MicroPython
Wifi Internal Error on ESP32C3 #10524
Replies: 17 comments 44 replies
-
|
Try this, first part of my connect(): def connect():
sta_if.active(True)
# seems more reliable to start with a fresh connect()
if sta_if.isconnected():
sta_if.disconnect()
print (f'started in the connected state, but now disconnected')
else:
print (f'started in the disconnected state')
utime.sleep(0.1) |
Beta Was this translation helpful? Give feedback.
-
|
Yeah, I have tried that too: |
Beta Was this translation helpful? Give feedback.
-
|
By way of example this eventually connects... mnost of the time, but NOT all of the time. Example: |
Beta Was this translation helpful? Give feedback.
-
|
Seeing the same behavior with Arduino IDE.... |
Beta Was this translation helpful? Give feedback.
-
|
Try catching the error as WiFi Internal error may not be very descriptive: try:
sta_if.connect(config.hotspot, config.password)
except OSError as error:
try:
with open('errors.txt', 'a') as outfile:
outfile.write(str(error) + '\n')
except OSError:
pass |
Beta Was this translation helpful? Give feedback.
-
|
There is another thread just down the page on issues with the C3 |
Beta Was this translation helpful? Give feedback.
-
|
|
Beta Was this translation helpful? Give feedback.
-
|
Just checking this also happens if you do a poweron reset, ie press the EN button. |
Beta Was this translation helpful? Give feedback.
-
|
Yep. I have 6 boards running right now and have two on the bench that I couldn't get going. I am going to wait and see if any of the 6 disconnect and get stuck at this point and hope I just have 2 or 3 bad ones |
Beta Was this translation helpful? Give feedback.
-
|
I went to Adafruit to look up your board. Is there a schematic somewhere?
current pulses during WiFi init can drop the 3V3 enough to cause issues. I noticed several comments about the regulator chips, ie different parts. |
Beta Was this translation helpful? Give feedback.
-
|
Found this for a similar product: |
Beta Was this translation helpful? Give feedback.
-
|
Here is how I test my connect function: from wifi_functions import connect
import utime
while True:
print ('looping')
utime.sleep(1)
connect()Note: I am not doing a disconnect and the connect function: import network
import utime
import config
import machine
sta_if = network.WLAN(network.STA_IF)
def connect():
count = 0
sta_if.active(True)
utime.sleep(0.1) # this could be important, recent posting by robert-hh
# seems more reliable to start with a fresh connect()
if sta_if.isconnected():
sta_if.disconnect()
print (f'started in the connected state, but now disconnected')
else:
print (f'started in the disconnected state')
utime.sleep(0.1) # this could be important
if not sta_if.isconnected():
print (f'connecting to hotspot...')
utime.sleep(0.1)
try:
# sta_if.ifconfig((config.WiFi_device, '255.255.255.0', config.gateway, '8.8.8.8'))
# another recent posting suggest that this should go after the call to sta_if.connect()
sta_if.connect(config.hotspot, config.password)
except OSError as error:
try:
with open('errors.txt', 'a') as outfile:
outfile.write(str(error) + '\n')
except OSError:
pass
while (count < 5):
count += 1
if (sta_if.isconnected()):
count = 0
print (f'network config: {sta_if.ifconfig()}')
break
print (f'. ')
utime.sleep(1)
if (count == 5):
try:
with open('errors.txt', 'a') as outfile:
outfile.write('failed to connect after 5 tries' + '\n')
except Exception as e:
pass
disconnect() # or you could get errors
print (f'machine reset')
utime.sleep(0.1)
machine.reset() # start from scratch
rssi = sta_if.status('rssi')
print (f'RSSI = {rssi} dBm')
utime.sleep(0.1)
if ((rssi < -95) or (rssi == 0)):
print (f'signal level is below -95dBm')
utime.sleep(0.1)
disconnect() # or you could get errors
print (f'machine reset')
utime.sleep(0.1)
machine.reset() # start from scratch
def disconnect():
sta_if.disconnect()
sta_if.active(False) |
Beta Was this translation helpful? Give feedback.
-
|
The PCB layout is here: |
Beta Was this translation helpful? Give feedback.
-
|
Very interesting... your code ran fine first time... and then: ont he next run fine again: Im going to try using your network functions and see how it goes |
Beta Was this translation helpful? Give feedback.
