• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Amplifiers Which Test Well, Under $1,000 USD

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,961
Likes
3,842
Location
Perth Western Australia
#41
Last edited:

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,190
Likes
2,281
Location
Zg, Cro
#42
Here's another good bit of info on the topic.
Any minimum-phase deviation from linearity in the magnitude response of a signal creates a related deviation in the phase response. This deviation in the phase response theoretically extends from DC (zero) to infinity, but of course becomes unmeasurable or at lest negligible at a certain point. This phase deviation is due to the relationship between phase and frequency. Unfortunately I've only ever seen this relationship explained in maths, which puts it in the category of things I accept, but can't fully understand :p
You are correct. If there is, as in your example, a capacitor in the signal path acting as AC coupler it may easilly affect phase in excess of 100Hz which will then affect how harmonics add up with the 100Hz base signal and the result may be as shown in the document you linked. Transfer function of such RC element would give exact figures and show what could be expected as an impact on output, but as I went digital at 3rd year of university education I forgot those things.. :p
 
Joined
Jan 11, 2019
Messages
79
Likes
76
#43
So that looks a lot like the nCore datasheet, which is good. In the real world, the amps always show a precipitous rise in THD+N. Well, okay, if precipitous is up to .15% at 20kHz. Still, a well-designed AB keeps this down to .01% or less. The rise normally starts right around 6kHz. I've never found a solid explanation of the divergence between datasheet and actual measurements, and I've always found that rise curious, since the amp generally has significantly better IMD performance than the THD rise would suggest. Is your measurement bandwith highly restricted here, or did you manage to do something different that solves this, or does everyone just measure them wrong?
 

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,961
Likes
3,842
Location
Perth Western Australia
#44
So that looks a lot like the nCore datasheet, which is good. In the real world, the amps always show a precipitous rise in THD+N. Well, okay, if precipitous is up to .15% at 20kHz. Still, a well-designed AB keeps this down to .01% or less. The rise normally starts right around 6kHz. I've never found a solid explanation of the divergence between datasheet and actual measurements, and I've always found that rise curious, since the amp generally has significantly better IMD performance than the THD rise would suggest. Is your measurement bandwith highly restricted here, or did you manage to do something different that solves this, or does everyone just measure them wrong?
This is Hypex data. Please note I mentioned that independent test data is coming soon :)

Regarding measurement bandwidth, we have had a good discussion on this elsewhere (NC400 test thread). My view is that using THD measurement bandwidth much beyond the audible range is misleading. As an example if you look at the NC400 test data Amir took you will see that with a fundamental of say 20kHz you will obviously get 2nd and 3rd harmonics at 40 and 60kHz. This is what tends to make class D look a bit worse than say typical A/B in terms of THD and THD+N if you measure to 96kHz bandwidth.

However, quite obviously, you can neither hear 40 or 60kHz nor can your speakers reproduce it at any significant level. Therefore in audible terms the measurement is erroneous. The inaudible and un-reproduceable harmonics have no audible detriment. If we want to see whats audible we need to look within the audible range. In which the P252 manages to stay around 0.001% THD + N (100dB SINAD) at half power.

So IMO THD and THD +N measurements should be limited to a bit beyond the audible range, say 24kHz which ties in with 48 kHz sample rate in test equipment.

Hypex also take this view and limit their measurements to the audible band. The cynical will say this is just to make the performance look better, however I have yet to hear a convincing argument (or any argument at all for that matter) against the fundamental principle I discuss above.

The real tough test is the 19+20kHz IMD test.
 
Last edited:
Joined
Jan 11, 2019
Messages
79
Likes
76
#45
This is Hypex data. Please note I mentioned that independent test data is coming soon :)

Hypex also take this view and limit their measurements to the audible band. The cynical will say this is just to make the performance look better, however I have yet to hear a convincing argument against the fundamental principle I discuss above.

