[16:10] "The best transistor for one use case is not the best for another use case" - general engineering rule applicable for pretty much everything! ❤ 🙏
You are right, however I met many people that found something that "worked perfectly once" and they believe its the solution to every problem. In this particular case, its easy to think that a low Rdson transistor will always be better than a transistor with higher Rdson.
Awesome work as ever! Very professional concise and great we get to see (a very much compressed version of) how you got to each step of the process of the performance dignosis
The shorter the switching time of the transistor, the more impulse noise is generated on the inductance. We reduce the losses of the transistor, but then we increase them on the filters in order to keep the noise level at an acceptable level. Look with a spectrum analyzer or sdr receiver on FM radio frequencies. What is happening on the air. Often, I was increased resistor on the base of transistors up to 60 Ohm to decrease EMC noises..
I love your videos on power electronics, hope you have something for us about Multiphase Buck Converters in the future. Thanks again for the time and effort.
Hey, FesZ. How are you doing? Hope you're doing great. I'm designing a SEPIC converter, but I'm having some difficulty smoothing out the current spikes generated when the MOSFET closes. It's like a shoot-through current that flows back from the output cap, back through the diode, back though the decoupling cap, and through the MOSFET. That should be a common-enough issue, and I'm sure people would love to learn - just like I would - about the techniques that can be used to mitigate it.
Nice video! Great way to analyse the losses and nice output. One realy inportant thing that could be add despite thermal measurement is EMC measurement. That will make the difference. May you use a near field probe or LISN or CP coil? This could be a nice content for a EMC debug video and a final decision for wich is the best choice. Congratulation!
Although I like the video series overall, it botheres me a bit on how the topic (switching and conduction losses) was shown at the real converter. I may be wrong, or misunderstood something, so please correct me if so. The efficiency takes the IC (and other components) into account. In your specific application this may be the best way to figure out the best FET, as you already pointed out yourself. However I was thinking that the whole topic is about the switching and conduction losses, which can't be determined exactly from the efficiency alone, since this is now a mixed bag. A seperate measurement of the IC power consumption would have clarifyed this. I belive that the result in the efficiency of the SUD50p06 at low power consumption could be a lot better. The warm IC is worth noting, but other then that just a unlucky byproduct of the converter for explaning this topic. That beeing said, when you need to switch a FET with a gate driver, you often take the needed gate currents and gate capacitance in to account, resulting in a gate driver, that doesn't get that warm under this conditions. Also by altering the gate resistor slow-switching (and also less ringing) with low rds_on and fast switching with high rds_on could have been compared, in order to point out conduction and switching losses (which I though was the topic). Also after watching the video twice, I found it a bit weird that you pointed out Radj that much, because (as far as I read the Datasheet) it shouldn't have any effect in your application other than I_ds over current limiting. By choosing Radj to a high value the I_ds current limit is theoretically disabled, and since you're not about to hit that current with that small load, this fact is now pretty much meaningless for most viewers. As I first watched the video, it was not cear to me, what Radj does, and thought from your commentary at 1:40 it had to do with the gate current (wich would have been very important to note, if it was the case). I think all of this could be explained more detailed for example in a future video about gate resistors and ringing, which is a important topic on its own. What do you think?
Hello Tim, I took the IC consumption into account since from my point of view, efficiency is at the whole SMPS level, not just the power stage. When analyzing a power supply, you are interested in the global behavior which usually has the highest impact from the power stage and the control IC is usually negligible; I wanted to point out that in certain cases this is not true - by not taking into account the limitations of the IC, a "good" transistor can still cause high global loses. I guess you are right, I did insist a bit too much on the Radj topic... The ringing problem is a separate topic - I do plan to cover this in the near future; I think these are separate and colliding topics, but both very interesting.
@@FesZElectronics Okay, I get your point :) About the ringing topic, I tought that slower switching reduces ringing but increases switching losses, so it would have been matching topics, but I'm no expert. I'm looking forward about your coverage of this topic!
For switching power-circuits, I've sometimes had ceramic capacitors (dielectric losses) being big culprits - switching from X7R to NP0 reduced the losses dramatically. Do you plan to touch upon this in a future video?
Did you use these capacitors as input/output buffers or? Sounds like a very interesting topic - maybe you can give a few more details on the actual usecase and benefits.
