Fried chip when i connected power the first time.

Preface
A couple years ago i ordered a mendel and a huxley from reprappro, where the plan was to print the parts for the huxley with the mendel. Two weeks later i had the mendel working, but never actually started the huxley build. (i had no problems building the mendel, it is still printing happily) About six months later i get the idea to build the huxley to be a displaypiece, so i print new pieces and finally open the bags to build the huxley. My modifications to make it look good enough to have on display is basically to put all wires in heat shrink tubing, creating a wire harness-ish thing, and cutting some metal/pvc plates to create more of a print chamber. While building and making the harness I checked everything as i went along and took my time, as this is a hobby project there is no rush. It looks great.

Problem
When i came to the comissioning step involving plugging in external power, I didnt actually see the smoke, but i could smell it. No more contact through USB. I put the project on the back shelf.

My working theory is that i screwed up, something is shorted, somewhere, and that fried the chip. To test this theory and get the project rolling i got my hands on two new atmega chips preloaded with marlin a year or so ago. Popped the chip and put a new one in last week, and it looks fine when connected to usb. I have not yet connected external power.
What should be my next steps at this point?

Solution?
I have checked the power cable, the switch wires and the thermo wires to the best of my ability and it all looks good. This leaves the heatbed and the motors.
My (uneducated) plan is to remove all connections to the sanguinolulu, connect external power, and try to boot, followed by the "safe" connections, followed by the motors, and then the heatbed, powercycling for every connection. The offending connection will fry my chip and i will replace it with my last spare, and try to figure out where the short or other error is. This plan is slightly high risk as i may have more than one short, and i am guaranteed to use at least one more atmega to figure it out, among other problems....
Visual inspection of the wires is at best reduced and at worst impossible with the shrink tubes in place, and i do not if at all possible want to pull it apart unless really necessary, as there is a risk of damaging the wires when removing the tubing...


RepRapPro Huxley kit bought ~2 years ago w/19v PSU
Sanguinolulu rev 1.3a

Edited 2 time(s). Last edit at 07/17/2014 06:55AM by ghazz.

Did you check the power adapter itself? And polarity?
I do not see how "the chip" could get fried due to connecting the various loads and sensors, even if they are wrong (well, nor fully messed up)).
It was either a bad power adapter, or wrong polarity, or a fabrication fault.
The later you can eventually investigate, but will not solve the problem, but the first two must be eliminated for sure, to avoid further problems.

Finally, there is an important note for jumpering the 5V power for the logic on Melzi. Did you follow it?

First of all; Thank you for answering.

This is great news. Yes, i have checked the power adapter, it delivers 19v as specified, and i have double checked the polarity. But this means that i have a focus to figure out where the error is. I will start pulling the power lines and rewiring them today.

I have to specify here that i have a huxley with a Sanguinolulu rev 1.3a, as delivered from reprappro shortly before their initial mendel rush. I preordered a mendel, and figured i'd buy a huxley as well

Ok, after testing and rerunning wires today;

The PSU delivers 19.47v, wich seems slightly high to me. The polarity of both the heatbed and the board match the plug.

The resistance over the power inputs on the board are as expected close to 4.7k, but on the heated bed it is in the megaohms. May this be the culprit? I notice the thermal grease step is removed in newer instructions, going to see if i decide to clean it off later today...

Yeah, turns out i used silver based thermal paste in the heated bed... When i measure the resistance over the power of the heatbed with the metal plate surface, it is 5-8 megaohms, without it there is no connection.

I am currently cleaning the heatbed pcb in hopes that this will give me better readings...

Ok, cleaning silver thermal paste is close to impossible.

Can i cover the pcb in kapton, maybe? or should i just bite the bullet and order a new heater?

To clean the paste use acetone or isopropyl alcohol. Then put a patch of Kapton over the thermistor and measure again. Though I don't think this is an issue. The thermistor has one wire to ground and one to an analog input where a 4.7 kohm resistor to +5V is also connected. Given the 100 kohm value of the thermistor, applying 19V through a x Mohm resistor is not something that could blow the chip, but rather affect the temperature indication.
You can test the bed itself very easy. The idea is to verify if the mosfet driver is good or, for any reason, fried. Let the aluminum plate aside, connect the power, then "command" the heater, from 19V through a 10 kohm resistor, on the corresponding control pin on the heated bed. In a very short time the track should warm up. Check if this is happening, and if it is uniform across the entire surface. If it's not working, it might be the mosfet is... gone.
If all is ok, then assemble the sandwich and test again.
By the end of these tests, you'll know if your HB is in good condition or not.