Spectrum analyser or a scope with good FFT?

[_Wim_]

dScope and Audio precision are indeed commonly used by profesionals. Apart from very good specs, these are mainly used because of features like automated testing and reporting, which is not so important for hobbyists. Also, as a professional, 10 of 20k for something you use on a daily basis is much more acceptable price wise then to a hobbiest who uses this max half a day a week and who gains no profit of it.

As to jitter, I do agree it is not so easy to measure accurately, but I do not agree this is very important for audability. I read of a controlled listening test were they introduced jitter in the audiochain, and asked people to indicate the level of jitter. In essense, people only started to recongnise the jitter because the DA was creating skipping sound when the jitter was so worse. I have searched for the reference in my archive, but could not find it. As far as I remember, the study was performed with members of a dutch audio forum.

Anyway, somehow in audio people tent to believe that our ears/brains are a much more sensitive instrument that high end measurement gear, and that stuff that can barely be measured accurately, make a "huge" difference in sound. My opinion is that the oposite is true. When it measures okay with reasonable quality gear, it is also ok. This means that is you built something with moden electronics (pre-amp, amp...) normally most well designed gear will "sound" exactly the same.

Things change however when you measure the combination of loudspeakers the acoustic environment they are in. When you start measuring reflections, power responces, decay times... you will see HUGE deviations from the "ideal" (currently there is no concensus about the "ideal" however, as to little is known about the phsycho-acoutics of the brain).

The point of all of this is that with quite low cost gear you will be able to measure the actual REAL problems in audio as a hobbyist, solving them is a lot more difficult however... Don't spent to much money on gear, but buy some good books on the subject instead.

Some good reading on the subject that is freely available:
http://www.davidgriesinger.com/
http://gedlee.azurewebsites.net/Books/AudioTransducers.aspx

[Muxr]

I think the problem with your standard audio PC hardware is jitter.

Care to expand on this?

I think it would be difficult to measure, and my understanding is that the commodity DACs and ADCs in computer hardware don't have terribly stable clocks, so your measurements might be off due to jitter/phase noise. There are some DACs for instance that do this well, but then again we're talking thousands in gear again.

I do agree though, with you and _Wim_, try the gear you have (there are some nifty open source tools as well), and invest in books. Also read Nwaguy's blog, it's full of good information regarding audio testing and measurements.

[_Wim_]

Another factastic book on the subject (and easily readable): http://www.amazon.co.uk/Auditory-Neuroscience-Making-Sense-Sound/dp/0262518023/ref=sr_1_1?ie=UTF8&qid=1352658057&sr=8-1

Also check out the wonderfull website:
https://auditoryneuroscience.com/

I do not want to turn this into an audioforum, but I only want to point out that there are real problems involved in audio, that are dictated by real laws in physics, and when you really make the effort to study the subject, you will understand it is not easy to solve (especially so if you want a commercial viable solution). The amount of BS that is floating around on the internet concerning this topic (in most cases thanks to marketing departments of audio companier) is enormous. So whenever you read something, please check actual physics and electronics books, and also do the calculations. Many marketing departments take real effects (for example skin effect in cables), but fail to tells you that the effect is soooooooooo small (at audio frequencies), that even with the best gear in the world you cannot measure it. And if you say it is unmeasureable, they will tell you that your ears are way more sensitive...

[amirm]

Just to provide a reference for what proper audio test equipment does as you look at Scope FFT, here are a couple of measurements I did measuring jitter in DACs/AVRs using Prism Sound Dscope:





Notice the noise floor (they do have processing gain).

And here are some with my Audio Precision Analyzer:




THe last measurement shows the effect of inserting a USB hub in the path of the USB connection to the DAC!

You can use sound cards but then you have to build your own analog front-end when testing amplifiers and such as you are not going to want to pump +- 50 volts into their inputs!

[_Wim_]

You can use sound cards but then you have to build your own analog front-end when testing amplifiers and such as you are not going to want to pump +- 50 volts into their inputs!

A simple resistor dividor in parallell with your dummyload is al that is required...

In my opinion the problem is really the other way around, how to measure the distorsion when the amp is running at low power (<100mW) Distorsion in that range (crossover distorsion) can be quite audable, so you need a pre-amp after the DUT before feeding it to your sound card...

But, if you really want to built yourself a good audio system, I would invest the sparse hobby time into building constant direcitivity speakers and optimizing room acoustics. Off course, then we have to move this topic to a different forum...

