MUFF WIGGLER Forum Index

 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler Blog & NewsMW News   Muff Wiggler StoreMW Store 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

Information
With the closure of the MW store the forum has gone back to being 100% community funded. To help us out the lovely Ross Lamond has put together an awesome raffle! You could win a complete modular system or one of many other great prizes. Please see the stickied thread in the For Sale subforum for details! It's my hope that we can raise our all our hosting costs through a once-yearly event like this. Thanks and good luck!

Audio test software for testing modular performance?
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author Audio test software for testing modular performance?

enoughstatic

I'm looking for advice on something other than an Audio Precision system (which can be very expensive), such as a SW App/Program and my external soundcard to test audio performance of prototype Eurorack module designs.

Any useful ideas?

(sorry, the search function is still not available in MW forum, and Google searches including MW aren't always helpful, so I apologize if this has already been discussed)


enoughstatic

Also, I do realize that you are limited to your hardware, including latency, noise floor, etc., but given a good external soundcard, I'm just looking to run frequency response curves on a filter. Perhaps phase shift, and maybe distortion. But Freq Resp. is the most important to me at this point.

And yes, I can plot it out using my signal generator and oscilloscope, point-by-point, but was hoping for something that will run sweeps and format plots into graphs and do the math for me.

Thanks!


flts

You didn't mention your computer operating system, but for Mac OS X my recommendation would go to http://supermegaultragroovy.com/products/fuzzmeasure/ in case it has all the features you need.


emmaker

So what do you want to measure?

I don't consider synths audiophile quality. Typically their distortion and noise levels aren't that great. Distortion I don't even think about, I'm more interested in the noise level and tuning. So for me other than a real good frequency counter, real good DVM, decent oscilloscope and my ears I don't need much more.


enoughstatic

flts wrote:
You didn't mention your computer operating system, but for Mac OS X my recommendation would go to http://supermegaultragroovy.com/products/fuzzmeasure/ in case it has all the features you need.


Thank you, this looks quite good!

Oops, forgot: I use Mac, but can also run Win 7 using Parallels for Win Apps


enoughstatic

emmaker wrote:
So what do you want to measure?

I don't consider synths audiophile quality. Typically their distortion and noise levels aren't that great. Distortion I don't even think about, I'm more interested in the noise level and tuning. So for me other than a real good frequency counter, real good DVM, decent oscilloscope and my ears I don't need much more.


I do have the above equipment (I've been servicing pro audio gear for over 20 years, so I have a nice collection of good test gear). As for what I'm wanting to test: mainly frequency response, phase relationship (lead/lag), and noise floor (not so much for SNR, but absolute noise floor levels, weighted or unweighted). My Rigol scope has an FFT, so that is a start.

Freq response and phase relationship are of most interest with filter modules.

Also, I build my own studio preamps, and effects pedals, so this is for a broader application than just eurorack / modular stuff.

BTW, I should mention that this is also for publishing specs. While not common in Eurorack, I'm interested in doing this, including prototype testing for evaluation / comparison of competitor products, etc.


kassu

A quick way to measure frequency response (amplitude, not phase) of a filter is to feed it a white noise signal, and look at the spectrum with the FFT of your scope (or with sound card, I sometimes use this program in windows for basic scope and fft function: https://www.zeitnitz.eu/scope_en).

Not as accurate as a proper "network analyzer" technique (feed a sinewave, measure output amplitude and phase, then repeat this for many different frequencies to build up the response curve), but very quick since you measure the whole spectrum at once.


Graham Hinton

enoughstatic wrote:
As for what I'm wanting to test: mainly frequency response, phase relationship (lead/lag), and noise floor (not so much for SNR, but absolute noise floor levels, weighted or unweighted). My Rigol scope has an FFT, so that is a start.

Freq response and phase relationship are of most interest with filter modules.


Well an FFT analyser won't do that, it measures the harmonic content of a signal, not the response of a device. You need to integrate a precision noise source aver a long time to get an accurate response curve. It is always speed vers. accuracy.

