- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I designed a board for a client of mine, and the board uses a PSoC 5 chip.
I got a whole load of boards made at a contract manufacturer and the client began testing them.
They all worked really well, except one small issue with one board, which I thought would be something simple.
The issue with the one board was that it could only be programmed with the 5-Pin header. When hooked up via USB windows did not detect the device at all, so it could not be programmed with BootloaderHost - and the customer needs to be able to flash via USB.
I got the board sent up to me for further investigation.
The board powered up OK, and ran though all the self tests perfectly, with the one exception that it did not appear as a USB device during boot. I thought it'd just be a connector or soldering issue.
When I looked a bit closer at the board I discovered that 7 capacitors were missing around the PSoC - in fact, all the 100nF caps were missing on the power rails going to the PSoC. For some reason yet to be determined, the manufacturer had missed off the decoupling caps at the PSOC.
Then I noticed somethint else, something far more interesting - there is a small, but deep hole in the top of the PSoC. Yip a small hole in the chip! It has a bit of charing around it. I assume something has gone POP inside the chip. It is strange that the only function not working is USB! I have attached a photo and you can just about make out the hole in the PSoC chip - and this board still works - all apart frm the USB functions!
This device only connects to GND, D+ and D- of the USB connector. The +V pin on the USB Type B connector is not used since we do not source any power from the USB host.
- Labels:
-
PSoC 5LP
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Could be a blown-up pin which was shorted or connected to an disadvantageous voltage-level. I would try to check (with a program) for dead or connected pins if the visual inspection of the board doesn't show any hint. I do not believe in chipworms or very small chipmunks. (;-)
Bob
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
That is interesting. We had some chips failed during protytping with 24V accidentally applied to the chip and the center part discolored. Never have a hole in the chip so far, 🙂
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Not sure if this works for todys plastic package material, but we used to use a
standard propane torch, take a part out to the parking lot, and torch it, plastic would
turn to ash, and we could examine die and leadframe for issues.
The principle way CMOS can develop high power shorts, localized heating, is the
parasitic SCR structure hung on I/O pins. If that gets triggered a short exists in that
region, between Vdd and Vss, usually blows out the bond wire on that pin. Or the
Vdd bond wire. Vss bond usually multiply bonded. Triggering mechanism is excess
current into a pin that has been driven below ground.
Regards, Dana.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I'm afraid that fabricating Crème brûlée out of an integrated circuit could destroy some of the structures you'd like to investigate. Have a look at http://en.wikipedia.org/wiki/File:Creme_brulee.jpg. I remember to have red that "they" use a mixture of concentrated sulfuric acid with 5% chrome (VI) oxide. This will remove all organic material (including fingers) and should not be handled by amateurs or outside of a laboratory. I presume that there are a few companies offering a service for that.
Bob
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Dana, Bob et al,
Do you think the missing decoupling capacitors could in any way be related to SCR latchup and subsequent 'venting' of the chip. Of course by 'venting' I really mean explosion/destruction. 🙂
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Short answer is yes.
Input excursions beyond Vdd or < Vss will do it, it only takes typically
a few 100 nS to a couple of uS to trigger the SCR, latchup.
Regards, Dana.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Sorry for the delay in replying!
(I got married last weekend, so I had to take a week off - you know how it is.)
Out of idle curiosity I wondered if any of the hardware team at Cypress could identify roughly which part of the chip that was which had vented - it's probably one of the VCC bond wires, or above an SCR pair I'd guess.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Most semiconductor operations have a Failure Analysis department that
can "de-lid" parts and establish what happened. Try contacting Cypress's
department on this.
Regards, Dana.