Mini-Noise Nuke Measurements

First, apologies to @atomicbob for using his naming. Maybe Noise Nuke JR, or Noise Nukelet? Anyway hopefully he will forgive.

This is about some testing I was doing with a new project. Sadly this product does not have the space for a large inductor and caps. So I set about trying to see what benefit a smaller setup would give me. The design is the same as the noise nuke, but smaller values as we only need to deal with 12V input.



The control signals are to allow the Raspberry PI to hold power after boot then release when safe to do so. This is what we have used on our Mercury Streamer. It's a bit imperfect since it requires software to be partially booted before you can safely "turn it off". Anyway that another topic!

Using an AP SYS-2722 Analyzer (second only to their new AP555) we looked at the input from a 12V Jameco PSU, then the output from the filter.

Input:


Output:


Output scaled 100x: (EDIT 10x)


So the 2.3mv peak is reduced to 20uv, greater than a 100:1 reduction! Note that the first spike is not relevant since we are measuring an AC signal riding on 12V DC. The analyzer does not let me filter out the DC component from the display. And what little noise there was in the rest of the frequencies is gone.

This is what the circuit looks like on the PCB:



L2 is the blue inductor in the upper right and the 4 caps are the left of that,

Total cost at qty 100 is ~ $10.

 

I am currently rocking with an older Pi2Aes streamer. I've been eyeing the Mercury streamer for a while, but how long until the new version 3 is out?

 

An ignorant question here.. but could you skip the soft ware and simply have it on a timer?

 

Waiting on the first proto due here soon. Although all the parts were available getting them all in house was still took a fair amount of time. But we did buy enough for 150 units so once we pass the first unit we can get the rest built up much faster. The plan is to start accepting orders once we have the tests complete, then delivery in 2-3 weeks after that. Likely late December into early January.

 

You're not wrong, but what would the right timeout be? It may be something we look at doing down the road if customers have issues with it.

 

It looks like you're running a common mode choke (the Triad CMF23H-103231) with the windings in series? I'd be cautious with this sort of arrangement as these chokes are designed so that DC flows through one winding and returns through the other to cancel out any flux generated by the DC. Here's a picture I stole from Coilcraft that explains it a bit better.



With the windings in series, the DC is able to generate flux which will likely saturate the core and cause the inductance to drop off.

I'd consider retesting with a 210uH inductor or a 200 milliohm resistor, to match the leakage inductance or DCR of the Triad part you're using. It would also be prudent to test with the same current draw that you expect your load to take.

 

we have tested at 5 amps. You are correct that the inductance drops a bit, but remember we don’t care about the exact value, only the reduction of noise. To us it’s a solid compromise between cost, space and results.

 

Is there that much variance in the boot-up times? If not, couldn't you just set it at the longest plus a second or two?

 

The new Mercury V3 with built in mini noise nuke is available for ordering now. V3 is now $599 which is $200 less than V2. Delivery is 2-3 weeks as we finally get the metal cases that caused the month long delay.

https://www.pi2design.com/store/p29/Mercury_V3_Streamer.html

 

Awesome! Any internal shots you can share?

Edit: Thank you!

 

Michael,

Very nice! I wish more manufacturers did something similar, but unfortunately, it's uncommon.

For the last several years, I have been using a DIY power supply designed by JeanPaul over at diyAudio. It uses CLC filters with common mode chokes before and after it's chip linear regulator and on both hot and ground. The chokes used are;

https://www.mouser.com/ProductDetail/TDK/ACM1211-701-2PL-TL01?qs=Y0gyZw6qHXe0cQ%2Brx8PtKw==

My reason for bringing this up is it gives similar performance with much smaller and, possibly cheaper, components. Something to play with for the future.

Keep developing and refining. We really appreciate you and your products.

Jac

 

They are cheaper, but they have much lower inductance. So when they reach their full saturation value, the inductance is even lower than what we used. The part that we used from Triad is pretty reasonably priced at a few dollars, so not a big money burden.

BTW I have used the TDK parts on other projects for Straight common mode use.

 

I actually do have some. Here: