I use multi tone tests on speakers. The best score -60dB at medium level over 50Hz.
Lowering the volume it gets better and raising the volume it gets worse. Why better electronics sound better on a speaker that is much worse is not totally clear. One explanation could be that good speakers have mostly second and third harmonic and the distortion goes down with level.
Lowering the volume it gets better and raising the volume it gets worse. Why better electronics sound better on a speaker that is much worse is not totally clear. One explanation could be that good speakers have mostly second and third harmonic and the distortion goes down with level.
Actually that is what it has been created for initially! Send a multitone from the studio desk out to the transmitter into the country, and from a receiver back to the analyzer portion. Fraction of a second off-air time and then you can analyze to your hearts desire. The AP marketing term was FastTest (italics AP).
The notion that it is a better audio test to correlate with listening perceptions came later, and reading the comments so far I'm not convinced it is.
But it is a nice design tool.
jan
If you have MathLab or MathCad or one of the open source versions you can create a .wav file with all the multitones you want and play that through the soundcard. No special 'precision' needed. The reason for the AP equipment is NOT that it can do multitones - multitone testing is mainly a matter of mathematics and a good generator and analyser, and the latter can be found in a good sound card.
jan
Janneman,
Could you draw a little picture of how you are arranging the connections to a simple sound card. Nothing intermediate is used between the outputs and inputs to the card? I understand what you are saying about creating a test tone with software, just how you are doing the comparison is what I am looking for here.
Steven
One option would be a floating supply with the mid-point driven from a HV opamp like the OPA453 IIRC.
One limitation to that would be the input stage; you can't lift the V- above ground because then the input stage gets into trouble.
But it should at least give you double the voltage range (also depends on your amp gain).
jan
The soundcard connection is straight forward: you play the multitone .wav from the output of the soundcard into your amp and take the amp output in through the soundcard input and do an FFT. The crux lies in the software that generates your multitone .wav and the measurement software you use for the soundcard. I was getting up to speed with ARTA before my house move so that's been on the backburner but i need to pick it up again.
But I guess apps like Audio Tester and others can do that too.
I will attach the ISO31 standard file as well as the .wav if I can find it.
jan
Last edited:
Yes, Octave and Sound Exchange come to mind, both free and heavily supported.
Some notes on the AP utility; It does not appear to build without an AP environment, from reading the .pdf they take the 31 tones you posted and pick the closest exact bin depending on record length and sample rate, and they simply minimize the crest factor randomly and keep the smallest if you want.
So here is what you do, pick a power of 2 record length (such as 16384) and make a table of exact bin frequencies from your sample rate (44,100,96000,...). For 16384 and 96000 it starts, by bin, 0,5.859Hz, 11.718Hz ,17.578Hz,..... keep as many digits as you can. Take the closeset 31 from the table and put the same amplitude tone in those bins with a random phase (0 to 2pi), the inverse FFT is then your time waveform (scale it to whatever you want). The crest factor will be a surprise so just reseed the random number and do it again if you want.
A comment on the frequency selection, I don't think the non-common factors of the bins is that important. An IM tone only needs to fall in a bin not exactly on the bins center frequency to go into a tone bin rather than a "noise" bin. If you go through this exercise at 16384 you will see the low frequency tones have no space (bin) between them, that is why I would do 65536 at minimum.
17.575
23.450
29.300
41.025
52.725
64.450
82.025
99.600
123.050
158.200
199.225
252.000
316.500
398.500
498.000
632.750
802.750
1,002.000
1,248.000
1,599.500
1,998.000
2,502.500
3,152.500
4,002.500
4,997.500
6,352.500
7,997.500
10,002.500
12,497.500
16,002.500
19,997.500
Last edited:
I dont doubt this. Broadcast industry has their station data on loop-back using multitones - which is both practicle for them but also gives an accumulated number that is useful for QC/QA and alignments etal. We (audiophiles) dont have such numbers (do we?). yes, we could use a 1-2 tone test thru the whole system and see the results and do it at several freqs and combo's of IM. Might just have to do that if we cant find an easy PC way to do it with one measurement - MultiTone. The system numbers could be more interesting as the distortion products add to one another to reach some interesting number levels. And, unlike measuring individual equipmnt and getting the same numbers as another person who has that piece of equipment... the total system numbers are unique to the combination used by a person in thier system.... if they use SET or analog or digital, class A or AB, tube, ss etc.
Even though the speakers are a disaster of distortion (most of them)... and dominate the sound (and acoustics).... the delta of a smaller added to a larger can be detected as a change... if above at least .03-.05% At least that is the official scientifically derived number for now -- we dont yet know the bottem number on hearing ability... so we need to keep an open mind when we make conclusions near these levels. We do know that the detectable limits under certain conditions are a lot lower now than we used to think from previous testing/assumptions. Ditto that re freq deviations and resonance detection etal.
Sorry, I know you know this stuff... its a prelude for anyone reading this to the system multiTone testing that audiophiles might find interesting results and maybe helpful in the total view of their music systems and designs. Thx - Dick
Last edited:
split 48vdc or +/- 24v. maybe slightly higher for losses due to heavy local nfb resistors used in amp circuit.
[using +/- 12v now]
Last edited:
The original AP utility came out with S1.exe and can run stand-alone from a DOS command line but I can't get it to work. With APwin it became integrated in the program.
jan
I see. In my example above you could also simply sum 31 sine waves with random phase at the right frequencies. The frequencies would be bin-number*96000/16384 up to 48000 kHz the Nyquist rate (65536 or larger wuld be preferable). To use the inverse FFT method you have to recall your complex math to build the frequency domain representation.
SW discrete opamp BF862 (post 135).
I have put in bootstrapped cascodes on the front end and 5401/5551 as the cascode BJT's in the "VAS", so there is no problem using the circuit at higher voltages.
As you can see in the attached file I have stepped the source resistance from 50ohm to 2kohm and there is almost no change in THD.
It is possible to improve the circuit, but it is rather pointless to design a perfect front without using a relevant output stage.
So in order to move forward we must, among other things therefore agree on the load.
600 ohm or 30 ohm?
BTW: Maybe the circuit needs a better name than "post 135" or maybe not?
Stein
I have put in bootstrapped cascodes on the front end and 5401/5551 as the cascode BJT's in the "VAS", so there is no problem using the circuit at higher voltages.
As you can see in the attached file I have stepped the source resistance from 50ohm to 2kohm and there is almost no change in THD.
It is possible to improve the circuit, but it is rather pointless to design a perfect front without using a relevant output stage.
So in order to move forward we must, among other things therefore agree on the load.
600 ohm or 30 ohm?
BTW: Maybe the circuit needs a better name than "post 135" or maybe not?
Stein
Attachments
30 ohms to me just seems brutal, other than for limited voltage swings into headphones. 600 is already pretty stiff, suitable for driving low-Z feedback networks I suppose and reasonable loads.
Ho Lee Chit says -->
You mean after all these years we cant just download a MultiTone software to a PC or iPOD thingy?
Little boy said about his dead grandfather -- " I LIKE BIRTHDAYS, but I think too many CAN KILL YOU."
What's another word for Synonym?
God put me here to accomplish certain things. Right now I'm so far behind I WILL NEVER DIE!
A good example is the best Sermon.
You mean after all these years we cant just download a MultiTone software to a PC or iPOD thingy?
Little boy said about his dead grandfather -- " I LIKE BIRTHDAYS, but I think too many CAN KILL YOU."
What's another word for Synonym?
God put me here to accomplish certain things. Right now I'm so far behind I WILL NEVER DIE!
A good example is the best Sermon.