AC Filtering, Grounding Boxes, Linear PSU and Balanced Power.

[gstew]

6 hours ago, YashN said:

 

@gstew Some pics added! 

And very nice pix and process. Thanks for sharing!

 

Greg in Mississippi

[gstew]

2 hours ago, YashN said:

A few other things I would have liked to do at the time I made the captures:

 

Since the custom USB connector  - by that time I couldn't even call it a USB 'cable' anymore, as there were components in a circuit computer-side as well as along the GND line - isolated the DAC quite a lot, and this was even enhanced by a custom shielding configuration - I started to think about what I was doing at the capture side.

 

The M-Audio can  be powered two ways: by its own external PSU or by USB (the same very noisy computer USB supply from the same iMac - oh no!).

 

So here, I chose not to use the USB power from the iMac through the M-Audio USB cable, but to power the soundcard by its external power supply.

 

However, this doesn't guarantee me that whatever noise is being transmitted via a generic USB cable from the Computer still doens't affect the card itself and therefore the measurements.

 

I could only assume (and here for lack of time and better equipment) that since the M-Audio is a card used for music composition, so also for capturing Microphones and other instruments ar 24-bit 96kHz max, that the manufacturer did some work for the isolation (its an older USB implementation so doesn't require the more recent and previously inexistent USB 2+ galvanic isolation implementations).

 

Secondly, the M-Audio external power supply I used was also not a Linear one...

 

To make things better for measurements, I would have liked to:

 

1. Have a clone of my Custom USB Connector to use that too for the M-Audio

 

2. Ideally power the M-Audio by an external, low-noise battery or something like that

 

 

Additionally, a few things to notice:

 

- I specifically did not work directly on isolating the data lines. However, filtering computer-side still improves SQ.

 

- The filtering computer-side still works when I A/B it (I installed a switch and a LED) even after I set up a custom shielding configuration for the USB Connector (which by itself already increased SQ too)

 

What this seems to suggest is that the data lines are still affected by whatever is happening computer-side, although I am very deliberately not powering from the computer, and although I also further isolated by working on the shielding configuration.

 

And finally this:

 

At this point in time, I thought I had a really good setup and that any leakage currents were not passing through my custom USB Connector, so I was thinking that if I re-connected my AC Filter in the system, it wouldn't make a big difference.

 

Here again, big difference!

 

This is what allowed me to say last year that I have a long-term plan of avoiding SMPSes in the listening setup: the work needed to really filter out all the nasties with SMPSes is extensive: you need to ensure the interconnects do not propagate unwanted things, you need filters post-SMPS as well as filters AC-side.

 

YashN,

 

All of this sounds good. AND should be  a good set of guidelines and areas of inquiry for those looking to optimize their audio setup.

 

AND to me one of the most important points is starting with really low-noise power supplies, at their output, on how much noise they feed into the AC line, and how they react with other gear (John Swenson's high-impedance SMPS leakage). 

 

One of my guiding principals has been to eliminate sources of noise rather then try to filter or shield from them. My thought is that once you have a particular noise source, filtering or shielding will only reduce its impacts, not eliminate it. The best way to have low noise is to not create or introduce it in the first place.

 

So your long-term plan of avoiding SMPSs in my setup is something I've worked hard to achieve... and with just recently moving to amplifiers that don't use SMPSs, I've gotten there. It was good. AND well worth the effort and expense to get there.

 

Greg in Mississippi