Posts made by greenleaf

For future viewers of this thread, the API key is the same as the MAC address. You can usually find this on your router config where it shows device info. The MAC is also shown in the serial output when you connect a burner device (or FTDI cable as mentioned above.

My problem was the device was stuck and needed reflashing. Once I did that it worked perfectly using the 12-digit MAC as the API key.

You may safely ignore the 8 digit "API Key" that is shown in M5burner or the command line, this was a red herring.

@felmue you are a god. I didn't want to order a $15 burning tool for my $5 m5stamp so I've been using a cheapo FTDI adapter. Connecting GPIO0 to GND worked like a charm.

When I boot in UIFlow mode this is how the serial output looks. But M5burner and esptool won't work at all...

@ajb2k3

I hooked up a FTDI adapter and am seeing these errors in the serial output. Did I get a bad board?

14:41:58.735 -> Start factory test
14:41:58.735 -> Start IO TEST
14:41:59.066 -> 18 >>>> 1
14:41:59.111 -> error HIGH-->>>OUT:26  IN:18
14:41:59.738 -> 18 >>>> 0
14:41:59.770 -> 21 >>>> 0
14:41:59.815 -> error HIGH-->>>OUT:19  IN:21
14:42:00.434 -> 18 >>>> 0
14:42:00.473 -> 21 >>>> 0

I also noticed every other device in UIFlow uses an 8 character API Key (MAC). The m5stamp is the only one that requires 12. Could this be a bug or unfinished configuration in the app?

As a last resort I tried the actual real MAC address of the device that I got from my router. That failed too, although I can see the device is definitely on the local network.

The docs suggest that the WiFi AP name doubles as your API key. But it's only 8 characters, UIFlow wants 12 characters:

I don't have the little USB adapter so was hoping to just fiddle around in UIflow to get started. Anyone know the secret to finding the API key on these? Did something change about the way it is configured?

I think I probably have a defective unit. I have tried with only a single tab open, latest version of Google Chrome, tried it incognito mode also just to be sure some extension wasn't getting in the way. Despite all efforts it never connects to UIFlow. I know it's on the network because my router is showing it connected and sending/receiving some packets.

I will probably stick with devices that have USB-C on them, this is way too much hassle and I don't want to have to rig up a programmer just to toy around with a prototyping device. Oh well, at least I'm only out $5.

Ok so the green pulsing light means "Normal, connected, ready for UIFlow"?

As you can see I'm not doing anything especially exotic or weird here:

How it looks when I click the Run button in UIFlow:

How it looks when I click the Download button and let the app run on its own:

I reset the unit, went to the captive portal, set my SSID and password and it connects successfully to my network. I can see that it is transmitting data on my router control panel. The light is pulsing green.

But when I try to connect in flow.m5stack.com it fails. I have carefully checked the MAC address, verified it on my router, and made sure I typed it in correctly. I've also tried resetting the device, reloading UIflow but nothing seems to help.

What's the trick to getting one of these m5stack esp32 pico devices to connect to UIFlow?

@rashedtalukder Congrats, hope you are on to somewhere that appreciates your skills!

@ajb2k3 It's super buggy. I tried getting this ATECC608 chip working with Arduino code and had very inconsistent results, kernel panics and crashes. The only way I've seen it run consistently is by running the RTOS code in the AWS edukit tutorials.

Wow, your balancing star is impressive!

I'm just playing around in UIFlow to see what's possible with this device.

I wrote a small program to print out the IMU X, Y, and Z accel values to the serial port.

It seems the readings can be pretty erratic, jumping from near zero to 0.4 or 0.5 even when the device is completely still.

How do you all deal with this? I tried taking the average of the last 5 readings which seems to help a little but it's still kind of jumpy.

Have a look. This is sitting completely flat on the desk:

X: -0.021 Y: -0.011 Z: -0.987
X: 0.421 Y: -0.014 Z: -0.986
X: 0.421 Y: -0.014 Z: -0.987
X: -0.018 Y: -0.016 Z: -0.986
X: -0.022 Y: -0.014 Z: -0.986
X: 0.423 Y: -0.013 Z: -0.987
X: 0.42 Y: -0.015 Z: -0.987
X: 0.417 Y: -0.013 Z: -0.988
X: 0.419 Y: -0.012 Z: -0.984
X: -0.015 Y: -0.011 Z: -0.984
X: -0.02 Y: -0.013 Z: -0.987
X: -0.017 Y: -0.014 Z: -0.987
X: 0.423 Y: -0.014 Z: -0.988
X: -0.019 Y: -0.012 Z: -0.986
X: -0.027 Y: -0.013 Z: -0.988
X: 0.421 Y: -0.013 Z: -0.99
X: 0.423 Y: -0.017 Z: -0.988
X: 0.421 Y: -0.016 Z: -0.988
X: 0.423 Y: -0.011 Z: -0.987
X: 0.42 Y: -0.008 Z: -0.987
X: -0.021 Y: -0.013 Z: -0.986
X: -0.018 Y: -0.014 Z: -0.99
X: -0.022 Y: -0.011 Z: -0.99
X: 0.42 Y: -0.013 Z: -0.984
X: 0.423 Y: -0.012 Z: -0.988
X: -0.023 Y: -0.014 Z: -0.984
X: 0.421 Y: -0.016 Z: -0.985
X: 0.417 Y: -0.015 Z: -0.985

When you copy this block and change the color, it makes pixels white:

When making animations you always want to copy the previous frame. This is inconvenient because every block must be created from scratch.

Has anyone else observed this? When I have a button handler the first time I click it doesn't get detected. But subsequent presses work fine. This happens for short, double, or long clicks consistently.

@felmue I think I found the issue; my deepsleep() call was inside a loop. Moving it outside the loop sorted it out.

This does not work:

But this does:

For some reason you have to hard-code the string of the local file path, this handler won't accept the variable and produces an error:

OSError [Errno 2] ENOENT