The real tough test is the 19+20kHz IMD test.
I would agree with that last comment 100%. Usually, though, the THD20 tells you if there is some sort of linearity issue with the amp that will show up elsewhere--presumably in the IMD. The way the Hypex measures, there should be a problem (using that term loosely). I know the designer has said that Class D amps can behave differently. I'm not sure I've ever seen a full explanation of why this is though. It would be interesting to see a power swept 19+20kHZ IMD, plus the square wave multitone at a variety of power levels. It's just a little discordant to see an amplifier that appears to have extremely low IMD on the datasheets have THD that is anything but.

Interestingly, 19kHz+20kHZ IMD measurements in the real world often haven't stacked up to the datasheet. All of them Stereophile has measured show significant sidebands in the area of -85 to -95dB--not the -110dB Hypex claims. Hypex takes the IMD measurement another 500Hz higher, and I have to wonder if there wasn't a sneaky reason for that.... I'm not saying they are not good amps, and perhaps perfectly transparent, but I would also question why they felt the need to "engineer" the measurements in a disingenuous--albeit disclosed--fashion almost no one else does.

I'm also aware that Hypex does not use the AUX-0025 filter while Stereophile does. I am not sure why that would cause an IMD difference, though. I suspect if you got down and dirty for awhile with some decent test gear you could figure out just what is going on, if anything. Suffice to say that I am a little troubled that an amp on a board never manages to measure like its datasheet... so far. Hopefully you'll crack that nut.
 
Last edited:

andreasmaaan

Major Contributor
Joined
Jun 19, 2018
Messages
3,517
Likes
2,767
#46
I would agree with that last comment 100%. Usually, though, the THD20 tells you if there is some sort of linearity issue with the amp that will show up elsewhere--presumably in the IMD. The way the Hypex measures, there should be a problem. I know the designer has said that Class D amps can behave differently. I'm not sure I've ever seen a full explanation of why this is though. It would be interesting to see a power swept 19+20kHZ IMD, plus the square wave multitone at a variety of power levels. It's just a little discordant to see an amplifier that appears to have extremely low IMD on the datasheets have THD that is not. Interestingly, though, 19kHz+20kHZ IMD measurements in the real world often haven't stacked up to the datasheet. All of them Stereophile has measured show significant sidebands. Hypex takes the measurement another 500Hz higher, and I have to wonder if there wasn't a sneaky reason for that.... I'm not saying they are not good amps, and perhaps perfectly transparent, but I would also question why they felt the need to "engineer" the measurements in a disingenuous--albeit disclosed--fashion almost no one else does.
Just eyeballing the factory measurements, THD @ 4Ohm looks to be a shade under 0.001% in the 18.5-19.5 KHz range.

The IMD test is also conducted into a 4Ohm load and shows IM3 distortion products at around the same level, i.e. (P[2F2-F1]+P[2F1-F2]) looks to be just under 0.001% of P[F1 + F2].

I'm sure I'm missing something? :)
 

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,961
Likes
3,842
Location
Perth Western Australia
#47
I would agree with that last comment 100%. Usually, though, the THD20 tells you if there is some sort of linearity issue with the amp that will show up elsewhere--presumably in the IMD. The way the Hypex measures, there should be a problem (using that term loosely). I know the designer has said that Class D amps can behave differently. I'm not sure I've ever seen a full explanation of why this is though. It would be interesting to see a power swept 19+20kHZ IMD, plus the square wave multitone at a variety of power levels. It's just a little discordant to see an amplifier that appears to have extremely low IMD on the datasheets have THD that is anything but.

Interestingly, 19kHz+20kHZ IMD measurements in the real world often haven't stacked up to the datasheet. All of them Stereophile has measured show significant sidebands in the area of -85 to -95dB--not the -110dB Hypex claims. Hypex takes the IMD measurement another 500Hz higher, and I have to wonder if there wasn't a sneaky reason for that.... I'm not saying they are not good amps, and perhaps perfectly transparent, but I would also question why they felt the need to "engineer" the measurements in a disingenuous--albeit disclosed--fashion almost no one else does.

