|
I ran into a research paper on measuring capacitors using a VNA that might be of interest. It analyzes capacitance and ESR accuracy for the S11 reflection, S21 series-through, and S21 shunt-through measurement methods. The paper covers connector parasitics and fixture deembedding. Lots of useful graphs.
Brian
|
|
|
Other sources that are in other countries than Bosnia Hertzegovina
The authors have a number of other interesting "measure component with a VNA"
Measuring impedances of DC-biased inductors by using vector network analyzers
|
Your first link redirects to the same url the OP posted but I got to ask: What is your issue with the paper being hosted in Bosnia&Hertzegovina, the paper originated there. On Mon, 31 Mar 2025 at 20:44, Jim Lux via groups.io <jimlux= [email protected]> wrote: Other sources that are in other countries than Bosnia Hertzegovina
The authors have a number of other interesting "measure component with a
VNA"
Measuring impedances of DC-biased inductors by using vector network
analyzers
|
Nothing.. just some places block accesses to various countries, so finding alternate ways to get there is useful.
There was a time when .int was resolved through a DNS server in China, which is blocked from government networks in the US. So if you had an alternate .org name, it was useful.
toggle quoted message
Show quoted text
On Mar 31, 2025, at 15:55, Dragan Milivojevic via groups.io <d.milivojevic@...> wrote:
?Your first link redirects to the same url the OP posted but I got to ask:
What is your issue with the paper being hosted in Bosnia&Hertzegovina,
the paper originated there.
On Mon, 31 Mar 2025 at 20:44, Jim Lux via groups.io <jimlux=
[email protected]> wrote:
Other sources that are in other countries than Bosnia Hertzegovina
The authors have a number of other interesting "measure component with a
VNA"
Measuring impedances of DC-biased inductors by using vector network
analyzers
|
Too bad it doesn't lead to any practical use. What do you expect me to do with a deadly formula like this?
--
F1AMM
Fran?ois
|
On 2025-04-02 01:03, Fran?ois via groups.io wrote: Too bad it doesn't lead to any practical use. What do you expect me to do with a deadly formula like this?
Fran?ois, on dit en anglais <<maths moche!>>, "ugly math". John, at radio station VE7AOV.
|
By coincidence, I just added an image showing capacitance and dissipation factor plotted by my S-parameter plotter. It was derived from an .s2p file provided by Kyocera. Scroll to the last image:
One thing I don't understand. Manufacturers specify useful frequencies several times the capacitor self-resonant frequency, as revealed by the .s2p files they supply. The 33 pF shown departs from 33 pF as it approaches its 1.6 GHz SRF, which is nowhere near the 10 GHz specified upper frequency limit. All the capacitor manufacturers seems to do this. Why?
Brian
|
The max frequency is probably more about where losses get too big.
And they’ll measure for a very wide range. Most design tools can deal with staying away from the SRF, if supplied with the part parameters.
After all, at microwave frequencies most parts have a very complex and varying impedance.
toggle quoted message
Show quoted text
On Apr 2, 2025, at 15:34, Brian Beezley <k6sti@...> wrote:
?By coincidence, I just added an image showing capacitance and dissipation factor plotted by my S-parameter plotter. It was derived from an .s2p file provided by Kyocera. Scroll to the last image:
One thing I don't understand. Manufacturers specify useful frequencies several times the capacitor self-resonant frequency, as revealed by the .s2p files they supply. The 33 pF shown departs from 33 pF as it approaches its 1.6 GHz SRF, which is nowhere near the 10 GHz specified upper frequency limit. All the capacitor manufacturers seems to do this. Why?
Brian
|
|
|
To elaborate on this a bit. Let’s think about a standard 0602 sized part - that’s 0.06” long or about 1.5 mm. A piece of wire that long has an inductance of about 1.5 nH.
With, say, 33 pF, that’s resonant at 715 MHz. And yet, such parts are regularly used in all sorts of circuits at higher frequencies. You just design for the parasitic values.
toggle quoted message
Show quoted text
On Apr 2, 2025, at 20:14, Jim Lux via groups.io <jimlux@...> wrote:
?The max frequency is probably more about where losses get too big.
