Arch Linux

kadet

Member

Registered: 2020-02-09

Posts: 2

[workaround] Lot of protocol errors when uploading to arduino nano

When trying to upload sketch into CH340G based (chineesee clone basically) arduino nano I got many errors regarding communication: 

avrdude: stk500_getparm(): (a) protocol error, expect=0x14, resp=0xfc

Most common responses are: 0x80, 0xfc and 0x90. Sometimes it somehow manages to correctly upload sketch into the board, but it's unlikely scenario. For now I've tested few boards and few usb cables - every has the same issue, so HW problem doesn't seem to be an issue here. I've also tested it on multiple devices (and multiple ports including usb2.0) with arch - same problem. Also, it worked fine few months ago. I tested it on void linux - works fine.
Full log from arduino-ide: https://bin.kadet.net/Nk5L

For now i suspect some problem with baud rate setting, but cannot confirm it or deny in any meaningful way. Has someone came across this issue?

Last edited by kadet (2020-02-10 08:09:25)

kadet

Member

Registered: 2020-02-09

Posts: 2

Re: [workaround] Lot of protocol errors when uploading to arduino nano

Okay, there is a problem with linux 5.5, which does not work particularly well with ch34x module. Today I tested it on void linux with 5.4.12 kernel and it worked fine, after upgrading to 5.5.2 same issue as arch. So problem is somewhere in the kernel. However self compiled module from https://github.com/juliagoda/CH341SER works fine - so there is working workaround. Issue seems to be related to https://github.com/torvalds/linux/commi … d#comments and should be fixed in 5.5.3. So, I just got really unlucky to hit spot when it does not work.

vargagab

Member

Registered: 2014-07-09

Posts: 5

Re: [workaround] Lot of protocol errors when uploading to arduino nano

Okay, there is a problem with linux 5.5, which does not work particularly well with ch34x module. Today I tested it on void linux with 5.4.12 kernel and it worked fine, after upgrading to 5.5.2 same issue as arch. So problem is somewhere in the kernel. However self compiled module from https://github.com/juliagoda/CH341SER works fine - so there is working workaround. Issue seems to be related to https://github.com/torvalds/linux/commi … d#comments and should be fixed in 5.5.3. So, I just got really unlucky to hit spot when it does not work.

The 5.5.3 kernel has the same issue. With LTE kernel all OK.