| Electronics > Projects, Designs, and Technical Stuff |
| (ultra)acoustic in circuit component level crack detector? |
| << < (5/12) > >> |
| rhb:
Thank you for the suggestion, but I spent 30 years doing time domain reflectrometry as a research scientist and programmer in the oil industry. Reflection seismology is just the low frequency, elastic wave version. This is the high frequency, elastic wave version. The only thing that changes is the decimal points. The hard part is the sensor and data acquisition. The rest I can do in my sleep. The thing that makes it hard is the dimensions. Ideally you want the dimensions to be much larger than 1/10 of a wavelength. When you're talking about 0204 parts that means very high frequencies and consequently fragile transducers. Fortunately, I have lots of xtals, so if I destroy a couple dozen it's under $3. |
| eliocor:
search for Confocal Scanning Acoustic Microscopy (CSAM). Some examples: https://sonoscan.com/resources/ami-overview/how-csam-technology-works https://www.tayloredge.com/reference/Electronics/Capacitors/MLC_FailureMechanisms.pdf http://www.us-tech.com/RelId/670224/ISvars/default/Finding_Singulation_Caused_Cracks_in_MLCCs.htm https://www.nanolabtechnologies.com/c-sam-imaging/ |
| coppercone2:
if there is a crack in the capacitor won't the magnitude just be less (high impedance connection)? Electrically/equivilantly what does the impedance look like here at the frequencies we are discussing? Like the solder joint/pcb under capacitor/solder joint interface has a impedance of X ohms and the ceramic bit has a impedance of Y ohms. Can you ball park the X and Y assuming its just simple resistance so I can get an idea of the dynamic range? If their similar maybe you can just do a scalar so its like 100 ohms in parallel vs 100 ohms or if its cracked 1000 ohms in parallel with 100 ohms. Then if its wrong you can maybe do something like resolder the joint then measure it again to make sure its not a solder joint fracture. a VNA measurement would be best but I also thought it would be hard. Is mechanical reflection just much more sensitive or are you thinking about trying to locate where the crack is rather then that there is a crack (for this kind of test we don't need to estimate where it is, the part is screwed if you touch it up and the crack is still there). If you can do it that way, and use cheap transducers, you could just use like three different frequencies, get RMS values through filters / analog network on a simple ADC, make a FFT, frequency shift it to like 3 diff frequencies (50Hz, 500Hz, 5000Hz) and play it, and it might be weird enough that if they dance around it gives you a different sound thats not really representative of whats going on but that something is off. I doubt you would want to trust tweezers to try to make a TDR of this thing, you could remove the part and analyze it under a specialist machine if you care or just replace it. I imagined a very fast tool for checking a buncha decoupling capacitors an stuff quick on a big PCB. I am no geophysicist but I thought that measurement relied on reflection because you can't get a receiver probe that can actually get you any scalar data unless you bury it real deep, in this case the part is part of a parallel circuit with the substrate. A S21 measurement rather then the S11 i am not saying not to shoot for the stars with a reflection measurement but i wanna know the options and electrical equivalents of what is happening/ I do not want to deal with a 200MSPS ADC either. At all. Or a cortex. |
| rhb:
I'd like to suggest we back up and start over. The fundamental problem is how does one efficiently identify SMD components with internal cracks or solder joint cracks. It's pretty clear that this is a significant cause of failures. And how does one do this without having service data for the board in question? SMD has given us the ability to put large complex systems on a single board. So they should be easier to repair at component level. And would be if schematics were available. |
| coppercone2:
I think knowing what the impedance/bode plots look like for cracked components could bring alot of light into this matter to see what the dynamic range requirements, loss, power, frequency, etc should be to aid in engineering some kind of sensor data at least the tool is a fun project and you could probably write a entire book on the test methodology but its not appealing to do so unless you have a toy to play with. too boring to work out that level of detail for me unless I know there is a method at least. this idea picked up alot of hype with alot of different engineers and sensors are fun my understanding of the mechanical process here is probably like an order of magnitude off which makes this extremely difficult to think about, i don't think many people know about ultrasonic wave propagation in ceramics or anything like that. i will rename this thread to mechanical crack detector because i am sure there are a slew of RF methods to maybe find it. i think this idea is more interesting to think about because its a cool gadget that people might find appealing to work on because you modify tweezers and stuff rather then more complex stuff that makes you feel under equipped. |
| Navigation |
| Message Index |
| Next page |
| Previous page |