|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
73.229.29.71
I'm looking at the wall wart power supply noise spec by an audio equipment manufacturer. It's on their website as well as on the retail box for their $50 SMPS wall wart. They claim:- Noise*: ~1uV
*Average audio band noise floor.
I assume they are specifying the RMS noise across a 20 Hz - 20 KHz "audio band" but they don't actually tells us. Reputable power supply manufacturers will often specify noise over a 20 Hz - 20 MHz (not KHz) range and tell us so. And they will also provide Pk-Pk noise specs as random noise spikes are easily hidden in the averaged RMS values.
I found an interesting application note by Keysight Technologies titled "How to Read Your Power Supply's Datasheet". For those who are not aware, Keysight Technologies was once Agilent and Agilent was the test equipment and instruments spin-off from Hewlett-Packard many years ago.
The complete Application Note is linked below. Here's an excerpt that might be of interest with respect to power supply noise:
"The output noise specification is labeled in different ways on a data sheet depending on the vendor. Output noise refers to the deviations of the DC output voltage from its average value over a specified bandwidth. It is typically measured in rms (root mean square) and peak-to-peak(p-p). It is important to note that there are two types of noise to consider: normal mode and common mode. If the specification does not explicitly state that it is common mode noise, then it is a specification of normal mode noise. Normal mode noise is the voltage deviation on the positive output terminal with respect to the negative output terminal.""The most important noise specification is the peak-to-peak voltage noise. If the peak-to-peak voltage noise specification indicates that the deviations are large, they could damage or destroy a sensitive device, such as a voltage controlled oscillator (VCO). The rms measurement is not an ideal representation of the noise and should not be considered a good representation of a power supply's noise performance. Fairly high output noise spikes of short duration could be present despite a low rms noise specification since they do not appreciably increase the rms value. "
"You should be wary of a power supply vendor that only specifies the rms noise value. Also, make sure you note the bandwidth over which the measurement was made. It should be in the range of 20 Hz to 20 MHz. Poor noise specifications can be hidden by changing the bandwidth over which it is measured."
Edits: 07/11/16Follow Ups:
I think we had that discussion at least half a year back.
Just try "search".
A year back? No many years back.
The guy still connects his Rendu to a fanned power supply which is known to be noisy using croc clips!
How are those massive blowers & fans sounding in that old Dell T5500 workstation you just bought? The one that you claim sounds pretty good from it's onboard audio?
nt
We've been discussing power supplies forever, not sure if I ever posted the Application Note. If you find it in a search, my apologies for posting it again.
Some digital voltmeters (For instance Agilent bench meters) can be setup as to provide a set of simple figures to sum up the noise (Min - Max - Average), according to the measurement parameters that have been choosen. This is convenient when the customer claims for a simple figure given precise measurement conditions
However, it would be useful if we could see a spectrum analysis of the power supply's noise, in average mode, or in peak mode, for longer or shorter periods, so that some damageable behaviours may be detected, particularily if there are strong harmonics of mains frequency, or switching frequency, the importance of whom may be hidden by a simple figure. Especially a simple peak figure may not tell the truth, for instance when several "noise" components are out of phese
Hi !
just to say that I agree with you.
Unfortunately spectra of power supplies are almost never provided.
I found an interesting article from Clifton Labs about this topic
I am attaching the link.
There is a very interesting picture of the 20 MHz spectra of different Power Supplies noise.
I honestly cannot image a more direct and better way to compare different power supplies.
Ok there is also reliability, MTBF, size, cost ... but speaking of residual noise/ripple ... I mean, I understand it even myself !
I love this kind of graphs.
What impresses me is that some very cheap smps have actually very decent performance.
But I am leaning towards good quality linear power supplies. They should have very low noise also beyond the audio band.
But a spectrum analyzer will tell almost everything about the quality of a power supply output. Very impressive instrument indeed.
I asked for some price and the good ones are very expensive.
If not I would buy one immediately.
If you have any source of spectral analysis on low voltage power supplies please share the link.
I would be very interested to know a little more.
Kind regards,
bg
Edits: 07/13/16 07/13/16
There are some possibilities to do similar measurements at a fraction of the cost, with a computer + soundcard + software (for instance, look for Spectraplus.com)
And, of course, enough knowledge about electricity
But, by no way you will get enough bandwith to comply with the 20 MHz requirement from the Keysight/Agilent/HP Application note
Which is advertisement directed toward the industry/laboratories, widely exceeding what would really be useful for average audiophiles
Except those who use switching power supplies, switching power amps, high definition digital sources, etc.
Hi ! thanks for the kind and valuable reply." There are some possibilities to do similar measurements at a fraction of the cost, with a computer + soundcard + software (for instance, look for Spectraplus.com)
And, of course, enough knowledge about electricity
But, by no way you will get enough bandwith to comply with the 20 MHz requirement from the Keysight/Agilent/HP Application note "Yes. It is what I am doing with a soundcard that samples up to 192 kHz with 24bit. I get noise spectra up to 96kHz. Beyond that the mistery.
" Which is advertisement directed toward the industry/laboratories, widely exceeding what would really be useful for average audiophiles
Except those who use switching power supplies, switching power amps, high definition digital sources, etc. "Yes. I have already been advised to use linear power supplies that should be more quiet at higher Hz.
It is just that smps are so small and convenient.
For instance I have done some noise floor tests on my soundcard.
Up to 96kHz the noise floor does not change much using a very cheap but genuine smps from a major brand (i.e. like Dell, Sony, Lenovo) or a decent quality linear psu.
But the problems maybe arise at higher Hz ?
Kind regards,
bg
Edits: 07/14/16 07/14/16
Generally, lab PSs have different requirements from audio PSs.
Study the posted supply circuit diagram. You'll see protection resistors inline with the output that increases output impedance in favour of current limit. You'll see fans because these guys are trying to save on heatsinks.
For audio purposes, the lower the output impedance and the faster the output tracks the demand the better. Thus excellent transient response is important.
They used to say that preamps and amps are really their power supplies and this appears to have been forgotten in digital audio.
A $50 PS that says it has fighter jet technology is not credible. What it has is feedforward elimination of certain peaks, or a narrow range of peaks. The technology is well known in noise elimination headphones and has been around since the 70s.
Hi thanks a lot for the very interesting advice.
Just for fun i am trying some regulator kits i have bought from ebay
Unfortunately i have no instrument like a spectrum analyzer that would be so useful.
I have to rely on what i see with a SA software (only up to 96kHz)
However using a R-core transformer most of the kits based on lt1083cp are not that bad. And very cheap.
I am using one to power with 12VDC a usb soundcard and i feel some improvements sound-wise from the stock cheap smps.
Maybe i am just hearing things.
Without proper instruments it is practically impossible to test these regulators objectively.
Thanks a lot again.
Kind regards,
bg
Edits: 07/16/16
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: