Experiences with software thd, sine generating

Status
Not open for further replies.
Jump to Latest
#101
Each modern sound card is able to work with fs 192kHz/24bit, most of them full duplex ( generating various signals and analyzing at the same time) so BW is to 95kHz, so to capture all distortion artefacts as with "normal" audio analyzer is no problem, also DIM100 or multitone spectral measurements are possible.. And limited BW to 80kHz is so to say a norm for THD measurement. So here is no real limitation.
In fact "vintage" audio analyzer use is more convenient, simpler , more robust to user mistakes, but far less sensitive and informative.
And look at thread name, what is here needed...some real experiences with software for THD capturing and signal generating.
 
Last edited:
#102
Yes real experiences. Imho these expectations are too optimistic. Otoh i have just build Elvee's undemanding Circlophone if one does not mess up too many components it works on first power on. No adjustment needed. If one does not want to be an audio pro doesn't want to spend money on precise test equipment he is right there with soundcard. The first step towards a pro is Peter Millet's interface. That is sound profound audio engineering.So i am quite satisfied with just the harmonics spectrum versus frequency and power, it has no x over distortion , and is excellent at low frequencies around 5 Hz which is hardly ever tested although such test unveils thermal instabilities. So after all i am satisfied but would not expect precision beyond that.
 
#103
scott wurcer said:
I have two, and I have used them for 35yr but when I needed ppm thd I used a passive LC ladder on a Sound Tech 1700 and twin T. The 3562 (at least my old one) BTW has issues with noise aliasing.
Click to expand...

To my limited experience with these old boat anchors the beauty is the ability to save the spectra data in vector format, that is, both the amplitude and the phase of the spectral components. It can also synchronously measure the gain/phase of a DUT, hence it is easy to cancel out the generator contribution to the distortion components. IIRC, since I was a student, the internal generator had some 0.04% distortions, but still could measure DUT distortions of one order of magnitude lower, without any extras or fancy ultra low distortion generators.

Mr. Wurcer got the point of the twin T (or double T) network in my previous post. A twin T network will add it's attenuation to the dynamic range, resulting in an apparent increase in the number of A/D bits. Example: 16bit is about 96dB, plus 48dB from a good twin T, is equivalent to the 24bit goal. Everything else can be averaged down to the background noise of the Universe, time permitting.

I'm afraid we don't have one of these my seniors in our lab, but I think I've seen a 35665A. I'll try to illustrate the above using this (not necessary better) instrument.
 
#104
hahfran said:
As an indisputable fact, developers of commercial amps , engineers of audio magazines , and you name it, have all sold their respective analyzers and turned to soundcard. 🙄 The reported data got much better since 🙂
Now, irony off. What is far more interesting is sort of batch mode. It may not be common knowledge , anyway , pro designers of commercial hi end amps give much on "psychoacoustics" , in plain words, what sounds good sells good, how it measures is a different pair of shoes. As a result of compiled insights from experiments in "psychoacoutics", advanced audio magazines report the harmonics spectrum versus frequency and power. A commercial analyzer can perform such measurements, it is a necessity in quality control and quality engineering. Even more advanced magazines go a step further, they report this for "critical" and "less critical" passive speakers as a reactive ( and hence, in practice "real" load). of course these measurements require support of "burst mode" otherwise ...good bye 10,000 bucks a pair speakers.
It has so far turned out, that per thd perfect amps sound nevertheless poor, if the harmonics spectrum has nasty spikes just at these power and frequencies which are in terms of amplitude and frequency spectrum the most frequently appearing in classical music. That becomes worse when driving "critical speakers". It turns out that many amps show this kind of irregularity.
Only some no feedback single ended triode amps appear to be free from this effect.
Click to expand...

What a pile of bio chips.
 
#109
Mr. Wurcer got the point of the twin T (or double T) network in my previous post
Click to expand...
Yes,double T network is good and necessary aid to make some improvement in resolution with vintage analyzers. But try to make some sweep measurements (e.g. THD vs. freq.), or CCIF, IMD , multitone...measurments with double T network.😱 No chance.
Modern souncard with good software is much more universal and accurate tool as any of those "oldies" (and I have some, too).. Price/performance ratio is what counts.
 
Last edited:
#110
Waly said:
To my limited experience with these old boat anchors the beauty is the ability to save the spectra data in vector format, that is, both the amplitude and the phase of the spectral components. It can also synchronously measure the gain/phase of a DUT, hence it is easy to cancel out the generator contribution to the distortion components. IIRC, since I was a student, the internal generator had some 0.04% distortions, but still could measure DUT distortions of one order of magnitude
Click to expand...

Python can be used as post processor for soundcard data and all this and more is possible. Arbitrary N FFT's are particularly useful in some cases.
 
#112
scott wurcer said:
Python can be used as post processor for soundcard data and all this and more is possible. Arbitrary N FFT's are particularly useful in some cases.
Click to expand...

I suppose that would work in any programming language, if access to the A/D raw data is provided. I wouldn't think this is a trivial task though, in particular when it comes to synchronizing the samples for spectrum and gain/phase measurements. Could be a nice feature for a new sound card software package, never seen such.
 
#115
google says:
 

Attachments

  • vessel.PNG
    vessel.PNG
    22.7 KB · Views: 153
#119
yah, delurked a pedant. You may find that your wikipedia notes on square brackets is for american usage, as OED is same form as I used.

BTW you need to translate your strapline to be inline with forum rules 😛 (sorry couldn't resist)
 
Status
Not open for further replies.
Top Bottom