Power supplies ripple _ does it must be really low ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi !
I have decided to post here because I have a very undefined question.
I see big efforts to reach the uV in ripple for power supplies for low signal units like dacs and preamps.
I wonder if this is really needed.
I am testing some smps Ac adapter that are likely to have quite high noise.
A figure of 1% of Vout for ripple could be quite typical.
For a 12VDC out unit this translates in about 120 mV (I guess peak to peak).
Nevertheless measuring the noise on a usb soundcard powered with them I see very low levels.
I understand that the PSRR of the soundcards acts suppressing the power supply noise.
My question is ...

what kind of improvement I could get from a high end psu considering also that usually these soundcards have dc converters on board ?

could I go with just one of these oem AC adapters as long as the measurements are ok ?
I cannot believe to what I am getting.
Also changing adapter has practically no impact on the resulting noise of the soundcard.
These adapters are very convenient because cheap and small.
Thanks a lot for any advice.
Kind regards, gino
 
Last edited:
Administrator
Joined 2007
Paid Member
The noise from an SMPS is likely to be in the tens of kHz region, probably far above what a sound card could capture... I'm assuming you mean you are using the soundcard to try and record and analyse the noise (in place of a scope).

Hard to say what improvements you may or may not get from a 'high end supply'. Each case would be different.
 
The noise from an SMPS is likely to be in the tens of kHz region, probably far above what a sound card could capture...
I'm assuming you mean you are using the soundcard to try and record and analyse the noise (in place of a scope).

Hi ! thanks a lot for the kind reply.
Yes ! I can see noise up to 96kHz selecting 192kHz sampling rate.
Actually I have no clue f the noise above that can have an impact on the sound quality. Someone says yes. That the noise also in the MHz range can be detrimental for sound quality of a digital device.

Hard to say what improvements you may or may not get from a 'high end supply'. Each case would be different

I see. Problem is that I am limited to 96kHz. The graphs look quite clean to me and the sound is ok.
As the space available inside the unit is limited using an ac adapter is the only way to avoid the need for an external power supply.
Thanks a lot again.
Kind regards, gino
 
The noise is likely to be so high that you can measure it on a scope.
Compare to the output of the reputedly noisy lm317/7805 to see just how bad an SMPS can be.
Compare to a Zener+EF to see what quiet looks like.
Then think what a LED+EF might look like. This and the previous will be very difficult to meausre with our normal amateur limited equipment. A very quiet LNA may be enough to give a decipherable scope pic.
 
The noise is likely to be so high that you can measure it on a scope. Compare to the output of the reputedly noisy lm317/7805 to see just how bad an SMPS can be.

Hi ! thanks for the useful advice.
what I am using is Arta software to check the cards noise floor.
So I am not measuring "directly" the psu noise but the noise floor of the card powered by the psu , that is what matters me more in the end.
A very cheap smps brick and a linear decent power supply with more or less 1mV RMS noise give the same noise floor. Same result.
My guess is that the soundcard has a very good PSRR.

Compare to a Zener+EF to see what quiet looks like.
Then think what a LED+EF might look like.

Are these psu ? are they described in the power supply section ?


This and the previous will be very difficult to meausre with our normal amateur limited equipment.
A very quiet LNA may be enough to give a decipherable scope pic

I know that low noise measurements are very challenging and out of my reach.
I suspect I am doing wrong measurements.
These adapters are very very cheap. One is a refurbished Dell unit.
It looks like it was recovered from a garbage bin.
Thanks a lot again.
gino
 
Last edited:
The noisiest linear PSU you are likely to build would be using a standard 3pin regulator.
Do one and compare the noise (scope trace) from that to any smps and see which is worse

Thanks a lot again for the very helpful advice.
My main doubt is about my testing procedure now, because I have already tested different PSUs that must have very different residual ripples, something like one 50 times more than the other.
However I will try others during the weekend and in the next days (I should receive others soon).
And the noise floor of the usb soundcard was looking the same.
To be perfectly honest all other cards noise floor looks quite independent from the psu used.
I have even compared a linear and decent 5VDC with a phone charger ... practically the same result.
Unfortunately and strangely enough the only one psu that has clearly more noise issues is the stock one :eek: .
From this the idea to replace it.
Even just a generic ac adapter has turned out to be better :eek:.
And this unit is intended for the professional use ... how strange.
I will attach 2 graphs obtained using different psu to explain my point better.

