SQ difference - Raspberry Pi vs x86_64?

[hkphantomgtr]

What have you read? I'm seriously considering it as a next stage for my server.

It's a great deal easier than messing with the clocks 

The Pi5 was designed to boot off microSD, requires a tweak to boot off USB if not using the official power supply, and requires a hat plus further config tweak for NVME. Maybe the issues were people's initial efforts, but I read that some NVME were not compatible ???

Some NVMe drives (mainly WD) are not compatible with most HATs.

---

What have you read? I'm seriously considering it as a next stage for my server.

It's a great deal easier than messing with the clocks 

Right, yet in my experience, the only significant improvement is replacing the crystal with ocxo clock. 

Battery power supply, adding power filter hat, vibration isolation, lan cable galvanic isolation, all help.  But none can compare the power of changing the crystal clock.  And 1 +1 is bigger than 2.

Got any good news? 

Thanks for that, though.

In recent days testing, I found that the sound quality varies with the source of power (of course)..... guess what?

Powered my pi 5 set solely by battery is the worse scenario.  It lost the most in density, the dynamic, and the focus.  It really took me by surprise.

The other 2 contesters are 1) powering by LPS directly to the filter hat, and 2) powering by LPS to the battery hat (including compulsory and automatic battery charging).  I think the result is needless to say.  The difference in sound is even more obvious.  Yet the sound of 1) is a bit strange to my ears.  I need to find out the problem and solution, e.g. detaching the battery hat, relocating the speakers.  I'm afraid I only can do this not earlier than this Sunday.

And powering the battery hat by 5V LPS and 12V LPS with same DC cables also make the difference.  5V wins quite clearly.

I know a few people who say battery power leads to disappointing SQ. I always use a Linear Power Supply of some sort. I've been scared off of batteries! 

I think we need some batteries which outputs 5.0v DC natively to re-write the story.

Btw, coincidentally, tonight I have measured the current my pi5 set draws.  In most of the time it's drawing around 3.0A.  Let's refresh, that pi5 set includes charging batteries, a 4G memory pi5, a 10M clock, a filter hat and a 2T NVMe hat.  Digital output is via USB.

[uglymusic]

The Pi5 was designed to boot off microSD, requires a tweak to boot off USB if not using the official power supply, and requires a hat plus further config tweak for NVME. Maybe the issues were people's initial efforts, but I read that some NVME were not compatible ???

Some NVMe drives (mainly WD) are not compatible with most HATs.

---

Right, yet in my experience, the only significant improvement is replacing the crystal with ocxo clock. 

Battery power supply, adding power filter hat, vibration isolation, lan cable galvanic isolation, all help.  But none can compare the power of changing the crystal clock.  And 1 +1 is bigger than 2.

Got any good news? 

Thanks for that, though.

In recent days testing, I found that the sound quality varies with the source of power (of course)..... guess what?

Powered my pi 5 set solely by battery is the worse scenario.  It lost the most in density, the dynamic, and the focus.  It really took me by surprise.

The other 2 contesters are 1) powering by LPS directly to the filter hat, and 2) powering by LPS to the battery hat (including compulsory and automatic battery charging).  I think the result is needless to say.  The difference in sound is even more obvious.  Yet the sound of 1) is a bit strange to my ears.  I need to find out the problem and solution, e.g. detaching the battery hat, relocating the speakers.  I'm afraid I only can do this not earlier than this Sunday.

And powering the battery hat by 5V LPS and 12V LPS with same DC cables also make the difference.  5V wins quite clearly.

I know a few people who say battery power leads to disappointing SQ. I always use a Linear Power Supply of some sort. I've been scared off of batteries! 

I think we need some batteries which outputs 5.0v DC natively to re-write the story.

Btw, coincidentally, tonight I have measured the current my pi5 set draws.  In most of the time it's drawing around 3.0A.  Let's refresh, that pi5 set includes charging batteries, a 4G memory pi5, a 10M clock, a filter hat and a 2T NVMe hat.  Digital output is via USB.

OK. That sounds about right, in a way. Rule of thumb and all that.

[Snoopy8]

@hkphantomgtr interestingly, Google AI gave me this reply

Batteries typically exhibit a slower dynamic response than linear power supplies in audio applications, leading to a perceived slower, less dynamic, and less articulate sound. This difference arises from the inherent characteristics of both power sources and how they interact with audio circuits

@uglymusic, previously, I was concentrating on improving the end point more so than the server.  But once the end point has reached a certain level, then changes to the server become more apparent.  And my reported comparisons between the Pi5 and NUC have all been as the server (running LMS).

Any comparison between my Pi5 and NUC as the end point will be unfair.  My end point NUC has a JCAT USB XE card, with its own linear supply, CamillaDSP running Audiolense convolution room correction.