Heatermeter death / blown fuses may be manifestation of corrupt eeprom


 
This is an informative post of an observed failure mode and the remediation that brought my heatermeter back.

While changing parameters via webui my heatermeter wigged out. Solid white backlight display with no text. Reboot showed "No Pit Probe" and then quickliy solid white backlight display with no text.

After searching forum posts here and learning what perhaps to check (spi pins in particular due to display involvement) I decided to force re-flash the AVR before breaking out the soldering iron.

Actually I first re-imaged the micro-sd. To no avail. So then:

I chose the local /lib/firmware version and it flashed to no avail - same symptom.

I did notice this was an apparently "older" version so decided to see if a newer firmware existsed that may have some bugs (?) fixed.

I chose the latest (non release) April and it indicated "Blown Fuses" (which the 2017 /lib/firmware didn't show).

But I also noticed a clear-eeprom hex file.

Interesting. No docs. No word. But - hey - I *was* tweaking my Thermocouple calibration in successive increments. Perhaps I corrupted the ee-prom or data cache?

So I flashed that clear-eeprom hex file.

I then, again, chose the latest (non release) April build and it fully flashed and my heatermeter came back!

So if your heatermeter flakes out, particularly during config changes, do try clearign the eeprom before re-flashing the avr firmware. It may be all you need. And also - when making webui changes - I'm thinking it takes a few more seconds for changes to take effect so don't make changes rapid fire.

Mike
 
Hello Mike - I'm seeing this same configuration issue - no display, machine completely wigged out. How exactly do I do clear the eeprom cache? Do I just copy over the previous eeProm file while loading the latest development release dated 4/20/18?

Thanks.

Andre Kvaternik
 
Last edited:
Yes that is exactly what mine did to me Friday night and I am struggling to try and fix it. Thinking it is a bad sad card because I can’t get my card to reload now though. How do we reflash the eprom?
 
Last edited:
Go to config/AVR Firmware and choose "Online Repository" and you should see Reset-eeprom there, you can flash from there to reset (I believe, never actually done it myself)
On the SD card, I have had a micro SD card lock up on me in the past, like getting stuck in write protect mode... I wasn't able to get it freed up no matter what I tried, just got another card and moved on since they are so cheap...
 
Ok, went to Linkmeter tab, AVR Firmware tab and then click on Online Repository button. Then I flashed the 4th tab down reset eeprom. The screen said flashing eeprom and done. My screen went dark and it seemed like it would not reset with the flashing eeprom done on the screen. So I went back to the AVR Firmware tab and selected bundled firmware button and flashed the first button which is hm.hex and my meter reset back to normal and I had to reset up my config tabs(thermocouple and probe types.) Seems like it is working normally also. I did have to replace the SD card and reinstall the stable version for the raspberry 3. Thanks for the help RalphTrimble!
 
Hello All - thanks so much for the collaboration. I followed Corbett's instructions and brought my HM back from the dead, just in time for tomorrow's Thanksgiving smoked turkey! I was sweating it a good bit as I thought this could be the end of the road for my HM if there was an electrical or other problem requiring a "trip back to the shop". Anyone have any idea what caused the eeprom error? My HM was working just fine, had previously powered it up 30 min prior to the crash and it worked fine. Did my not-so-graceful shutdown (pull the plug) contribute to this problem?

Just trying to prevent it from happening again.

Thanks again.
 
Hello All - an update from my last post - when I went to use my HM again I got the same dark screen with square dark characters, but no backlight on the LED. I tried flashing as above, but could not get it to reset. I tried a bunch of stuff, and eventually I got it to revive by these steps 1) Restoring a backup configuration I had saved when I got it working last, flashing reset.eeprom.hex and hm.hex.

I powered it up the next day and had to go thru this same sequence again to revive it, which is a bit much. Any idea how to fix this?
 
Last edited:
I've never been able to reproduce the EEPROM corruption that a few others have experienced so I haven't ever been able to find a cause. I just pull the power on all my HeaterMeters and they should be totally fine with it. They only write to the EEPROM when a setting is changed so it isn't like it is always in flux. I think the most common cause of corruption is unstable power when doing the write so you might want to verify that your 100uF/10V capacitor on the 3.3V line is soldered in well, and the 0.1uF capacitors connected to the AVCC and VCC pins on the ATmega chip are as well (all of those are between 3.3V and GND).

Also, the way that the EEPROM was written to was changed to be the recommended method by Atmel this year sometime in the snapshot firmware, so if you're on the last release version you may not have the updated code. You can also just use "snapshots/trunk/heatermeter.hex" from the online repository if you don't want to overwrite your whole image, because that has the new code as well.
 
I'm a software guy, so I don't know how to validate the capacitor as you suggested. When I booted up my HM today, I got the same error - no lights in the LED display, square characters displaying. There was no configuration loaded when I went in thru the web interface and checked it. I applied the "snapshots/trunk/heatermeter.hex" and it fixed it.

I guess I could do this every time I start up, but is there something I can try to fix this permanently?

Thanks.
 
I can't say what is causing it, but you definitely shouldn't have to re-flash the firmware every time you plug it in. If you can, I'd try swapping the ATmega chip with another one? I'm not sure what to tell you to try on account of it isn't something I've run across.

Do you have a multimeter? To validate what I listed above, everything that I said "is connected to" you set the meter to "continuity" mode (icon usually looks like a speaker beeping) and test with a probe on each thing and make sure it beeps to say they are connected. To check voltage, put one lead on GND and the other on AVCC / VCC / 3.3V and make sure they all say 3.3V. The only thing I can think of is that the voltage is too low or something and the chip is barely functional.
 

 

Back
Top