Linear PSU, significant improvements or...?

With a bit of luck I’ll have my custom built linear PSU’s tomorrow. There are two identical units mounted on top of one another using standoffs. The pic is of a test build. If testing shows it makes a definite and appreciable difference it’ll be put into a case and it’ll become a permanent member of my audio equipment.

Following the build it was tested to ensure it conforms to specification. Whilst it’ll only be required to deliver 5V 2A it’s capable of a lot more, so for the purposes of testing performance the output was set to 25V DC using a 25 turn trimpot. The unit was given 15 mins to warm up and then measured using an Agilent / Keysight 34461A multimeter.

Measurements over a period of 5 minutes:
Minimum Voltage: 25.0342
Maximum Voltage: 25.0350
Average Voltage: 25.03457
Standard Deviation: 0.00018
Measured over 1101 samples

The difference between the minimum and maximum is 0.0008 Volt. Divided by the average voltage of 25.03457 that gives an absolute regulation figure of around 0.0032%

A Keithley (Tektronix) 2231A triple output precision power supply (list price $687) measured using the same method and equipment yields:

Minimum Voltage: 24.9877
Maximum Voltage: 24.9898
Average Voltage: 24.98928
Standard Deviation: 0.00046
Measured over 1007 samples

The difference between the minimum and maximum is 0.0021 Volt. Divided by the average voltage of 24.98928 that gives an absolute regulation figure of around 0.0084%

That’s a 262.5% improvement on the Keithley precision bench supply.

Whilst the designer has some pretty serious measurement equipment, he was unable to measure ripple and noise on the line with the equipment on hand - it is below the noise-floor of the measurement equipment… which is to say that it is impressively tiny.

Now that I’ve established and confirmed the LPSU’s credentials I’ll be using it to ascertain whether there is indeed any merit in adding a really, really good LPSU to the audio mix.

One will power the Roon Bridge device and the other a custom USB cable that carries data only from the source and requires independent power in the event a DAC’s USB port requires power from source.

As my listening room has been decommissioned I won’t be able to use my ATC’s, so I’ll resort to the following:
Source: Roon Core > Roon Bridge on ODROID-C2 > USB > DAC
DAC: Yggdrasil
AMP: Schiit Jotunheim HPA
Cans: Sennheiser HD800 with balanced cable.

Hopefully I’ll be able to do some testing on the weekend, and with a bit more luck I’ll have some exotic USB transports to test a little later as a comparison. If I get to compare some exotic USB transports I’ll measure what comes out the business end of the DAC.


my lab measuring tools consist of:

  • a mirror
  • a permanent felt tip marker
  • my grin

whenever I add/change something in my system I:

  • play some some music I like
  • take the mirror and look at my grin
  • using the marker I mark on the mirror how large my grin is

if it’s larger than previous marks the change stays, if it doesn’t it goes
and, back IT… every time I changed to a better PS my grin was definitely larger :slight_smile:

YMMV :stuck_out_tongue:


Yip, expectation bias always creates a bigger and better smile.


I’m very interested to hear about your findings.

Too bad it’s not so easy (or doesn’t appear so to me) to hook an external LPS up to the Yggdrasil. Both my main endpoint and my DAC have provision for easy attachment of an external power supply, so I’m in the position of contemplating where an LPS would provide more bang for the buck.

The part I’m most interested in is whether powering my transport using the LPSU instead of the supplied switching PSU is going to make any audible difference when listening via the HD800’s. Thereafter I’m hoping to compare some “audiophile grade” SOC’s against my ODROID-C2. I’ll listen first, and if there’s time before I leave my current home for good I’ll measure the output of the business end of the DAC.

so… basically you will use, first and maybe only, your own variant of my “mirror and felt tip marker” method :wink:

think I should patent it :stuck_out_tongue:

1 Like

Evan, what kind of measurements will you make?

