A noise floor below 140dB should be adequate to detect any relevant noise in the audio band, imho. No need to buy expensive pro-stuff!
Hi and thanks a lot for the helpful advice.
My question is how much is important for a usb soundcard (i.e. AD/DA usb converter) what happens beyond 96Khz ?I cannot see anything of what happens above this limit
Could it be that the power supplies emit a lot of garbage in the MHz region and this has a very bad impact on the soundcard performance ?
Thanks a lot, gino
Last edited:
Noise above 96kHz is outside the passband of your soundcard. This noise either is low-pass filtered by the system - thus irrelevant. Or by some kind of demodulation it is transformed into the audio band. In that case you will see it in your spectrum anyway.
All in all anything audible will be visible in the audio spectrum - it as is simple like this.
And the audio resolution of a good 24bit soundcard is outstanding, there is not much on the market to top that.
Last edited:
Thanks a lot again for the very helpful explanation. This is a very good news.
I am very ignorant on the topic.
Maybe MHz noise can have some impact on dac clock performance instead ?
Presently I am using a Tascam uh7000 that provides sampling rate up to 192kHz and 24 bit resolution.
I am experimenting with different power supplies to see if anything changes in the noise floor.
I have already advised to try batteries (i.e. zero ripple psu) to establish the top performance obtainable from the soundcard, at least in terms of noise floor.
These soundcards usually take an external DC voltage, often 12VDC, and generate the needed voltages for the circuits with on-board dc to dc converters.
The quality of these converters can vary depending on the quality of the design and construction of the soundcard.
What I have already found is that different psu even with ripple 20 times different give almost the same overall noise floor.
I mean psu with some 100mV pk2pk and 5 mV pk2pk give similar noise floor graphs.
But I still have to find and possibly try a really low ripple psu.
Thanks a lot again, gino
Last edited:
So are you then expecting that with a ripple of say 1mV you all of a sudden have a lower noise floor?
That is not to be expected - your measurements show that the DAC noise floor will not go lower with ripple lower than 100mV pk-pk.
Jan
Hi and yes exactly. To be reasonably sure that the psu is not a limiting factor in any way.
I think I have to accept this thing, i.e. i have already hit the bottom noise floor.
And that is not possible to start "digging" by using a better psu ... 😱
However it is a very interesting topic this of power supplies.
I read too many discussions and I have become literally obsessed by noise and ripple.
Until now the only time I had really different results has been when using a faulty psu.
Thanks a lot for your valuable advice.
kind regards, gino
Last edited:
Why not? You already use a much more expensive but limited performance substitute for spectrum analyzer, your ears.
Hi, Velleman makes PC based scopes that might do for you. They have an external box and connect via USB. They are dual trace and will do spectrum analyzer functions as well. The cheaper one has less sensitivity and less bandwidth (25meg as I recall) buty will do Bode plots. The better one I seem to recall will do 100meg. The cheaper one is often on sale for under $300US and the better one for around $500. There are additional brand of PC scopes and they have various functions. A good moderately cost solution IMO.
Do I remember right, you are already using a linear PSU? If so why do you think you have high frequency noise that you can't measure?
A spectrum analyzer can be a useful tool, but if you don't have a good theoretical grounding it is easy to misuse it and pick up noise from other sources.
If you want to start to experiment and improve your understanding and technique, perhaps an Analog Discovery would be a good choice. I've seen used ones on on ebay for $150. They can function as an osciloscope, a spectrum analyzer, a function generator, and a logical analyzer
A spectrum analyzer can be a useful tool, but if you don't have a good theoretical grounding it is easy to misuse it and pick up noise from other sources.
If you want to start to experiment and improve your understanding and technique, perhaps an Analog Discovery would be a good choice. I've seen used ones on on ebay for $150. They can function as an osciloscope, a spectrum analyzer, a function generator, and a logical analyzer
cheap scopes with FFT option incorporate 8 bit ADC - thus resolution is poor compared to any cheap onboard soundcard.
Used Advantest SAs are fine, but they were designed for high frequency measurements with 50Ohm RF inputs. 8bits of resolution and unsuitable bandwidth - no way for audio measurements.
Used Advantest SAs are fine, but they were designed for high frequency measurements with 50Ohm RF inputs. 8bits of resolution and unsuitable bandwidth - no way for audio measurements.
FYI, There is an Advantest FFT analyzer series R9211x with 16-bit resolution and 90dB dynamic range. The main advantage of these "ancient" boxes is their auto-ranging input. For measurement of line level, performance wise, a decent PC soundcard and free FFT software combo is a better option.
Good morning ! thanks a lot for the very helpful advice.
Honestly I thought much more. They are very complex and high end instruments indeed. I watched some videos on Youtube about maintenance and I am amazed. Moreover while I understand clearly the output of a SA the output of a scope is a mistery to me ... I understand the Y axis but not the X axis ... response in time ??? 😕
I much much prefer the SA. Very much indeed.
Thanks again, gino
Hi ! well I did not have to buy them 🙂
Maybe I am overestimating the importance of this instrument.
For instance I wonder how many here have one.
Is it really needed in the end ?
Hi ! well yes but i am not completely set on the solution.
As i said before i am not seeing big difference between different psu as long as they are decent in quality.
And i have here nice little smps that are half a pocket of cigarettes ...
I am still experimenting with different solutions.
If i understand well you mean that if i use a linear psu i will be reasonably sure about not having noise let's say above 96kHz ? 🙄
that would be a strong point to go linear indeed 🙂
Are you referring to this one here ?
Analog Discovery Portable Analog Circuit Design Kit by Digilent Inc Lghtly Used | eBay
Let's say that up to 96kHz i could do fine with the SW.
I am now not so sure about the need to measure beyond that.
Thanks again for the valuable advice.
Hi and thanks a lot for the advice.
The price is good and also it is a portable solution that i like.
I will look for reviews and try to understand something.
Thanks again.
Hi and thanks a lot for the advice.
The whole point is ... how much important is to test above 96kHz ?
considering that my almost only interest are power supplies and dacs.
And if i have to choose one i will say power supplies.
So is it really important to see a psu ripple above 96kHz ?
Thanks again.
If you measure a DAC up in the hundreds of megahertz, you'll see that DAC's that boast distortion under 0.00xx were actually much higher distortion.
You say not hear, but actually following modules will generate intermodulation distortion in the audible area.
Because of this, DAC's with identical parameters (sometime even the same chip) still sounds differentand and because of these components you can hear the cablessssss for example!
I measured my DAC with a 10 to 4GHz Spectrum analyzer (at headphone output because of 50Ω input) and I read a lot of components over audio band until hundreds of MHz's, and the level was not insignificant, I read levels over -90db.
You say not hear, but actually following modules will generate intermodulation distortion in the audible area.
Because of this, DAC's with identical parameters (sometime even the same chip) still sounds differentand and because of these components you can hear the cablessssss for example!
I measured my DAC with a 10 to 4GHz Spectrum analyzer (at headphone output because of 50Ω input) and I read a lot of components over audio band until hundreds of MHz's, and the level was not insignificant, I read levels over -90db.
Last edited:
Hi and thanks a lot for the valuable advice.
So there is a good reason to be curious about the MHz range 🙄
I read about digital being sensitive to noise in the RF range.
Very very interesting digital and also challenging good digital.
I will keep on reading for sure.
I am 100% digital. No analog here.
Thanks a lot again, gino
- Status
- Not open for further replies.