Fuzzmeasure looks like it does the job, but is no better than the sound interface on your computer and there needs to be a way of calibrating it. Don't forget that you lose a bit resolution for every 6dB lower you measure and are limited by the sample rate and anti aliasing filters. When I contacted SMUG about it it soon became apparent that I was talking to Mac heads generating eye candy with no experience of proper test equipment or any interest in making any. I was almost tempted though, but decided I was better off with some decent plotting software and correct data.

The industry standard before AP came along was the HP 8903B and you can pick these up relatively cheaply now. If you think $1k is expensive check what they used to cost new in 1980s money. They will do everything you need except phase and they have two advantages over modern audio analysers: NO FANS and they work standalone without Windows.

Quote:

BTW, I should mention that this is also for publishing specs. While not common in Eurorack, I'm interested in doing this, including prototype testing for evaluation / comparison of competitor products, etc.


Good for you. It's about time more people did this properly.


tarandfeathers

You could look at the Lindos range. The LA100 is probably out of reach, price wise, but the MS10 is only about the cost of a decent soundcard, and has moderate automation capability so you can sequence frequency sweeps etc and log the results. An LA100 is still our main audio test set at work (though the last place I worked had a shedload of Prism dScopes which are similar price range to the AP stuff and I'd rather have one of those if we could afford it!)


JohnLRice

Maybe not as advanced as FuzzMeasure (?) but SignalScope Pro is pretty nice. I have the iOS version. Maybe worth a look?
http://www.faberacoustical.com/apps/mac/signalscope/

Actually their most advanced package would be better:
http://www.faberacoustical.com/apps/mac/electroacoustics_toolbox/

Measurements:
• Frequency response magnitude and phase
• Impulse response
• Auto and cross correlation
• Time delay (latency)
• Group delay
• Signal-to-noise ratio (SNR)
• Total harmonic distortion (THD, THD+N)
• Coherence
• Crest factor and rms, peak, and peak-to-peak signal levels
• Signal mean, variance, and standard deviation values
• Flat, A, and C-weighted equivalent (Leq), time-weighted (Lp), and percentile-exceeded (Lx) sound levels
• And more!

Built-in Tools:
• Dual FFT Analyzer (TF/IR analyzer)
• FFT Analyzer
• Meter Bridge
• Octave Analyzer (RTA)
• Oscilloscope
• Signal Generator
• Sound Level Meter
• Spectrogram
• X/Y Plotter

Available via in-app purchase:
• Room Analyzer tool for estimating room acoustics parameters from impulse response data (T20, T30, T60, EDT, Clarity, Definition, Center Time).
• Noise Dosimeter tool for measuring noise dose with various threshold and criterion levels, with 3, 4, and 5 dB exchange rates.


benbradley

There are two popular and free software packages, just google ARTA and RMAA. Either one and a good audio interface should be just what you need, - even a modern laptop's built-in sound interface should be good enough for synth module development.

DSOs like the Rigol are great for what they do (like checking the rise and fall times of square and sawtooth waves), but the sample depth is only 8 bits (maybe 10 or 12 bits on a high-end DSO), just good enough for a 256-pixel high display, but certainly not good enough for "CD quality."

If ARTA or RMAA and your sound interface really aren't good enough (you did mention studio mixers), there's the QA-401 which (from everything I've read) does very well for "only" a few hundred dollars, a lot less than an Audio Precision. This model is brand new this year, and it's a refinement of the same company's model QA-400 that came out three years ago, so I don't think you could get a standard sound interface in the same price range that works as well.


BugBrand

tarandfeathers wrote:
You could look at the Lindos range. The LA100 is probably out of reach, price wise, but the MS10 is only about the cost of a decent soundcard, and has moderate automation capability so you can sequence frequency sweeps etc and log the results. An LA100 is still our main audio test set at work (though the last place I worked had a shedload of Prism dScopes which are similar price range to the AP stuff and I'd rather have one of those if we could afford it!)


Thanks - I'd not heard of these before. The MS10 looks surprisingly affordable (well, certainly not cheap).

[what ball park price is the LA100? The web says 'contact for price'.. which I guess I could do but..]


emmaker

benbradley wrote:
DSOs like the Rigol are great for what they do (like checking the rise and fall times of square and sawtooth waves), but the sample depth is only 8 bits (maybe 10 or 12 bits on a high-end DSO), just good enough for a 256-pixel high display, but certainly not good enough for "CD quality."


