Potential problem with Arduino 1.0 and USBtinyISP

[BillO]

While investigating yet another quirk in Arduino IDE 1.0, I came across the following.

While trying to burn an Arduino Mini 328 (16Mhz) with the new bootloader (presumably optiboot_atmega328-Mini.hex) using a USBtinyIISP, I get the following:

avrdude: verification error, first mismatch at byte 0x7e81
0x93 != 0xe0
avrdude: verification error; content mismatch

This happens repeatedly with several different Arduino Minis, but does not happen with Arduino IDE 0023.

Has anyone else seen this, or would have an opportunity to test on their end?

[BillO]

Wow, it amazes me that this of no concern.

You'd think the Adafruit folks at least might be a bit curious and either verify, deny or ask for more information.

[adafruit_support_bill]

The Mini is not one of the boards we sell. Both boards.txt and the bootloader binary for the mini changed substantially in 1.0. Have you posted this on the Arduino.cc forums?

[BillO]

Yes, I have. That's another matter though.

It has nothing to do with the Mini as such. It has to do with the Arduino IDE not working with USBtinyISP. It should not matter which board causes the problem. It's still a problem for your customers that might be using the USBtinyISP with a mini. Like me.

[BillO]

Okay, just an update on this. It sometimes seems to work. Once in a while everything will verify and the bootloader gets loaded. Most of the time. no luck.

It works more often if done from the command line, but will still fail more often than not. So it seems that it's a problem with avrdude 5.11. I'm trying to get a compiled version of avrdude 5.11.1 to see if it is any better. avrdude 5.4 is flawless and never fails... ever.

[BillO]

Not that anyone here cares, but the problem also exists in avrdude 5.11.1. It affects other programmers too and other code files (not just bootloaders). At this point is seems to be a timing problem and really comes into play when large writes are made to flash. The errors usually begin to occur at 0x7800 and beyond.

Personally, I'm a little disappointed with the support offered here and that USBtinyISP is the last thing I'll buy from Adafruit.

Not that it's my thing to do, but I'm going to try to work with Joerg Wunsch on this as I need my tools to work properly. Apparently adafruit don't give a ghuck. This will be my last post on the subject, so unfortunately no one here will get the benefit of a solution when one is found.

You can feel free to delete/lock/block my account.

[adafruit_support_bill]

Sorry that you feel that way. The Arduino universe is quite large these days and a the 1.0 release affected many things - including breaking virtually every single library out there. We do what we can to keep things current, but it is impossible for us to test, much less diagnose and debug every possible combination out there.

As it turns out, the problem is not caused by, or even unique to the USBTinyISP. Thank you for letting us know that it is a problem with avrdude. We hope that you will continue to share what you learn with the community.

[adafruit]

Not that anyone here cares, but the problem also exists in avrdude 5.11.1. It affects other programmers too and other code files (not just bootloaders). At this point is seems to be a timing problem and really comes into play when large writes are made to flash. The errors usually begin to occur at 0x7800 and beyond.

Personally, I'm a little disappointed with the support offered here and that USBtinyISP is the last thing I'll buy from Adafruit.

Not that it's my thing to do, but I'm going to try to work with Joerg Wunsch on this as I need my tools to work properly. Apparently adafruit don't give a ghuck. This will be my last post on the subject, so unfortunately no one here will get the benefit of a solution when one is found.

You can feel free to delete/lock/block my account.

Hi BillO, apologies if you don't feel we helped you sufficiently, we were trying but could not reproduce this error here and we burn -tens of thousands- of chips using the USBtiny so we were not sure what it could be. Often times these kinds of bugs are user errors (bad wiring, power, clock, etc) and so the problem gets fixed on its own

The USBtiny is a surprisingly 'basic' programmer - it just does what Avrdude tells it to. It sounds like a bug was recently introduced in 5.11.1 which we have not heard about. Thanks for letting us know as it will help identify this bug in the future. We'll also alert the Arduino team