Ghost circuit shorts under smt resisters only when they are located on solder side of pcb (glued down)

We are having a process related problem with pullup resistors on a controller card which has a I960 micro-processor. The controller monitors signal inputs and it's own resources then blinks a LED to indicate its working. The problem is that after a period of time that varies from 1 to 2 weeks the uP will hangup and it is always found that a pullup's value measures below 1k ohm...sometimes less than 10 ohms. The problem doesn't exist initially because each pcb is probed for circuit impedance as part of the assembled pcbs final test. Also a power "on" test proves the operation of the uP and it's resources. The fix of this problem is to measure the pullups until you find the values that are less than 10k ohms then remove the resistor and scrape off "contamination" and replace the part. The Assembly House has had the contamination tested, has analysized their process, and tried several things to prove what is causing this without success. We usually see the problem on pcbs back from the field and it is showing up on at least one other product in a unrelated circuit. It seems to always be on the bottom of the pcb so it must be related to the epoxy process and subsequent flow solder and aqueusient cleaning operation. However the epoxy vendor and labs say they have never seen this problem. Meanwhile we are having to Burn-in product for at least a month to prove the pcb quality.

reply »

| We are having a process related problem with pullup resistors on a controller card which has a I960 micro-processor. | The controller monitors signal inputs and it's own resources then blinks a LED to indicate its working. The problem is that after a period of time that varies from 1 to 2 weeks the uP will hangup and it is always found that a pullup's value measures below 1k ohm...sometimes less than 10 ohms. The problem doesn't exist initially because each pcb is probed for circuit impedance as part of the assembled pcbs final test. Also a power "on" test proves the operation of the uP and it's resources. The fix of this problem is to measure the pullups until you find the values that are less than 10k ohms then remove the resistor and scrape off "contamination" and replace the part. The Assembly House has had the contamination tested, has analysized their process, and tried several things to prove what is causing this without success. We usually see the problem on pcbs back from the field and it is showing up on at least one other product in a unrelated circuit. It seems to always be on the bottom of the pcb so it must be related to the epoxy process and subsequent flow solder and aqueusient cleaning operation. However the epoxy vendor and labs say they have never seen this problem. Meanwhile we are having to Burn-in product for at least a month to prove the pcb quality. | I have seen shorts on bottomside SMD parts before and found that they were due to the glue process or specifically the glue cure process. If the glue is hygroscopic and absorbs moisture then porosity/voiding will occcur when cured, especially if the cure profile is aggfressive. THen when going over the way, flux and/or solder may migrate into the adhesive, reducing the resisitance across the part.

Pry off a suspect resistor and examine under magnification for flux residue or solder shorts. Also get some glass slide and place a dime sized thin section of adhesive and let it stand out for a couple of days. Then put another glass slide over the ahesive and run it through yur cure profile and look for porosity in the cured glue.

reply »

| | We are having a process related problem with pullup resistors on a controller card which has a I960 micro-processor. | | The controller monitors signal inputs and it's own resources then blinks a LED to indicate its working. The problem is that after a period of time that varies from 1 to 2 weeks the uP will hangup and it is always found that a pullup's value measures below 1k ohm...sometimes less than 10 ohms. The problem doesn't exist initially because each pcb is probed for circuit impedance as part of the assembled pcbs final test. Also a power "on" test proves the operation of the uP and it's resources. The fix of this problem is to measure the pullups until you find the values that are less than 10k ohms then remove the resistor and scrape off "contamination" and replace the part. The Assembly House has had the contamination tested, has analysized their process, and tried several things to prove what is causing this without success. We usually see the problem on pcbs back from the field and it is showing up on at least one other product in a unrelated circuit. It seems to always be on the bottom of the pcb so it must be related to the epoxy process and subsequent flow solder and aqueusient cleaning operation. However the epoxy vendor and labs say they have never seen this problem. Meanwhile we are having to Burn-in product for at least a month to prove the pcb quality. | | | I have seen shorts on bottomside SMD parts before and found that they were due to the glue process or specifically the glue cure process. If the glue is hygroscopic and absorbs moisture then porosity/voiding will occcur when cured, especially if the cure profile is aggfressive. THen when going over the way, flux and/or solder may migrate into the adhesive, reducing the resisitance across the part. | | Pry off a suspect resistor and examine under magnification for flux residue or solder shorts. Also get some glass slide and place a dime sized thin section of adhesive and let it stand out for a couple of days. Then put another glass slide over the ahesive and run it through yur cure profile and look for porosity in the cured glue. | It sounds more like a cleaning problem than a glue problem. If an aggressive flux is used during the wave soldering process and it is not properly cleaned off, it can become conductive with time, temperature and humitity. How does the assembly house confirm that the cleaning process is working adequately? They should be sampling the cleaning process and at least testing some product for residual ionic contamination.