[amirm]

You can use sound cards but then you have to build your own analog front-end when testing amplifiers and such as you are not going to want to pump +- 50 volts into their inputs!

A simple resistor dividor in parallell with your dummyload is al that is required...

Not if you are measuring THD+N relative to power. You need auto-scaling or sit there and screw around with segmented measurements.

[markone]

And here are some with my Audio Precision Analyzer:

How can a jitter problem produce 60dBc frequency tones distant 8Khz from an 11KHz carrier ?

What i can see in this picture is  AM modulated carrier.

Jitter produces phase noise components placed nearby central freq.

[FrankD]

A 2015THD can measure distortion down to 0.002 percent but the build in signal generator has about 0.02% distortion. You can use a state variable filter at 1Khz on the siggen output to get a lower distortion sine. If your 2015THD has a rs232 input you can control it with an Arduino or other uController to automate the sweeps to your liking and tune the state variable filter to the outputs frequency.

To view a spectrum you can use a 9KHz spectrum analyzer when your test signal is 10KHz instead of 1kHz. (At one 1kHz the distortion will be lower. )

By using a square wave you can use a scope alone simple by observing the resulting square wave from the dut.

[Loboscope]

I could recommend the analysis-software "HpW-Works" (http://www.hpw-works.com/).
It is used used normally with a soundcard and for reliable measuring it is obvious that a really good one is necessary. Therefore I would recommend the interfaces from RME (http://www.rme-audio.de/), especially the "PCI-Card HDSP Hammerfall 9632" or the new "Babyface Pro". These interfaces have not only dgital IN/OUT (SPDIF / ADAT) but extremely high performance analog IN-/OUT-Connections.

The software can connect to the interfaces up to 32 bit and max. samplerate. The software limit of samplerate will be 1 GHz, but there is no soundcard available capable of more than round about 200 KHz. So this will be the real limitation in frequency range, normally ca. 100 KHz.
Because the software is able to manage multiclient ASIO, it can handle separate interfaces for Input/Output. This means, you can measure a DUT, for example a Digital Mixer-Desk by sending the stimulus signal via PC-Interface to the analog-In of the Desk and use the digital interface-connection of the desk, for example USB to receive the signal from the Desk. PC-interface and Desk are connected via its own ASIO-driver parallel to the software and so there will be no additional AD/DA-converter in the signal chain.

HpW-Works has its own signal source witch can produce absolutely clean test signals who are a must for doing any reliable testing. The test signal can be adjusted in 0,5 dB steps.

In my opinion HpW-Works is fantastic analyser tool what produces solid and professional results - at a reasonable price.
Looking at the main user list (http://www.hpw-works.com/index.php/2015-08-09-08-41-49/users) you will find quite a lot of the "who-is-who" in the professional audio world.

Greetings, Jürgen

[_Wim_]

Not if you are measuring THD+N relative to power. You need auto-scaling or sit there and screw around with segmented measurements.

As explained before, measurement automation is indeed a big advantange of the profesional gear. As a hobbyist, measurements take indeed a little longer. My sparse free time is very valuable to me, but not enough so to spend >10k$ on a single piece of measurement gear...

[_Wim_]

I could recommend the analysis-software "HpW-Works" (http://www.hpw-works.com/).

Thanks for this, I will give this a try also.

[amirm]

And here are some with my Audio Precision Analyzer:

How can a jitter problem produce 60dBc frequency tones distant 8Khz from an 11KHz carrier ?

What i can see in this picture is  AM modulated carrier.

At low amounts that we see in audio products, jitter can be approximated to be AM modulation with just a pair of modulated sidebands (modulation index is so low that the rest of the FM sidebands disappear in the sidebands). 

That said, further analysis needs to be made to make sure that is not Vref modulation by changing the main tone to see if the amplitude of the sidebands remains the same or not.

Jitter produces phase noise components placed nearby central freq.

Deterministic jitter can have whatever frequency/spectrum it wants.  Such was the case in the one DAC measurement you quoted.  All you need is your clock polluted by that frequency, say USB packet timing, and it will pop up in the analog output of the DAC.  Its frequency will depend on the source of pollution, only its amplitude will depend on the frequency of the main tone.

I did show however another case where we get narrow-band but random jitter such as this other example:



Here you see two measurements, one of a $1,000 Audio/Video Receiver from Pioneer and the other my dedicated Mark Levinson DAC.  In both cases, *identical* digital bit stream is sent to them.  What you see is sharply different outputs with the AVR clearly suffering from spreading of the J-test tone due to narrow-band random noise and deterministic jitter.  The Mark Levinson DAC despite being nearly 15 years older, is free of both.

