[Power Supply] Switching/hybrid vs Full Linear

[dougduck]

You can test this out.
Get a high performing device with an external power supply and swap supplies. You can try linear, low cost SMPS, batteries, etc. RME has several supplies available for their gear, people have even tested them even if I can't find the test someone posted at ASR, there are differences but both definitely not in a transformative way, especially in the audio range. The device retains it's limitations irrespective of the power supply. I imagine some devices will be much more sensitive to power supply than an RME.

Exactly, but the test involves a buck converter -> LDO -> AD/DA converter. It is not simply replacing the external power supply of a device.
We can develop a PCB to test this, but I need free time.
But I'm also inclined to this conclusion, the filter stage of these recent AD/DA converter models should solve most of this issue.

Op amps are another topic. [offtopic]
These tests generally consider a pure, constant sinusoidal signal... music isn't like that...
this is not a criticism of your test, okay? it's generalizing.
a guitar chord has much more of a saturated wave signal! And here the required parameters of the op amp are much more critical!

I am guessing that a lot of the members have a fetish for beefy reserves (aka big*ss Capacitors), as an integral part of our power-sources.

I think everyone at some point has been attracted to beautiful capacitors, in my case, Nichicon MUSE
For the beginner, everything is beautiful, lovely and enchanting.
But with the passing of time and the accumulation of knowledge, the charm ends.

 

[antcollinet]

Among the best DAC models reviewed on the forum is the TOPPING D90 III Sabre, a switching-mode power supply-based/hybrid device that performs as well as the SMSL SU-10, which is a full linear PS.
Is the performance of DC-DC converters like the TPS5430 in the D90 really so good that a single LDO in the last stage can suppress all switching noise?
What do you guys think?

The measurements don't lie. Obviously the noise is suppressed since it doesn't appear at the output of the DAC. Regardless of how that is achieved.

 

[MaxwellsEq]

I know this is off-topic for an audio forum but can you explain this? Is there something specific about the signals or even the noise being processed in medical gear? Why might this translate to audio?

They also use AD/DA converters, but for higher frequencies and lower-amplitude signals, which makes implementation more difficult, as small uV impairs the conversion and alters the result.
With this type of equipment, signal fidelity is more important than the best-looking signal

Which might make designing a suitable SMPS not economically viable for this specific use-case. I'm sure it could be done, but probably not worth the effort.

 

[antcollinet]

These tests generally consider a pure, constant sinusoidal signal... music isn't like that...
this is not a criticism of your test, okay? it's generalizing.
a guitar chord has much more of a saturated wave signal! And here the required parameters of the op amp are much more critical!

Have a look at the multi-tone test in the time domain - and tell me how it is sufficiently different from music to impact any test results. And that is *without* considering the superposition principle.

 

[MAB]

These tests generally consider a pure, constant sinusoidal signal... music isn't like that...
this is not a criticism of your test, okay? it's generalizing.
a guitar chord has much more of a saturated wave signal! And here the required parameters of the op amp are much more critical!

This is often claimed. And if true, I would expect to commonly see stuff in multitone tests that I do not see in tests with pure tones at various frequencies. But I don't, barring rare corner cases. So I typically don't post that data. In fact, pure tone is more discriminating, more brutal on the gear, and easier to hear if that tone is in the middle of our sensitive hearing band, like 1 kHz or so.

Edit: Fixed a sentence.

 

[Mnyb]

You can also clearly see in the measurements of the Topping D90 that there is no ”noise” to combat , what there is is exceptionally low in level

 

[JSmith]

These tests generally consider a pure, constant sinusoidal signal... music isn't like that...

... link is timestamped @35.26, where multitone testing discussion starts.

Multitone Analysis with APx500

Discover how multitone testing with APx500 works, offering fast, accurate audio measurements, and its advantages for production, broadcast, and internet testing

www.ap.com

JSmith

 

[Snickers-is]

yep! agree.
But I don't think I expressed myself correctly.
Are you a designer for any brand?
At no point did my question mean to impose a topology as good or bad; that's not the point.

Yes, I work for a consulting company, mostly involved in loudspeakers, amplifiers and DSP.

I understand, I just think of this as useful generic info for people, a kind of counterweight to all the messed up ideas about old fashion power supplies somehow being superior.

