2

We have been developing a product to work outdoors.

During the winter months we have had no problems on our product and now in the summer the following problem came up:

Our product charges in a vehicle and we discovered that when the vehicle is outside and the product is charging the pcb resets. Once cooled down the problem is away.

Now all our microchips on the pcb is rated to work at 80 - 85 degrees celcius.

We took out the pcb and tested it again with a hairdryer. Heated up the product resets and the with more heat the pcb shuts down took it to under an aircon and while cooling down it started to work again.

Our PCB has the following circuits on Main Control Microchip IC which communicates with bluetooth chip. Also we have a MP3 decoder chip STA013 which communicates with an amp. Further more we have a lithium Ion charging circuit for the battery.

As all the components are rated for high ambient temperature we get resets of the pcb at around 50 - 60 degrees which is normal in a vehicle as our daily summer temperatures is around 30-40 degrees celcius.

With the scope we picked up that the MP3 and Microchip IC is normal. But on the MUte pin on the op amp we saw a drop once heated

I hope someone can help with this problem.

JJ Swart
  • 21
  • 1
  • 1
    Redo the soldering – jippie Nov 09 '15 at 17:30
  • 1
    The issue only happens when the device is charging? Did you de-rate all your capacitors for high temperatures? – The Photon Nov 09 '15 at 17:30
  • Components generally heat up above ambient. if the air is 50C the components will be higher. Have you actually measured the board and component temperature or just ambient? The IC may be hotter than you expect. – crasic Nov 09 '15 at 19:46
  • If you are using an Li charging chip could this be disabling or resetting some of the circuitry at a high temp? There is often a thermal shut off limit on such chips to prevent battery charging at high temps. – Nedd Nov 10 '15 at 03:54
  • Showing us a full schematic or at least the circuitry near this suspected MUte pin might be helpful. – Nedd Nov 10 '15 at 03:59
  • Hi Everyone we have made some ground in isolating the problem. The boards are manufactured in a factory. We isolated that when we bring heat to the main PIC32MX270F256D then the problem starts. We then wrote a toggle program for LED from PIC and heated and had no problem. But as soon as we run everything and heat the pic we get the problem. And its not only when charging – JJ Swart Nov 10 '15 at 11:59

1 Answers1

1

As jippie mentioned, reflow everything. Cold solder joints can wreck thermal efficiency.

Beyond that, lithium batteries are prone to fire hazards so they usually have thermal safety cut offs built in.

My assumption is that you are probably having a "thermal resistance" (Rth) issue. While from a distance, your board probably appears safe, up close, individual components are heating up on a much larger scale. Heatsinks and larger ground planes can improve this, as well as, thicker copper pours for the planes.

If you know the power draw of the failing components, you can easily calculate their temp by multiplying the power by the thermal resistance and adding it to the ambient temp. If you look at the attached datasheet, it gives a thermal resistance of 417 C/W. That means that with a 20 degree safety region between ambient and cutoff, you could only run it at 47mW before failure...

http://www.onsemi.com/pub_link/Collateral/MMBT3904LT1-D.PDF

SpaceCowboyMDK
  • 375
  • 1
  • 7
  • I don't think reflow is necessary, I have done thermal testing with devices cycling from below freezing to 80C in the chamber and it is not normal for components to fail if they are operating within temperature norms, this means proper headsinking for critical components. If you have solder joint issues when the ambient temp changes by 40C I would go yell at the manufacturer. I suspect the device runs hotter than OP expects and components are going into thermal shutdown – crasic Nov 09 '15 at 20:01
  • Depends on who soldered it... If OP is dealing with the issue on his initial design/prototype he may have soldered it himself... While I'm sure he's got steady uncaffeinated hands ;-) he's probably nowhere near as efficient as a certified board house. – SpaceCowboyMDK Nov 09 '15 at 20:15
  • Agreed, in that case I would consider a production assembly to be the only reliable way to test environmental stability. Hand soldered LQFP or hand balled BGA (judging by the STA chip) is not a reliable production method. – crasic Nov 09 '15 at 20:20
  • Hi Everyone we have made some ground in isolating the problem. The boards are manufactured in a factory. We isolated that when we bring heat to the main PIC32MX270F256D then the problem starts. We then wrote a toggle program for LED from PIC and heated and had no problem. But as soon as we run everything and heat the pic we get the problem. And its not only when charging. – JJ Swart Nov 10 '15 at 11:54
  • @JJSwart put a thermocouple on it and report the temperature on the case – crasic Nov 11 '15 at 07:10
  • The toggle LED program probably uses far less processing power (less heat) than your normal program... are you sure you've provided sufficient copper on your ground plane to heat sink the IC? Have you considered adding a heatsink on the PIC? – SpaceCowboyMDK Nov 20 '15 at 17:59