"Resistance” between BGA ball pairs?

[jackyzhangsh]

Hi,

A PCB assembly factory told me that they require the relative voltage between BGA balls, which I don’t understand.
For example, for a 256-ball BGA chip:

1. They would first set the multi-meter mode to “diodes”.
2. Then they measures the relative “resistance” (what the technicians said, which I am not sure) between balls. Of course it doesn’t have to be done for each (ball1, ball2) pair, but would suffice if they fix one ball as a reference, and measures the “resistance” between all other balls to the reference, from which the relative “resistance” between any pair can be calculated.
3. Small values can be as low as zero, and large values can be more than 300Ω, such as 310, or 400+ Ω. I am also not quite sure about whether the exact unit is Ω (ohm) here.

The purpose of doing this, according to a technician, is to make sure the BGA chip is in good state before soldering. Without this, they cannot know whether there are erroneous short-circuits between the balls or whether the chip has already been damaged by electrostatic shocks due to improper handling.

My question is that what are they actually measuring? Does it relate to the IBIS (Input/output Buffer Information Specification) or SPICE model?

When I asked them how did they get the correct reference data for a specific chip, they told me that the data has been accumulated. For example, many of their customers used a common type of MPUs, so after they measure and solder the first such MPU from the 1^st customer and had verified that the chip was good, they immediately save the measurements into an Excel file which would be stored in their database. For later customers which used the same type of MPU, they simply measure the chip and compare with their previous “known-good reference” measurements.

Does anyone know the theory behind this practice, and what is the quantity they are actually measuring? If I am their first customer using a particular chip which means they don’t have previously recorded measurement, where can I know what the “resistance” between balls are supposed to be so I can compare with the technician’s measurements to verify the “good” status of the chip? Is this information obtainable from the IC factory?


Jack

 

[FvM]

The practice seems absurd for devices sealed in original manufacturer trays. I also wonder how the measurement will be practically performed? Do they use a flying probe tester for it? Who's paying the effort?

It's very uncommon to have full shorts (zapped protections diodes) from electrostatic events, if they are likely at all, see above. So the significance of the test must be doubted in addition.

If at all, it may make sense for devices from dubious sources, or "sweeped up" from the factory floor.

Measuring the forward voltage of substrate diodes is a standard means of in-circuit test and well suited to check the continuity of solder connections and bond wires. It tests the health of protection circuits to some extent.

 

[jackyzhangsh]

The practice seems absurd for devices sealed in original manufacturer trays. I also wonder how the measurement will be practically performed? Do they use a flying probe tester for it? Who's paying the effort?

It is not in original manufacturer's tray. I am making a prototyping board so I ordered only one chip from a distributor which was sent to me wrapped in anti-electrostatic package. They don't charge for this test for small-quantity prototyping boards.

I had no knowledge of electrostatic shock when handling the chip, and I don't know if the chip really got damaged due to the very limited number of times that I touched it. Nevertheless, It seems to me that comparing the ball-to-ball "resistance" with a previously verified chip of the same type is a reasonable idea for knowing whether the present chip has defects. If the chip passed the test, any problem discovered later should most likely be attributed to soldering process or circuit design.

It's very uncommon to have full shorts (zapped protections diodes) from electrostatic events, if they are likely at all, see above. So the significance of the test must be doubted in addition.

What is "zapped protection diodes"? Could you explain a bit more and give a link to it?

If at all, it may make sense for devices from dubious sources, or "sweeped up" from the factory floor.

By "sweeped up", do you mean "swept up", such as the chip fell to the floor due to uncareful handling and after picking up they want to verify that the chip has not been damaged during the falling and picking up?

Measuring the forward voltage of substrate diodes is a standard means of in-circuit test and well suited to check the continuity of solder connections and bond wires. It tests the health of protection circuits to some extent.

I am somehow confused. Does "forward voltage of substrate diodes" refer to their measuring method I outlined in the 1st post? At the beginning you call it "absurd" for original manufacturer trays but now you seem to approve of it.


Jack

 

[FvM]

I mean, measurement of the diode forward voltage is a suitable test method. So it can be used in this test, if anyone thinks it has a purpose. But I don't think so.

Zapped diode means, the pin is shorted or has a considerably reduced resistance against ground or supply terminal(s). That's the only defect that can be easily detected without setting the chip into operation. I expect, that only a small part of possible defects is covered this way.

Todays chips have a basic protection against ESD, e.g. 8 kV standard test

Delamination due to moisture absorption is a much more likely faiilure meachanism, if devices have been repacked without baking them before soldering.