Clarifications: 1. @29:00 I meant to say 'Number of measurement points increases resolution' (receiver bandwidth decreases noise). 2. Discrepancies in MLCC self-resonant frequency (measurement vs datasheet): I assume this is likely due to the limitations of the impedance adapter (sub 20mOhm measurement, 10% error already at 0.5Ohms, inductance limits already at 200kHz, etc etc - see the B-SMC user manual). Also how much force is applied to the mechanism to improve the contact (during measurement & calibration) will influence the result slightly - like I mentioned in the video, I'm solely relying on the clamping force of the spring: e.g. for a 'short' calibration at around the resonant frequency of the MLCC cap my measured impedance is well within the range of the ESR.) Also, looking at the impedance vs frequency curve given by Murata, the SRF dip is actually quite broad (hits 20mOhms and below at 4-10MHz). During making of this video I've recalibrated numerous times to check for any calibration errors, checked multiple different caps, put two caps in series, and am consistently getting the same results. In future videos, we'll look at other measurements techniques (e.g. 2-port) and compare the measurements (I measured the same cap with 2-port and am getting far closer to the correct SRF and inductance).
Phil, this is beyond a masterclass, theory, experience, pitfalls, data collection, simulation and comparative analysis (this is my career for the past 30yrs). I hope your viewers take away that a capacitor isn't really a capacitor and numerous reasons why simulation incorrectly predicts circuit performance. Thanks and Merry Christmas to all over there at Phil's Lab.
Very happy to see videos other than high speed and analog design on this channel. Please make more videos on SMPS design, loop compensation, high power converter design, control system related topics....
Thank you for the very informative video! It would be neat to see the Bode results with a DC voltage applied to the capacitor using your DC Block and Bias adapter.
super interesting! thank you! without wanting to be impertinent, how about a follow-up on capacitor measurements for audio filter applications, in particular distortion of the different types?
Thanks Phil for the great detailed video. Out of curiosity have you tried to use the "seek passivity"(45:46) feature when extracting a model? I was also wondering if you had any plans to show on-board PDN impedance measurements in the future? I'm currently looking to do this for some of my higher-end boards where the processor/FPGA have explicit target impedance requirements. Cheers.
Thank you! I wanted to test the passivity feature out and compare - I've only recently updated to this new software version with circuit fit, so haven't played around with it terribly much. And yes, I've had some PCBs manufactured where we'll measure the PDN profiles in real-time - in my eyes, super cool stuff to see in the 'real world'!
Hello Phil. 29:00 I'm not familiar with the device, but it seems to me that if you want to increase the resolution then it would be necessary to increase the number of measurement points rather than narrowing the filter. Narrowing the filter will reduce the noise level, not the resolution. 29:30 The resonant frequency error is very large. 4.5 MHz measured and in the datasheet 7 MHz. Converting this to inductance, we get 516 picohenrys according to datasheet or 1250 picohenrys according to measurement. This is a very large measurement error. You performed the calibration and justify this error with an uncalibrated inductance of some kind? So You did the calibration wrong? Why didn't you do this calibration again already correctly? or maybe it is the instrument that measures so inaccurately? I did such measurements with the NanoVNA-H4 on 0805 100nF capacitors and got near 800 picohenrys parasitic inductance. Error much smaller than with a device for thousands of dollars. I also measured 2 pieces in series, in which case the inductance adds up, and it more or less agreed with the measurement of 1 piece.
Thanks for your comment! Regarding resolution and measurement points - you're completely right, I'm not sure why I didn't mention that. Regarding possible measurement error - I assume this is likely due to the limitations of the impedance adapter (sub 20mOhm measurement, 10% error already at 0.5Ohms, inductance limits already at 200kHz, etc etc - see the B-SMC user manual). Also how much force is applied to the mechanism to improve the contact (during measurement & calibration) will influence the result slightly - like I mentioned in the video, I'm solely relying on the clamping force of the spring: e.g. for a 'short' calibration at around the resonant frequency of the MLCC cap my measured impedance is well within the range of the ESR.) Also, looking at the impedance vs frequency curve given by Murata, the SRF dip is actually quite broad (hits 20mOhms and below at 4-10MHz). During making of this video I've recalibrated numerous time to check for any calibration errors, checked multiple different caps, put two caps in series, and am consistently getting the same results. In future videos, we'll look at other measurements techniques (e.g. 2-port) and compare the measurements.
