Looking for an audio analyzer

[1audio]

For FFT if your doing more than testing analog amplifiers accuracy (i.e. looking at musical waveforms) there is software http://www.qsl.net/dl4yhf/spectra1.html Spectrum Lab that will show much more than a typical FFT. if you are looking at acoustics there are several with ARTA being more versatile for both electronics and acoustics. REW is a good room and speaker equalizer package and there are many more. You can do well with any of a number of soundcards, the best are in the $600-1K range or something like the new standalone stereo boxes from Lynx or RME for around $1500.

I have a QA401 and find it does what its intended to do quite well. You can get about 10 dB better for about 50X the price with the latest APX555 but that seems like a small ROI. Any of these are far beyond anything you can encounter acoustically.

You can find more on these specifics at http://www.diyaudio.com/forums/design-build/

[DaJMasta]

Spectrum Lab looks like a really great visualizer for FFTs, and I'm similarly surprised that it's free!

I think my aim is gradually being swayed towards a really high quality audio interface over a dedicated audio analyzer.  The analyzer will have its specified software suite, will have better input protection, and will probably have more output level flexibility whereas going with an interface means wider software compatibility (much more likely to just be an ASIO driver or similar), potentially built in mic preamps, and a lower noise floor for a given price.

While I do intend to be testing some of my own designs, I won't be testing in an industrial environment so input protection may not be a top concern, and having a preamp and phantom supply built in takes another part out of the chain for doing the microphone measurements.  Something like the RME Babyface Pro, for as ridiculous as its name is, offers very impressive noise floor, on par with the AP 2700 series analyzers according to their numbers, and has built in preamps for something like $750 - which may be still a bit high vs the QA401 given that the a preamp box probably won't cost the remainder, and either is considerably under a "full on" analyzer's price.

[deadlylover]

Look up on the old audio analysers made by Panasonic and ShibaSoku, there are a few models that hold up today in terms of residual THD+N like the Panasonic VP-7722A which gets about -112dB(0.00023%) THD+N @ 1kHz 2.5Vrms, 30kHz BW, you'll have a tough time approaching that with a studio audio interface + software combo due to the limitations of our current ADCs.

Those old instruments can be controlled by GPIB so you can get some pretty graphs that way and they have a monitor port in which you can perform FFT on the post-notch signal, I've attached an example below, the second harmonic is at -139dBr but I couldn't quite figure out how to make it scale properly in ARTA. 

Great thing about these old analysers is that if you only need quick and dirty distortion results, there's no need to boot up a PC or anything and they can read individual harmonics up to the 5th at the touch of a button, too easy.

The ShibaSoku 725 series don't include the generator, but in return you can get -120dB THD+N @ 30kHz BW performance with some easy modding so they are still very capable machines today. I think unmodded they did around -116dB, which still eats anything alive except the AP flagships, not too bad for maybe $500 used eh?

The dScope started at around $3000 or so IIRC, it'll probably do everything you need but it's a biiit out of reach for hobbyists IMO (which I'm assuming this is more for hobby/learning use as otherwise you'd already have gone for an Audio Precision ahahah). 

[_Wim_]

The OP wrote:

"I'm a musician with a bit of a fascination with audio gear, and it's been a fairly longstanding goal of mine to do some measurements to characterize the sound of different instruments, techniques, spaces, etc..  I'm also interested in being able to characterize the performance of my gear - frequency response, THD+N, etc - both as a reference for further experiments and so I can see the effect of different configurations or modifications."

The way I read this, is he wants to tweak his equipment to a get better sound, and understand the effect of the tweaks. His goal is not to get better numbers  only (he does not work for a commercial business where it can be interesting for marketing to build the first 32 ENOB-dac with an SN of 194db :-) so all audiophiles consider this the next Walhalla and sell their house to buy this new gear).
So we should advice gear that is :
-   capable of measuring “audible” differences (and a bit below audioble so the cumulative effect of multiple changes is also possible)
-   relatively easy to use and to understand (he will not  use this 8 hours a days, 5 days a week…)
-   affordable (so the OP can spent the rest of his available budget on other nice gear for an electronics lab :-) )

So I think the recommended QA401 will be perfect for measuring the electronics (differential inputs are a very big plus), and use a sound card with ARTA or equivalent software (he can play around with as many free tools he wants) will get him going on the acoustics side.

