In Intel's official guide for PSU testing, here:
https://cdrdv2.intel.com/v1/dl/getContent/338448
There is "I_PSU% Delay test", which is on page 51. They say to test using the capacitive load described in table 4-7 (page 31). Any capacitive load to which you apply DC will finish filling up the capacitors rather quickly, baring that the capacitors are very big*.

So why do they say to use a capacitive load? Wouldn't a resistive load work well enough?

Thanks!

* Or that you have a high value resistor in between the capacitor and one of the legs, but that's not really applicable here based on the current flow -- 36ohms for the 12v rail.

 

This may be related to inrush. With capacitive loads, there is a high inrush current that goes through the system.

 

Because a capacitive load affects the frequency response of the control loop, and is likely to cause instability in a power supply which has inadequate phase margin.

 

It is not necessarily required. It depends upon the particular power supply and the expected load

 

As a first-order approximation, a real world power supply consists of a theoretically perfect power supply in series with a small, fixed impedance, dominated by its resistance. The resistor is internal to the supply assembly, between the output of the (perfect) regulator and the output connector. The important part is that the point at which the output is sampled and fed back too the control loop amplifier is *after* the resistor. Thus, any external capacitance forms a single-pole lowpass filter with the supply's output impedance, with attendant phase shift and group delay, and the output of that filter is the input to the error amplifier.

This phase shift can cause instability in the control loop, which is why external capacitance is important in power supply testing. If you spend waaaay too much time reading datasheets, you will see that some regulator ICs, both linear and switching (and some opamps), have a maximum output-side capacitance spec. I think it is Linear Tech that has a line of chips that brag about how large of a load capacitance they can handle.

"This phase shift can cause instability in the control loop." How? Well, there is an entire branch of Electrical Engineering devoted to feedback and control systems, and those guys are just dying for someone to ask them. Buckle up; the math is not fun.

ak

 

Hendrik Bode Network Analysis and Feedback Amplifier Design published 1945.

 

Hendrik Bode Network Analysis and Feedback Amplifier Design published 1945.

1945! Is it still relevant? That's even before color TV!
An even better question is, shouldn't I be reading something more up-to-date?

 

Hendrik Bode Network Analysis and Feedback Amplifier Design published 1945.

1945! Is it still relevant?

I really hope you're kidding.

ak

 

1945! Is it still relevant? That's even before color TV!
An even better question is, shouldn't I be reading something more up-to-date?

Dr. Bode was one of the many pioneers [along with H. S. Black & Harry Nyquist] of circuit analysis, all working at Bell Labs. Black invented feedback amplifier designs in the late-1920s which made long-distance telephony reliable, and wrote a famous paper on the subject in 1934. [The early 1950s explosion of so-called "high fidelity" home audio systems happened because Black's patent on feedback amplifiers expired]. Nyquist was a mathematician who assisted Black with rigorous expression of these concepts. Bode contributed a graphical technique of analysing for amplifier stability. Reading their work in their own words is superior to learning from modern summaries of these ideas, but you may have trouble with the high-level concepts involved until you have gotten a better idea from modern textbooks. Keep reading around - no one reference is sufficient - the field of electronics is very broad, and you need to have exposure to both historical & modern approaches. Best wishes for your studies! There are no shortcuts to avoid long and difficult study. Read the following carefully for some history of these gentlemen:
https://en.wikipedia.org/wiki/Hendrik_Wade_Bode
https://en.wikipedia.org/wiki/Harold_Stephen_Black
https://en.wikipedia.org/wiki/Harry_Nyquist

 

I really hope you're kidding.

ak

Just to put my comment in context, when I was trying to learn about PCs, almost* every book from the libraries was outdated. They were about Win98, DOS3.1, etc, when we were all running on Vista and Win7 was just out. There was also one outdated book on Linux mainframe administration from Redhat.
So I ask this question out of necessity.

* By "almost" I mean that in the whole library system, after extensive help and searching through multiple library systems and with multiple librarians, I found exactly 1 up-to-date book on PCs -- outside of the official windows manuals.

 

The effects of a capacitive load on the output of a feedback circuit are many, and many of them are not specific to one type of application, such as a switching power supply, let alone one sub-specific application such as a "modern" CPU board versus an "ancient" one from 1990. All regulated power supplies have a feedback mechanism and a control amplifier. The math behind that has not changed in over 100 years. Bode's work is the gold standard for frequency response, stability, etc., and is taught in every competent electronics school on this planet.

It is true that some concepts and methods have been improved or updated over the decades. Bode is *not* one of them; you picked the wrong one to tangle with. Despite all of the noise surrounding the latest wiz-bangy toys, the core concepts of electricity and Electrical Engineering are rooted in the 1700's and 1800's, and will never change. Compared to Faraday's Law of Induction, Bode is an up-and-coming young whippersnapper.

And if you think Bode's work is old, consider that Laplace died in 1827.

https://en.wikipedia.org/wiki/Laplace_transform

ak

BTW, Ohm's Law also still finds some use.
(OK, that was piling on.)