What's the secret of high frequency harmonics?

[Spogg]

Spogg ... where did you get the A-weight module?

I don't know! I think it came from the old Synthmaker site and was made by a guy calling himself DWB.

I'm like a magpie. Anything shiny and interesting goes into my toolbox

Cheers

Spogg

[rocknrollkat]

I checked this with the spectrum analyser and the square wave has a lower and higher harmonic than the fundamental in addition, but not the sine.

My ears can hear the lower harmonic cut in with the square wave selected.

Martin’s oscillators are brilliant, but at this frequency I think there’s a wee bit of aliasing.

Schematic with the analyser attached.

Cheers

Spogg

Interesting observation, Spogg !
I tuned out the subharmonic that you discovered and the results are the same.
Yes, I should have done this first but I trusted the Flowstone Square wave generator,which is obviously not quite accurate !
The sine and square are subtly but distinctly different.
Here's the hookup.

Enjoy !

ROXY

[RJHollins]

Spogg ... where did you get the A-weight module?

I don't know! I think it came from the old Synthmaker site and was made by a guy calling himself DWB.

I'm like a magpie. Anything shiny and interesting goes into my toolbox

Cheers

Spogg

I do the same thing. I'll then, one night, go looking through the various schematics.

Again ... thanks for your post.

[ChrisHooker]

I tuned out the subharmonic that you discovered and the results are the same.

Roxy, you may not be able to see the aliased (lower) frequency on the graph anymore, but although it is greatly reduced, it is not gone entirely. The underlying issue of the alias frequencies being created is still there, uncorrected. Sweeping the square wave starting from a base freq with harmonics below the Nyquist and then increasing the base freq, you can see the alias frequencies moving down the graph, as they fold-over.

Do you have the original/stock SM/FS square wave that is made of 2 sawtooth waves? That version does not generate alias frequencies and would be a better comparison against sine, and I suspect you'll be harder-pressed to hear a difference. Even at a slightly lower frequency, say around 8K.

[rocknrollkat]

I tuned out the subharmonic that you discovered and the results are the same.

Roxy, you may not be able to see the aliased (lower) frequency on the graph anymore, but although it is greatly reduced, it is not gone entirely. The underlying issue of the alias frequencies being created is still there, uncorrected. Sweeping the square wave starting from a base freq with harmonics below the Nyquist and then increasing the base freq, you can see the alias frequencies moving down the graph, as they fold-over.

Do you have the original/stock SM/FS square wave that is made of 2 sawtooth waves? That version does not generate alias frequencies and would be a better comparison against sine, and I suspect you'll be harder-pressed to hear a difference. Even at a slightly lower frequency, say around 8K.

Hello Chris,
I just saw this post today, January 1, 2018.
I will recreate the experiment using my Wavetable function generator and get back to you.
As I mentioned at the beginning, the anomaly was discussed with Rupert Neve a few years ago.
I will get back to you.
Beyond that, I am lurking.
ROXY

[rocknrollkat]

Chris,
As per my last post a few minutes ago....
"....I will recreate the experiment using my Wavetable function generator and get back to you....."
Due to Flowstone's inability to reproduce higher frequencies with any degree of accuracy, a wavetable square wave is seen as a sine wave at 10kHz.
My spectrum analyzer shows no harmonics, as would be expected for a sine wave.
Martin's oscillator is displayed as a nonsymmetrical stairstep at 10 kHz, hence the subharmonics, etc.
I'll rely on Rupert's experiment in the analog realm, as described in my comments months ago.

Happy New Year,

ROXY

I'm gone.............

[ChrisHooker]

Definitely, the analog realm is a better place for this auditory test, but it is not for any lack of FlowStone to be able to reproduce high frequencies with accuracy (I'm not clear on why you think this), I would say the argument in favor of analog because all digital systems incorporate a lowpass filter on the output D-A converters, plus digital representations of audio waves become error-prone at frequencies approaching the nyquist. (A 22K sin wave is represented as 1,0,1,0,1,0... at 44.1K SR.)
Even if there is no aliasing in a digitally-generated signal (such as in the oscillator you used), or in the conversion from D to A, the harmonics of a 10K or higher square wave (whether distorted from digital sampling or not) will likely be cutoff from those D-A filters, preventing you from hearing any difference between that and a sin wave.

In that respect, a fully analog test system will provide a pure reference, to allow you to make an unclouded judgement.

Once again, though, I'd like to point out that this is not a limitation internal to FlowStone. (I don't want people thinking FlowStone is worse at producing high frequencies than any other software.) Attached is a schematic that utilizes a non-aliasing oscillator, so you can see that the harmonics of the square wave do not get aliased (folded over the nyquist and reflected back to be reproduced as lower frequencies).

[rocknrollkat]

"...Definitely, the analog realm is a better place for this auditory test, but it is not for any lack of FlowStone to be able to reproduce high frequencies with accuracy (I'm not clear on why you think this)...."

Hi Chris,
Here's a screenshot of the hookup you included, which is Martin's sine wave osc.
At 11kHz. this is what Flowstone delivers.
There are "shortcomings" to digital audio, sure, and I can't say for sure if this is a Flowstone issue, soundcard issue, etc., but this is what my 'scope shows.
I could discuss this further but I don't want to attract any 'experts' and have an instant replay of last week.
Bottom line ?
I wish I had my analog lab gear from decades ago, we could run the experiment that way.
But my 700 sq. ft. apartment says "so more room, son".

Best regards,
ROXY

"Poof !"
".... vaporized........"

[rocknrollkat]

Addendum....

Hi Chris,
Here's a screenshot for comparison, my Wavetable generator processing a 1024 sample sine wave at 11 kHz., 12 dB down.
Quite a difference.
This waveform shows odd and even harmonics, <80 dB down.
Flowstone ?
Soundacard ?
Not perfect, but at 11kHz, not too usable musically, so it's of no concern to me.
(The triangle wave at the beginning of the trace is caused by the one shot 'grab' made by the 'scope.)

ROXY

"..............vaporized.................................."

[adamszabo]

Showing the details of a waveformat high frequency has nothing to do with Flowstone. This is no mystery, it has to do with samplerate. I recorded a square wave at 10Khz at 192000 samplerate, then at 44100. You can see that 44100 has less samples per second so the data points show less of a square. Simple as that.

http://www.adamszabo.com/internet/samplerate.jpg