I'm also aware that Hypex does not use the AUX-0025 filter while Stereophile does. I am not sure why that would cause an IMD difference, though. I suspect if you got down and dirty for awhile with some decent test gear you could figure out just what is going on, if anything. Suffice to say that I am a little troubled that an amp on a board never manages to measure like its datasheet... so far. Hopefully you'll crack that nut.
Sorry Im puzzled why you think the THD is "anything but" low.

Your previous post quoted THD+N at 20kHz as 0.15% when it is actually more like 0.0015% (4ohm 1w or Pr/2)

I know this isnt the same amp but Amirs independent measurements of the NC400 show strong correlation with Hypex measurements (note that Amirs test amp was PSU limited at high powers) THD V Power. Admittedly less so with THD V Freq, but it is stated as THD not THD +N.

Also I beleive he used 90kHz BW for that test, which demonstrates my point.. So Im not sure there is reason for any major suspicion regarding Hypex data..

Capturenc400.PNG


Capture 2 nc400.PNG


Hypex nc400 amplifier distortion vs frequency at different levels measurements.png




CaptureF.PNG
 
Last edited:

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,961
Likes
3,842
Location
Perth Western Australia
#48
Just eyeballing the factory measurements, THD @ 4Ohm looks to be a shade under 0.001% in the 18.5-19.5 KHz range.

The IMD test is also conducted into a 4Ohm load and shows IM3 distortion products at around the same level, i.e. (P[2F2-F1]+P[2F1-F2]) looks to be just under 0.001% of P[F1 + F2].

I'm sure I'm missing something? :)
Im not sure quite what you are getting at, (it was a very late night, bit fuzzy this morning so excuse me if being dim :) )
 
Last edited:

andreasmaaan

Major Contributor
Joined
Jun 19, 2018
Messages
3,517
Likes
2,767
#49
Im not sure quite what you are getting at, (it was a very late night, bit fuzzy this morning so excuse me if being dim :) )
Late at night here too, and I was lazy finding the right keys on the keyboard :) Just that IMD and THD seem to both be in the same ballpark in the 18.5-19.5KHz range (just under 0.001%).

Still can't find the right keys, but was trying to get at this kinda thing (and P was meant to be "power"):

1550024789040.png
 

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,961
Likes
3,842
Location
Perth Western Australia
#50
Late at night here too, and I was lazy finding the right keys on the keyboard :) Just that IMD and THD seem to both be in the same ballpark in the 18.5-19.5KHz range (just under 0.001%).

Still can't find the right keys, but was trying to get at this kinda thing (and P was meant to be "power"):

View attachment 21826
Guess what, I was just looking at the same picture hahahaha

Is that fundamentally wrong?

AES17

8.2 Intermodulation (IM) measurements
IM measurements shall be performed with a twin tone signal, one tone at the upper band-edge frequency and
one tone 2 kHz below this frequency. The amplitudes shall be in a 1:1 amplitude ratio, the peak amplitude
being adjusted to equal the peak amplitude of an equivalent sine wave at the full-scale level. The output signal
shall be measured with a spectrum analyzer or narrow band-pass filter to obtain the level of the second- and
third-order difference frequency components. Their r.m.s. sum shall be reported in decibels relative to the
output level.

so we need to sum the RMS of the (in band) 2nd and 3rd order difference signals an ref the overall level.
 

andreasmaaan

Major Contributor
Joined
Jun 19, 2018
Messages
3,517
Likes
2,767
#51
Guess what, I was just looking at the same picture hahahaha

Is that fundamentally wrong?

AES17

8.2 Intermodulation (IM) measurements
IM measurements shall be performed with a twin tone signal, one tone at the upper band-edge frequency and
one tone 2 kHz below this frequency. The amplitudes shall be in a 1:1 amplitude ratio, the peak amplitude
being adjusted to equal the peak amplitude of an equivalent sine wave at the full-scale level. The output signal
shall be measured with a spectrum analyzer or narrow band-pass filter to obtain the level of the second- and
third-order difference frequency components. Their r.m.s. sum shall be reported in decibels relative to the
output level.

so we need to sum the RMS of the (in band) 2nd and 3rd order difference signals an ref the overall level.
Exactly :)
 

