Hi all, you’re probably going to say this is a dumb idea, but please instead consider it a challenge...

I’m trying to build a 12V power supply to use with RF projects. The catch is that I want to try to make it run off of usb-pd, from a usb-pd decoy board. I’m trying to get as clean of a power rail as I can before regulating it down further to supply my RF circuits (HF radio stuff which I’m making to run off 9V).

I think I have the following noise sources based on measurements / tracking things down:

1. USB switching power supply, around 500khz w awful harmonics
2. Mains noise (from fluorescent lights in other parts of the house), around 50khz w awful harmonics
3. Digital noise transients from the usb-pd board

My plan is to 1) shield the PD board, and 2) filter its output. But the devil is in the details, of course. I wanted to run some thoughts by y’all to see if I’m thinking about things correctly.

First, the output filter design. From the PD output I’m planning to use a 100u toroid choke along with about 5 caps from 470u to 470p (the shotgun approach) to make a lowpass filter.

Question 1: When would I need a common-mode choke for something like this? What noise sources would a common-mode choke help inhibit?

Now, construction. I’m planning to have a division in my metal box (two compartments) to separate the usb-pd board from the output filter.

Question 2: Where is the best place to put the output filter inductor? Should it go in the same compartment as the usb-pd board (and then have its output immediately go thru the wall into the output filter side), or should it be placed on the same side as the output filter? Or does it matter?

Finally, measurement. To measure noise on the output rail I add a 50-ohm output resistance, then AC couple to a spectrum analyzer.

Question 3: Is this a correct measurement setup? Because I’m matching my output/impedance to the (shielded) coax cable, is it (relatively) protected from outside rf interference? (In particular, I’m imagining the usb-c cable is radiating a fair amount of transients). If I was not matched (like connected to a high-z scope), would the signal be more vulnerable to outside transients?

That’s all for now, thanks ahead of time for your thoughts!

(Side note: I’m new to RF and dealing with EMI is a bit of a black magic to me. This video by w2aew was really helpful for me to understand why ground planes are important and how HF goes thru them. Specifically, at 6:14 where he talked about HF current following the path of least inductance (to minimize loop area) was SUPER helpful. If anyone has any other good resources to recommend, I’d appreciate it!)

 

Could you please elaborate on the method of following noise sources based on your measurements.

Sure -- as I briefly mentioned above, to measure noise on the output rail I add a 50-ohm output resistance, then AC couple to a 50-ohm spectrum analyzer (or scope w termination). I see impulses at about 50khz when I turn the fluorescent light on that's in another room. Then, in my spectrum analyzer I'm seeing a train of -60dbm harmonics spaced around 1mhz apart, which I'm guessing are harmonics from a ~500khz switching frequency of the usb wall-wart.