Yes very true, most digital scopes only have 8 bit ADCs. In an ideal world 1 bit of data equals 6 db of dynamic range. So theoretically 8 bits gives you 48 db, 12 bits 72 db and 16 bits would be 96 db. That's in an ideal world. Even though you can buy 24 bit ADCs/DACs they really only go to about 20-21 bits so ideally they'd give you about 120 db of dynamic range. In reality most 24 bit sound cards will run with about 90-95 db of dynamic range. The really high end/expensive ones can get that around 100 db.

So if you want to measure base noise below about -90 db (assuming 0 db is max input) you might have to spend some bucks for a high end ADC. Also if you can don't go by the manufacture specs on noise and dynamic range. See if you can find a unit that someone that knows what they are doing measure it. Most the time card manufactures just quote the converter specs. This does not take into account op-amps, power supplies, PCB layout and so on.

Another thing to check is if you use a FFT on a scope make sure it's OK for audio. All the scopes I know about are setup to measure RF and not audio.


tarandfeathers

BugBrand wrote:
[what ball park price is the LA100? The web says 'contact for price'.. which I guess I could do but..]


IIRC, the last time I checked (about 3 or 4 years ago) they were around £3.5k for the generator/analyser set. They're still quite widely used in local broadcast (BBC & ITV both have a lot of them), though not so much for national/network stuff where most of the audio is now digital.


BugBrand

tarandfeathers wrote:
BugBrand wrote:
[what ball park price is the LA100? The web says 'contact for price'.. which I guess I could do but..]


IIRC, the last time I checked (about 3 or 4 years ago) they were around £3.5k for the generator/analyser set. They're still quite widely used in local broadcast (BBC & ITV both have a lot of them), though not so much for national/network stuff where most of the audio is now digital.


Cheers! Yeah, that'd be too big a push for me.. I'm pretty sure I'll get ahold of an MS10 before too long now. Thanks for the tip! [or if you ever happen to see a 2nd-hand one..]


slow_riot

Depending on the specific model, SPICE analysis can give some very useable measurements. I just used AC sweep mode to get some figures for resistor values to compensate for integrator phase shift.

Keep an eye on Ebay for old test gear. I use a 1980s analyser that was being thrown out of a university lab, £50 plus petrol money to pick it up. Will do a lot of very very complex analysis functions (including phase vs frequency) although I've just been using it to see power bands in the frequency spectrum. It's made by Schlumberger and I think they had a few models. It is for DC to 30khz, unlike a lot of the low end spectrum analysers that are for RF (and often just give power measurements only).

These tools are however just for "looking" (where SPICE is so ideal it is only relevant as theory, and the ancient hulking analyser is a struggle to keep working properly, let alone in cal) , if you want to publish figures as part of a product specification I suspect you will need to cough up for a modern industry standard reference.


sempervirent

Fuzz Measure and Signal Scope both look really interesting, thanks for the links guys.


Troubleshooter

I don't undestand what you want to do or prove with those measurements.

A module for music-synthesis should only perform as somthing that sounds good in combination with other modules.
Good sounding is not the same as having perfect electrical properties you can measure. Think about the very un-square TB-303 square VCO...

The only usefull thing about measuring equipment is trimming/calibration of:

Trimming Tuning and scaling (V/oct).
Trimming Waveform symetry (DC offset).
Trimming Wave shapers like SAW 2 TRI/SIN converters.

But where possible I use my ears for the most pleasing setting.

I understand it is very tempting to for example test the audiospectrum a vca for example passes trough its omapms but in the end a lower spec module offenly sound better for some bizarre reason.

If you are still interested and definitely want some measurements software based I would recommend a good lenear DC-coupled audiocard with 192Khz 24bit.

Great for tuning are the AudioPhonics tuners
Any scope software with build in blows and wissles like spectrum analysers and fuction generators will do the rest.