@@PhilsLab It seems to me that a weak clamp would increase the ESR measurement result, and here it came out very close to what is in the datasheet, so rather the clamp is ok. I wonder how the signal enters the adapter's pin with the spring. Perhaps there is something going on here and depending on the position of the adapter's pin the phase of the signal changes which affects the measurement error. This error is in my opinion so big that it actually disqualifies this adapter for such measurements. And when you measured 2 capacitors in series, what kind of inductance results came out? did they add up? or were they shifted by some fixed value? Have you tried perhaps to measure a piece of thick copper wire, so that this pin is in the same offset as with the capacitors? A piece of wire should have a predictable inductance. The measurement with this adapter is a three-port measurement, so it seems like it should be the most accurate. But the design of this pin seems to me that it should be split, so that the signal to CH1 is taken as directly as possible from the DUT. I look forward to your next videos, with dual-port measurements. I find such measurements very interesting. Thank You.
Thanks for the ideas! With 2 caps in series the measured inductance approx. doubled (although it is still outside (too low) the 'rated' range for this adapter). I'll try your idea of measuring with a thick copper wire. Trying a two port measurement with the same cap gives the correct SRF + inductance with the Bode 100.
Ooh, capacitor parasitics, my special interest! :D Have you found any good modelling solutions for DC bias derating in class II MLCCs? I've managed to do it manually in SPICE with behavioural elements, but it's very cumbersome and the simulation ends up being rather slow and often unstable due to the solver struggling to converge.
I'm afraid my method is also rather manual and often I just set a very specific operating point in terms of bias. Do let me know if you find a good way of tackling this though - I'd definitely be interested in hearing more about that!
Clarifications: 1. @29:00 I meant to say 'Number of measurement points increases resolution' (receiver bandwidth decreases noise).
2. Discrepancies in MLCC self-resonant frequency (measurement vs datasheet): I assume this is likely due to the limitations of the impedance adapter (sub 20mOhm measurement, 10% error already at 0.5Ohms, inductance limits already at 200kHz, etc etc - see the B-SMC user manual).
Also how much force is applied to the mechanism to improve the contact (during measurement & calibration) will influence the result slightly - like I mentioned in the video, I'm solely relying on the clamping force of the spring: e.g. for a 'short' calibration at around the resonant frequency of the MLCC cap my measured impedance is well within the range of the ESR.) Also, looking at the impedance vs frequency curve given by Murata, the SRF dip is actually quite broad (hits 20mOhms and below at 4-10MHz). During making of this video I've recalibrated numerous times to check for any calibration errors, checked multiple different caps, put two caps in series, and am consistently getting the same results. In future videos, we'll look at other measurements techniques (e.g. 2-port) and compare the measurements (I measured the same cap with 2-port and am getting far closer to the correct SRF and inductance).
For some reason, you always drop a video covering topics I was wondering about a few days before. You're the best. Merry Christmas!
Great minds think alike :D Thanks, Mark - happy holidays!
Phil, this is beyond a masterclass, theory, experience, pitfalls, data collection, simulation and comparative analysis (this is my career for the past 30yrs). I hope your viewers take away that a capacitor isn't really a capacitor and numerous reasons why simulation incorrectly predicts circuit performance. Thanks and Merry Christmas to all over there at Phil's Lab.
Thank you so much, Craig! Merry Xmas and all the best for 2025 :)
@PhilsLab oh, I forgot to mention a tight script and excellent video editing.
Absolutely amazing and perfect timing as usual Phil! It is the best Christmas present ever. Wish you a merry Christmas and a Happy New Year!
Thank you very much - too kind! Merry Xmas and a happy new year to you too!
Very happy to see videos other than high speed and analog design on this channel. Please make more videos on SMPS design, loop compensation, high power converter design, control system related topics....
The level of detail in this video is amazing. Amazing! Thank you for the video.
I'm glad to hear that - thank you!
Thanks a lot for making this video. I can't wait for the series. Would it be possible to make a video on EMC testing or signal integrity?
Thank you for the very informative video! It would be neat to see the Bode results with a DC voltage applied to the capacitor using your DC Block and Bias adapter.
super interesting! thank you! without wanting to be impertinent, how about a follow-up on capacitor measurements for audio filter applications, in particular distortion of the different types?
Thank you! That's definitely a cool idea - I'll probably use my QA403 audio analyser for that!
Thanks Phil for the great detailed video. Out of curiosity have you tried to use the "seek passivity"(45:46) feature when extracting a model? I was also wondering if you had any plans to show on-board PDN impedance measurements in the future? I'm currently looking to do this for some of my higher-end boards where the processor/FPGA have explicit target impedance requirements. Cheers.