Ilkless

Active Member
Joined
Jan 26, 2019
Messages
224
Likes
305
#52
For members with 230V in power line :)

CustomWorks HYPA 10 power amplifier (915 EUR) offers superb measurements into 3.9 Ohm providing 60W:
- https://www.customworks.cz/amplifiers/hypa-10 (for measurements open "Parametry" tab and scroll down)
Very interesting to hear about Central and Eastern European manufacturers that go under the radar - Poland, Austria, the Czech Republic, Serbia and Ukraine off the top of my head have a surprising amount of audio manufacturing going on and I'm always intrigued by the well-engineered stuff that gets mentioned occasionally (not so much the Lampizators of the world).

Germany also has many more manufacturers focusing on the regional and domestic market - especially in (semi-)pro audio - than I expected.
 
Joined
Jan 11, 2019
Messages
79
Likes
76
#54
Just eyeballing the factory measurements, THD @ 4Ohm looks to be a shade under 0.001% in the 18.5-19.5 KHz range.

The IMD test is also conducted into a 4Ohm load and shows IM3 distortion products at around the same level, i.e. (P[2F2-F1]+P[2F1-F2]) looks to be just under 0.001% of P[F1 + F2].

I'm sure I'm missing something? :)
Read the asterisk on the datasheets. They disclose on at least one of them that after 6kHZ or so, they substitute in their IMD measurements. Review by separate persons have all measured at least .1% to .2% at 20kHz, which is not SOTA by a mile. The nc400 measurements you posted above that Amir did show the same thing, albeit not as drastic. Look at the NAD M22 for a representative IMD measurements. Posted third party measurements of Hypex products have never (to date) measured to the datasheets. They are still good, but I find it odd that a product that is little more than an amp-on-a-board has that issue, repeatedly (including in the designers own Mola Mola amps): https://www.soundstage.com/index.ph...la-kaluga-mono-amplifiers&catid=97&Itemid=154. I'd love it if someone would spend some serious time measuring one of these things to figure out why. Are there repeated implementation errors, measurement errors, or are the datasheets just a bit optimistic?
 

andreasmaaan

Major Contributor
Joined
Jun 19, 2018
Messages
3,517
Likes
2,767
#55
Read the asterisk on the datasheets. They disclose on at least one of them that after 6kHZ or so, they substitute in their IMD measurements. Review by separate persons have all measured at least .1% to .2% at 20kHz, which is not SOTA by a mile. The nc400 measurements you posted above that Amir did show the same thing, albeit not as drastic. Look at the NAD M22 for a representative IMD measurements. Posted third party measurements of Hypex products have never (to date) measured to the datasheets. They are still good, but I find it odd that a product that is little more than an amp-on-a-board has that issue, repeatedly (including in the designers own Mola Mola amps): https://www.soundstage.com/index.ph...la-kaluga-mono-amplifiers&catid=97&Itemid=154. I'd love it if someone would spend some serious time measuring one of these things to figure out why. Are there repeated implementation errors, measurement errors, or are the datasheets just a bit optimistic?
Ok, wasn't aware of this.

I've found this in the NC400 datasheet:

1550092470644.png


I presume that's what you're talking about?

My interpretation of this is that the two-tone IMD measurement is an actual measurement of the device (albeit using atypical frequencies of 18.5 & 19.5KHz), but that the THD vs frequency measurement is not. This would be consistent with the Soundstage measurements, which show rising THD above 3-5KHz as power increases beyond 1W, but which also show an SMPTE vs power measurement that fairly closely tracks the THD vs power measurement (which I presume was taken at 1KHz).

Personally I wouldn't be worried about harmonics outside the audio band in any case, but it is a little strange that they don't use standard measurements in their datasheets.

It's also strange that this note appears only in the consumer-level NC400 datasheet, but not that of any of the OEM NCore products (eg NC252, NC500). Does this mean it doesn't apply to those amps?

Also puzzling me atm is that Soundstage states they are using an AUX-0025 to do their measurements, which is spec'd as having a 20-20,000Hz passband only. Yet looking at the AP website, I'm struggling to understand what this filter's cutoff is. There are two separate and very different frequency response measurements shown. If the latter is the filters actual response, I can't see how higher order harmonics could even make it into the measurement.

