search

jomjol / AI-on-the-edge-device

Easy to use device for connecting "old" measuring units (water, power, gas, ...) to the digital world
https://jomjol.github.io/AI-on-the-edge-device-docs/
5.97k stars 648 forks source link

Failed to read existing file #646

Closed cismarine closed 2 years ago

cismarine commented 2 years ago

Hi, Thanks for the amazing tool. Unfortunately after flashing, esp32cam is connecting to network, but getting a message as in subject. Multiple erases and flashing attempts with expresiff software and esptool command line. I can flash different image without issues.

jomjol commented 2 years ago

Looks like a problem with the sd-card - ESP32CAM is very sensitive to this. Try another one (ideally about 4 GB).

cismarine commented 2 years ago

I have tried quite a few. Still same error. It does read the card as It is connecting to network. Ok, It is sorted. Error while copying files to sd card. Sorry for false alarm.

j3sk0 commented 2 years ago

i have the same error how did you solve it?

infa-presales commented 2 years ago

Same here for me. I successfully flashed with a 32GB Intensio SD Card and fetched an initial image, but the lens focus needed to be changed. So I stopped the ESP32, try to correct lens focus and reconnected. The ESP32 get the IP address (I see the DHCP log) but any attempt to reconnect was failing with error Failed to read existing file I also tried a brand new Samsung Evo Plus 32GB but get the same error. I will now check if I have a 4GB SD in my fundus to test with it. However it would also be great which file might be missing when accessing the SD Card

infa-presales commented 2 years ago

Same here for me. I successfully flashed with a 32GB Intensio SD Card and fetched an initial image, but the lens focus needed to be changed. So I stopped the ESP32, try to correct lens focus and reconnected. The ESP32 get the IP address (I see the DHCP log) but any attempt to reconnect was failing with error Failed to read existing file I also tried a brand new Samsung Evo Plus 32GB but get the same error. I will now check if I have a 4GB SD in my fundus to test with it. However it would also be great which file might be missing when accessing the SD Card

Ah, I guess I know what might happen: I decided to re-flash and so I connected the device with cable back to my Linux Server USB3 port which I was using when flashing the device. And I realized that device was working again. So it looks as if the USB port I used to power up the ESP32 onsite just seem not to have enough power !

caco3 commented 2 years ago

That's a bit odd, I have it running on my docking station as well as a battery powered labtop. In both cases it works fine! Note that 32bit might be too large. I use 16GB with different cards and never had an issue.

infa-presales commented 2 years ago

Note that USB ports at PC usually allow to attach hard drives, which even might have a high current / power need if spin on . In the past old externel USB HDD drives had a Y cable. So USB ports are usually quite powerful. However the cheap China USB power plugs might not be good on peak power need, not large enough ("expensive" in cent area) capacitors are used... When flashing the device I also used a PC USB3 port, which have enough Power.

Unfortunately I cannot find the discussion at Github this mail thread was created of. But I have in the moment a quite strange behavior in number recognition, which might be caused by the poor adjustment of wheels and digits. Here my current picture

which shows an error

Neg. Rate - Read: - Raw: 00367.1413 - Pre: 368.0970

as it looks suddenly the lowest value digit is again recognized as 7 and not as 8, so that the former value 368.0970 is larger then the actual when raw string 00367.1413. is converted

Is there a parameter to avoid that? regards     Detlef (aka infa-presales)

PS:  Nevertheless that I a really great peace of code. I was positively surprised, as at University I once  worked in the late 80's on Neuronal Networks (topic: learning behavior of back-propagation networks). The inherent limitation seem never allow such results, but the pure computational power gain we had since then brought NN into a region that this kind of tasks can be done. I  never expected that this will over the years reach a state that NN it can be really used in practice, but obviously I was wrong.

Zitat von CaCO3 @.***>:

That's a bit odd, I have it running on my docking station as well as a battery powered labtop. In both cases it works fine! Note that 32bit might be too large. I use 16GB with different cards and never had an issue.

-- Reply to this email directly or view it on GitHub:

https://github.com/jomjol/AI-on-the-edge-device/issues/646#issuecomment-1275164800 You are receiving this because you commented.

Message ID: @.***>

caco3 commented 2 years ago

Note that USB ports at PC usually allow to attach hard drives, which even might have a high current / power need if spin on . In the past old externel USB HDD drives had a Y cable. So USB ports are usually quite powerful.

Yes, I am aware of this, that's why I mentioned that it successfully runs from a battery powered laptop. In oposite to a desktup, they usually do not allow pulling more than he 500 mA.

If power is too weak, I would expect a Brown Out Reset! If this is not the case on your setup, then it most likely is not a power issue!