I’m curious only in that of the articles I’ve read, decent measurements that could reveal the differences need pretty high end kit (and training on it) - like an Audio Precision. I know dCS were allegedly pioneers in Fourier analysis type measurements, and I found some of their discussions mind-boggling, like how do you take that analysis and know what’s important in terms of what we hear.

I’ve often wondered about measuring (never hurts to cross check the ears with a bit if science), but just assumed it was impossible without super-expensive kit and lots of time…

Certainly it would be nice to see lots more measurements to back up claims, and give consumers a bit more info.

I use KECES linear power supply DC-116 to power the NUC loaded with Roon and the other NAS. There’s some improvement compared with brick wall switching power supply that come with the NUC. Noticeablely, I heard a clearer presentation overall; with more details in a more quiet background.

Same multimeter. What I’d be curious to see is whether given the same input there is any appreciable change in voltage output from the DAC, with the only change between measurements being use of a different SOC. Similarly, using same SOC, does changing PSU from the wall wart to the lpsu yield any appreciable change in voltage output from the DAC. I’ll defer to the chap that designed my last custom DAC but I suspect it should suffice as I’m not trying to ascertain DAC linearity, merely looking at what it’s sending to the preamp

I’d love to see meaningful specs published for LPSU’s and SOC devices claiming to deliver significant improvements. Unfortunately they’re rarely forthcoming.

Didn’t get around to collecting LPSU, but here’s the USB header that doesn’t carry power from the source. Add USB cable of your choosing.

Would you be able to share your plans or supplier? I’m looking to put together a 20VDC supply for my phono amp.

At present there’s only the single prototype as as a standalone PSU. It’s been designed and built by a friend who is using it in a commercial application so I suspect he’d be reluctant to share the design.

For comparison here’s the standard ODROID smps under no load then with a 1A and 1.5A load respectively . It’s highly unlikely it would be capable of delivering 2A on a sustained basis lest the magic smoke be released.

Back to no load:

We decided we had a better use for the SMPS pigtail

Prototype ready for deployment (the USB connectors will remain unused at this stage):

@evand where are you sourcing the paladin boards from?

The design is a friend’s. Not sure who manufactures the PCBs for him, it’ll be a local crowd. He’s toying with the idea of commercialising completed LPSU’s so you may be able to purchase in time. The board you see is used in an amplifier design and is completely over the top for powering USB and/ or a SOC.

I’ve had a little time to do some listening, so I’ve started with using the LPSU to power only the USB input on one of my Vivere DACs whilst powering the other Vivere’s USB input directly from the ODROID C2.

Given the Jotunheim only has one SE input and one XLR input pair, I had to connect one using SE inputs and the other XLR.

Config was as follows:
Wall wart > ODROID > Vivere1 USB input > Vivere1 balanced output > Jotunheim
Wall wart > ODROID > USB header powered by LPSU > Vivere2 USB input > Vivere2 SE output > Jotunheim

Volume level was matched by lowering Vivere1’s output.

Listened to redbook, 24/96 and DSD, switching inputs on the Joty between balanced and SE. For the life of me I could not hear any difference at all, it’s like the input switch on the Joty is a placebo.

Interesting and disappointing at the same time. :confused:

What if you leave the two chains the same but switch just the balanced vs. single-ended part? (I’m hypothesizing that the LPSU could be bringing the SQ of the SE output more in line with the balanced, but if you switched, there might be a more noticeable difference.)

… not sure your friendship with the guy who makes those LPSs will last :laughing:

That’s the next thing I’m going to try. Balanced isn’t necessarily better by definition, especially not on short runs. Nevertheless, I’m keen to ascertain whether I can discern any difference. Dropping the output of the Vivere using balanced outputs may have put it at a disadvantage. I’ve got some inline attenuators I’ll try for volume matching as well…

Later I’ll test with the LPSU powering the ODROID and ODROID powering the USB and 1 x LPSU powering ODROID and another LPSU powering USB.

1 Like