I recommended this as an “ex-audiophile”, as once upon a time I also really believed in the benefits of using ultra expensive audiophile capacitors, ultra-low jitter clocks, buying expensive signal cables…  and wasted quite some money doing that until I started reading.

One of the things that cured me from my “audiophile illness” was the following software tool:http://www.foobar2000.org/components/view/foo_abx
It allows you perform blind ABX comparison between two music tracks. For example you can test if you can hear the difference between a compression less music and MP3@320, the effect of 0.01% THD added, the effect of some phase distortions,the effect of adding some reflections …) For manipulating the audio tracks (THD distortion, phase distortion, low and high pass filters, adding reflections) use the typical musician tools which I suspect will be not an issue for the OP.

By doing lots and lot of ABX testing (use a good headphone) you will be amazed how difficult it is to actually hear some differences. For example I was really sure I would  be able on any music to recognize MP3 (even at 320kbps) from compression less, but this proved to be not so easy. 

Combine this with a good book on psychoacoustics  and acoustics, and you soon realize you will have been wasting your time on the wrong things (and the wrong audiophile gear). Your new goals will be to manage early reflections in your listing room, having a speaker with constant directivity and having multiple subs in many positions to get a reasonably flat frequency response in this modal sound region. All of these are dictated by the way sound waves travel (and the wavelengths involved) from the speaker to your ear, and the way our hearing works. The electronics cannot alter these new goals (only exception, having a speaker with steers the sound beams using DSP and multiple speakers). So once you know your electronics are “good enough” (which should not be a problem for modern affordable gear), you can  start tinkering on how the fool the laws of physics…

[Ice-Tea]

In the interest of learning how to crawl before you can run... have you considered an Analog Discovery (2)? All the instruments mentioned before in this thread are several order of magnitude better than this but you may find it easier to guage what you need when you have worked with something that fails your criteria...

http://store.digilentinc.com/analog-discovery-2-100msps-usb-oscilloscope-logic-analyzer-and-variable-power-supply/

[_Wim_]

What also helped me in my cure for audiophile illness was something our friend Albert Einstein once said: If the fact’s don’t fit the theory, change the facts!


What I mean by this, if the following:
Once I was invited by our local high end hifi store for a listening session to hear the sound difference between a reasonably prices solid state amp, and a high end valve amp, both on the same high end speaker. And for sure the high end valve amp sound much better, so after that I knew for a FACT that valve amps were much better, and I wanted (lusted) for a high end valve amp from then on.
What I did not realize, is that the high end speaker used during the demo had a minimalistic passive crossover specifically tuned for an amp with a high output impedance (=valve amp), which caused the solid state amp to sound worse. This was not the fault of the solid state amp (in the contrary, the solid state amp showed the faults of the high end speaker more clearly). This kind of thing unfortunately happens a lot in the business of high end audio, making it very  difficult to convince people some differences are not real.
Just take this into account whenever you “hear” a difference, there might be another reason then the better “quality” of the equipment…

[jackenhack]

I'm currently finishing up a headphone amplifier build and I got the QuantAsylum QA401 when I started developing the amp. With todays extremely low noise and distortion figures on op amps, You'll hit the bottom of what the analyser can measure very quickly. I had to build a twin-T notch filter and get hold of a extremely low THD signal generator to be able to measure below -108dB THD. Getting a industry standard Audio Precision is the dream, but they are way to expensive for me...

But the question is, why would we need to see lower than that? It can be fun as a technical exercise, but it will not improve the sound quality we hear any more…

Going down the rabbit hole is part of the fun. I want it to measure as good as possible.


Sent from my iPad using Tapatalk

[DaJMasta]

Haven't been considering the Analog Discovery because I have a nice scope and a nice function gen - that level of generation and data capture I can already achieve.  Specifically with the FFT performance, I can get much better results with a sound card's extra resolution than with the scope, even in high res mode with as many bins as possible.