And they’ll measure for a very wide range. Most design tools can deal with staying away from the SRF, if supplied with the part parameters.
After all, at microwave frequencies most parts have a very complex and varying impedance.
On Apr 2, 2025, at 15:34, Brian Beezley <k6sti@...> wrote:
?By coincidence, I just added an image showing capacitance and dissipation factor plotted by my S-parameter plotter. It was derived from an .s2p file provided by Kyocera. Scroll to the last image:
One thing I don't understand. Manufacturers specify useful frequencies several times the capacitor self-resonant frequency, as revealed by the .s2p files they supply. The 33 pF shown departs from 33 pF as it approaches its 1.6 GHz SRF, which is nowhere near the 10 GHz specified upper frequency limit. All the capacitor manufacturers seems to do this. Why?
Brian
|
On Wed, Apr 2, 2025 at 08:14 PM, Jim Lux wrote:
And they’ll measure for a very wide range.
After looking at more .s2p capacitor files, I've concluded that the file comments refer to the measurement frequency range, not the recommended frequency range for the part. I've attached S-parameters for an 82 pF porcelain capacitor that are pretty wild. Capacitance calculated with the S21 and Y21 series-through methods differ greatly. Brian
|
On Thu, Apr 3, 2025 at 06:21 AM, Jim Lux wrote:
You just design for the parasitic values.
That's what I finally guessed designers must do. Check out the Y21 method plot for capacitance I just posted for a different capacitor. Above the first two resonances, capacitance returns to a positive value, though not the LF or rated value. Brian
|
On Wed, Apr 2, 2025 at 03:34 PM, Brian Beezley wrote:
One thing I don't understand. Manufacturers specify useful frequencies several
times the capacitor self-resonant frequency, as revealed by the .s2p files
they supply. The 33 pF shown departs from 33 pF as it approaches its 1.6 GHz
SRF, which is nowhere near the 10 GHz specified upper frequency limit. All the
capacitor manufacturers seems to do this. Why?
This is common practice in microwave power amplifiers and and some MMIC designs as one component brings in two. The R and L for above SRF of the C serves as a DC block while adding a stabilizing R and an L for the match. The same tactic is used with distributive L. First glance at some PA designs will most likely not make sense without recognizing that the parasitic is used to advantage. A complete sim is required adding in all device models including the passives.
|
Well.....here's another one for you: SM inductors. Most of the suppliers quote Q at 1 MHz. 1) The RF "resistance" (due to skin effect) should be used instead of the DC resistance. 2) Outside of RFI suppression, who really uses them at 1 MHz. Where are the plots of Q vs. frequency? Or RF "resistance" vs. frequency? This is why I always source MuRata who gets it right. CoilCraft? No. Dave - W?LEV On Wed, Apr 2, 2025 at 10:34?PM Brian Beezley via groups.io <k6sti= [email protected]> wrote: By coincidence, I just added an image showing capacitance and dissipation
factor plotted by my S-parameter plotter. It was derived from an .s2p file
provided by Kyocera. Scroll to the last image:
One thing I don't understand. Manufacturers specify useful frequencies
several times the capacitor self-resonant frequency, as revealed by the
.s2p files they supply. The 33 pF shown departs from 33 pF as it approaches
its 1.6 GHz SRF, which is nowhere near the 10 GHz specified upper frequency
limit. All the capacitor manufacturers seems to do this. Why?
Brian
-- *Dave - W?LEV* -- Dave - W?LEV
|
On Thu, Apr 3, 2025 at 08:11 AM, alan victor wrote:
The R and L for above SRF of the C serves as a DC block
while adding a stabilizing R and an L for the match.
I had guessed the DC block above SRF might be useful even when C was not the LF value. Thanks for the insight on R and L, Alan. Brian
|
Internally-matched power transistors use the internal bond wires as high-Q inductive elements. In conjunction with MOS capacitors they are used to build matching networks inside the package to bring the sub- to few-Ohm transistor impedances to something more manageable, or in some cases directly to 50 Ohm, at the device leads. The number of wires and their length, height and spacing determine their values, so are tweaked for tuning.
The devices' parasitic capacitances are absorbed into the matching network, making a useful element out of what was a hindrance.
73, Don N2VGU
|
Donald S Brant Jr