Works for a while, then burns in a nebulous way...

Discussion in 'The Projects Forum' started by Erratum, Aug 25, 2015.

  1. Erratum

    Thread Starter New Member

    Aug 24, 2015
    6
    0
    Hi guys,

    I'm working on a industrial prototype (composed of several modules) where I use arduini and fairly basic electronics to automatize the whole thing.
    But I encounter a big issue on one of them, as it seems to work but always comes to an inopportune arduino burning...

    By the way I can't tell if it's the 5V regulator on the board or the atmega chip itself which burns, as it only produce very little smoke and no odor, but it can't work anymore after that.
    It happens (sometimes) to burn when it comes to power the steppers (Nema23, unipolar). But it also works great when it decides to...

    I upload the typon and two board views, with and without the arduino, to let you see any details that may be the cause !
    Globally : The steppers run until they meet their respective limit switch, the solenoid actuator is powered, and then the steppers run backwards.

    I've been working on this problem for about a month, I burned at least 15 arduini and I changed the design of the board 3 times... it still doesn't work. And it drives me really crazy.
    If anyone can help me on that, (s)he will have my biggest thanks ever..!

    Nicolas
     
  2. bertus

    Administrator

    Apr 5, 2008
    15,645
    2,344
    Hello,

    Can you post a schematic of the circuit, in stead of a board layout?

    Bertus
     
  3. Erratum

    Thread Starter New Member

    Aug 24, 2015
    6
    0
    Hello Bertus,

    Unfortunately I didn't have the time to create the schematics, and I directly drew the board in Eagle...
    Do you know if there is a mean to get the schematics from the board in Eagle ?

    If there isn't, I will make it asap,
    Thanks for your reply :)
     
  4. blocco a spirale

    AAC Fanatic!

    Jun 18, 2008
    1,438
    368
    The obvious method for determining if the 5V regulator is faulty is to test it; is there 5V at the output?

    If you'd had a schematic and applied some logical thought to the process you wouldn't have destroyed so many boards.
     
  5. Erratum

    Thread Starter New Member

    Aug 24, 2015
    6
    0
    Hi,

    I'm building the schematic at the moment, I'll post it as soon as it is finished.
    Yeah, obvious could be the right term to use here, but I didn't tested it because, as one of the two is destroyed, the second burns too. I meant that I couldn't observe which one burn first, sorry if I was not clear.

    About the number of board destroyed, I didn't mentioned several things : I made changes on the board every single time I had to replace the arduino. An other arduino was on the same power circuitry and linked to the first through optocouplers (inputs and outputs mentioned on the board layout) : when the first one burns, the second happens to burn too. So you quickly get 10 or 20 destroyed boards EVEN if you apply some logical thought to the process...
     
  6. Kermit2

    AAC Fanatic!

    Feb 5, 2010
    3,771
    930
    If you are satisfied the mechanical portions of your design are working as planned, replace all LOADS (outputs) from your circuit board with dummy loads and all INPUTS with switches or other signal generators and run it. Hook the loads back up in single steps, one at a time, then run again, and when you burn a chip you will know where to start looking for the problem.
     
    atferrari likes this.
  7. mcgyvr

    AAC Fanatic!

    Oct 15, 2009
    4,769
    969
    Have you checked to ensure you aren't simply pulling more current from the Arduino board than you are supposed to?
     
  8. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,773
    1,103
    Do your steppers and the Arduino share the same supply?
    Do you have a star ground system so that large ground rail currents don't affect the Arduino?
    Do the steppers have diodes and/or transient suppressors to kill back-emf spikes?
     
  9. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    So let me recapitualte, you burned 15 arduinos, redesigned the board three times, and yet you don´t have the time to make a schematic that would ensure everything is wired correctly?
    Sounds like the old “There’s never enough time to do it right, but there is always enough time to do it again”.
     
    JohnInTX, absf, KJ6EAD and 2 others like this.
  10. blocco a spirale

    AAC Fanatic!

    Jun 18, 2008
    1,438
    368
    I totally agree. This could be a long thread.:rolleyes:
     
  11. Erratum

    Thread Starter New Member

    Aug 24, 2015
    6
    0
    Thank you everyone for your wise answers,

    Here is the schematic.

