How do you measure the capacitance of a bondpad? I know it doesn't really matter as much for digital but its important when dealing with RF. I Googled but got a lot of patents
Thanks in advance.
Thanks in advance.
Bondpad Capacitance
- Thread starter Management
- Start date
You will need to model it to extract the C-parasitics at the pad. Industry would use something like Ansoft's Q3D Extractor to model the device and extract the C (and R and L if you like) parasitics by placing a sink and source around the bondpad contact. The simulator would then electrically model the bond pad behaviour according to the source/sink arrangement. If you assume uniform source and sink area across the bond pad you will get the pure-C of the bond pad.
It is very difficult to empirically measure the RLC parasitics of bond pads because of the negligible small values of these elements. I did quite a lot of work of these some years back.
Dave
It is very difficult to empirically measure the RLC parasitics of bond pads because of the negligible small values of these elements. I did quite a lot of work of these some years back.
Dave
Now I just have to figure out how to model it. I have a boards with space for a few MEMs devices but there are only the pads there now. I just wanted to know if it was possible to do something in the lab to measure the capacitance.
I don't know if I can get my hands on Ansoft though. What would I need to do it empirically and/or do it in the lab some how?
Thanks for the response Dave.
thingmaker3
- Joined May 16, 2005
- 5,083
Would the "Fastcap" software work for modeling this?
No its not raw chip die pads. It is a small board that has 3 bondpads on it. Similar to the bondpads in this picture but only 3.
http://www.chipworks.com/uploadedImages/INFAIM200611-019-s.jpg
Can I even measure the capacitance of a bondpad like this? If so, how would some go about it. If there is a website I can read to figure it out then that would be helpful as well.
Thank you for the replies.
Distort10n
- Joined Dec 25, 2006
- 429
What exactly is that? When you say bond pads of a board are the bonding wires from the die pads being connected to these bond pads? I am confused. 
As to making measurements, if you do not have parameter extraction software then be prepared to have the greatest fun in your life.
There is a way to use a VNA to do paremeter extraction by emperical measurements. There is actually a JEDEC paper that discusses this:
http://www.jedec.org/download/search/JEP147.pdf
You would have to design a fixture and measure the parasitic capacitance first, then place your device into the fixture and make another measurement. The delta change will be your result, but this may me a little clumbsy I don't know.
You could also use an LCR meter, which suprisingly is pretty damn accurate down to single digit pF, but this will also be clumbsy for such small devices. You would not use an LCR meter to measure capactiances or indutances of a die, IMO.
As to making measurements, if you do not have parameter extraction software then be prepared to have the greatest fun in your life.
There is a way to use a VNA to do paremeter extraction by emperical measurements. There is actually a JEDEC paper that discusses this:
http://www.jedec.org/download/search/JEP147.pdf
You would have to design a fixture and measure the parasitic capacitance first, then place your device into the fixture and make another measurement. The delta change will be your result, but this may me a little clumbsy I don't know.
You could also use an LCR meter, which suprisingly is pretty damn accurate down to single digit pF, but this will also be clumbsy for such small devices. You would not use an LCR meter to measure capactiances or indutances of a die, IMO.
My inferral is that it is the bonding to the lead-frame or die.
Distorti0n's suggestion of measuring the delta-C between a know config and that with the bond-pad is, in principle, the right one. However, having tried this sort of thing before, it is difficult because the negligibly small parasitic capacitance (could be down to fF - 10^-15) due to the bond pads will be swamped by the measurement system. I would also suggest that the contact impedance will have a factor.
If you want a realistic answer you are going to need to model it. As suggested previously Ansoft's Q3D Extractor is a Finite Element and MoM modeller that solves for parasitic RLC values. They also have the free version of Ansoft Designer (ref. http://www.ansoft.com/ansoftdesignersv/) which, from what I can tell, also extracts parasitics - don't count me on this one. Otherwise it is back to first principles - you could calculate this using a Transmission Line Matrix, although you will need to get your coding head on.
Dave
Distorti0n's suggestion of measuring the delta-C between a know config and that with the bond-pad is, in principle, the right one. However, having tried this sort of thing before, it is difficult because the negligibly small parasitic capacitance (could be down to fF - 10^-15) due to the bond pads will be swamped by the measurement system. I would also suggest that the contact impedance will have a factor.
If you want a realistic answer you are going to need to model it. As suggested previously Ansoft's Q3D Extractor is a Finite Element and MoM modeller that solves for parasitic RLC values. They also have the free version of Ansoft Designer (ref. http://www.ansoft.com/ansoftdesignersv/) which, from what I can tell, also extracts parasitics - don't count me on this one. Otherwise it is back to first principles - you could calculate this using a Transmission Line Matrix, although you will need to get your coding head on.
Dave