I think I have to agree with the last post, I enjoy being able to measure precisely and I really like being able to notice things I couldn't without the hardware.  For the sake of being sure I won't be able to notice any difference on my own, I want something with significantly better fidelity than I can discern.  Most of the testing I have in mind is either strictly electrical or acoustic with as much of the room taken out as possible (a source and a mic in a small isolation chamber in a quiet room), so my goal isn't trying to setup a perfect listening room or evaluating speaker setups.  I don't consider myself an audiophile, I'm perfectly satisfied with high bitrate mp3s for listening, and I don't believe in tube, capacitor, or oxygen-free copper voodoo.  If I'm going to buy an expensive audio cable, it's because I want one with lots of EMI rejection because I have it routed through a nest of digital signal cables and want it to effect my noise floor an absolute minimum (and for whatever reason I couldn't just untangle the nest).

The QA401 seems like the right approach, but I don't know if it's the ideal unit for my applications.  The BNC interfaces just mean I need a bunch of adapters and no mic preamps mean extra stuff in the signal chain and extra expense.  While the software may be good, it seems there's no shortage of good audio analysis software that will run with sound card devices, so i don't think that's a particular selling point.  It's price point is good, but if I can get a deal on a good audio interface, I can get the connectors and preamp integrated and can get a bit lower THD+N baseline on everything.  I don't think the QA401 offers any special input protection that studio gear wouldn't match.

[ci11]

The advatnage of using an audio interface is its relatively low initial cost, the con is the opamp many of them have in the signal chain were not made specifically for sound and vibration measurement - but "good sound", whatever that means to the designer.

Bruel & Kjaer makes measurement mics, some people recorded music with them (including Mark Levinson at one point) and subjectively, they were never the equal of Neumann. The same goes for audio interfaces. It's the combination of the input opamps, ADC's and PSU that make it all work, or not. The one that sits in this sweet spot is the E-MU 1616m, which uses good quality NJM2068s and near state-of-the-art AK5394 ADCs. Using the supplied PatchMix software, a 24 192 session could be setup to feed analysis software through ASIO drivers.

Like many good things,  they are out of production and since E-MU has been acquired by Creative, they no longer evolve drivers for the new OSs coming on stream or supply schematics for repairs. The 1616m itself does show up on eBay but since these are all at least 10 years old, a thorough recap is likely needed, even if it's "new" -  since this was made to sell at a pricepoint, the caps they used were not the best and there are many of them. Attached are pictures of the bottom PCBs, before and after modications. The top PCB is modded too, a little less complicated but not much.

This route is definitely for the curious and patient. It's a fascinating trip and very educational. But the dollars and cents do add up in the end.

[_Wim_]

Going down the rabbit hole is part of the fun. I want it to measure as good as possible.

I also agree this can be a lot of fun, and if this is the goal, then buy all means buy the gear necesarry for it. But I just wanted to avoid people thinking it is essential to get good sound...

[Loboscope]

If you have a audio-interface like the RME Babyface or you intend to buy it or a similar high-quality interface like the Fireface UC, then I will consider again "hpw-works". I am musician too and my approach  to electronics is to control, maintain and if necessary repair my audio gear. Therefore I have some signal-generators, a Rigol scope (DS1104Z) and of course some multimeters. But if I want to check or verify the performance of modern half-decent and decent preamps, ADC´s and so on more in deep, this stuff will not be accurate and sensitive enough.
But I never would pay a fortune for one of the (really superb!) Audio-Precision devices. If you will buy expensive stuff like this, you must make money with it.

So I looked for am reasonable and payable alternative. The QA401 looked also good to me, but I had already some very good RME-Interfaces and finally I discovered the software "hpw-works" and I considered that this software-solution would give me not only a more than sufficient but also a really professional test-suite together with my RME´s. So i purchased it.

You can download a 14-day test-version of hpw-works here: http://hpw-works.com/index.php/download/evaluation-edition-sw-kit
There you will find also the pdf-tutorial of hpw-works. The fact that it has an amount of 475 pages may give a hint of the capabilitys of this software.

My experience is, that if troubles, failures and bugs will become audible, you never have to look as deep as -120 dB into a spectrum, audible trash will be located far above this level. But if your test-suite is capable to look as deep as -120 dB (or even deeper how my suite can do) and you will not find any "dirtiness", than you can put your mind at rest and you can assume that you audio-gear will work perfectly.

[_Wim_]

For the sake of being sure I won't be able to notice any difference on my own, I want something with significantly better fidelity than I can discern.