    Kermit2 :
    Thank your for your answer, which confirmed that I did the right tests :)
    I got rid of the second arduino for the latest tests, so it didn't come from that part of the circuitry. I also tested with external supply for inputs, and it burned as well.
    About the loads, I didn't tested with weak loads (tiny steppers..) but with no loads it never burned (although it could be a simple chance as it works most of the time, and suddenly... burned).
    I did isolate every test, replaced all the components each time etc... but the only coincidence was connecting the steppers (which I changed too).

    mcgyvr:
    I use 2 kind of resistors (500R and 10k) so even the 9*500R at the same time would give at 5V, 90mA and the others are negligible... So we are far from the 800mA the regulator is supposed to support...
    Do I make a mistake somewhere ?

    Alec_t :
    They do share the same supply. This is the reason why I put TVS diode to protect the board, but they seem to have no effect...
    I don't know what a star ground system is, I'm gonna make further researches about it !
    I saw other people having issues with shared power and logic circuitry, I think it is the most probable reason, even if I can't see why...

    kubeek :
    Let me recapitulate. The board works perfectly (100 % of its functionalities) during 90% of the time : so the components ARE wired correctly.
    As it makes no difference to read a schematic or a board layout, I usually don't do the first one, and it always work as it should. This case is a first, I've never burned an arduino before.

    My problem is probably due to a bad conception, but I don't see where :/

    Thank you all again for your answers,
    Nicolas
     
  12. djsfantasi

    AAC Fanatic!

    Apr 11, 2010
    2,795
    830
    What I don't understand is how/why another Arduino was used, when the failure wasn't identified? If the TS doesn't know if it's the regulator or ATMega328 (or something else?) which failed, how can he redesign the board and believe it has fixed the problem? If something burnt, one would think it could be identified visually (scorched), thermally ( touch or thermometer) or electrically (as blocco described).
     
  13. bertus

    Administrator

    Apr 5, 2008
    15,645
    2,344
    Hello,

    The used fets have a rather high gate threshhold voltage.
    Better try to use some logic gate version.

    Also there are no decoupling capacitors in the schematic and on the board.

    Bertus
     
  14. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    A board that works 90% of time and burns out 10% of time is still bad, the 90% just tells you that nothing is blatantly wrong and it doesn´t immediately set on fire.
    What are the two outputs 1, 2 and 3 connected to? How close is S1 and S2 to the board, are they running along with power wires? Those seem like the places where voltage spikes could come back to the board and either upset the power supply voltage or destroy the arduino.
    The no decoupling and no bulk capacitance present is also a good point.
     
  15. Erratum

    Thread Starter New Member

    Aug 24, 2015
    6
    0
    Thank you Bertus for your quiet and neat answer.
    Good point about the threshold voltage of 2-4V of the mosfets used, although they should be sufficient at 5V for my application. I'll try to find logic level ones.
    I've never used decoupling capacitors (nor studied), I'll have to check the method for dimensioning these. If I correctly understand their role, they keep steady the power supply of the arduino (mostly absorbing Voltage losses) while heavy loads induce fluctuation on the power supply output ?
    So voltage losses could damage the atmega chip ?

    I totally agree with you Kubeek, 90% is far from sufficient, in fact it's really bad. And that is the exact reason I'm here for ;)
    What really confused me is that in a normal functioning, nothing heats up, and the complete cycle was run.
    When it happened to burn, it was at the very beginning of the routine, never while running.

    Thank you for your advice about S1 and S2, it could really be the key as the steppers are running or not in relation to their state. But I can hardly visualize how they could be the place of spikes, why ? Because of their connection to the ground ? How to get rid of this ?
    They are connected with 24awg wire, about 20cm long.

    Bulk capacitance, one more thing I need to study...

    Thank you very much again, some of you really help :)
    Nicolas
     
  16. bertus

    Administrator

    Apr 5, 2008
    15,645
    2,344
  17. JohnInTX

    Moderator

    Jun 26, 2012
    2,340
    1,022
    You are getting some good help here but I'll add that the way to use Eagle is to draw the schematic using library symbols and packages and THEN click on 'BOARD' and see all of the components with airwires just ready to be routed. As you detail the design, the schematic and board will back-annotate themselves so that they are always consistent. You can run ERC and DRC's on the board to identify a lot of problems. The big payoff is that when you post a circuit, it will be THE circuit that's on the board.

