Back in 2001 , while Design Service Mgr at C-MAC, I was awarded a contract to design and produce 100 prototype head mounted video LCD's with lens and walk-man like interface for VGA and NTSC signal from an Ultrasonic scanner. We were given the Kopen AMLCD chip as the best source for high resolution tiny chips to use as the video display engine.
LCD chips of this kind require row and column scanning with analog signals sent to the active matrix which acts like a sample and hold for each pixel and the reference voltage for row and column switches must be well regulated and can be used to adjust black level and brightness of the display.
THe circuit considered of PLL for horizontal sync and logic for vertical sync and adjustment of pixels for pan and zoom using an Altera chip with discrete logic. There was a single 9V input given and a requirement for about 6 regulated voltages including 3.3 for for logic and 9, 12 15V for the AMLCD chip.
So I designed to use a step down DC-DC SMPS for the bulk power of the logic chip and step up SMPS for the analog circuits which were low power. The 1st prototype worked and had video on the lens close to the eye like a 15" monityor at arms length. with 640x480 VGa resolution ony avail at that time.
My problem in this case was where was that sound of a babling brook or running water coming from? Turned out to be an SMD choke used in the SMPS circuit. But how could that vibrate enough to cause a slight audble noise and sound like random noise? (water flow)
After a scratching my head I remember hearing about Chaos Theory which is basically anything with randomness in an orderly control system. My research concluded that was in deed what was happening. The PWM of the primary step down regulator was being loaded by the FM of the secondary regulator, both running at ultrasonic 100KHz ~500Khz rates but the modulation was effectively the noise in the control circuit causing an instability with high current pulses going thru the choke and hence mechnalical induce vibrations on the micro coils in the choke.
I wasn't interested in the Theory as much as a solution, so I experimented with sensitivity tests on loading factors filter ripple with response times. and settled on a capaitor value that was a tradeoff between minimum ripple and adequate margin away from of a 2nd order system loading another 2nd order system with " coupling on the interact of each.
Typical linear amplifiers which might oscillate when they do not have enough phase margin with load step response. In this case to self-regulated SMPS interact weakly to create Chaos or random noise within the loop bandwidth.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.