Power Supply EMI + BW Woes – #31

EEs Talk Tech - An Electrical Engineering Podcast
EEs Talk Tech - An Electrical Engineering Podcast
Power Supply EMI + BW Woes - #31

Kenny shares his experience debugging 800 MHz EMC issues at an unnamed engineering site. The culprit? A power supply! Sometimes, that 1:1 probe just isn’t enough…

Daniel Bogdanoff and guest host Erin chat with Kenny Johnson about the impact of power supplies on conducted and radiated emissions.


Links to discussed topics:

Decoupling Capacitor Optimization for Power Integrity Webcast:

Click to access 29March2018WebcastSlides.pdf

How to Design for Power Integrity Video Series:


Click to access 5_Power_Integrity_Ecosystem.pdf

Kenny’s Favorite Probe


Kenny likes textbooks

Kenny is a power integrity expert

Mobile device design is hard, Kenny feels bad for designers

Power integrity is coupled in with their radio, and makes it hard to pass EMI and EMC

EMI/EMC is failing, but:
Hardware guy has good data
Software guy has good software
Power guy looks to have no issues

4:45 FCC, ETSI

Types of EMI and EMC are:
Conducted emissions
Radiated emissions

Example: The IoT processor is only clocking at 5 MHz, but the EMC engineer is picking up noise up to 750-800 MHz. And, the system is dropping bits.

The 1:1 passive probe was hiding the higher frequency noise.

Then, they were able to trigger on the power supply and see the noise in the data line – power supply induced jitter.

A common rule of thumb is to have 20 MHz of bandwidth, but that’s not always enough!

Optimizing decoupling capacitors. How to choose the right capacitors? Where to place decoupling capacitors?

Many complex components come with design guidelines (voltage regulators, capacitors, etc.). But, it shouldn’t be treated as law.

Helpful resources

If you’re working on more prosaic devices (they aren’t crazy fast), even if you aren’t having an EMI issue, the same part of the board that’s having the EMI issues can also pollute the antennas.

How much bandwidth should you get?

Kenny connects to his device at full bandwidth, then pulls up an FFT. Then, he bandwidth limits to where the FFT rolls off.

A new power rail probe goes out to 6 GHz. Why do we need this much bandwidth? Higher BW noise!

Kenny saw a startup hub in Boston. It had a lot of different startups that pooled their collective resources to get access to higher end test equipment.

Kenny feels like the free tools are good for qualitative measurements, but not for quantitative measurements.

20:46 – Adam Savage – “Buy the cheapest tool first. If you break it, go buy a nice one.”

21:30 Kenny is part of the inspiration for this podcast.

Stupid Questions:
What’s the worst possible power integrity advice you could possibly give to someone?

What’s your favorite probe and why?


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.