Re:Exploring Power Distribution Networks
by csense on Friday March 31, 2023 @11:52AM
(#35388962)
This is way too much jargon and not enough background to be useful for someone who isn't already an experienced RF engineer. I've dabbled in basic electronics and I understand maybe 5% of what is being said.
One thing that strikes me is it looks like the test load seems to be just a resistor, and the input is a sine wave. Real practical IC's (especially CPU's) have variable loads that switch at unpredictable times, is a constant resistance and frequency really a good model of that?
Also it says the "DC bias problem" is "a trap for new designers", but instead of explaining what the DC bias problem is and how to solve it, it instead jumps immediately into talking about an experiment with a "bias tee RF+DC", where you have to set the frequency has "to be a decade below the SRF to avoid the effects of the resonance". I don't know what any of those words mean, or under what kind of circumstances I should avoid "class II" ceramic capacitors (whatever class II is).
OP is talking about vector network analyzers, spectrum analyzers and tracking generators. What is that equipment, how does it work, how expensive is it? What does it tell you about capacitor placement that you wouldn't be able to find by, say, setting an ordinary, run-of-the-mill oscilloscope to measure voltage on your microcontroller's power pin, then programming the microcontroller to wake from a deep sleep power state, max out its clock frequency and pump as much current into its GPIO's as the datasheet allows, and checking the voltage doesn't drop below the microcontroller's rated input?
Re:Exploring Power Distribution Networks
by hollandays on Friday March 31, 2023 @04:11PM
(#35392865)
If you're interested, you can learn enough to follow along by watching a lot of YouTube videos. I don't think it would be practical to explain all of the background in one article or one video, so you'll have to be tenacious and chase down whatever piques your curiosity. But I'll still try to get you started down the path of answering some of your questions.
The capacitors supplement the power supply by handling sudden, short changes in the load. The tests sweep through a range of frequencies one at a time. It's a powerful technique, but it has some limitations. For example, you have to know that the measurements will change if you change the DC voltage.
Determining if a physical system will work under all conditions and tolerances requires a model and some analysis techniques.
Re:Exploring Power Distribution Networks
by hollandays on Friday March 31, 2023 @07:24PM
(#35395142)
You ask good questions so I'll give very short "answers" to get you started in your search for good explanations.
The tests are frequency sweeps, one frequency at a time over multiple frequencies. Linear models can be decomposed into individual sine waves via Laplace/Fourier transform, so it's valid as long as the model is valid. The input voltage looks like a DC (bias) plus ripple (RF). The frequency response of some capacitors depends strongly on the DC bias (the trap), and flips to being inductive above the self-resonant frequency (SRF). Good test equipment is expensive, but there are options for hobbyists. One test like that can't predict the response for all inputs and conditions, so you need a model of the dynamics.
One thing that strikes me is it looks like the test load seems to be just a resistor, and the input is a sine wave. Real practical IC's (especially CPU's) have variable loads that switch at unpredictable times, is a constant resistance and frequency really a good model of that?
Also it says the "DC bias problem" is "a trap for new designers", but instead of explaining what the DC bias problem is and how to solve it, it instead jumps immediately into talking about an experiment with a "bias tee RF+DC", where you have to set the frequency has "to be a decade below the SRF to avoid the effects of the resonance". I don't know what any of those words mean, or under what kind of circumstances I should avoid "class II" ceramic capacitors (whatever class II is).
OP is talking about vector network analyzers, spectrum analyzers and tracking generators. What is that equipment, how does it work, how expensive is it? What does it tell you about capacitor placement that you wouldn't be able to find by, say, setting an ordinary, run-of-the-mill oscilloscope to measure voltage on your microcontroller's power pin, then programming the microcontroller to wake from a deep sleep power state, max out its clock frequency and pump as much current into its GPIO's as the datasheet allows, and checking the voltage doesn't drop below the microcontroller's rated input?
The capacitors supplement the power supply by handling sudden, short changes in the load. The tests sweep through a range of frequencies one at a time. It's a powerful technique, but it has some limitations. For example, you have to know that the measurements will change if you change the DC voltage.
Determining if a physical system will work under all conditions and tolerances requires a model and some analysis techniques.
The tests are frequency sweeps, one frequency at a time over multiple frequencies. Linear models can be decomposed into individual sine waves via Laplace/Fourier transform, so it's valid as long as the model is valid. The input voltage looks like a DC (bias) plus ripple (RF). The frequency response of some capacitors depends strongly on the DC bias (the trap), and flips to being inductive above the self-resonant frequency (SRF). Good test equipment is expensive, but there are options for hobbyists. One test like that can't predict the response for all inputs and conditions, so you need a model of the dynamics.