CT scan analysis of various DAC chips

[weissi]

Hello fellows,

I'll have the possibilites to access a CT scan at a local research facility. The CT is said to offer details in the 0.000001m range.

One scan run takes about 45min, in total I'll have about 3h access to the scanner.

Already set are scans of:
- a defective TDA1541A
- defective/faked AD1862
- re-labeled PCM58

In my inventory I further have:
- PCM69
- PCM67
- PCM1794a
- AD1865
- AD1955
- CS4328 (my personal favorite)
- TDA1540
- TDA1387
- TDA1305
- various digital filter ICs (DF1706, SM5813, SM5842, YSF210, YM3434

There is one requirement that I was ordered to: To describe the test objectives....
I thought I can argue about the defective parts (identify the failure mode) and the old-school "hybrid", e.g. Bipolar/CMOS/CrNi parts.
I have no real arguments about the latter standard-CMOS parts, so maybe one of you fellows have some input...

Cheers, Markus

 

[miro1360]

PCM1704 is missing (or PCM1702)

 

[weissi]

I don't have PCM1704(2) and I'm not planning to get some.

 

[Ken Newton]

I suggest that there's no sense in consuming any part of your 3 hour access time scanning known defective devices. FMA (failure mode analysis) is valuable for design and production improvement, but are of little use otherwise. It seems to me that the most benefit to be had here is in identifying counterfeit devices, assuming that CT scanning would reveal that. Identifying which area of an device is defective still requires knowledge of the circuit funtioning of that area even to assist in design and production. Identifying which devices are counterfeit, and how generally widespread is that problem, is useful information.

 

[weissi]

Well, it basically is just for the pure fun and because I have the opportunity.
I shall repeat that I *need* to provide some background why I want to scan a particular part, so a defective one is the easiest approach (?) - that's because of my non-professional backround in chip design.
If I could provide more substance why I want to scan a particular part ... that is why I started this thread!

I'm more or less set for the CS4328 'cause I really want to see if that's a dual chip assembly.

 

[werewolf]

Well, it basically is just for the pure fun and because I have the opportunity.
I shall repeat that I *need* to provide some background why I want to scan a particular part, so a defective one is the easiest approach (?) - that's because of my non-professional backround in chip design.
If I could provide more substance why I want to scan a particular part ... that is why I started this thread!

I'm more or less set for the CS4328 'cause I really want to see if that's a dual chip assembly.

Yes, the CS4328 is a dual-chip package.

Please see: AES Convention Papers Forum >> 18-Bit Stereo D/A Converter with Integrated Digital and Analog Filters

 

[esl 63]

And the result is?

 

[weissi]

Not done yet, have to wait another 2 or so weeks...

 

[IVX]

I think such old parts had >>1um process and easily visible by optical microscope without scanning by X-rays. However, the method is destructive because of the need to open die.