• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Do you need linear power supply for DACs?

bennetng

Senior Member
Joined
Nov 15, 2017
Messages
464
Likes
312
#81

miero

Active Member
Joined
Aug 1, 2018
Messages
114
Likes
81
#87
Although ultrasonic cannot be detected by human, but these signal will be passed to amplifier and then loudspeaker, would they have any negative effect on the final signal?

At the end of the day, we don't hear the waveforms the DAC produced. We hear the waveforms loudspeaker produce. We don't know whether these ultrasonic peaks will affect speaker yet.
Yes, it might have a negative impact. For example Hypex nCore class-D amplifier is sensitive to frequencies close to multiples of its switching frequency:
- https://www.diyaudio.com/forums/class-d/321632-hypex-ncore-nc400-input-anti-alias-filter.html
 

Purité Audio

Major Contributor
Industry Insider
Barrowmaster
Patreon Donor
Joined
Feb 29, 2016
Messages
2,847
Likes
742
Location
London
#88
I agree but, everyone on this forum enjoys good engineering and Miska’s filters are good, look at the latest Archimago blog.
It is just that the improvements are inaudible.
Keith
 

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,182
Likes
2,449
Location
Perth Western Australia
#91

miero

Active Member
Joined
Aug 1, 2018
Messages
114
Likes
81
#92
@March Audio Yes, but a bare NC400 module already contains a buffer, which does not filter this out properly. And I don't expect that many DIY nCore builders implemented a custom buffer for NC400 (me neither).

I expect that in your amplifiers the custom buffer filters this out properly, because for OEM modules it is required. Maybe you can show us some measurements in your thread... :)
 

bennetng

Senior Member
Joined
Nov 15, 2017
Messages
464
Likes
312
#93
It seems that no one read Archimago's graphs clearly.

No upsampling:
44.png


DSD256:
dsd256.png


There were discussions about the noise are caused by ADC aliasing. But think about it, if DSD256 improves out of band noise then aliasing should be minimal, recording at 768k should ameliorate aliasing as well. Then why the noise floor is still so high?
 
Last edited:
Joined
Feb 24, 2019
Messages
93
Likes
122
#94
I have to be honest and write I’m having trouble keeping a straight face when I read about these minuscule differences in signal processing. Just from an engineering point of view it’s great to have some measurements.
Apart from those with exceptional hearing and apparently trained listeners (really, who wants to listen to their stereo to pick out differences in signal processing. Most people listen to the illusion of music.
Anyway, just say you manage to get an electrical signal that perfectly represents whatever is on the chosen media; at some point in order to be audible there has to be a conversion from electrical energy to mechanical energy. We do this by making a suspended cone, or plate vibrate in most instances. These vibrations create a pressure wave and when it reaches our ears and then our brain it may sound like what we call music.
So for example, take a bass driver. Usually it’s a cone made of a material that needs to be light enough to move without massive amounts of power. A perfect cone would not flex or distort on any part of it’s surface. They all do though. We can make reasonably uniform cones but they all are imperfect. These cones are suspended, in a frame on rolls of rubber at the front and often what is in effect a paper spring at the back. I’ll ignore the problems associated with the magnet, the pole piece and the air gap and the varying impedance of each drive unit. There are good matches but I have yet to see a perfect pair. The boxes we enclose these drivers in vibrate, act as capacitors in some cases, the internal pressure waves often interfere with each other. With some the baffles flex, the material densities of the materials used have variation. The temperature of the air effects the suspension of the cones etc etc and then we put these boxes in a room which sets up all sorts of nodes and refections that effect what we hear.
Yet, here we are, or some are, worrying about whether what to most is an inaudible amount of jitter and distortion effects the sound we hear.
It doesn’t matter how perfect you get the electrical signal once it reaches the conventional loudspeaker in a room and the conversion from electrical energy to sound waves happens all that perfection in the signal gets swamped by the imperfections in the loudspeaker.
Thankfully the brain can still piece the vibrations together and make it sound like music.
So, for Miska in particular who recounts years of listening for submarines, I’ve got something for you to think about. I used to work in R&D that made avionic equipment and the problem was rarely distortion or imperfect signals, the problem was how the brain interpreted what it received. Basically the brain ‘hears’ what it wants to or expects to hear. It’s well know problem in military communications and is equally applicable to audio reproduction for music. You hear what you expect to. Ask your wife if you have one; she’ll explain it to you.
 