I think this holds true for a QA401 also, and even for a sound-card based system.

Most of the testing I have in mind is either strictly electrical or acoustic with as much of the room taken out as possible (a source and a mic in a small isolation chamber in a quiet room)

Depends on what frequencies you want to measure, but going below 1000Hz it becomes increasingly more difficult to have “an isolation chamber” and take out the room, and you can forget about”small”. Outside is than the most feasible option, and for low frequencies you can use ground plane measurements.

I don't consider myself an audiophile, I'm perfectly satisfied with high bitrate mp3s for listening, and I don't believe in tube, capacitor, or oxygen-free copper voodoo.  If I'm going to buy an expensive audio cable, it's because I want one with lots of EMI rejection because I have it routed through a nest of digital signal cables and want it to effect my noise floor an absolute minimum (and for whatever reason I couldn't just untangle the nest).

Than you are already on the right track ?

The QA401 seems like the right approach, but I don't know if it's the ideal unit for my applications.  The BNC interfaces just mean I need a bunch of adapters and no mic preamps mean extra stuff in the signal chain and extra expense.  While the software may be good, it seems there's no shortage of good audio analysis software that will run with sound card devices, so i don't think that's a particular selling point.  It's price point is good, but if I can get a deal on a good audio interface, I can get the connectors and preamp integrated and can get a bit lower THD+N baseline on everything.  I don't think the QA401 offers any special input protection that studio gear wouldn't match.

That is a reasonable point to which I can fully agree , but I think buying AP or dScope like gear will still be a lot more expensive unfortunately . But if you can catch a nice deal on one of these, it is for sure more convenient to play around with, and the extra THD+N headroom  is a nice bonus…
It is indeed unfortunate that the QA401 still has no ASIO drivers available (it on their to-do list), that would make this device absolutely fantastic.

[_Wim_]

The advatnage of using an audio interface is its relatively low initial cost, the con is the opamp many of them have in the signal chain were not made specifically for sound and vibration measurement - but "good sound", whatever that means to the designer.

This is not true, I have this card, and it frequency responce is rules flat, so no sound tailoring. See meausurment I made in loopback (output connected to input)

Edit: addad summary

[_Wim_]

Although the EMU-1616m is very good performance wise, the big disadvantage are its drivers, which reguraly cause a BSOD and other nasty issues

[ci11]

The advatnage of using an audio interface is its relatively low initial cost, the con is the opamp many of them have in the signal chain were not made specifically for sound and vibration measurement - but "good sound", whatever that means to the designer.

This is not true, I have this card, and it frequency responce is rules flat, so no sound tailoring. See meausurment I made in loopback (output connected to input)

Edit: addad summary

I think there is a misunderstanding. What I said is "The one that sits in this sweet spot is the E-MU 1616m". This means that the 1616m is good, not bad. My comment is about 2 points:

1. Not every audio interface is a good candidate for objective measurement. The circuit, parts and implementation must be carefully evaluated. The 1616m is a winner, many are not.

2, Many audio interfaces cost well under $1000, some costing just $100 to $200 retail. At these prices, parts quality are not always the best. And that means they do not last, or they are not always linear. Even if the circuit, parts and implementation are all correct, if the parts quality do not stand up to critical use, then it's only as strong as the weakest link. In the case of the 1616m, the problems are in the caps count and their low quality. You can see bulges in many of the caps in the pictures listed in the post.

As to the drivers, there has been no issues with Win7 X64 or Win10 Pro. Touch wood - I have not seen a BSOD ever.

[_Wim_]

The advatnage of using an audio interface is its relatively low initial cost, the con is the opamp many of them have in the signal chain were not made specifically for sound and vibration measurement - but "good sound", whatever that means to the designer.

This is not true, I have this card, and it frequency responce is rules flat, so no sound tailoring. See meausurment I made in loopback (output connected to input)

Edit: addad summary

I think there is a misunderstanding. What I said is "The one that sits in this sweet spot is the E-MU 1616m". This means that the 1616m is good, not bad. My comment is about 2 points:

1. Not every audio interface is a good candidate for objective measurement. The circuit, parts and implementation must be carefully evaluated. The 1616m is a winner, many are not.

