Missing solder point on pcba assembly. 3

[mooimooi]

Hi there!

Need a bit of some guidance from pcba assly gurus;
You see, I have double sided pcb and on this i mount only the following components.
a) resistors - 3 pieces , total 6 solder points.
b) Photo interruptors - 3 pieces, total 12 points.
c) Hook up wires - 12 points

In total I have 30 points to manually solder. After solder,I go through a functional test. The functional test will also "pass" the points that were missed out during manual solder. This is because the component lead still makes contact with the pth(plated through hole) conductor.

Is there any simple way to check effectively whether all the 30 points have been soldered?

I tried using connection tester that detects resistance in the order of 0.4 ohm to 4 ohm, but am not very successful.

Another option is to make a template to check the 30 points, but this too can cause escapee to go through as we know that 100% visual inspection is at best 80% effective.

Another expensive way is probably using PRS ( Pattern recognition system ) , but such systems are beyond my means to buy.

Would appreciate anyone out there can provide some simple solutions to this simple problem. I am not too sure whether this is a simple problem or not!

Thanks and Regards

 

[melone]

Temperature testing is often an effective method for checking solder joints. Place your PCB in a temp chamber and drop the temp to -20°C (providing your devices are rated that low), and perform the funcitonal test. Then, turn your temp chamber up to 105°C (providing your devices are rated that high), and perform the functional test. If it passes those 2 tests, and a quick visual inspeciton of commonly missed solder joints, you should get a warm and fuzzy feeling that your assembled PCB is "good".

Hope this helps!

 

[mooimooi]

Hi Melone,
Thanks for the quick response.
I agree with you this Temp cyle test is probably the most effective test method for weeding out such kind of defects.
But our volume does not justify the purchase of a TC chamber.
I was looking for some kind of simple electrical test circuits that might be able to detect such circuits. Well if testing does not work, maybe some "poke Yoke' ie error proofing methods might be deployed to ensure all 30 points are soldered, and if any points is missed out, then an audible or visual annuciator is activated.

Thanks

 

[dspDad]

Visual inspection is much superior to any functional testing method if your failure is missed solder points.

If this is a production process, consider wave-soldering which has other failure modes, but missed points is not among them.

Another thought is to get your manual processes under control by counting the solder points and make sure that 30 have been hit.

Finding an error of this type is far more expensive than preventing it.

DspDad

 

[melone]

Or add a second operator to resolder the devices. Hopefully, both operators will not miss the same points.....

 

[nbucska]

These may not be high tech but one may be good enough:

1.) Make a template which covers everything but the solder points. Make a fixture which allows the pressurization
of one side of the board. Wet the surface with IPA
and look for bubbles.

2.) Same template. Make fixture which illuminates all solder
joints with plastic optical fiber -- look for light.

<nbucska@pcperipherals.com>

 

[mooimooi]

Hi all,

When i review all the valuble inputs from all of you, i conclude this is problem is quite a tough nut to crack.

I must admit that electrical testing might not be the &quot;catch all&quot; for this kind of problem.

I think visual inspection might be the best solution, as susggested by dspDad and nbcska. What i plan is to make template to cover everything esle except the solder points and shine some very bright light source and on the component side of the pcb put a shrouded dummy pcb with an array of photodiodes to detect for any light leakage.
Of course there might be some ambient light leakage going through , but i think this can be compensated by the gain of the amplifier. This is also quite cheap in terms of cost.
In this case, i think i can make the process operator independent.


Joy to all of you and wishing you a very happy Christmas and a prosperous 2004 and once again thanks to all of you for the help rendered.

Mooimooi

 

[Madcow]

Try a functional tests during vibration.

 

[BrianR]

If you want reliable solder joints then do not resolder them. That is a shortcut to failed joints. For small production runs it is probably just as cheap to inspect them using a 2-4x magnifier. This is certainly cheaper than faultfinding completed assemblies.

 

[steveowens]

I agree with the temperature testing idea – it is the most effective way to find bad solder joints. However, I would add that you need to change the temperature rapidly (>20C/min) in order to do much good. BTW – this called HASS testing – Highly Accelerated Stress Screening, or ESS – Environmental Stress Screening. The are several standard methods. We use 20 cycles, and then do a function test at low and high temperatures. If you only have a few PCBAs, then we could do them for you for free – send me an email.

Steve Owens
Steve.Owens@FinsihLinePDS.com

 

[mooimooi]

Hi Steve Owens,
We are running mass pro around 5k boards per day and i do not think it is possible to subject all of them into this accelarated environmental test methods, as this is introducing another process in the process plan and moreover my hookup wires in the pcba cannot withstand more than 80 Deg C. Also under such test methods, there will be insulation shrikage on the insulation crimp of the wire to board terminals, thus exposing the conductors. But i do agree with you that this is the most effective method to catch such intermittent failures. We are already doing a functional test now after assembly and what i was looking was to do a one pass test along with the functional testing test that would catch the &quot;no solder&quot; points using the same hookup, and optical methods looks like more promising. What would be your opinion Steve?
Regards

 

[nbucska]

Mooi**2

If you are using light, try to select a color for which
the board is opaque. For ambient light: use e.g. 40 KHz
modulation. 2500/480=5 sec/PCB -- this gives 6600 cycles
of the 40K, more than enough to filter & detect even allowing for brd changing time.


<nbucska@pcperipherals.com>

 

[cbarn24050]

Hi, there are many professional optical inspection around, with that kind of volume it sould be worth the investment. If you want to make your own it's quite straight forward. Set up two cameras to a monitor with a circuit that switchs between the two cameras each second. You put a good board under 1 and the test board under the other. If they are the same then the image stays constant, any differences flash as the cameras switch over.

 

[mooimooi]

Hi Nbucska,
I see the point of using a color that is opaque to the board, but i do not understand the your calculations, ie the 40Khz and 2500/4870 , and 6600 cycles. Would appreciate if you could explain in simpler terms. My knowledge of optical physics is quite bad..
Thanks.

 

[mooimooi]

Hi Charn24050,

Yes I think that would be a smashing idea if it works!
Just need more info on the system. Do you use B/W or color CCTV system? and how do you trigger the mechanism of flash when the two photos mismatch?. Sorry I may be asking some dumb questions, as I have not used such systems b4
Joy to you





!

 

[nbucska]

Mooi**2:

480 minutes=8 hours -- a shift
5sec * 40,000 cycles/sec /30 points = 6666 cycles/point
testing sequentially -- so you have enough time for the
filter to settle etc.

<nbucska@pcperipherals.com>

 

[cbarn24050]

Hi, there's not much BW around these days. You don't need to trigger anything, the differences show up themselves.

 

[zeitghost]

I think that the two cameras would require a common sync feed (genlock) so that the camera switching minimizes transients on the video going to the monitor.

rgds
Zeit.

 

[Madcow]

Try searching National Instruments site

http://www.ni.com

They have tons of info on imaging.

http://zone.ni.com/devzone/devzone.nsf/webcategories/B3A95D63D5EB5631862569390072EBED?opendocument

 

[markls8]

Hi BrianR -
(Not to hijack the thread.)
You wrote &quot;If you want reliable solder joints then do not resolder them. That is a shortcut to failed joints.&quot;
What is your experience in this? Is this a mil spec or aerospace requirement?
I dip solder small PCBA's and have often removed and resoldered individual components (resistors, triodes, capacitors, microswitches, terminal blocks etc.). I consider that the resolder joint, using standard technique, and removing any residual solder, is at least as good as the dip solder joint as it has received direct visual inspection from both sides of the board. Is it the presence of rosin residue which is the concern?
Just curious.
Thanks - Brian.