Would be interested to know if others know more...
 
Joined
Jan 11, 2019
Messages
79
Likes
76
#56
Ok, wasn't aware of this.

I've found this in the NC400 datasheet:

View attachment 21881

I presume that's what you're talking about?

Would be interested to know if others know more...
That's it. I dislike that practice, as well as the sketchy 18.5kHz+19.5kHz IMD test points. Couple that with the real world 19+20kHz measurements that have never yet approached those taken 500Hz lower--and which aren't all that impressive, really--and I'm left with the conclusion that Hypex is intentionally shading the truth. Still, given that they still seem to perform well--miraculously so for a Class D--this arguable deception and obfuscation was quite unnecessary. I think they wanted to portray it as being as clean as the best Class AB amplifiers, which it is not. Well, at least not based on a those particular measurements. I would love to see a squarewave style multitone of a Hypex versus something like the Cambridge amps, or a Bryston. That would go a long way toward showing whether Hypex is really all its cracked up to be. For now, though, for under a grand, I would search out a Cambridge.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
21,325
Likes
28,902
Location
Seattle Area
#57
Also puzzling me atm is that Soundstage states they are using an AUX-0025 to do their measurements, which is spec'd as having a 20-20,000Hz passband only. Yet looking at the AP website, I'm struggling to understand what this filter's cutoff is. There are two separate and very different frequency response measurements shown. If the latter is the filters actual response, I can't see how higher order harmonics could even make it into the measurement.
There are two versions of that filter. One that is flat to 20 kHz, and another that is flat to 40 kHz. The latter is what I have and use.

The second set of graphs are highly amplified versions of the same (0.02 dB accuracy).
 

andreasmaaan

Major Contributor
Joined
Jun 19, 2018
Messages
3,517
Likes
2,767
#58
There are two versions of that filter. One that is flat to 20 kHz, and another that is flat to 40 kHz. The latter is what I have and use.

The second set of graphs are highly amplified versions of the same (0.02 dB accuracy).
Ok that clarifies it, thanks. In the first set of graphs, the scale is so large that I couldn't even see the knee that is so visible in the zoomed graphs.

In any case, both the 20K and 40K versions allow enough supersonic content in that you'd expect any harmonics to register in the measurement (as it does in Soundstage's graph, and in your measurement of the DIY NCore you measured).
 

andreasmaaan

Major Contributor
Joined
Jun 19, 2018
Messages
3,517
Likes
2,767
#59
That's it. I dislike that practice, as well as the sketchy 18.5kHz+19.5kHz IMD test points. Couple that with the real world 19+20kHz measurements that have never yet approached those taken 500Hz lower--and which aren't all that impressive, really--and I'm left with the conclusion that Hypex is intentionally shading the truth. Still, given that they still seem to perform well--miraculously so for a Class D--this arguable deception and obfuscation was quite unnecessary. I think they wanted to portray it as being as clean as the best Class AB amplifiers, which it is not. Well, at least not based on a those particular measurements. I would love to see a squarewave style multitone of a Hypex versus something like the Cambridge amps, or a Bryston. That would go a long way toward showing whether Hypex is really all its cracked up to be. For now, though, for under a grand, I would search out a Cambridge.
I share your disenchantment at their sneakily not including a true THD+N measurement in their spec sheets, although if we look at Amir's measurements of the MC400, you see that distortion doesn't begin to rise below 15KHz until more than 8W is put out. Even at higher powers, those harmonics are beyond the audible range and don't bother me. It's the IMD measurement that matters at those frequencies, and that appears to be very clean. But although I don't think it's likely the amp will ever be asked to put out much more than 8W at those frequencies, nor that even .01% IMD would be audible at those frequencies, I'm still a little sceptical about their choice to break from convention and measure IMD at 18.5 and 19.5KHz...
 

Ilkless

Active Member
Joined
Jan 26, 2019
Messages
224
Likes
305
#60
Top Bottom