The Yxlon X-ray or another equivalent really good X-ray is a good first step, a non-destructional method to rule out internal registration issues. I have been able to see CAF with it, but you need to know your target first.
And I agree with you, normally it is very hard to detect.
Thermal imaging is also good, but can cause further damage to the CCAs because it requires full power functional testing to generate the thermal image.

-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Yuan-chia Joyce Koo
Sent: Monday, April 11, 2016 10:34 AM
To: [log in to unmask]
Subject: Re: [TN] Recommendations for Failure Analysis / Root Cause of PCBA Failure...

CAF is hard to detect using X-ray.  cross section is also a hit and missing game.  Unless you have thermal imaging data.
not easy.
    jk
On Apr 11, 2016, at 10:04 AM, Stadem, Richard D. wrote:

> If you have determined that the via in the picture is the culprit, or 
> is very likely the issue, I would start by having a set of really good 
> X-rays performed with a machine such as an Yxlon.
> Iff those X-rays prove the internal short condition, then you have 
> justification for microsectional analysis by a testing house that has 
> experience in this.
> Concurrently, electrical analysis using PWB layer artwork should be 
> able to determine if there is +12V and -12V both in the same area with 
> close approximation to each other. I am assuming the via in the 
> picture is either one or the other.
> If that is the case, it may be a situation of CAF internal to the PWB, 
> which is the only scenario that I can see that would develop over 
> time. But that is speculation at this point.
> To answer your question, the more samples you can provide for the 
> initial X-ray analysis, the better. Two may be enough, but perhaps
> 5 or 6 will provide more data; at least enough to justify sacrificing 
> one for sectioning. Try to find one that displays gross shorting.
> An initial X-ray survey should cost less, perhaps 2-3K.
> If the X-Rays provide justification, then microsectioning would go to 
> another $3K, depending on the number of cuts/mounts required.
> A good analysis house can give you a good estimate if you were to 
> forward the pictures to them and explain what you are experiencing.
> They may have other ideas for confirming that the via is the culprit, 
> and in determining the cause. Be prepared to provide the layer 
> artwork. I would get approval from the customer first; there may be 
> security or ITAR concerns.
> And I am assuming the large probe mark in the via is from 
> troubleshooting to locate the short. The problem is, the short could 
> be anywhere. How was it determined that that particular via is the 
> source?
> dean
> -----Original Message-----
> From: TechNet [mailto:[log in to unmask]] On Behalf Of Steve Gregory
> Sent: Friday, April 08, 2016 11:00 AM
> To: [log in to unmask]
> Subject: [TN] Recommendations for Failure Analysis / Root Cause of 
> PCBA Failure...
>
> Good Morning All,
>
> We have an assembly that we build here for a customer that we built 
> for some time, over 2,000 assemblies. There have been some failures 
> that our customer has had field returns with that are beginning to 
> escalate. There have been 12 field failures that we know of. The 
> failures are avionic assemblies that have been in the field ranging 
> from 111 days to 472 days.
> The actual failures are internal +12V to -12V shorts and we need to 
> know why. I have two failed assemblies here. Here are a couple of 
> external photos of a via that shows the short:
>
> http://stevezeva.homestead.com/Internal_Short.jpg
>
> http://stevezeva.homestead.com/Internal_Short_Close.jpg
>
> So my question is, can a lab take these boards and examine them to 
> determine what the root cause actually is from the two assemblies that 
> I have, or does there need to be more samples? What is the range we 
> should expect to pay for this type of analysis?
>
> We're trying to work with our PCB vendor on this issue, but I have 
> also been asked by my boss to pursue this with a third party.
>
> Thanks!
>
> Steve
>
>
> ______________________________________________________________________
> This email has been scanned by the Symantec Email Security.cloud 
> service.
> For more information please contact helpdesk at x2960 or 
> [log in to unmask] 
> ______________________________________________________________________
> --
>
>
> This email and any attachments are only for use by the intended
> recipient(s) and may contain legally privileged, confidential, 
> proprietary or otherwise private information. Any unauthorized use, 
> reproduction, dissemination, distribution or other disclosure of the 
> contents of this e-mail or its attachments is strictly prohibited. If 
> you have received this email in error, please notify the sender 
> immediately and delete the original.
>
> ______________________________________________________________________
> This email has been scanned by the Symantec Email Security.cloud 
> service.
> For more information please contact helpdesk at x2960 or 
> [log in to unmask] 
> ______________________________________________________________________


______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please contact helpdesk at x2960 or [log in to unmask] 
______________________________________________________________________

______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please contact helpdesk at x2960 or [log in to unmask] 
______________________________________________________________________