-
|
C8 22uF won't do much for current peaks. |
Beta Was this translation helpful? Give feedback.
-
|
Probably need to put a test so that it stops when actually connected as that program will just keep looping forever. Errors should show-up in errors.txt |
Beta Was this translation helpful? Give feedback.
-
|
So I have a current stubborn board and its basically just not connecting. Is there a way to turn on internal ESP debugging? I'm thinking non-zero values for osdebug, though I dont know what those should be: |
Beta Was this translation helpful? Give feedback.
-
|
Hey folks, Attached is a 4 minute Wireshark capture of all traffic to and from one of the boards running the connect code below: the log: I am seeing a lot of malformed packets sent by the ESP32:
There is some sort of bug int he stack, and I need to figure out who/how to fix. Attached is a pcap (I am using a test network so passwords in this stuff dont matter): ESP32-C3-Micropython-Capture.pcap.zip The ESP and the access point are about 7 feet apart. |
Beta Was this translation helpful? Give feedback.
-
|
Here is an updated video showing the unit flashed, and how the basic out-the-box code does not allow it to connect: |
Beta Was this translation helpful? Give feedback.
-
|
Try taking out: sta_if.ifconfig((secrets["IPAddress"],secrets["SubnetMask"],secrets["GatewayIP"],secrets["DNSIP"]))There has also been a recent comment that maybe this should after your connect(). Both seem to work for me. And no errors printed out, maybe try: except OSError as error:
print(f'{error}') |
Beta Was this translation helpful? Give feedback.
-
|
Yep, been there and done that. The reason I have this hardcoded if I wanted to take DHCP out of the equation. |
Beta Was this translation helpful? Give feedback.
-
|
Works like a dream on an RPi board.. :( |
Beta Was this translation helpful? Give feedback.
-
|
Took 5 attempts. Try it a few more times, I usually see one attempt. |
Beta Was this translation helpful? Give feedback.
-
|
I have left it running for hours |
Beta Was this translation helpful? Give feedback.
-
|
Took 5 attempts on a RPi board?
On a RPi board and without any problems? |
Beta Was this translation helpful? Give feedback.
-
|
Hey Jeremy, I have the same problem with Adafruis QT PY ESP32C3 as you described. It seems to be related to strength and quality of wi-fi connection. In my case OSError: Wifi Internal Error occurs every time I move the board more than 2 meters away from my router. Then the script always fails. Other devices, like phone and PC, have no issues with wi-fi connection. Also, using the same code, there are absolutely no such problems with Pico W, regardless from the distance from the router (within a rather small flat). Simple code snippet used: I don't really know what range should I expect, but for now it's a disappointment. Maybe it's due to power source (USB from PC), some wi-fi configuration? I haven't dived into the subject yet. |
Beta Was this translation helpful? Give feedback.
-
|
Yeah, I have returned 13 of these and switched to RPis. Pretty silly that there is no antenna on the QT |
Beta Was this translation helpful? Give feedback.
-
|
Went to their site and looking at the product information I see a white block beside that 4 pin I2C connector. That is most likely a chip antenna. Whether or not it is connected or matched properly to the ESP32 RF interface is unknown. The older RaspberryPi 3B had a chip antenna on them, so I would expect similar range. I never did range testing with the ones I had. |
Beta Was this translation helpful? Give feedback.
-
|
Try to add : ` |
Beta Was this translation helpful? Give feedback.
-
|
I noticed during tests with the esp_hosted solution that the STA does not connect if the ifconfig() is called before connect(). But the traffic between the MCU board and the WiFi board indicates that there is communication between the MCU board and the AP. For now, I set the IP address after connect. |
Beta Was this translation helpful? Give feedback.
-
|
This solution worked from the first attempt on ESP32-c |
Beta Was this translation helpful? Give feedback.
-
|
Recently I have been having problems with WiFi on the esp32-xx. One way to improve connectivity is to disable power management. import network
wlan = network.WLAN(network.STA_IF)
wlan.active(True)
wlan.config(pm=wlan.PM_NONE) # disable power managementThe default after boot is 1, setting it to 0 will disable power management. |
Beta Was this translation helpful? Give feedback.
-
|
What version of esp's idf are you using? I started running into a very similar exception ( |
Beta Was this translation helpful? Give feedback.
