I believe in objective evidence.
Do you really not understand the difference between "audible" and "measurable"?
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Review and Measurements of Hypex NC400 DIY Amp
- Thread starter amirm
- Start date
I believe in objective evidence.
Do you really not understand the difference between "audible" and "measurable"?
This statement is not from BP, it's from somebody else.
This is the statement from BP about hysteresis distortion : https://purifi-audio.com/2020/05/13/combating-hysteresis-distortion-part-1-amplifiers/
Hypex with Ucd and Ncore being part of the "very small coterie"
Feyire
Active Member
Alright, so here is where it looks like our interpretations differ.
Referring to the Purifi article:
So what I understand, based on the statements in the audioexpress article and in the Purifi article, is that the increased feedback loop in the Eigentakt design (i.e. the greater than 10x better error attenuation compared to other designs) is required to push the hysteresis distortion to inaudible levels. i.e. the feedback loop in the previous designs (UcD, NCORE) was not enough to push hysteresis to inaudible levels in all scenarios:
Purifi Audio - A Straight Wire to the Soul of Music said:
So it was audible before (i.e. in previous designs), and now in the Eigentak design, is it attenuated way down, to inaudible levels - thus resolving the issue.Combating Hysteresis Distortion (part 1: Amplifiers) said:Here’s the response of the Eigentakt circuit to a transient disturbance. It’s plotted as a spectrum so: give the amp a kick, record the signal as the amplifier recoils, then plot the spectrum of the kick and the recoil combined, because that’s what’s coming out of the speaker terminals.
This graph assumes the worst possible kind of kick: a single, infinitely short pulse, the kind you’d least expect feedback to be able to do anything about. Its spectrum is the red curve and is flat all the way to infinity. The surprise outcome is shown in blue: the pulse is effectively attenuated all the way up to 200kHz. The audible fraction is knocked down by all of 75dB. If hysteresis distortion was audible before, it certainly isn’t anymore.
This is my interpretation of the presented information from the various articles.
"if it was audible before" is a pretty non-commitical statement. If it was obvious and provable that it was audible, he would have worded it differently
Thread Notice: Please please please stop feeding the Trolls. They can’t Troll a thread if you don’t reply to their nonsensical statements. Warning was issued and a 14 day reply ban. This is an Official Review Thread. Keep your comments and conversation relevant to the review.
Please and thank you .
Please and thank you .
Hello Everyone:
My first post. I read through most of this thread with interest because I hate Class D amps after two decades running a concert production company where they made everything sound worse, but were lovely for subs. Less weight, less power requirement, fewer amp racks (but you had to keep them away from the wireless mic/IEM racks).
Now I'm retired and finally putting together a nice home audio system and keep reading raves about this amp and its designer, so I'm going to give a pair of NC400's a shot. I have a pair of AHB2's powering modified Klipschorns in a large room with proper acoustics, though the end game for these Class D amps, if they actually sound good, is in a home theater setup.
I'm a measurement nut too, though certainly not in Amir's league and have a humble APx515 with pretty much all the options. Yesterday I took delivery of an AUX-0040 switching amp filter and know about this thread's discussion concerning its transfer function. Forgive me if this is old news, but the APx software's "Compare Ratio" plot function is an ideal way to characterize an unknown device (the AUX-0040 in this case), that requires another device for it to function correctly (an amplifier in this case). This function allows you to measure the amp (AHB2 in this case), and divide following measurements of magnitude and phase, etc., by the initial measurement, resulting in a lovely flat line.
Example 1
I made a 5 second 2Hz - 80.1kHz continuous sweep measurement of the AHB2 with an 8Ω resistive load. Then I setup "Compare Ratio" plots of both magnitude and phase that divides future measurements by that first sweep. Simply remeasuring the AHB2 into 8Ω produced these plots as expected:
Example 2
Then I simply inserted the AUX-0040 between the AHB2 with 8Ω load and the analyzer. The result is the switching amp filter's transfer function alone:
Bruce Hofer's work was amazing, the AUX-0040 was his last design prior to retirement. I hope Audio Precision can carry on without him.
God bless you and your precious family - Langston
My first post. I read through most of this thread with interest because I hate Class D amps after two decades running a concert production company where they made everything sound worse, but were lovely for subs. Less weight, less power requirement, fewer amp racks (but you had to keep them away from the wireless mic/IEM racks).
Now I'm retired and finally putting together a nice home audio system and keep reading raves about this amp and its designer, so I'm going to give a pair of NC400's a shot. I have a pair of AHB2's powering modified Klipschorns in a large room with proper acoustics, though the end game for these Class D amps, if they actually sound good, is in a home theater setup.
I'm a measurement nut too, though certainly not in Amir's league and have a humble APx515 with pretty much all the options. Yesterday I took delivery of an AUX-0040 switching amp filter and know about this thread's discussion concerning its transfer function. Forgive me if this is old news, but the APx software's "Compare Ratio" plot function is an ideal way to characterize an unknown device (the AUX-0040 in this case), that requires another device for it to function correctly (an amplifier in this case). This function allows you to measure the amp (AHB2 in this case), and divide following measurements of magnitude and phase, etc., by the initial measurement, resulting in a lovely flat line.
