0.5mm Pitch LGA Soldering

Status
Not open for further replies.
joe_prince

joe_prince

Well-known member
I'm eventually going to make a flight controller board for a quadcopter using the BNO055 for inertial measurement, and will have to solder the components to the board myself. The BNO055 is a 0.5mm pitch LGA chip (http://www.mouser.com/ProductDetail/Bosch-Sensortec/BNO055/?qs=QhAb4EtQfbV8Z2YmISucWw==), and so I will be using a hot air rework station to solder it (I don't own a reflow oven).

I have soldered TQFP chips in the past, by hand and with the hot air rework station, and it is visibly easy to see if there are any solder bridges or bad solder joints that need to be fixed. However, for an LGA chip the solder joints underneath aren't visible, so how would one go about testing whether the solder joints are good, no bridges have formed, etc.?

I suppose I can just "assume" everything went well, and give it power and see if the chip is behaving as it should, but I would like to be able to fix issues before potentially frying the board and components.
 
You could place testing points on each track going to a LGA pin and then measure if the signal is between expected levels (e.g. not floating, pulled high when programmed to go high etc). Of course this isn't a fail safe test because you can't put all the pins in a determined state, but at least it will give you a general view on how good your solder joints are.

An alternative to hot air can be a simple (kitchen) hot plate, or both to reduce thermal stresses.
 
Epyon said:
You could place testing points on each track going to a LGA pin and then measure if the signal is between expected levels (e.g. not floating, pulled high when programmed to go high etc). Of course this isn't a fail safe test because you can't put all the pins in a determined state, but at least it will give you a general view on how good your solder joints are.

An alternative to hot air can be a simple (kitchen) hot plate, or both to reduce thermal stresses.
Click to expand...

Thanks for the helpful tip, I might go ahead and try that.

Constantin said:
Eventually, consider building or buying a reflow oven...
Click to expand...

I will eventually get a reflow oven, but it still begs the question of how to test the quality/connections of the joints...
 
joe_prince said:
I will eventually get a reflow oven, but it still begs the question of how to test the quality/connections of the joints...
Click to expand...
The only way to check the quality of all joints is an X-ray photograph. Which isn't really practical. In volume production the applied solder paste is checked per JEDEC specs, and when properly reflowed will almost always result in good solder joints.
 
I have soldered more than 400 of these sensors and one thing you have to be careful of is not to put too much solder paste on the pads that the solder makes contact with the metal "wire" ends protruding through the package at the bottom sides of the package. I think Bosch did a bad job of designing this package; having exposed metal contacts outside on the package surface is not a great idea. At the same time, you have to make sure you have enough solder paste in each pad that you don't get cold joints. The only way to test this other than xray is to run a program that tests the functioning of all the connected pins. I have several times had a mag or gyro not ACK and discovered the solder did not "wet" on the corresponding pad that supplies power or GND to the package for this sensor. I use low temperature solder paste and highly recommend it partly because MEMS sensors do not like heat but mostly because rework is pretty easy. In those cases where I had a cold joint, I simply reheated, lifted the sensor of the board, applied solder paste directly to the (nearly) bare pad(s), replace the sensor on the land pad and reheated. Voila, everything works!

I also found it to be advantageous to very lightly tap the top of the sensors after the solder starts to flow to make sure there is good uniform contact on all the pads. You'll understand what very lightly means the first time your sensor package goes sliding half way across your board because you shoved it with your tweezers! Just pick it up and put it back, and nudge it (very lightly) into place if this happens. The reflow soldering process is really pretty forgiving.

The last thing to remember is that the art is to recognize the symptoms when the sensor is not working after assembly. You will make mistakes, but an expert is someone who has already made all the mistakes possible, so this is the only way to achieve mastery in this endeavor. Just start assembling using your best judgement and after a while you will (make enough mistakes to) know what works for you and your equipment.
 
