
Thanks

[helmut64]

Messure via a scope the level and timing of VDD, EN and IO0 during power on. This may explains it.

[dtboy1995]

@helmut64 Thank you for your reply.
Is there any way to start the program without pressing the en button?
I really need help.

[helmut64]

@dtboy1995 do you I recommended, show me the scope picture of the startup and I may can tell you what is wrong.

[ukpiglet]

I'm replying to an old post I know, but may be useful for someone else like me with this problem and finding this thread. When getting the same issue it was resolved by increasing the input voltage from 5v to to 5.3v

[helmut64]

@ukpiglet I fixed the problem by adding a 10k pullup to IO0 (PROG) on my own board, the internal MCU pullup does not work reliable on reset. On EN I have also a 10k Pullup and a 100nF EN to GND. With this combination the power-on reset timing works for me.

[JohnGENZ]

I found I needed to hold EN low while the 5V power supply capacitors charged - a capacitor from EN to GND solved the problem. In my case 47uF - but a smaller capacitor is probably fine.