Example 1
I made a 5 second 2Hz - 80.1kHz continuous sweep measurement of the AHB2 with an 8Ω resistive load. Then I setup "Compare Ratio" plots of both magnitude and phase that divides future measurements by that first sweep. Simply remeasuring the AHB2 into 8Ω produced these plots as expected:
Example 2
Then I simply inserted the AUX-0040 between the AHB2 with 8Ω load and the analyzer. The result is the switching amp filter's transfer function alone:
Bruce Hofer's work was amazing, the AUX-0040 was his last design prior to retirement. I hope Audio Precision can carry on without him.
God bless you and your precious family - Langston
Last edited:
Not quite sure I followed you, but the Benchmark AHB2 is not a class-D amplifier... The NC400 is, however.
So true! Sorry for the confusion. I definitely did not want to use a Class D amp to characterize the transfer function of the AUX-0040 filter because the necessary first step is to derive a reference measurement of an amp without the filter involved. A Class D amp without this filter would of course exceed the slew rate limitations of the input of my analyzer, thus would not be a proper choice for this procedure.
Actually, any non-switch mode amp (with an output impedance less than 2.5Ω per the AUX-0040's requirements) would be fine for the reference since it's divided out to zero (flat) anyway.
Please fire away with questions if I'm still muddying up the waters. : )
God bless you and your precious family - Langston
What a great support for the point I'm making elsewhere that the measurements we see here on ASR, while exceptional and unique, don't in all cases completely define the quality of a component's output. Hysteresis distortion isn't something @amirm tests for, AFAIK.
Hysteresis shows up in distortion measurements. But, while it exists, the big breakthrough years ago in class D amplifiers were schemes to include the output filter inductor in the feedback loop, which suppresses that (and all other) distortion. Or you could do as another poster and replace it with an air-core inductor...
Yes, I don't mean to imply that hysteresis distortion doesn't show up in the measurements. But the fact that the excellent measuring Hypex amp displayed an unpleasant audible consequence ("graininess") notwithstanding its superlow THD suggests that different flavors of distortion can have adverse impacts on sound quality even if they are at "inaudible" levels.
Putzeys was very clear that reports of audible nonideality were the reason he investigated the problem to begin with, even in Class D amps that measured very well.
‘You say: “ I don't mean to imply that hysteresis distortion doesn't show up in the measurements.” That is exactly what you Are implying!
You say: “ the fact that the excellent measuring Hypex amp displayed an unpleasant audible consequence ("graininess")”.
As this is an official Review thread. Either provide proof/evidence of this “graininess” you claim exists but can’t be measured. In this Specific product, not some generalized unmeasurable anti matter reality. Not more words, actual evidence or proof that other Scientists/Engineers have detected and published these findings. Otherwise I suggest you move on and quickly.
Last edited:
The "graininess" wasn't my comment; it came from Mr. Putzeys. But I don't claim that it related to this specific amplifier. If anyone inferred that to be my claim, I apologize.
Can you help here and post a link to this referenced comment by Mr. Putzeys please.
He's talked about it many times; here are a few examples.
https://audioxpress.com/news/purifi-audio-unveils-investigation-about-hysteresis-distortion
https://www.purifi-audio.com/wp-con...on-The-Sound-That-Dare-Not-Speak-Its-Name.pdf
https://repforums.prosoundweb.com/index.php?topic=16944.15
https://audioxpress.com/article/purifi-audio-a-straight-wire-to-the-soul-of-music
Last edited:
Forgive my ignorance, but since the Putzeys quote in Audioexpress mentions that ferromagnetic material in speakers (presumably drivers, possibly also binding posts etc) could also be problematic, is there any point in worrying about eliminating them from one link further back in the audio chain (the amp)? Do the inductors in eg preamps also matter?
Seems to me that you're anything but ignorant. Bruce Hofer, the guy that made Audio Precision what it is today focused on Class D amplifier measurement and distortion just before retirement. The external low-pass filter that Amir uses when he measures Class D amps was Mr. Hofer's final design there. I'm saying all this because he's one of the most studied people on the subject and he thinks the key to Class D's problems lay with inductors and inductances in audio equipment.
Check out what Hofer says on page 3 under the section titled "Switch-Mode Amplifier Review".
God bless you and your precious family - Langston
Last edited:
Thanks for the kind words @Langston Holland .
It seems most arguments about audibility are doomed to go nowhere, which is why I appreciate ASR and its focus on objective assessments (thanks for getting this thread back on track @AdamG247 by the way).
That Hofer paper is very interesting. Apart from enlightening me as to the significant HF output of switching amps, it makes a refreshing change from the undergrad law papers I have to mark/grade at the moment.
Greetings from Melbourne (Australia, not Florida).
It seems most arguments about audibility are doomed to go nowhere, which is why I appreciate ASR and its focus on objective assessments (thanks for getting this thread back on track @AdamG247 by the way).
That Hofer paper is very interesting. Apart from enlightening me as to the significant HF output of switching amps, it makes a refreshing change from the undergrad law papers I have to mark/grade at the moment.
Greetings from Melbourne (Australia, not Florida).
Similar threads
