Hi man, have more time today... I'm thinking of compiling the tools I created, giving access to the Patrons... this should deliver value, would work as a very cheap software subscription... what do you think?
Good stuff Greg. Good stuff Greg. In the past I've also used non-gradient based optimization methods to tune ceramic resonator filters. Hint: You can not get trapped into a local minima when the loss function is quadratic, may be you can exploit this fact. Concerning impedance matching I'm doing a personal research on analytic gain-bandwith theory.
Interesting, I have never seen the current vs voltage vs frequency chart you showed but it makes total sense. I'm more used to see everything only in the power domain in RF design. Did you also take a look at QUCS or RFsim? Both are somewhat old but both work pretty well for this kind of stuff. Do you plan to measure the S11 paramters of the LDMOS or will you just take the value given in the datasheet? I guess if you want to measure it, you would have to do a full load pull table.
Hey Michael! Never used QUCS or RFsim.. The S11 probably need to be measured or the datasheet values taken... It is a good first step for the input match. S22 no. The output needs a loadline match, not a return loss match, as S22 is a small signal parameter that is not considering the device's physical limits. A very interesting fact: loadpull works because it is measuring the loadline match after impedance transformation from Cds and package inductance. Have you ever though about this? We have the right loadline resistive impedance in parallel with Cds, and in series the package inductance. This is a LC network that, unfortunately, transformers the impedance down (from outside we look at the inductance). This is why the impedance are so low.. like 1 + 0.5j.. You can see in the video, when I fixed the Cds and package L, the smith chart first goes to a lower impedance, and after comes to the specified impedance. The specified impedance is actually what we want the ideal transistor current source to see. =]
Support the channel becoming a Patron patreon.com/allelectronics
Hello Gregory, Wow your Smith Chart tool is incredible! The algorithm used for tuning works really well. Excellent job!! RF Man
Wow, thanks!
Hi man, have more time today... I'm thinking of compiling the tools I created, giving access to the Patrons... this should deliver value, would work as a very cheap software subscription... what do you think?
Good stuff Greg. Good stuff Greg.
In the past I've also used non-gradient based optimization methods to tune ceramic resonator filters. Hint: You can not get trapped into a local minima when the loss function is quadratic, may be you can exploit this fact. Concerning impedance matching I'm doing a personal research on analytic gain-bandwith theory.
Yep!! Probably a more useful cost function is something non-linear.. considering tolerances, temperature sweep and commercial component values!
Interesting, I have never seen the current vs voltage vs frequency chart you showed but it makes total sense.
I'm more used to see everything only in the power domain in RF design.
Did you also take a look at QUCS or RFsim?
Both are somewhat old but both work pretty well for this kind of stuff.
Do you plan to measure the S11 paramters of the LDMOS or will you just take the value given in the datasheet?
I guess if you want to measure it, you would have to do a full load pull table.
Hey Michael!
Never used QUCS or RFsim..
The S11 probably need to be measured or the datasheet values taken... It is a good first step for the input match.
S22 no.
The output needs a loadline match, not a return loss match, as S22 is a small signal parameter that is not considering the device's physical limits.
A very interesting fact:
loadpull works because it is measuring the loadline match after impedance transformation from Cds and package inductance. Have you ever though about this?
We have the right loadline resistive impedance in parallel with Cds, and in series the package inductance.
This is a LC network that, unfortunately, transformers the impedance down (from outside we look at the inductance).
This is why the impedance are so low.. like 1 + 0.5j..
You can see in the video, when I fixed the Cds and package L, the smith chart first goes to a lower impedance, and after comes to the specified impedance.
The specified impedance is actually what we want the ideal transistor current source to see.
=]
Awesome work!
Thanks man!! Welcome to the channel
Awesome job!
Thanks!!
How Can i get your Code? Do i need to be registered at all electronics? If yes, which membership is rewuired
Cool. In reality the inductor and capacitor has a resistive part.
😁😁
Amazing 👏👏👏
🤘🤘🤘
Wow! 🤯
👍