    While Eagle can be a pain to find and/or create the various objects that make up a design, once its done you get the payoff. Shortcuts can be very expensive. I fear that you are fixing to find that out very soon.

    Good luck though!

    EDIT: That said, does your stepper ever get mechanically rotated i.e. its off but the shaft gets turned by other mechanical forces? How are the P6KE27A's holding up? See if this fits:
    I was called in to fix a stepper driver setup where a roller would contact another (printing press) when the stepper was idle. The design had some diodes much like your schematic shows. When the roller was back-spun, the stepper motor generated over 600Vpk-pk! back into the circuit. The 200V diodes got cooked and caused the rest of the circuit to fry.

    Aha! I replaced the diodes with 1KV Shottkys and that stopped the diode failure BUT now, the power generated by the stepper had to be absorbed by the power bus - which also powered the various regulators for the rest of the circuit. Look at yours, same thing. When the stepper was back-spun, those 1KV diodes did their job and routed the HV to the busses where the power could not be consumed by the rest of the circuit. That raised the bus voltage to way above the 35V input limit of the other regulators that were on the bus. Hilarity ensued. To get rid of the power, I used a big, well heatsinked PNP shunt regulator across the busses so that when the back-spun motor powered it, it clipped the voltage to keep the regulators safe and dissipated the rest as heat. No failures in 15 years.

    Even a single cog can cause big spikes. Your diodes are really avalanche devices designed to work by breaking down in the reverse direction. I think they are being forward biased... just looking at it.

    FWIW: I proved out my solution by taking a motor, chucking the shaft in a drill press, anchoring the other end to the drill table with double backed tape and let 'er rip. I don't know if your problem is the same as mine, but it sounds like it.

    Good luck.
     
    Last edited: Aug 25, 2015
    kubeek likes this.
  18. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    If the wires to those swithces are long and run parallel to other wires like the power wires for the steppers, you could be getting induced spikes in them from the large current that runs the steppers.
    You didn´t say where the outputs 1-3 go?
     
  19. takao21203

    Distinguished Member

    Apr 28, 2012
    3,577
    463
    you arent by chance working for the govt. ?

    Just a joke. Seriously, 7805 arent good ICs except for very small currents.

    2. Separate the power supply rails when you drive motors.
    Use switching regulators the modules are cheap and do the job.
    There are high efficiency kinds dont even turn warm.

    You can use 470uH chokes the small kinds have some DC resistance, or higher, like 10mH or so.
    You'd have to try a little which kinds solve your problems.

    Try capacitors across the motor windings to absorb spikes.
     
  20. Erratum

    Thread Starter New Member

    Aug 24, 2015
    6
    0
    Bertus :
    thank you again, I'll have a look at this thread for sure !

    JohnInTX :
    Fortunately, my steppers aren't back spun, so I don't have the problem of gigantic spikes that would burn the diodes.
    Just a question, if you finally placed a PNP shunt regulator to limit under 35V the generated current, were the 1kV shottky diodes still necessary or the first ones would have been sufficient ?
    Thank you anyway for your impressive and detailed answer, I did burn an arduino when I hand back spun the motors, but it was an incident : it isn't supposed to happen in normal operation.
    However the pnp shunt regulator is one more thing I need to get interested in ;)
    I'm a fairly poor skilled user of eagle, I'll take some time to get more insight into ERC and DRC abilities of the software.

    kubeek :
    I misunderstood you first question, sorry for that ! They are not that long as I said, about 20cm long. And their are quite far from the power wires of the steppers, lets say 6 or 7cm, so I suppose that induction is negligible ?
    Yes I forgot that one : output 3 is the ground, while Output 1 and 2 are respectively connected to an optocoupler (with a 500ohms resistor to limit current).

    As everyone answer, I see there are many possible causes to my problem... and I'm afraid not to be skilled enough to solve them all (if I didn't fix the good one I would burn another arduino, and djsfantasi wouldn't understand why ;) )

    So, I thought to an alternative solution (which I doubt is clean and tidy...) but that catch up with your piece of advice Takao21203 : separating the board in two circuitries, one for the arduino and the second for the power components, linked thanks to optocouplers. Each would have their own power supply, of course, and I would also make all the modifications you told me (if possible because I've serious limitations on available space).
    Is that the better attempt I can consider ?

    Thank you all, again !
     
Loading...