Joined
Jan 21, 2019
Messages
13
Likes
3
#95
Do you think the difference in distortion of the 2 plots you posted is audible?

Yes or no?
I think it is audible. The second plot has distorsion artefacts that resemble some form of clipping. Higher order harmonica is bever a good sign. Especially when they do not show an even decay.
 

johan

Member
Manufacturer
Joined
Feb 23, 2018
Messages
64
Likes
80
#96
Hi all

I will throw my hat into the ring.

I just finished testing our LPS and SMPS and we get very good numbers (of course we will kindly ask Amir for outside testing)

My question is , at what power was the testing done ? There is always a difference in noise depending on power.

Second , in house testing we are able to see a difference of at least 1db in THD+N of DAC (tested on Katana stock) . Mostly its because the dominant N in THD+N (Katana)got reduced and same effect is seen on our other DACs (Boss etc) but THD+N is unchanged since THD is predominant .

Our testing is done with single ended cables (for power) we are making setup to test with balanced .
 

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,182
Likes
2,449
Location
Perth Western Australia
#97
@March Audio Yes, but a bare NC400 module already contains a buffer, which does not filter this out properly. And I don't expect that many DIY nCore builders implemented a custom buffer for NC400 (me neither).

I expect that in your amplifiers the custom buffer filters this out properly, because for OEM modules it is required. Maybe you can show us some measurements in your thread... :)
Diyers can still add input filtering. Plus you must remember that the signals shown here have been at - 100dB. So you need to put the potential impact into perspective.

A theoretical problem that can be stimulated by contrived conditions doesn't mean that in normal use a problem is encountered.
 
Last edited:

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,182
Likes
2,449
Location
Perth Western Australia
#98
I think it is audible. The second plot has distorsion artefacts that resemble some form of clipping. Higher order harmonica is bever a good sign. Especially when they do not show an even decay.
It isn't. Those higher harmonics are at around - 130 dB. Totally inaudible. I can make a file with just those harmonics and no fundamental for you if you want to test it out for yourself.
 

Superdad

Active Member
Joined
Aug 6, 2017
Messages
102
Likes
48
#99
Do you need an external power supply for dac.jpg

The supplies don't have current meters so I don't know what they put out. I could try to tap in the middle but have to wire up a fixture and I am too lazy to do so. :)
You're kidding right? Because that DC electronic load box in the above pic (exact same model sold by the original mnfr. ITECH for about half the price than BK charges) makes load capability and voltage measurements child's play. No middle of the cable tapping--simply plug the output of the power supply into your BK box.
See the two oval buttons marked ENTER and ON/OFF? Dial in--or direct entry on the numeric keypad--some current amount, press ENTER, then press ON. With the load in CC (constant current) mode as you show in the pic, the machine will read out voltage and current. Dial up the load until the voltage starts to drop and you will have your answer as to what the DUT is able to handle.

I know that to you, nearly everything is about measuring noise (though you are not equipped to properly directly measure PS noise down into the µV range), but the much more important performance figures of merit for power supplies used for digital components is output impedance, plotted broadband--from about 5Hz to 700KHz. Of course performing that measurement is rarely very straightforward either. :cool:
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
17,430
Likes
14,906
Location
Seattle Area
You're kidding right? Because that DC electronic load box in the above pic (exact same model sold by the original mnfr. ITECH for about half the price than BK charges) makes load capability and voltage measurements child's play. No middle of the cable tapping--simply plug the output of the power supply into your BK box.
I did not have time to use the dc load for these tests. I originally wanted to stress test the supplies with it but then worried it could damage them so I did not go there. At some point with my own gear will put it to use.

Until then, I appreciate keeping these pedantic and insulting remarks to yourself.
 
Top Bottom