1
\$\begingroup\$

I am developing a board that has a couple of small SMD chips. I have successfully made a couple of working boards.

Unfortunately I found out that one of the chips in my original design is going to be phased out soon (LSM330DLC), so I'm replacing it with manufacturer's suggested part (LSM6DS3).

I ordered a test batch with this new chip and it has a couple of problems.

  1. On all 4 of the boards that I ordered the LSM6DS3 chip is somewhat more shiny than other chips (see attached image)
  2. 2 of my 4 boards have electrical short somewhere that I am not able to pinpoint visually. I tried removing the LSM6DS3 from one of the shorted boards and suddenly there is no more short. It is worth noting, that in the process of removing the chip I also accidentally removed some of the solder pads.
  3. The weirdest thing is the data that I receive from the motion chip (LSM6DS3). I couldn't get anything from the shorted boards. Board number 3. outputs constant max value on all accelerometer axes. Board number 4 outputs good data only on accel Y and Z axes - the X axis always shows a value near 0 no matter how i turn the sensor chip. I tried soldering leads to the hand removed part from one of the shorted boards and wired it by jumpers back to the board that I removed it from and suddenly it also started outputting good data on all 3 axes and the short was still gone.

I have tried using leaded evaluation board to develop my code and it works fine with it (I got the good data reference from it). I also triple checked my design and everything seems to be OK. I still kept the old chip (LSM330DLC) on the I2C line and on all of the 4 boards it works flawlessly.

I am now suspecting manufacturer's fault. Either they managed to burn the chips during the reflowing or they somehow used some quick test-run lower quality setting on PCB production. I'm just guessing.

So my question is: Are there any telltale signs, that a chip might be burnt or damaged during reflow? Would the shininess of chips be any good way to tell whether the chip is burnt? Any other possible problem causes?

Board image My board with LSM6DS3 Eval Board. Notice the different chip color

\$\endgroup\$
  • \$\begingroup\$ It is vital that you never blindly accept recommendations from a supplier for a replacement part - you have to do your homework and go thru everything in the two data sheets that might be relevant. It's painful to do but less painful than the mess you can find yourself in. \$\endgroup\$ – Andy aka Dec 10 '15 at 14:15
  • \$\begingroup\$ I'd guess the shininess is caused by flux residue (you could try to remove it with ethanol or something) and the chip might have been hand placed on the board (with all the possible complications). I found it pretty hard to kill an unpowered IC with heat. \$\endgroup\$ – Arsenal Dec 10 '15 at 14:21
  • \$\begingroup\$ I did check the part (I bought the evaluation board and tested it and concluded that the part is suitable for my needs) \$\endgroup\$ – Roberts Gotlaufs Dec 10 '15 at 14:41
  • \$\begingroup\$ I also already tried removing the shininess as you suggested with ethanol and some other agent (anti drinking agent :) ) mix. It did not help. The shininess almost seems like some kind of coating (Like one of these acrylic clear sprays). I am only focusing on the shininess because the chip on the eval board looks different (as well as every other chip on my bard) \$\endgroup\$ – Roberts Gotlaufs Dec 10 '15 at 14:59
  • \$\begingroup\$ So, is it fair to say that your only relevant question is contained in the eight paragraph and nothing much of any relevance is contained in the previous 7 paragraphs? \$\endgroup\$ – Andy aka Dec 10 '15 at 14:59
2
\$\begingroup\$

I will focus on two specific topics of the original question,

I am now suspecting manufacturer's fault. Either they managed to burn the chips during the reflowing or they somehow used some quick test-run lower quality setting on PCB production. I'm just guessing.

So my question is: Are there any telltale signs, that a chip might be burnt or damaged during reflow? Would the shininess of chips be any good way to tell whether the chip is burnt? Any other possible problem causes?

Regarding the first (PCB assembler's fault), it is really difficult to know, however, I would certainly try to unsolder one of the suspicious chips (with a hot air pencil-like gun) and resolder it. It is not as difficult as it may seem, however, you obviously take the risk of not being able to solder back the chip in place.

As per your second question... It is much more difficult that it seems to damage a plastic or ceramic packaged chip during reflow. Unless the assembler's oven is faulty or the operator was really sloppy setting the heating curve.

I would definitely try to clean the residues on top of the cheap, as the shininess looks like flux remains. If the flux is of the water-soluble type, then you have a chance of solving the issue by cleaning thoroughly the top of the IC and also the lateral and other shiny remains on the PCB. This is because water-soluble flux is usually conductive.

How to properly clean the flux? The standard way in the industry is using isopropyl alcohol, commonly known as IPA. It is a good starting point as it is a good general solvent, with good cleaning properties, leaveing no residues behing. Beware of its use, as it is highly flammable and emits vapors. I would also use an anti-static brush.

https://en.wikipedia.org/wiki/Isopropyl_alcohol

|improve this answer|||||
\$\endgroup\$
  • \$\begingroup\$ I have already tried to clean the board with alcohol mix (C3H8O + C2H6O) that I dot at chemist's store. it did not solve my problem ( the shorted board is still shorted and the bad-x-axis board is still very bad-x-axis-data). It also did not remove the shininess. Today I got the heat gun and desoldered one of the chips from the other board that had a short and now the short is gone. I noticed that some of the chips' pads are not covered in solder, but are in golden-coppery color. I will try to cover all the pads with solder and resolder it. \$\endgroup\$ – Roberts Gotlaufs Dec 11 '15 at 8:43
  • \$\begingroup\$ I managed to take a picture of the chip. See the link: goo.gl/photos/iJ3Ud2W15dsafzZN6 \$\endgroup\$ – Roberts Gotlaufs Dec 11 '15 at 8:52
  • \$\begingroup\$ It looks good! I have in your same situation several times. I guess the pin pitch is 0.3-0.4mm. However the IC has a relatively number of pins, making things easier. You can clean the pads of the PCB and surroundings, do not use flux (may be conductive, and too thick). For me, it usually works simply re-applying clean solder (not paste) with a fine tip iron. Remove the excess and leave only a tiny amount of Sn alloy attached to the pads. Reposition the chip, heat with the gun and let the chip align by itself. You may have to retry the procedure a few times until you get all connections OK. \$\endgroup\$ – jose.angel.jimenez Dec 11 '15 at 21:34
1
\$\begingroup\$

As I look to the picture the footprint on the PCA is not according to the design guidelines of the chip manufacturer. The pads with solder paste are not covered by the device. According to the data sheet there is a metal part on the bottom of the chip which can cause a short.

See the Technical Note of ST on soldering and designing MEMS devices.

|improve this answer|||||
\$\endgroup\$
  • \$\begingroup\$ I am aware that the pads are not according to guidelines. When I made the design I made the pads slightly longer (outwards from the chip) to be able to more easily try and use heat gun to resolder the chip if it was absolutely necessary. There is no metal part underneath this particular chip that could cause a short. \$\endgroup\$ – Roberts Gotlaufs Dec 11 '15 at 8:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.