Help!! MCU dead after changing fuse bits!

[apakhira]

Hi
I had been using an Atmega32L in internal RC oscillator mode. The fuses were:
Lfuse: 0xE3
Hfuse: 0xD9

Now i needed to use a crystal because i needed it for UART. I changed the fusebits to:
Lfuse: 0xEB
Hfuse: 0xC9

And bam! The MCU stops working! The USBasp pogrammer can't recognize it now! Help! What should i do? I have connected the crystal between the Xtal1 and Xtal2 pins and used 22pF caps. One thing i think i need to mention is due to shortage of space in the board i had to solder the crystal at a different place in the board(away from the MCU). The crystal is connected to the MCU through short wires.

 

[Sceadwian]

Do you have a function generator? Feeding a square wave into the XTAL2 pin will allow you to clock it long enough to reprogram it just make sure the voltage isn't too high, you'll need to determine what's wrong with your crystal setup though. How far away from the chip is the xtal, and where are the caps? A quick picture might help.

If you don't have a function generator just grab a 555, or make a quick oscilator using an inverting schmit trigger. Xtal1 is the output from the AVR's clock amplifier and Xtal2 is the input.

 

[apakhira]

Here's the pic. Now, are the fuse settings correct? Are they posing a problem? The datasheet says that for >=1Mhz crystals the CKSEL 3:1 should be 101, 110 or 111. I didn't knw which one to use and one randomly.

 

[ericgibbs]

Here's the pic. Now, are the fuse settings correct? Are they posing a problem? The datasheet says that for >=1Mhz crystals the CKSEL 3:1 should be 101, 110 or 111. I didn't knw which one to use and one randomly.

hi,
1. That crystal & caps should be close to the PIC pins.
2. I dont see any power rail decoupling caps on the pcb.
3. The LED's appear to be connected directly to the PIC without series resistors.

Please post your FULL code and PIC type so that we can advise.

 

[Sceadwian]

Why are all those wires going over the top of the board, typically when you do perf board like that the components are on one side and the wiring is on the other, except for perhaps a few jumpers.

 

[ericgibbs]

Why are all those wires going over the top of the board, typically when you do perf board like that the components are on one side and the wiring is on the other, except for perhaps a few jumpers.

How would he solder wires to the copper side.?

They way that he has the wiring on the component side is perfectly OK for that pcb.

The wiring has not been routed neatly and looks like a birds nest, he should relocate the components in order to minimise the connecting wire lengths.

 

[Sceadwian]

I don't see a copper side... Perhaps I missed the image? And generaly you solder wires on the copper side of a board to the copper traces....

 

[ericgibbs]

I don't see a copper side... Perhaps I missed the image? And generaly you solder wires on the copper side of a board to the copper traces....

The copper tracks of that perf pcb are on the other side as shown in his photo.

Its not generally that you solder wires on the copper side of the pcb, I dont know where you got that idea from.??

Consider that you are using insulated stranded wire, you couldnt easily solder to holes in the copper tracks, unless you left enough bare/tinned copper wires to apply solder.

The board would look a mess and be prone to soldering shorts.

 

[mvs sarma]

one can always wire on solder side also, though little laborious. It would look neat
many people do that type of wiring with thinner wire. The type of board used is like bred board pattern on copper clad (solderable bredboard). this is of late being made in India.

 

[ericgibbs]

one can always wire on solder side also, though little laborious. It would look neat
many people do that type of wiring with thinner wire. The type of board used is like bred board pattern on copper clad (solderable bredboard). this is of late being made in India.

hi sarma,
The OP is using what looks like stranded insulated wire, NOT solid Kynar wire which is used for wire wrap and can be used on the copper side.
If you have a complex pcb with wires crossing each other on the copperside, you quickly get to a point where its a problem soldering in new wires.

My advice would be, where possible trace the wires neatly on the component side and use solid Kynar wire

 

[Sceadwian]

Sorry eric, I wasn't thinking when I posted that apparently, as I have two circuits sitting on my desk I wired the same way <cough>
I spent 20 dollars at the store today on my way home, and forgot the one thing I went to the store for =\ I've had an off last few days.

 

[ericgibbs]

Sorry eric, I wasn't thinking when I posted that apparently, as I have two circuits sitting on my desk I wired the same way <cough>
I spent 20 dollars at the store today on my way home, and forgot the one thing I went to the store for =\ I've had an off last few days.

hi,
Sorry to hear you are not 100%, hope you get well soon.

 

[mvs sarma]

Hi
I had been using an Atmega32L in internal RC oscillator mode. The fuses were:
Lfuse: 0xE3
Hfuse: 0xD9

Now i needed to use a crystal because i needed it for UART. I changed the fusebits to:
Lfuse: 0xEB
Hfuse: 0xC9

And bam! The MCU stops working! The USBasp pogrammer can't recognize it now! Help! What should i do? I have connected the crystal between the Xtal1 and Xtal2 pins and used 22pF caps. One thing i think i need to mention is due to shortage of space in the board i had to solder the crystal at a different place in the board(away from the MCU). The crystal is connected to the MCU through short wires.

you can restore it be erasing and loading the fuse bits back using a regular parallel programmer that works using Vpp. please read the datasheet.
no point in worrying.

 

[apakhira]

OK guys...sorry for the extremely late reply. Been out of town. Turns out it was just a case of dry solder!!
@ericgibbs: The power decoupling cps haven't been shown in thepic. The LEDs have 330 ohm resistors soldered, on the copper side(lack of space on this side). And I have used solid (single strand) wire.
@mvs sarma: parallel prorammer will work despite a screwed up clock source? Why? What about serial?

 

[mvs sarma]

OK guys...sorry for the extremely late reply. Been out of town. Turns out it was just a case of dry solder!!
@ericgibbs: The power decoupling cps haven't been shown in thepic. The LEDs have 330 ohm resistors soldered, on the copper side(lack of space on this side). And I have used solid (single strand) wire.
@mvs sarma: parallel prorammer will work despite a screwed up clock source? Why? What about serial?

Nice to know that it is resolved.
In case you need to reset the fuses , it is possible only with parallel programming and with Vpp enabled. please see the datasheet.

 

[apakhira]

Ok. Thanks...