@@FesZElectronics The PWM filter-capacitors - so, for example, C1 and C4 in the schematic visible at 5:09 . The AC charge/discharge current into the capacitors led to them getting hot. Temperature is, of course, not a direct measure of dissipation, but a good first indicator (observable, as you did, with a thermal camera).
Sorry, one addition - we had a wireless power-transfer system (~100kHz), so the capacitors saw a large dV/dt. In a DC-DC convertor, this will be limited, as the capacitor-voltage is fairly constant. (My pain-point may be true for the e.g. class-D RF applications, though.. In your circuits you seem to have used polymer capacitors which are OK, I suppose)
I think the main reason why NP0 is not commonly used is price vs capacitance; Especially at 100KHz I think you where using relatively large values so the NP0 was more expensive and larger (from a size point of view) compared to the equivalent X7R. Since you mentioned power transfer, I guess the value was also critical if you relied on resonance; X7R has quite a large capacitance value variation both from initial tolerance and temperature as well as from DC biasing. I think that in general when a precise capacitance value is needed, X7R is not to be used - either NP0 or film would be better.
As usual a very good and informational video. You and www.youtube.com/@w2aew are definitely in the absolute top tier when it comes to electronics tutorials.
Can you do a video on MOSFETs and combining and load share with multiple binoculars for a linear amplifier I found videos of the irfp260n but the videos are in a different language other than English it's hard to understand what is being done thank you for your videos
In other words, if you want high power switching, go for the lowest possible resistance FET and make up with it using 10A+ gate drive ICs? Ok cool will do.
Vgs normally should be around 8-10 volts to achieve maximum efficiency. Also, compare parameters with cheap counterfeit Chinese MOSFETS, as it is not reasonable to pay too much, if you can just take several of them and get them work. It is not fair to compare same category of price (high priced) components. Also, there should be NO diodes in DC DC converters except the protection ones. Voltage drop on them is a pure waste of efficiency.
What a thrill when I open my phone and see your video in subscriptions. Thanks!
PCB date: 09.2022 !
He has been working on that for months... This makes us realize the time and dedication needed to make those videos! Thanks a lot!
Every video starts many months before the publishing date :D
[16:10] "The best transistor for one use case is not the best for another use case" - general engineering rule applicable for pretty much everything! ❤ 🙏
You are right, however I met many people that found something that "worked perfectly once" and they believe its the solution to every problem. In this particular case, its easy to think that a low Rdson transistor will always be better than a transistor with higher Rdson.
This channel is criminally underrated. Keep up the good work.
Thank you for the kind words! I'm happy you are enjoying the content.
Awesome work as ever! Very professional concise and great we get to see (a very much compressed version of) how you got to each step of the process of the performance dignosis
Thank you so much for the time spent and the efforts put into making this very educational video.
The shorter the switching time of the transistor, the more impulse noise is generated on the inductance. We reduce the losses of the transistor, but then we increase them on the filters in order to keep the noise level at an acceptable level.
Look with a spectrum analyzer or sdr receiver on FM radio frequencies. What is happening on the air.
Often, I was increased resistor on the base of transistors up to 60 Ohm to decrease EMC noises..
You are My Online Lecturer🔥
Great explanation and practical verification. Many thanks for your efforts.
I love your videos on power electronics, hope you have something for us about Multiphase Buck Converters in the future. Thanks again for the time and effort.
Hey, FesZ. How are you doing? Hope you're doing great.
I'm designing a SEPIC converter, but I'm having some difficulty smoothing out the current spikes generated when the MOSFET closes.
It's like a shoot-through current that flows back from the output cap, back through the diode, back though the decoupling cap, and through the MOSFET.
That should be a common-enough issue, and I'm sure people would love to learn - just like I would - about the techniques that can be used to mitigate it.
Very informative video on transistors that I never knew about! Thank you for the education :)
Nice video! Great way to analyse the losses and nice output. One realy inportant thing that could be add despite thermal measurement is EMC measurement. That will make the difference. May you use a near field probe or LISN or CP coil? This could be a nice content for a EMC debug video and a final decision for wich is the best choice. Congratulation!
Great video thanks!
With you very professional and very informative videos, you deserve millions of subs! :) Love the way you do it all! :)
Great video, like always.
Although I like the video series overall, it botheres me a bit on how the topic (switching and conduction losses) was shown at the real converter.
I may be wrong, or misunderstood something, so please correct me if so.
