DIY Audio Analyzer with AK5397/AK5394A and AK4490

Here an own measurement, done with the RTX6001, from an genuine 24V laptop SMPS followed by a LM317 regulator circuit bringing it down to 21V. So definitely nothing special.
The RTX input was set to 0dBV:
NT 100OhmLast 0dbV.jpg
 
Hi Demian, Chris,
I have to agree. Switching supplies can be made very quiet indeed. Using test equipment as an example couldn't be more effective.

-Chris

As with many things these days it is also convergence of technologies, namely that modern LDO regs also have much better HF PSRR.

For example, a low noise SMPS feeding LT3045 is a very effective, ultra low noise combination.

T
 
Administrator
Joined 2004
Paid Member
Hi Terry,
For sure! You won't get top performance without proper technique and component parts.

You can seal the switcher inside a box and bring out your filtered DC using feedthrough capacitors (old technology) and directly into your linear regulators. That ought to get you there.

-Chris
 
Member
Joined 2004
Paid Member
Implementation is always the hard part. Typically some small detail in the grounding and return current paths is overlooked and injects noise across the system. Split supplies can be the worst since that center tap represents multiple source and return currents. In one otherwise well designed system I had to reroute a connection around a transformer the long way or stray fields were coupled into the system. All of this was well below the spec performance floor but made a difference in the phase noise.
 
Administrator
Joined 2004
Paid Member
Hi Demian,
Stuff like that isn't normally self evident until you spark a new design up. At least it wasn't several problems all contributing to the mess that is the output. Those designs are best simply simpler to begin again from scratch. That would include most power supplies I run across in "high end" audio equipment. A few will be so bad that all you can do is move on to the next piece.

In a nutshell, I agree with you completely!

-Chris
 
I finally found a Windows PC that I could borrow for a moment, so I could upgrade the firmware of my RTX analyser a few days ago. When I started using the RTX again today, I ran into a few unexpected things:

  • When the software connects to the RTX, the relays selecting the output levels go crazy. They seem to run through a sequence of 5x to 10x switching the the level, until they stop at the level that is set using the front-panel knob. This was not the case with the old firmware.
  • When the software (MATAA) tries to determine the level switch settings, the RTX sometimes returns the wrong level settings. This has worked fine before.

Any ideas what's going on?

Ok, I "fixed" this. The RTX firmware has changed such that it broke the interface to the MATAA software. Jens helped me out by explaining some of the details of the firmware change. I changed MATAA's interface to the RTX accordingly, and all is well now.