So I found out that the design of the power supply has to improve.
I already have the design in mind and in basically improves on PSRR and output impedance over a factor 10 and keeping it’s linearity over the frequencies.
What happened? After working flawlessly on the mini server and sounding better than the LHY pro power supply I now tried it on the RIVO+ where over all the LHY was clearly the winner as the more overall balanced and getting it right.
So it’s (subjective) performance differs greatly between the server and the RIVO+.
The power supply is to blame, not the devices it is feeding.
The difficulty lies in the fact that only with experience it is possible to get a grasp on how the all parameters of the power supply works in combination with what it is feeding because as in the server, it is followed with a switching power supply and also in the RIVO+ I assume, but I don’t have the schematics of the RIVO of course, from the server I do.
I am gaining experience here with a analog (iinear) power supply feeding a switching power supply, and the assumptions I had turned out to be not adequate.
Eating humble pie, learning as we go along…
So how the the LHY compare to the own design on the RIVO+?
The own design made it sound way more analog, very musical.
On the other hand it seemed that the volume was turned down, the bass was less & bad, and the highs seemed a tad grainy and it sounded less dynamic, and last but not least: compared to the vinyl version, but also on the experience how things should sound, and DO sound with the LHY: the sound was completely off with the own design.
the 4th version of the power supply is ready and tested and it is AWESOME!
Both on RIVO+ and the small dedicated server it is outperforming the LHY pro power supply by a large margin.
However I did found out that due to wrong adjustment of bias at some place of the design in the 3rd version, one of the 4 filtering stages wasn’t working (nor doing harm to the rest of the working of the rest of the power supply)
This was an happy accident: if made me strive for a better topology which I found.
The 4th has 10 times lower output impedance while maintaining the wide linear frequency range (which translates of course to the large open-loop bandwidth) and ditto for the PSRR.
Besides that is has some nice small benefits that makes the design more elegant: for instance the way the HF compensation is doing it’s work.
So I am now finally ready to design the PCB for the power supply.
The crux of the whole process is that I found out that the way the quality of design, parts, cables and what have you works out the same on the sound quality as in analog, which I kinda expected/knew some odd 20 years ago.
The reason I ditched 2 times my expensive digital rig was a.o. because this wasn’t understood (enough) by the audio industry. Even the common mode problems, which I am harping about for decades, is more often mentioned and even to a lesser or larger degree addressed in some cases.
So: the time is now better to step into digital, even when it took me only some weeks to design a fully discrete 5V power supply that is as I honestly believe second to none to any of the commercial available units.
Let this sink in: a power supply is nothing else than a power amplifier. Who would accept an high-end power amplifier that uses integrated circuits??
There is a nice and simple abstract way to explain the importance of a power supply, but it is only till so far that I want to share with AI.
Thanks for your attention to this matter.
If I see it fit to do those measurements I will post them.
I always say" I have an Audio Precision, and Audio Precision is what I have.
(because it sounds silly and it is true)
Honestly: perhaps I am too lazy, but I only do measurements if I feel I should do them.
Designing a power amp I do very frequently do measurement with the powerful Audio Precision and other equipment for instance.
With measurements it is only as useful as the interpretation is. Better specs could very well be worse. For me most often measuring only is used to see if something is off, and if everything is working like it is intended or expected to work.
If I may say so you seem to be a bit fixated on measurements. I use it only as one of the tools in the toolbox and I hope to know what en when I should measure and what it means.
Yes i always needs to know if something measures good.
And ‘good’ is when there is no oscillation or something wierd going on in the system.
Maybe not hearable but i wanna know.
And as Power Supplies related, it’s not rocket sience, you only want to know if the output is free of ripple and stays on the output voltage you designed it.
And of course what it does with fluctuation (load), but that depends fully for what it is used to.
So to me the best power supplies have the least ripple/noise floor and handles load very well within the best time frame.
That’s something you only can measure, not with your ears.
But that’s me.
Of course after each modification I use the oscilloscope, precision voltage meter, and different loads and capacitance on the output if it is not oscillating, and if the output is stiff enough.
As said: sure you have to check if everything is working OK.
Ripple will be not measurable in my power supply, it will be way down the noise floor.
And these measurements where taken at the end of the DC Cable with a load.
This supply was fed with an graphene lipo connected to some parallel LT3045.
I think it’s not gonna be better than this.
Sadly i disassembled it and reused some components to make it for my Rivo+.
After several years the lipo was bolted … and you know what that means, although i yearly checked it, the risks are to high if it explodes.
Keep up the good work Frans.
Nice to hear you are happy now into the digital area.
I spend 3 freaking days to design 4 different PCB’s for the power supply.
Still only a second phase prototype setup will be realized, which will be much neater and making further experiments way more easy, and might well be the final setup: no need for making it ultra slick.
They are on their way from China right now.
I received a Plixir 5V 3A power supply for testing that was broken in already: the experience was pretty much the same as with the 5A not broken in power supply.
I’d much rather would at the end buy a power supply than using my own developed power supply, believe it or not. Only of course if the factory build would be as good or better.
I had today for the first time reboot the small server twice because it wasn’t to be found anymore on the network. Arghh!!
measured the power supply: voltage spot on, stable.
Well, they never promised me a rose garden…