2, Many audio interfaces cost well under $1000, some costing just $100 to $200 retail. At these prices, parts quality are not always the best. And that means they do not last, or they are not always linear. Even if the circuit, parts and implementation are all correct, if the parts quality do not stand up to critical use, then it's only as strong as the weakest link. In the case of the 1616m, the problems are in the caps count and their low quality. You can see bulges in many of the caps in the pictures listed in the post.

As to the drivers, there has been no issues with Win7 X64 or Win10 Pro. Touch wood - I have not seen a BSOD ever.

Ok, I misunderstood, sorry.

About point 1, I agree they are not all as good as the EMU, but many are more then good enough.
About point 2, haven't checked the caps, but so far the card perform normal
The drivers, I run on win7 32, and gives sometimes touble on pc, and from googling for a solution, I was for sure not alone (but havent found a solution)

[ci11]

The drivers, I run on win7 32, and gives sometimes touble on pc, and from googling for a solution, I was for sure not alone (but havent found a solution)

This works for me - but be careful these are the exact drivers you need, which may not be the ones on the Creative support site.

1. Install first the driver (EmuPMX_PCDrv_US_2_30_00_BETA.exe),
2. Reboot,
3. Then the app (EmuPMX_PCApp_US_2_20_00.exe)

Good luck. It may or may not work, so YMMV.

[_Wim_]

This works for me - but be careful these are the exact drivers you need, which may not be the ones on the Creative support site.

1. Install first the driver (EmuPMX_PCDrv_US_2_30_00_BETA.exe),
2. Reboot,
3. Then the app (EmuPMX_PCApp_US_2_20_00.exe)

Good luck. It may or may not work, so YMMV.

Thanks. If it start acting up again, will give that a try.

[ruairi]

DajMasta,

A $200 soundcard, $100 measurement mic and free software will not be the limiting factor in your tests. Measuring audio equipment and running acoustical testing accurately is not a trivial business.  Relating measurements to subjective observations is even trickier.

Take for example what you said below.

"Most of the testing I have in mind is either strictly electrical or acoustic with as much of the room taken out as possible (a source and a mic in a small isolation chamber in a quiet room)"

The space you measure in will dominate any tests at real audio frequencies.  Are there things you can do?  Sure, windowing for example is very powerful and can create quasi anechoic measurements limited in low frequency range only by the size of your room (which determines when surface reflections make it back to your mic).

Get an affordable sound card, a cheap calibrated mic and start measuring.  You will learn a lot by doing. 

[alex89]

I am in a similar situation but my budget is currently limited around 1K. So far these are the unit I'm considering:

• QA401
• AP SYS-222
• JensH Audio Analyzer

The QA401 seems a very nice product. Yes it comes with its own software but as far as i know it works well and the software team seem responsive to user request. I don't like the BNC connectors and the fact you need to make your own adapters cable but it is probably the best bang per buck. I'm very tempted to buy it. The AP system one is a very old unit but it is the only option if you want to bring the best brand audio analyzer in you shop. The performance I guess are in the same neighborhood of the QA401 (?) but you need an old pc with ISA card and that doesn't seem very attractive to me.
The last one is not yet a product available in the market but it seems the more promising to me. I haven't followed the entire discussion on diyaudio.com but it looks like will give better performance compared to a QA401 and also ASIO support. I'd like to hear your thoughts.

Alessio

[ruairi]

I haven't used the Quant, but will probably pick one up for fun this year. 

I have owned an AP system one and it is a wonderful, huge and powerful beast.  But it is unsupported, and not trivial to set up. FFT is very powerful and missing on the older S1.

Jens' project looks great but again learning what to do is more important than waiting for "better". 

[_Wim_]

• JensH Audio Analyzer

The last one is not yet a product available in the market but it seems the more promising to me. I haven't followed the entire discussion on diyaudio.com but it looks like will give better performance compared to a QA401 and also ASIO support. I'd like to hear your thoughts.

Looks at first glance like a very nice produc, but would like to see a performance summary first. The hardware is certainly up to the task, and the fact that is supports ASIO will make it much more attractable then the QA401... Thanks for pointing this unit out to us.