Boca

reply »

| | | We are having a process related problem with pullup resistors on a controller card which has a I960 micro-processor. | | | The controller monitors signal inputs and it's own resources then blinks a LED to indicate its working. The problem is that after a period of time that varies from 1 to 2 weeks the uP will hangup and it is always found that a pullup's value measures below 1k ohm...sometimes less than 10 ohms. The problem doesn't exist initially because each pcb is probed for circuit impedance as part of the assembled pcbs final test. Also a power "on" test proves the operation of the uP and it's resources. The fix of this problem is to measure the pullups until you find the values that are less than 10k ohms then remove the resistor and scrape off "contamination" and replace the part. The Assembly House has had the contamination tested, has analysized their process, and tried several things to prove what is causing this without success. We usually see the problem on pcbs back from the field and it is showing up on at least one other product in a unrelated circuit. It seems to always be on the bottom of the pcb so it must be related to the epoxy process and subsequent flow solder and aqueusient cleaning operation. However the epoxy vendor and labs say they have never seen this problem. Meanwhile we are having to Burn-in product for at least a month to prove the pcb quality. | | | | | I have seen shorts on bottomside SMD parts before and found that they were due to the glue process or specifically the glue cure process. If the glue is hygroscopic and absorbs moisture then porosity/voiding will occcur when cured, especially if the cure profile is aggfressive. THen when going over the way, flux and/or solder may migrate into the adhesive, reducing the resisitance across the part. | | | | Pry off a suspect resistor and examine under magnification for flux residue or solder shorts. Also get some glass slide and place a dime sized thin section of adhesive and let it stand out for a couple of days. Then put another glass slide over the ahesive and run it through yur cure profile and look for porosity in the cured glue. | | | It sounds more like a cleaning problem than a glue problem. If an aggressive flux is used during the wave soldering process and it is not properly cleaned off, it can become conductive with time, temperature and humitity. How does the assembly house confirm that the cleaning process is working adequately? They should be sampling the cleaning process and at least testing some product for residual ionic contamination. | | Boca |

One other thing to look at besides how hydroscopic the epoxy is, is the type of pattern that the glue is being applied with...if it's being dispensed by a glue dispenser, double-dots will be more prone to trapping contamination and allowing the shorts to occur. I prefer to stencil glue myself (much faster and more consistent). A single dot, or a stripe between the pads acts like more of a "dam" if you will, to help prevent that...

-Steve Gregory-

reply »

| We are having a process related problem with pullup resistors on a controller card which has a I960 micro-processor. | The controller monitors signal inputs and it's own resources then blinks a LED to indicate its working. The problem is that after a period of time that varies from 1 to 2 weeks the uP will hangup and it is always found that a pullup's value measures below 1k ohm...sometimes less than 10 ohms. The problem doesn't exist initially because each pcb is probed for circuit impedance as part of the assembled pcbs final test. Also a power "on" test proves the operation of the uP and it's resources. The fix of this problem is to measure the pullups until you find the values that are less than 10k ohms then remove the resistor and scrape off "contamination" and replace the part. The Assembly House has had the contamination tested, has analysized their process, and tried several things to prove what is causing this without success. We usually see the problem on pcbs back from the field and it is showing up on at least one other product in a unrelated circuit. It seems to always be on the bottom of the pcb so it must be related to the epoxy process and subsequent flow solder and aqueusient cleaning operation. However the epoxy vendor and labs say they have never seen this problem. Meanwhile we are having to Burn-in product for at least a month to prove the pcb quality. |

It sounds like flux residue to me - either electromigration or ionic contamination. Usually, if the board washing process is out-of-control or inadequate, then some contamination usually doesn't get cleaned off, especially in a hard-to-reach area.

I also used to work at a place where certain high-impedance parts of the circuit were very sensitive to contamination and would fail over time (i.e. field failures would get returned).

Do what the others suggested, and stress test the boards under high temp/high RH or temperature cycle them to check the robustness of the cleaning process.

reply »