For precise voltage measurement skip the software solutions and buy a descent fluke MM.


enoughstatic

JohnLRice wrote:
Maybe not as advanced as FuzzMeasure (?) but SignalScope Pro is pretty nice. I have the iOS version. Maybe worth a look?
http://www.faberacoustical.com/apps/mac/signalscope/

Actually their most advanced package would be better:
http://www.faberacoustical.com/apps/mac/electroacoustics_toolbox/

Measurements:
• Frequency response magnitude and phase
• Impulse response
• Auto and cross correlation
• Time delay (latency)
• Group delay
• Signal-to-noise ratio (SNR)
• Total harmonic distortion (THD, THD+N)
• Coherence
• Crest factor and rms, peak, and peak-to-peak signal levels
• Signal mean, variance, and standard deviation values
• Flat, A, and C-weighted equivalent (Leq), time-weighted (Lp), and percentile-exceeded (Lx) sound levels
• And more!

Built-in Tools:
• Dual FFT Analyzer (TF/IR analyzer)
• FFT Analyzer
• Meter Bridge
• Octave Analyzer (RTA)
• Oscilloscope
• Signal Generator
• Sound Level Meter
• Spectrogram
• X/Y Plotter

Available via in-app purchase:
• Room Analyzer tool for estimating room acoustics parameters from impulse response data (T20, T30, T60, EDT, Clarity, Definition, Center Time).
• Noise Dosimeter tool for measuring noise dose with various threshold and criterion levels, with 3, 4, and 5 dB exchange rates.


Nice. This is good to know. Thank you!


enoughstatic

Troubleshooter wrote:
I don't undestand what you want to do or prove with those measurements.

A module for music-synthesis should only perform as somthing that sounds good in combination with other modules.
Good sounding is not the same as having perfect electrical properties you can measure. Think about the very un-square TB-303 square VCO...

The only usefull thing about measuring equipment is trimming/calibration of:

Trimming Tuning and scaling (V/oct).
Trimming Waveform symetry (DC offset).
Trimming Wave shapers like SAW 2 TRI/SIN converters.

But where possible I use my ears for the most pleasing setting.

I understand it is very tempting to for example test the audiospectrum a vca for example passes trough its omapms but in the end a lower spec module offenly sound better for some bizarre reason.

If you are still interested and definitely want some measurements software based I would recommend a good lenear DC-coupled audiocard with 192Khz 24bit.

Great for tuning are the AudioPhonics tuners
Any scope software with build in blows and wissles like spectrum analysers and fuction generators will do the rest.

For precise voltage measurement skip the software solutions and buy a descent fluke MM.


Troubleshooter, I'm not looking to prove anything to the general public. This is more about quantifying performance (electrically) as a designer/developer so that I have a concrete idea as to what changes result in what performance. That's all.

While it may sound like more of a personal interest, this is certainly for validating my prototypes and characterizing their performance so that I can make educated decisions when implementing revision changes. And perhaps being able to publish the specs if and when I choose to do so.

I do have a precise meter, but taking 100 points of data, one-by-one, then entering data into a spreadsheet, then plotting graphs (post math operation, etc.) is a bit laborious when there appears to be a good selection of audio test SW, assuming a good sound card and the ability for the SW/App to null out or normalize it's own base-line performance (calibration routine).


enoughstatic

JohnLRice wrote:
Maybe not as advanced as FuzzMeasure (?) but SignalScope Pro is pretty nice. I have the iOS version. Maybe worth a look?
http://www.faberacoustical.com/apps/mac/signalscope/


So btw, I did purchase SignalScope Pro and it is an awesome tool. The sig gen and the FFT alone work quite well. This is pretty much what I was looking for!


enoughstatic

[duplicate post deleted]


forbin

Whilst I agree with troubleshooter that your ears should be the first call on the performance of your gear, if you are designing something it is still nice to be able to quantify how well it is working...

I am sure that my ears (which are rather old now and have visited too many loud concerts in their youth) will hear things differently to how other ears will perceive things. I often use Adobe Audition for simple checks and also Spectra Lab (http://www.qsl.net/dl4yhf/spectra1.html).

I am a bit spoilt by having access to a Stanford Research Dynamic Signal Analyzer at work. It has an amazing front end but about the worst UI on any test equipment ever...

An example where I used it is: http://galon-oscillator.blogspot.com.au/2016/04/output-quality.html

MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 1
Powered by phpBB © phpBB Group