-
|
I ended up here because I have the same problem (OSError: Wifi Internal Error). I don't have a solution or specific question at this point but maybe this info could help. Board: Xiao ESP32-S3 Sense with camera I downloaded this firmware linked on this page: https://wiki.seeedstudio.com/XIAO_ESP32S3_Micropython/ So I visited: https://github.com/shariltumin
esptool gives me this message when trying to flash the firmware: Abou the wifi problem, it seems hard to reproduce / irregular. But from my experience it happens after some/several socket errors or camera capture errors. Sorry if this doesn't add anything useful, just sharing |
Beta Was this translation helpful? Give feedback.
-
|
Please find the newest firmware here |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the reply. At first I kept getting this error when flashing: I downloaded the bin file by right clicking, save as. Also I had the wifi error again once or twice. |
Beta Was this translation helpful? Give feedback.
-
|
Any news here? |
Beta Was this translation helpful? Give feedback.
-
|
From the earlier comments it looks like a hardware issue (power drops on that board) and/or a bug in the IDF. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks, but that is a bit above my skill level. |
Beta Was this translation helpful? Give feedback.
-
|
I have same problem, the problem is fixed if the USB not connected. I suspect the problem is due to interference from USB3 as the ping TTL is fairly stable around 1ms, if I plug in the board I get packet loss and TTL all over the place (>100ms). Sometimes it completely drops out. Update: It is very inconsistent.
All of the above tests had intermittent results (connect), sometimes the thing would refuse to connect no matter what. UPDATE2: Somehow different version of IDFs can mask the issue. |
Beta Was this translation helpful? Give feedback.
-
|
I have soldered a piece of wire to one of the terminals (the output end) on the ceramic antenna and it made the problem worse. Interestingly enough before the antenna mod I had RSSI of -56 with the device under 1m away from the AP. Now with the antenna mod I have RSSI of -20. Dropping TX power to 5 fixes the issue when anatenna mod is in place. This points to a mismatched RF section. Next step is to desolder the ceramic antenna and testing it only with the wire antenna. |
Beta Was this translation helpful? Give feedback.
-
|
Unless that "piece of wire" is a correct load (50 Ohms? at 2.45GHz) for the WiFi module you might damage the WiFi module. On re-reading I see you mention 61mm. If that is for a 1/4 wavelength ground plane antenna you will be missing the ground plane. If you suspect WiFi RF energy is affecting the USB I would find the 50 Ohm point on the WiFi module output matching connect a 50 Ohm coax cable terminated in a 50 Ohm dummy load say 1-2 meters away from the dev board. Back in 8 hours to see how you are doing. |
Beta Was this translation helpful? Give feedback.
-
The problem is not USB is being affected but WiFi is being affected by having USB plugged in. So the dummy load test is not relevant. I do not have required RF equipment to test, so I am resorting to trial and error. If I blow the RF section, it is fine, in current state these modules are junk, and I have whole bunch of them to go through (idiot me bought more without testing the wifi). Of course that piece of wire is not 50 Ohm, but I am running with assumption that the whole CLC+Ceramic antenna setup is not 50 Ohms either. The USB/no USB affected the badly tuned system just enough to push it over the threshold of not connecting. The fact that RSSI dropped from -56 to -20 with a mismatched 61mm long wire, suggest that the RF section is junk. There is no can on the SoC either so that can also affect a lot of thing. I will scour my junk pile of old Wifi routers and see if I can an usable "50Ohm" antenna off them. EDIT: Here is a relevant doc on the RF tuning for C3 : https://docs.espressif.com/projects/esp-hardware-design-guidelines/en/latest/esp32c3/schematic-checklist.html#fig-rf-tuning Looking at the actual board there is only 1 capacitor in series with the the ceramic antenna, as well as the antenna is floating (other end is not grounded). There is nothing else coming from pin 1 on the SoC. The whole thing seems to be botched up and amazingly it works at all (although terribly). EDIT2: I did further inspection on the boards I have and all of them are missing CLC filter and instead the L part is replaced with a single 9.6nF capacitor and two other caps are left unpopulated. The value of the capacitor is confirmed by measuring in circuit with LCR so might not be 9.6nF (it measured 9.670nF by LCR meter DE-5000, so fairly accurate).