As you mentioned: "However, if what you are looking for is the best possible performance, you would not make an SMPS at a budget."
I agree, and that's precisely what impresses me when I see these published results.
Seemingly simple implementation of traditional converters.
So what's truly responsible for these excellent results? Just a good PCB layout? That's what I want to understand.

Oh, that is a really big question. But if you just need something for a low consumption circuit, I would just use an off the shelf converter with low radiation and high impedance between input and output. You typically get that from almost any of the major brands these days. Then you should focus on the output impedance to load, because the noise that components see are mostly generated by themselves. If you can add the right set of capacitors close to all the parts on the PCB you improve overall performance. All the consumers will then have their own tiny capacitor bank, and the requirements for the PSU itself will not be that critical. If you need something with stable low noise regulators, use some of those with separate probe pins and make sure the PCB paths have low LC so you do not end up with oscillation.

 

[JSmith]

Yes, I work for a consulting company, mostly involved in loudspeakers, amplifiers and DSP.

So do you post here as that consultant for the Company, or as an private individual?

Rules for Audio Dealers, Manufacturers, Members with Commercial Interest. This includes any Guest Contributors/Reviewer with Commercial endorsement

Audio Science Review Rules for Dealers, Manufacturers, Reviewers and Members of Industry. We encourage participation from members of industry in the forum. So please feel welcome here. We expect you to post with highest levels of professionalism. You are representing an industry and it is...

www.audiosciencereview.com

JSmith

 

[KSTR]

Both switching regulators and linear regulators have evolved over time and when the circuits are designed properly there is zero noise penalty. Competent PCB layout is one key point and one that goes wrong often. Low power switchers run at MHz frequencies these days and that requires expert layout to avoid EMI/EMC issues.

 

[Snickers-is]

So do you post here as that consultant for the Company, or as an private individual?

Rules for Audio Dealers, Manufacturers, Members with Commercial Interest. This includes any Guest Contributors/Reviewer with Commercial endorsement

Audio Science Review Rules for Dealers, Manufacturers, Reviewers and Members of Industry. We encourage participation from members of industry in the forum. So please feel welcome here. We expect you to post with highest levels of professionalism. You are representing an industry and it is...

www.audiosciencereview.com

JSmith

I post as a private individual.

 

[dougduck]

Yes, I work for a consulting company, mostly involved in loudspeakers, amplifiers and DSP.

I understand, I just think of this as useful generic info for people, a kind of counterweight to all the messed up ideas about old fashion power supplies somehow being superior.



Oh, that is a really big question. But if you just need something for a low consumption circuit, I would just use an off the shelf converter with low radiation and high impedance between input and output. You typically get that from almost any of the major brands these days. Then you should focus on the output impedance to load, because the noise that components see are mostly generated by themselves. If you can add the right set of capacitors close to all the parts on the PCB you improve overall performance. All the consumers will then have their own tiny capacitor bank, and the requirements for the PSU itself will not be that critical. If you need something with stable low noise regulators, use some of those with separate probe pins and make sure the PCB paths have low LC so you do not end up with oscillation.

Somehow, I noticed
perhaps because of your politeness and congruence.

Thank's for your attention and answer.
That was a very helpful explanation.

 

[Sokel]

Somehow, I noticed
perhaps because of your politeness and congruence.

Thank's for your attention and answer.
That was a very helpful explanation.

If you're interested using actual music (any kind) or noise or anything close, apart from the suggestion above for multitone signals you can also use the FSAF measuring method, here's a couple of links:

FSAF Measurement

REW Fast Subband Adaptive Filtering (FSAF) measurement

I just discovered this new measurement option in REW. AV Nirvana Thread REW Guide I have not been able to study enough yet to understand what it is for and what advantages it can provide compared to the sweep. It would seem that it can produce valid TD+N measurements for complex signals such...

www.audiosciencereview.com

It's fun using it but results need proper interpretation as always when various levels are involved.

 

[Nzama]

This may come as a shock to you, but with audio devices, we don't listen to the power supply.

Oh, and by the way, let me extend a warm, Hank Scorpio welcome to ASR.

View attachment 483393

Indeed. We measure 1khz SINAD
AND THAT IS ALL YOU WILL EVER NEED
/sarcasm off