Thank you! I wanted to test the passivity feature out and compare - I've only recently updated to this new software version with circuit fit, so haven't played around with it terribly much.
And yes, I've had some PCBs manufactured where we'll measure the PDN profiles in real-time - in my eyes, super cool stuff to see in the 'real world'!
Hello Phil.
29:00 I'm not familiar with the device, but it seems to me that if you want to increase the resolution then it would be necessary to increase the number of measurement points rather than narrowing the filter. Narrowing the filter will reduce the noise level, not the resolution.
29:30 The resonant frequency error is very large. 4.5 MHz measured and in the datasheet 7 MHz. Converting this to inductance, we get 516 picohenrys according to datasheet or 1250 picohenrys according to measurement. This is a very large measurement error. You performed the calibration and justify this error with an uncalibrated inductance of some kind? So You did the calibration wrong? Why didn't you do this calibration again already correctly? or maybe it is the instrument that measures so inaccurately? I did such measurements with the NanoVNA-H4 on 0805 100nF capacitors and got near 800 picohenrys parasitic inductance. Error much smaller than with a device for thousands of dollars. I also measured 2 pieces in series, in which case the inductance adds up, and it more or less agreed with the measurement of 1 piece.
Thanks for your comment! Regarding resolution and measurement points - you're completely right, I'm not sure why I didn't mention that. Regarding possible measurement error - I assume this is likely due to the limitations of the impedance adapter (sub 20mOhm measurement, 10% error already at 0.5Ohms, inductance limits already at 200kHz, etc etc - see the B-SMC user manual).
Also how much force is applied to the mechanism to improve the contact (during measurement & calibration) will influence the result slightly - like I mentioned in the video, I'm solely relying on the clamping force of the spring: e.g. for a 'short' calibration at around the resonant frequency of the MLCC cap my measured impedance is well within the range of the ESR.) Also, looking at the impedance vs frequency curve given by Murata, the SRF dip is actually quite broad (hits 20mOhms and below at 4-10MHz). During making of this video I've recalibrated numerous time to check for any calibration errors, checked multiple different caps, put two caps in series, and am consistently getting the same results. In future videos, we'll look at other measurements techniques (e.g. 2-port) and compare the measurements.
@@PhilsLab It seems to me that a weak clamp would increase the ESR measurement result, and here it came out very close to what is in the datasheet, so rather the clamp is ok. I wonder how the signal enters the adapter's pin with the spring. Perhaps there is something going on here and depending on the position of the adapter's pin the phase of the signal changes which affects the measurement error. This error is in my opinion so big that it actually disqualifies this adapter for such measurements. And when you measured 2 capacitors in series, what kind of inductance results came out? did they add up? or were they shifted by some fixed value? Have you tried perhaps to measure a piece of thick copper wire, so that this pin is in the same offset as with the capacitors? A piece of wire should have a predictable inductance. The measurement with this adapter is a three-port measurement, so it seems like it should be the most accurate. But the design of this pin seems to me that it should be split, so that the signal to CH1 is taken as directly as possible from the DUT.
I look forward to your next videos, with dual-port measurements. I find such measurements very interesting. Thank You.
Thanks for the ideas! With 2 caps in series the measured inductance approx. doubled (although it is still outside (too low) the 'rated' range for this adapter). I'll try your idea of measuring with a thick copper wire. Trying a two port measurement with the same cap gives the correct SRF + inductance with the Bode 100.
Ooh, capacitor parasitics, my special interest! :D
Have you found any good modelling solutions for DC bias derating in class II MLCCs? I've managed to do it manually in SPICE with behavioural elements, but it's very cumbersome and the simulation ends up being rather slow and often unstable due to the solver struggling to converge.
I'm afraid my method is also rather manual and often I just set a very specific operating point in terms of bias. Do let me know if you find a good way of tackling this though - I'd definitely be interested in hearing more about that!
Is the load resistance 100 ohms across the whole range with no capacitance?
awesome stuff!
Thanks, Matt!
Amazing content from the best channel
Thank you!
Amazing like always ... Great in detailed component physics... Lovefrom india... Thanks a lot sir for this wonderful content
Thank you, Prakash - greetings from Germany :)
When he mentions the Bode 100 being expensive, the only price I could find was around $6k USD.
Very nice
Thank you, Michael!
That looks like an expensive piece of kit...
Yeah, certainly not inexpensive - but it's essentially a one-of-a-kind bit of lab equipment, and very useful.
👍😍🙏