Thankfully from psychoacoustics point of view, random narrow-band jitter is not very audible so the measured difference doesn't translate into audible difference.  The measurements do dismiss the myth that "bits are bits" in audio.  Clearly the identical set of bits generated two distinct analog outputs.

[amirm]

To expand on the last point, here are two measurements on the exact Audio Video Receiver, one using its HDMI input and the other, its S/PDIF digital audio input.  In both cases the source device is my computer and the bit streams identical:



So with the identical DAC inside the same device we get two different analog outputs.  The reason of course is that with HDMI audio is slaved to video so to receive it, you have to light up the whole video subsystem.  And that subsystem is bleeding noise and deterministic jitter into the DAC clock, resulting in far degraded performance versus much simpler S/PDIF that doesn't need the video subsystem to do its thing.

Going back to OP, if you are going to do such analysis, then you need a dedicated audio measurement system that is trusted and people can replicate your results.

[amirm]

Not if you are measuring THD+N relative to power. You need auto-scaling or sit there and screw around with segmented measurements.

As explained before, measurement automation is indeed a big advantange of the profesional gear. As a hobbyist, measurements take indeed a little longer. My sparse free time is very valuable to me, but not enough so to spend >10k$ on a single piece of measurement gear...

Agreed.  Peace .

[_Wim_]

Not if you are measuring THD+N relative to power. You need auto-scaling or sit there and screw around with segmented measurements.

As explained before, measurement automation is indeed a big advantange of the profesional gear. As a hobbyist, measurements take indeed a little longer. My sparse free time is very valuable to me, but not enough so to spend >10k$ on a single piece of measurement gear...

Agreed.  Peace .

Peace off course! If I came over I as being on a war, for sure that was not my intention... Just wanted to explain to people here that you can do an awful lot with low cost gear when it comes to audio.
This makes this in my opinion a fantastic hobby, the only real problem is the amount of misinformation and audio foolery around on the net.  But when reading the right books, making good measurements (don’t be an audiophile who only “trust” his ears) and maybe do some software simulation (I can recommend ABEC3 for this, free licence available for hobby users), it is possible to build a fantastic sound system for a reasonable amount of money…

[markone]

At low amounts that we see in audio products, jitter can be approximated to be AM modulation with just a pair of modulated sidebands (modulation index is so low that the rest of the FM sidebands disappear in the sidebands). 

I was expecting more spurious spreaded with lower amplitude level, those side band are quite strong, are you sure that you do not have any sampling rate conversion in you PCM stream chain ?

[amirm]

At low amounts that we see in audio products, jitter can be approximated to be AM modulation with just a pair of modulated sidebands (modulation index is so low that the rest of the FM sidebands disappear in the sidebands). 

I was expecting more spurious spreaded with lower amplitude level, those side band are quite strong, are you sure that you do not have any sampling rate conversion in you PCM stream chain ?

No sample rate conversion.  I don't own a $20,000 instrument without knowing how to use it .

That is the measured performance of a $400 DAC that clearly has design flaws.  The same sideband does not show up when another digital input is used.

Here is its USB input:



The sideband is gone.  But see what happens when I change the media player used to send out the exact bit stream:



Yup, the DAC is sensitive to activities in the PC!  WMP in the second example pre-buffers the file and that extra activity bleeds into the DAC due to lack of isolation.  Media Player classic reads the bits as it goes along so doesn't generate the same noise profile.

Again, this demonstrates that "bits are not bits."  Just because there is a digital stream being played, it doesn't mean the output will be the same regardless of how you get it there.

My apology to OP for continued off-topic posts .

[hibone]

What about analog discovery?

It has 14-bit adc and 30 Mhz bandwidth, I don't know if FFT runs on board or on the PC though

[markone]

No sample rate conversion.  I don't own a $20,000 instrument without knowing how to use it .

Surely it was not my intention to say that, anyway i'm still convinced that those clean cut high level (for the context) lateral side bands are not due to a jitter problem,  at least in the sense in which it is normally meant.
All the other spectrum graphs that you showed are of course compatible with jitter effect.

That is the measured performance of a $400 DAC that clearly has design flaws.  The same sideband does not show up when another digital input is used.

S/PDIF perfomance wise is quite a rip-off for a $400 device, but i bet that lot of its proud owners did not even noticed that in listening sessions ;-)