This doesn't match the schematic found on the internet when searched for ESP32-C3 Supermini. I am tempted to restore the CLC filter with reference component values form the Espressif documentation and see if there is any improvement. I suspect the accountants had a go at the board and further reduced the cost by saving 10c on the components. I will give my thoughts to the seller about the situation. At least I have only bought 7 of the things, given the cost of these things not a great loss, except of the huge waste of time. Given the issue with the RF and not the power section, it is beyond my skill/experience/tooling. I have no means to confirm the center frequency, Q factor, or even impedance of the RF section. The hardest part of sourcing/building a 2.4-3.0nH inductor, given my LCR meter has resolution of around 0.001μH. Experimenting with a piece of wire it seems that inductor is pretty much a 3-4mm long jumper, so I suspect a 0Ohm resistor would the job. |
Beta Was this translation helpful? Give feedback.
-
|
First off: 35 Ohms to 50 Ohms go like this: Another way to make a 2.7nH is to use a defined length of PCB track. 50 Ohm track about 8mm long gives you 2.7nH. Without the proper gear it is a very difficult task. |
Beta Was this translation helpful? Give feedback.
-
|
Looking at the antenna sheet. 20dB return loss with just a 3nH (XL at 2.45GHz is 46 Ohms) shunt inductor tell me that they think the chip antenna is 50//-j46. So, as well as the Pi- matching as above you also need 3nH shunt inductor at the chip antenna. There are other ways of matching the chip antenna to 35 Ohms, give me some more time. |
Beta Was this translation helpful? Give feedback.
-
|
Doing a parallel to series conversion the chip antenna is 30-j30, so you would get a pretty good match with only a +j30 inductor (1.9nH) between the ESP32 RF output and the chip antenna. Good luck! |
Beta Was this translation helpful? Give feedback.
-
|
Awesome! I will try and source the 1.9nH in that form factor. I am located in a far corner of the world so I will have to wait a week or two for the inductors to show up. I certainly don't have tools to measure that low, so making one by guessing the length of a wire is not going to be fun. |
Beta Was this translation helpful? Give feedback.
-
|
I think that there will be bigger strays then getting the inductor wrong by 50-100%. The difference between "slapping" the chip antenna directly on the ESP32 output and doing a proper matching job is maybe worth a few dB. You have to many big variables in your environment to tell if things are better by a few dB. |
Beta Was this translation helpful? Give feedback.
-
|
Yet another update. I have managed to salvage a 3nH inductor (maybe the value is not exactly 3nH) in correct form factor from really old Linksys ADSL router (funnily enough labelled as C5 on the silkscreen). After that modification I get rssi around -30 (-50 or worse in the original configuration), and connection on first try. Unfortunately the ICMP is still over the place, but changing txpower now has no effect. I might order bunch of inductors (2.2nH, 3.6nH and 3.9nH) and experiment. This is a somewhat success, the mod improved connectivity but reduced signal quality (perhaps the inductor is not of the right value). |
Beta Was this translation helpful? Give feedback.
-
|
It is brand specific, I use Mikrotik gear, so the command to get the stats is: It also available in the web UI. |
Beta Was this translation helpful? Give feedback.
-
|
OK, I have done IPERF testing on digital radio systems, but no expert. |
Beta Was this translation helpful? Give feedback.
-
|
I noticed something peculiar about the specification of the antenna vs how it is implemented on the super mini: Interestingly enough, there is more separation on almost exactly same boards on the internet: BTW moving the parallel shunt to series has worse results, but still connects instantly. The original setup is just flawed in many ways. |
Beta Was this translation helpful? Give feedback.
-
|
That 3.8mm without ground plane beside it or under is not 50 Ohm track. I am guessing but if you drop a big cap or 0 Ohm resistor on the first component pads you will end up with a 2-4nH inductance between the chip antenna and the green ground plane. The chip antenna is very close to the ground plane on the 1st board image. Is there ground on the bottom-side under the chip antenna? There might even be internal ground layers which will not be visible. |
Beta Was this translation helpful? Give feedback.
-
|
There is no ground plane below it it ends at the via stitching. To improve the current design I have tried so far:
I am going to move the antenna 90 degrees on the plane of the board (stick out), and orthogonal to it (chimney). I have following options:
|
Beta Was this translation helpful? Give feedback.
-
I have a bunch of Adafruis QT PY ESP32C3 boards flashed with the latest 1.19.1 code and I am having trouble with them consistently connecting to WiFi:
How can I consistently get connected as opposed to getting these internal errors. (Which on some boards happens for a few attempts if I try: except: and then connects)
returns:
Beta Was this translation helpful? Give feedback.
All reactions