However I cannot say anything of the noise above 96kHz.
And this could be the range where the better psu give a benefit.
This is outside the audio range but the question is how much this high Hz noise (i.e. above 100kHz) can impact the sound.
I do not have a clue.
Thanks a lot again, gino
 
Last edited:
Hi !
just to add that I think I have found an important confirmation about the suitability of smps for digital units.
This is a high quality/precision master clock generator by Apogee and
it is powered by what looks like a MeanWell smps (I do not know which series by the way).

DSC_0082+%25282%2529.JPG


If smps are fine for them they can be very fine also for me.
I have just to check the grounding of the stock one. I read on a Meanwell datasheet

for better EMC performance please secure a good connection with the chassis grounding

it would be funny if it were only a smps grounding issue.
Thanks again for all the valuable advice.
Kind regards, gino
 
Last edited:
..............
If smps are fine for them they can be very fine also for me.
I have just to check the grounding of the stock one. I read on a Meanwell datasheet



it would be funny if it were only a smps grounding issue.
Thanks again for all the valuable advice.
Kind regards, gino
The smps should be enclosed inside a screening box and that box should be directly attached to the enclosure.
The filter between the mains input and the SMPS should also be inside a screening box and that too should be directly attached to the enclosure.

Apogee have omitted both !
This will allow emi to back feed into the mains and contaminate all your other equipment.
It also allows the unscreened emi generators to spray everything inside the enclosure with interference.
Don't just copy them because they have a "name".
Find out how to do it right and implement that.
 
Last edited:
Then get the scope switched on and have a look.

Hi! yes but ... I really need to have a look ? I mean, will noise in this range (>100kHz) impact on the card performance ?
because someone say absolutely yes and other say absolutely not. With digital I mean.

It will show noise upto and a little beyond it's specified bandwidth.
Do you have a 30MHz, or 50MHz, or 100MHz scope?

No but as I am very interested I could search for one to buy.
But I would like to be sure that I have really to look above 100 KHz. My only present interest are usb soundcard or dac.
And I am already in a mess ...

The smps should be enclosed inside a screening box and that box should be directly attached to the enclosure.
The filter between the mains input and the SMPS should also be inside a screening box and that too should be directly attached to the enclosure.
Apogee have omitted both !
This will allow emi to back feed into the mains and contaminate all your other equipment.
It also allows the unscreened emi generators to spray everything inside the enclosure with interference.
Don't just copy them because they have a "name".
Find out how to do it right and implement that

Thanks a lot ! I feel this is a very important point. Also in my unit the smps is completely exposed (I do not see a filter by the way). Not the right installation I understand.
Very important for me to fix some basic rules.

Going back at the scope issue again I am not sure that looking above 100kHz is really needed.
This is also another very important point for me to understand. In the case I will look for a scope.
I would like to be able at least to compare and select power supplies.
It should not be extremely difficult with the right equipment.
Thanks a lot again for the very valuable advice.
Kind regards, gino
 
Last edited:
Not difficult??!

I spent a week (and it cold have gone longer) qualifying a replacement 12V 5A line lump for a fairly run of the mill product we manufacture, power supplies are in no way as simple as they appear.

One of the issues I had to deal with was a flyback type supply where the ringing occurred at close to the frequency of an on board clock feeding the audio converters, it showed up very clearly on the DScope III and was actually audible (Note the problem was NOT in the audio band, it was up around 25MHz), THD and silence looked fine, two unequal tones not so much.

So yes, sometimes the stuff above the audio band DOES matter.

A scope is more or less mandatory if you want to do this stuff, but like all tools you also need to know how to interpret what it is showing you, there is no substitute for understanding.

Regards, Dan.
 