Last edited:
onehorse said:
I have soldered more than 400 of these sensors and one thing you have to be careful of is not to put too much solder paste on the pads that the solder makes contact with the metal "wire" ends protruding through the package at the bottom sides of the package. I think Bosch did a bad job of designing this package; having exposed metal contacts outside on the package surface is not a great idea. At the same time, you have to make sure you have enough solder paste in each pad that you don't get cold joints. The only way to test this other than xray is to run a program that tests the functioning of all the connected pins. I have several times had a mag or gyro not ACK and discovered the solder did not "wet" on the corresponding pad that supplies power or GND to the package for this sensor. I use low temperature solder paste and highly recommend it partly because MEMS sensors do not like heat but mostly because rework is pretty easy. In those cases where I had a cold joint, I simply reheated, lifted the sensor of the board, applied solder paste directly to the (nearly) bare pad(s), replace the sensor on the land pad and reheated. Voila, everything works!

I also found it to be advantageous to very lightly tap the top of the sensors after the solder starts to flow to make sure there is good uniform contact on all the pads. You'll understand what very lightly means the first time your sensor package goes sliding half way across your board because you shoved it with your tweezers! Just pick it up and put it back, and nudge it (very lightly) into place if this happens. The reflow soldering process is really pretty forgiving.

The last thing to remember is that the art is to recognize the symptoms when the sensor is not working after assembly. You will make mistakes, but an expert is someone who has already made all the mistakes possible, so this is the only way to achieve mastery in this endeavor. Just start assembling using your best judgement and after a while you will (make enough mistakes to) know what works for you and your equipment.
Click to expand...

Fantastic information Kris, thanks so much for sharing! I'm going to try to document the process when I get there so if I run into issues I can trace where any problems might have occurred. I'll be sure to take all of these helpful tips into consideration.

I've been admiring both of your BNO055 boards you have designed - real beauties! Thanks for making them available on OSH Park as well. I have the files downloaded and have been reverse engineering them in Eagle so I can see how a "professional" board would be laid out. For my project interrupts aren't necessary, and don't have a need to change the I2C address since I'll only be using one IMU on the bus, so I won't be hooking up the INT or COM3 (I2C address select) pins. My goal is to just have 3V3, GND, SDA, and SCL broken out since I'll be continuously polling the data.

I'll let you know when I have the board designed - your trained eye will spot any errors from a mile away :)
 
Last edited:
My advice based on BMP280 experience is don't leave those pins floating.
My BMP280 sporadically changed I2C address until I figured out that was the problem and tied it down.
 
Wozzy said:
My advice based on BMP280 experience is don't leave those pins floating.
My BMP280 sporadically changed I2C address until I figured out that was the problem and tied it down.
Click to expand...

Thanks Wozzy and Kris - that's a very good point, you just saved me one round of revisions ;)
 
I glanced at the BNO055 data sheet.
If it's anything like the BMP280, the caps on VDD and VDDIO are absolutly necessary for stable output.
Don't be tempted to skimp on those. I can't speak for any oftheother BNO055 circuit details
 
Wozzy said:
I glanced at the BNO055 data sheet.
If it's anything like the BMP280, the caps on VDD and VDDIO are absolutly necessary for stable output.
Don't be tempted to skimp on those. I can't speak for any oftheother BNO055 circuit details
Click to expand...

Absolutely I agree. I'll be including all the necessary caps and pull up resistors on the board, just only breaking out the power, ground, and I2C pins.
 
IMO, the keys to the right amount of solder paste are the right stencil thickness, the right pad size reduction and clean edge cuts (not melted blobs from some laser cutter settings/plastics).
 
jonr said:
IMO, the keys to the right amount of solder paste are the right stencil thickness, the right pad size reduction and clean edge cuts (not melted blobs from some laser cutter settings/plastics).
Click to expand...

I was planning on applying the solder paste by hand (very carefully)...do you feel this won't work? I've never had a stencil made before, but I suppose there's no reason why I couldn't. Is that a service OSH Park offers that I can request when ordering a board?
 
I think it's much better with a stencil. But maybe I just need more practice with a syringe. I'd try one and see how you like it.
 
Status
Not open for further replies.
Back
Top