Efficient 12v to 5v step down converter without noise?

[hoffsta]

Here is what I'm trying to do:
Add a wi-fi enabled Raspberry Pi (5v) into an existing boombox (12v) for storing and streaming music. This boombox will spend time plugged in to AC at home but also see a good amount of use running off batteries.

I tried adding a cheap LM2596 buck step-down converter board between the 12v PS and the boombox electronics. Like this one:
Amazon.com: LM2596 DC-DC Buck Converter Step Down Module Power Supply Output 1.23V-30V: Car Electronics

This design resulted in god-awful buzzing & screeching when the Pi was powered up. I went for this unit because of the increased energy efficiency over a traditional LM7805 chip (an important consideration when running on battery power) but didn't anticipate the noise!

I later tested a few things such as powering the LM2596 from a separate AC-DC PS and using a 5v cell phone charger. This resulted in clean output, great when plugged into the wall, but not a solution for battery power.

I want to know what would be my best solution for this. Is there a different way of wiring the LM2596 to isolate the noise from the amp circuits?

I am a total electronics noob so please hit me up in laymans terms! Thanks

 

[dfy]

As your buck converter is adjustable, a simple low-cost solution you could try is setting the buck converter to 7.5 or 8V, then running an 7805 off that. That's a lot more efficient than running the 7805 off 12V. Granted, it's still not ideal, but you probably already have all the parts and you can simply try to see if it's good enough.

/edit: Also, keep the Pi and the audio interconnects away from the inductor on your converter board.

 

[hoffsta]

That's an interesting idea I hadn't considered. Thanks! Off hand, do you know how much less energy I'll lose to heat in that configuration? Is there some formula to calculate it?

 

[dfy]

Yeah:

Let's say your RasPi draws 500mA (not unreasonable). If you use the LM7805 on 12V, you will lose (12V-5V)*500mA = 3.5W dissipated as heat on the 7805. With the buck as a preregulator, it goes down to (7.5V-5V)*500mA = 1.25W (plus whatever your buck converter dissipates, but that should be negligible in comparison).

 

[hoffsta]

Thanks, that is a significant improvement.

I found this thread on Stack Exchange that seems to address my exact situation.
audio - Ground loop? Switch mode power supply noise - Electrical Engineering Stack Exchange

Unfortunately, I don't really understand all of it yet. It looks like the poster solved his problem by placing a differential amp between the computer and the power amp... still trying to figure this out.

If anyone has a comment to make this easier for me to understand, I'm all ears!

 

[benb]

A linear regulator for the final drop to 5V may help, and so may some added filtering on the 5V, but that (the +12 amd +5 power connections) but I'm thinking that's not be the only path for the noise.

You've essentially got two grounds from the boom box, one from the power ground to the step-down switcher and then to the R-Pi. You've then got the ground from the R-Pi's audio out to the ground at the audio input of the boom box. You might try "lifting" (disconnecting) one ground, then reconnecting it and lifting the other, and see if either of those helps substantially.

Otherwise, I think the solution would be a 12v-to-5V switching supply with isolated output. It's the only way to keep the switcher's and/or R-Pi's significant and varying ground currents out of the audio path (without a lot of playing around with ground paths).