Electronics > Projects, Designs, and Technical Stuff

Surface Mount Reliabily Questions

(1/5) > >>

mribble:
I have two questions related to issues I'm having with an open source project I'm working on.  The project is the next version of the Camera Axe for those interested.  www.CameraAxe.com

Images:
Bad Oscilloscope Image: http://www.glacialwanderer.com/_blog/blog2011/10_Oct/b_bad.jpg
Good Oscilloscope Image http://www.glacialwanderer.com/_blog/blog2011/10_Oct/v_good.jpg
Schematic: http://www.glacialwanderer.com/_blog/blog2011/10_Oct/ca_schematic.png
PCB Board Layout: http://www.glacialwanderer.com/_blog/blog2011/10_Oct/ca_layout.png

1) I have 4 test units and one of them doesn't startup correctly.  If it has been sitting for more than 3 seconds I need to turn it on, then turn it off, and then turn it on again.  If I just turn it on the micro doesn't initialize correctly.  Thinking about this some I realize I probably should have the on/off switch tied to the enable pin on the main boost converter instead of powering up the whole boost converter, but I generally like my on/off switch to be a true on/off so there is no leakage.  The two oscilloscope images above are the output voltages from a good unit and the bad unit.  All the good units looked like the good graph and I only had one bad unit to test.  The bad unit shows the same graph whether turn it turns on correctly or doesn't boot the micro.  Comments on this?

2) When I mount these boards to the case the operations becomes flaky.  I am using metal standoffs that I put on the PCB and then use more screws to put those on the case.  If I don't tighten the screws to the case as much the problem isn't as bad.  One of the boards I did this with actually failed.  I looked into it and it was the SON-10 package on my boost converter (http://www.mouser.com/ProductDetail/Texas-Instruments/TPS61200DRCR/?qs=sGAEpiMZZMtY9G8Xaw%2fcnvufCtuHXcIe).  It formed a short between some of the pins on this package.  I reflowed the solder with a hot air solderer and the problem went away.  My guess is that the problem is causes by board flex?  I had this boards assembled by the company who will be assembling the final product so it makes me nervous if just tightening some screws causes reliability issues.  Is this a common problem with this package type?  Is there things I can do to lessen the issue?  Perhaps some kind of flexible washers or some board change?

Thanks for any help!

amspire:
Just had a quick look.

Don't really like the way the reset pin is connected to the FS232's DTR via a capacitor. What happens at start up?

I am assuming DTR is normally high, so it is going from low to high at start up at the same time the AVR is starting.

Can you put one channel of the oscilloscope on the 5V and the other on the AVR reset pin, and capture a startup?

Look for any reset pulses shorter then the specified minimum AVR reset pulse length, or the reset pin going higher then the 5V rail by more then 0.5V. Or lower then the 0v rail by more then -0.5V. 0.6V is too much. Don't know the particular AVR chip you are using, but minumum low pulse on the reset line should be something like 2uSecs or more, so anything shorter may be a problem.

If the DTR looks at all ugly, you could remove the capacitor to DTR and see how it goes.

Richard

mribble:
Richard, you were correct!  Here is the graphs of the 5V bus and the reset pin.  http://www.glacialwanderer.com/_blog/blog2011/10_Oct/IMG_2558.jpg

When the micro fails to start correctly it has the dip in the reset pin.  When it works correctly there is no drop and the reset and 5V lines match.  I removed that C1 capacitor and it boots fine every time.  However, I need that capacitor since it is how the chip is reset so a new program can be loaded onto the micro (using an ATMega368 with the Arduino bootloader).  Any ideas how to fix this without removing that capacitor?

I'm also still looking for ideas on #2.

Thanks!

amspire:
You could try adding a 0.01 uF cap from reset to ground, and replace the DTR cap with a diode to allow the DTR to pull Reset low.

You may have to add a 100 ohm resistor in series with the diode if discharging the cap is a problem for the 232 chip.

Richard.

mribble:
Richard, I don't think the diode will work.  DTR gets pulled low while the program is loading so if I put a diode there the mirco will keep getting reset which isn't what I want.  I want a single reset when the pin is first pulled low.

I did some looking around and it seems like there are a lot of projects that use a cap like I did to reset the micro such as all the arduino boards I looked at.  I don't understand why it would be causing me a problem.

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version