What's wrong with my C-notes?

[joshshetter]

Thanks? Man this forum is kinda weird lol.

[norman bates]

That is a very good statement.

[joshshetter]

I sent my findings to PE but never got a response, don't think they really care too much. I gave my c-notes to a friend who owned the Audiosource amp I borrowed, he should be set. I wonder how many people out there are using cheap class d amps that have this issue and don't even know it.

[Steve Lee]

Thanks for the info, guys.

[joshshetter]

Here is the thread the ASR member linked me that prompted me to test different amplifiers.

https://www.audiosciencereview.com/forum/index.php?threads/comparison-of-power-amplifier-distortions-and-fr-into-complex-speaker-dummy-load.21682/

I don't really think it's class D in general that has this issue with these speakers, I have measured many speakers that use class D including genelec 8030 and there is no issue. I also have Amiga kit that does not exhibit this issue, and my current project that involves a DX25 + WG300 does not exhibit the issue with A07 amp, so I believe it's down to a complex interaction between the c notes xover and particular class D designs that lack post filter feedback as stated by the ASR member (I don't know what that is though, that is beyond me).

I guess at this point my options are to attempt to redesign the crossover for these drivers or buy a new amplifier. Both of those options kind of suck but it is what it is. I took several measurements just to make sure the results were consistent and they were, there was even a very audible difference between the two amps used (A07 and old audiosource amp 100). I hope this marks the end of my troubles.

[JRT]

I am not familiar with the amplifiers mentioned above, and didn't bother to dig further. Maybe this does not apply.

Here is an oldie but goodie... see sect 7.1.4

https://www.ti.com/lit/pdf/slaa117

...which pertains to a type of class D amplifier that has a low pass PWM demodulation filter (aka reconstruction filter) affected by load impedance, does not have the PWM demodulation filter inside of a control loop, and so load impedance at high audio frequencies affects that filter, with linear distortion effects in frequency response at those high audio frequencies (top octave).

For an example different from that, Purifi modules are largely independent from load impedance within the constraints of maximum peak current output.

[a4eaudio]

I have generally heard not to use Class D amps for measurements. I was recently reading a thread from back in 2014 and the HDMI input on a Class A/B amp, even in Direct mode, was messing up the measurements. I think this just goes to show how important it is to really spell out the measurement process when posting, just in case someone recognizes a potential problem.

[Steve Lee]

^ I'd like to hear more about this - its kind-of disturbing if true . . .

[joshshetter]

So I think this is finally solved. A user at ASR shared some info on load dependency with certain class D amplifiers. I borrowed a class A/B amplifier and measured. Orange is Class A/B, green is Class D that I was using before.

[AdrianD]

I was playing with the xo in vituixcad using your FRD and the PE supplied ZMA files. Have you tried swapping C2 and C1 around? When i do that in my simulation it looks like the transform function of the filter creates a higher Q value and makes the rising response less dramatic. I'm not super versed with crossover design, i was just poking around seeing what would happen.


Filter function with C2 and C1 swapped

Filter function with your current crossover.
​​

[joshshetter]

I went ahead and disassembled the crossover and added each component one by one. Here is the .mdat for that session including my mics calibration file.

https://drive.google.com/file/d/1kg0O3cXV8qkI5IKsI3tRA0zQ_b9orS34/view?usp=share_link

[joshshetter]

Do we need to start a gofundme account? Not sure what the endgame is here...

Please try to keep comment construcive. My goal here is obvious, figure out why my cnotes are so off in my own data vs PE's.

Here's the Kali Lp6 overlaid with the C-notes so they're on the same window settings.

[davidroberts]

Here is with and without cal file from PE. Maybe 2db difference, overall rise is still there. I kinda just want my money back at this point so I can invest in a better mic.

Do we need to start a gofundme account? Not sure what the endgame is here...

[joshshetter]

Here is the kali lp6 v2, no HF rise. I also did a loopback test to confirm the pre's on the motu m4 I used are ok, the loopback test was basically a straight line. This raises some questions as to how PE is measuring their speakers. I have no idea what to make of any of this. My C-notes certainly sound like there is a large HF rise, measurements show it, but it doesn't show up in PE data? I am quite confused.

[joshshetter]

Usually the calibration adjustment is pretty minor on the low and especially high ends, which is consistent with what you show. For my first speaker design I struggled for weeks, into months, getting just bizarre results. It was the laptop! I had a laptop from work and was able to install DATS and REW but I didn't have admin privileges so I never could quite figure out what audio settings were set wrong. But I finally was ready to give up and tried my wife's MacBook and BOOM all of my measurments looked like what I expected.

Also, I had a pair of C-Notes and a pair of commercial speakers at the time. One sanity check was to measure them all, as well as the ones I was trying to design. The strange results were consistent across all speakers, suggesting it was something in the audio settings. Do you have any other speakers you can measure and see if the same rising top end shows up there?

My friend is kind enough to lend me a Kali LP6 v2 monitor that I'll be picking up today. We have an erins audio corner review to show how that speaker should measure so hopefully this sheds some more light on things.