[Assafl]

I'm currently finishing up a headphone amplifier build and I got the QuantAsylum QA401 when I started developing the amp. With todays extremely low noise and distortion figures on op amps, You'll hit the bottom of what the analyser can measure very quickly. I had to build a twin-T notch filter and get hold of a extremely low THD signal generator to be able to measure below -108dB THD. Getting a industry standard Audio Precision is the dream, but they are way to expensive for me...

But the question is, why would we need to see lower than that? It can be fun as a technical exercise, but it will not improve the sound quality we hear any more…

Going down the rabbit hole is part of the fun. I want it to measure as good as possible.


Sent from my iPad using Tapatalk

There is also the QA aspect of this - if you designed your channels to measure very low, any noisy component or problems in the amplification stages will pop up in the THD+N. They may not be audible, but you production run has a problem. It is no longer a Blameless amp.

I think that is the reason companies like Bryston measure very low and add a cert sheet to every amp - QA. I do not have the ability to measure (nor hear) that low - but their amplifiers are very nice (every bit as nice as every other blameless amplifier).....

[ci11]

Going back to the OP's 2 applications, monitoring a FFT from complex audio signals in real time is quite a different matter than using an ultra pure sine wave to measure the THD+N of a circuit. These are very different needs,  if only one devices is used to satisfy, that device would need to be of instrument grade and capable of doing so.

For the first applicataion, B&K at the high end and AudioControl at "everyday" price levels come to mind. The signals come from microphones and go into a device that "analyzes" and displays the level in each frequency band. They respond fast but depending on the sophistication of the device, price varies. AudioControl RTA boxes sold for years around $200 with a microphone, but there are no bins or frames to poke into. B&K, the Danish sound & vibration expert that supplies everyone from NASA to Mercedes Benz made very competent microphones, mic pres and analyzers, but again, their output reflects the need to characterize a complex audio signal and sort out that hairball in a meaningful way. Their tools weed out the chafe.

For developing circuits, every "chafe" is meaningful, because some glitch that happens at -100dB has a root cause, perhaps originating from somewhere else at -110dB. In the time domain, scopes that see reliably see these glitches are highly prized, and many tools have been developed over the years to catch these elusive glitches more reliably and visibly, precisely because they need to find the root cause. In the frequency domain, the need and technique is no different - just because it's not evident at -60dB where it is clearly audible does not mean there is no problem at -100dB. Simply putting on a set of good headphones will reveal significantly more normally inaudible glitches - that's why recording studios use all 3 ways to monitor a session - the soffit-mounted 15" monitors, the mini's on the console meter bridge and headphones. Any engineer who lets a mix go to tape without first getting all 3 right would not have a job for long.

An expensive analyzer like an AP or Rohde & Schwarz can do both, but they are not cheap and is not an casual purchase, unlike a USB soundcard or a freeware FFT. But these instruments are made by seriously competent people with very demanding customers, and their findings are not compromised by a budgetary constraint or makeshift construction. Of course, there is a component of brand value in their price, but in the long run, it really comes down on how much the user needs to rely on their findings, and what value they place on these findings.

[ruairi]

Going back to the OP's 2 applications, monitoring a FFT from complex audio signals in real time is quite a different matter than using an ultra pure sine wave to measure the THD+N of a circuit. These are very different needs,  if only one devices is used to satisfy, that device would need to be of instrument grade and capable of doing so.

Why are they different?  REW, a free and very well supported piece of software can do both with it's signal generator and FFT.  It won't do stereo but I find that it's rarely a limitation. 

Your point does get to the heart of something though - that everyone has a slightly different approach to measurement.  I for example am not very interested in THD+N as a metric, with out proper filtering PSU hum (especially in tube gear) can be a contributor that doesn't have a huge subjective effect.  I'm more interested in the harmonics, when I see higher order harmonics I am more concerned.

The O.P. needs to start his journey of measuring to understanding what it is that he needs. 

Simply putting on a set of good headphones will reveal significantly more normally inaudible glitches - that's why recording studios use all 3 ways to monitor a session - the soffit-mounted 15" monitors, the mini's on the console meter bridge and headphones. Any engineer who lets a mix go to tape without first getting all 3 right would not have a job for long.

This is not the case.  I know many very famous engineers who never use headphones and who never check on the mains (lots of newer studios don't have mains and in older spaces they often sounded dreadful)."