Not difficult??! I spent a week (and it cold have gone longer) qualifying a replacement 12V 5A line lump for a fairly run of the mill product we manufacture, power supplies are in no way as simple as they appear.

Hi ! thanks a lot for the very kind and valuable advice.
At least i have established one point ... they are indeed very difficult :eek:
To be honest i suspected that.
I have no hope to design anything. A very small one to build something.
And a little bigger one to use one.
I have a usb soundcard with a noisy smps and this is confirmed using a spectrum analyzer software.
I am almost sure i will use an external one (this is a very common solution and now i understand very well why).
Placing a smps (also not shielded) very close to the circuits is a recipe for disaster.
I will place a dc socket on the back and i will do with that.
I have already noticed that changing external psu has little to no effect on the noise floor of the soundcard.
As i am using cheap smps with ripple around 1% of the Vout=12V this should mean that the circuit has a very good PSRR.
I am getting low noise floor when input gain is set to zero (graph attached).

One of the issues I had to deal with was a flyback type supply where the ringing occurred at close to the frequency of an on board clock feeding the audio converters, it showed up very clearly on the DScope III and was actually audible (Note the problem was NOT in the audio band, it was up around 25MHz), THD and silence looked fine, two unequal tones not so much.
So yes, sometimes the stuff above the audio band DOES matter.

This is a very important point for me to understand.
When you had the audible issue did you see anything in the audio band ?
What i am trying to understand is if audible problems show also on the instruments in the audio band. I think they should.
If i hear distortion i should also see distortion products in the audio band with the analyzer.
Maybe the opposite is not always true.

A scope is more or less mandatory if you want to do this stuff, but like all tools you also need to know how to interpret what it is showing you, there is no substitute for understanding.
Regards, Dan.

I think that for now this is outside my reach. What i would like to do is to perform some distortion measurements in the future.
I am convinced that low noise and low distortion are essential requirements for good sound.
Thanks a lot again.
gino
 

Attachments

  • noise ext psu.png
    noise ext psu.png
    23.7 KB · Views: 151
Last edited:
In my own humble view, a power supply is critical to the operation of a system, so the better the power supply the better things will be overall. And here I do not mean linear is better than SMPS, both can be excellent and both can be a pain.
But as well as the power supply you have to consider the whole power delivery system these days especially with a digital design and that means proper decoupling and filtering.
 
In my own humble view, a power supply is critical to the operation of a system, so the better the power supply the better things will be overall.

Thanks for the valuable advice. I have no problem with this. I agree 100%. A very good psu will never be detrimental for sound.

And here I do not mean linear is better than SMPS, both can be excellent and both can be a pain.
But as well as the power supply you have to consider the whole power delivery system these days especially with a digital design and that means proper decoupling and filtering

I understand. My problem is ... is there any kind of measurement that ... I rephrase. There is a guy who says that his dac sound without any doubt better with a high end linear power supply (very expensive) than with the stock 12V ac smps adapter.
And he says this on the basis of listening tests.
Can I see this superior quality with any measurement instead ?
My guess is that noise spectrum analysis could be that measurement.
In the graph I attached above I can see for instance that the linear unit is quite worse than the smps ones.
Clearly a case of poorly designed/built linear psu.
Maybe I am wrong.
Or let's assume that I have two different kits of regulators. Same mains transformer.
Which kind of measurement can tell me in a reliable way which is the better of the two ?
Spectra are very rare but are they really the most telling test ? I tend to think so.
Moreover ... as a spectrum analyzer is too expensive which is the second best and accessible test ?
I cannot believe that a psu designer relies on listening test ...
Thanks a lot again, gino
 
Last edited:
And he says this on the basis of listening tests. Most audiophiles don't do listening tests! They do self fulfilling seeing tests.

Hi I agree ! I know many cases of friends who select units by the look.
For now i have at hand only a software Spectrum Analyzer called Arta.
The more I know it the more I love it.
I am about to receive some cables to test the best interface I have.
If swapping psu (it comes with a cheap 12V adapter ) will not have effect It will be already a good outcome. I am very curious.
I hope to be able to run the test this next weekend.
Thanks a lot, gino
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.