The efficiency takes the IC (and other components) into account. In your specific application this may be the best way to figure out the best FET, as you already pointed out yourself. However I was thinking that the whole topic is about the switching and conduction losses, which can't be determined exactly from the efficiency alone, since this is now a mixed bag. A seperate measurement of the IC power consumption would have clarifyed this. I belive that the result in the efficiency of the SUD50p06 at low power consumption could be a lot better. The warm IC is worth noting, but other then that just a unlucky byproduct of the converter for explaning this topic. That beeing said, when you need to switch a FET with a gate driver, you often take the needed gate currents and gate capacitance in to account, resulting in a gate driver, that doesn't get that warm under this conditions. Also by altering the gate resistor slow-switching (and also less ringing) with low rds_on and fast switching with high rds_on could have been compared, in order to point out conduction and switching losses (which I though was the topic).
Also after watching the video twice, I found it a bit weird that you pointed out Radj that much, because (as far as I read the Datasheet) it shouldn't have any effect in your application other than I_ds over current limiting. By choosing Radj to a high value the I_ds current limit is theoretically disabled, and since you're not about to hit that current with that small load, this fact is now pretty much meaningless for most viewers. As I first watched the video, it was not cear to me, what Radj does, and thought from your commentary at 1:40 it had to do with the gate current (wich would have been very important to note, if it was the case).
I think all of this could be explained more detailed for example in a future video about gate resistors and ringing, which is a important topic on its own.
What do you think?
Hello Tim,
I took the IC consumption into account since from my point of view, efficiency is at the whole SMPS level, not just the power stage. When analyzing a power supply, you are interested in the global behavior which usually has the highest impact from the power stage and the control IC is usually negligible; I wanted to point out that in certain cases this is not true - by not taking into account the limitations of the IC, a "good" transistor can still cause high global loses.
I guess you are right, I did insist a bit too much on the Radj topic...
The ringing problem is a separate topic - I do plan to cover this in the near future; I think these are separate and colliding topics, but both very interesting.
@@FesZElectronics Okay, I get your point :)
About the ringing topic, I tought that slower switching reduces ringing but increases switching losses, so it would have been matching topics, but I'm no expert. I'm looking forward about your coverage of this topic!
Thanks 😊 great video!
For switching power-circuits, I've sometimes had ceramic capacitors (dielectric losses) being big culprits - switching from X7R to NP0 reduced the losses dramatically. Do you plan to touch upon this in a future video?
Did you use these capacitors as input/output buffers or? Sounds like a very interesting topic - maybe you can give a few more details on the actual usecase and benefits.
@@FesZElectronics The PWM filter-capacitors - so, for example, C1 and C4 in the schematic visible at 5:09 . The AC charge/discharge current into the capacitors led to them getting hot. Temperature is, of course, not a direct measure of dissipation, but a good first indicator (observable, as you did, with a thermal camera).
Sorry, one addition - we had a wireless power-transfer system (~100kHz), so the capacitors saw a large dV/dt. In a DC-DC convertor, this will be limited, as the capacitor-voltage is fairly constant. (My pain-point may be true for the e.g. class-D RF applications, though.. In your circuits you seem to have used polymer capacitors which are OK, I suppose)
I think the main reason why NP0 is not commonly used is price vs capacitance; Especially at 100KHz I think you where using relatively large values so the NP0 was more expensive and larger (from a size point of view) compared to the equivalent X7R.
Since you mentioned power transfer, I guess the value was also critical if you relied on resonance; X7R has quite a large capacitance value variation both from initial tolerance and temperature as well as from DC biasing. I think that in general when a precise capacitance value is needed, X7R is not to be used - either NP0 or film would be better.
Thank you for thes videos.... They help me so much 😃
As usual a very good and informational video. You and www.youtube.com/@w2aew are definitely in the absolute top tier when it comes to electronics tutorials.
Can you do a video on MOSFETs and combining and load share with multiple binoculars for a linear amplifier I found videos of the irfp260n but the videos are in a different language other than English it's hard to understand what is being done thank you for your videos
In other words, if you want high power switching, go for the lowest possible resistance FET and make up with it using 10A+ gate drive ICs? Ok cool will do.
Vgs normally should be around 8-10 volts to achieve maximum efficiency. Also, compare parameters with cheap counterfeit Chinese MOSFETS, as it is not reasonable to pay too much, if you can just take several of them and get them work. It is not fair to compare same category of price (high priced) components.
Also, there should be NO diodes in DC DC converters except the protection ones. Voltage drop on them is a pure waste of efficiency.