AudioCircle
Other Stuff => Archived Circles => Hypex Owners Circle => Topic started by: mgalusha on 4 Jun 2012, 02:08 am
-
For anyone who is interested, the FFT plot below if from an NC400/SMPS600 combination, 1W output power into an 8 ohm load.
The 2nd order harmonic is about 110dB down, 3rd and 4th harmonic at perhaps -115dB and higher orders even lower or in the noise floor.
Plot generated using Prism Sound dScope III.
(http://www.audiocircle.com/image.php?id=63435)
-
:o
-
:o :o
:D
:thumb:
Anand.
-
Mike, the FFT looks pretty good, the higher harmonics should be attenuated by the output filter so they will look increasingly better as the frequency goes up.
It's certainly hard to find fault with how this amp measures.
Thanks for the new data, this is the first FFT I have seen of Brunos's new baby.
Scotty
-
Whoa.
-
Thanks Mike!
-
you know, you can't say hypex without...hype :green:
-
The distribution of the distortion harmonics looks favorable, if the relation of harmonics remains as output level is reduced that could explain a lot of good sound people report.
-
WOW!!! Thats an incredibly cool new toy tool.
http://www.prismsound.com/test_measure/products_subs/dscope/dscope_functions.php
(http://www.audiocircle.com/image.php?id=63472)
-
WOW!!! Thats an incredibly cool new toy tool.
http://www.prismsound.com/test_measure/products_subs/dscope/dscope_functions.php
(http://www.audiocircle.com/image.php?id=63472)
I'd be totally geeked out just looking at it, much less trying to figure out how to use it. There was a happy time in my engineering career when the boss would come in some time in December and say "we still have $xyz budget left to spend for the year, so what should we buy?". This would definitely make that kind of list.
So many uses for it and, well, it's just way cool :thumb:
-
Very cool, Mike!
Let me know if you run out of things to measure... I'd be happy to bring random pieces of gear by to be measured just for the curiosity and coolness of it! :)
-
So do the plots that Mike posted surprise anyone? Are they difft than what the mfgr has posted?
-
Moreso for speakers, but I think a lot of folks regard manufacturers' plots as suspect. Data can be goosed to look favorable by interested parties...
-
Data can be goosed to look favorable by interested parties...
I have no interest other than seeking the facts. That is as raw as it gets, I have just barely figured out how to measure even that, no diddling with the number of data points, no smoothing, just a raw trace exported to a jpg. Given than Jan Didden posted a couple weeks ago about attending a seminar where Bruno demonstrated the measurements were indeed real and Jan made strong mention of the fact that the distortion actually dropped at higher frequencies vs rising as do almost all amplifiers, I am pretty certain the Hypex are not tweaking the data to make it look good.
mike
Edit, btw, not trying to hammer on roscoeiii, just trying to be straight.
-
(http://www.prismsound.com/test_measure/graphics/prod_photos/ds3_rp_sm.jpg)
So what kinda aftermarket power cord are you using on that ? :lol:
-
Yeah.... the one that came in the box. :)
-
I have no interest other than seeking the facts.
Not to mention that spending that much dough on that piece of test equipment to muck with the numbers would be.....
Not so sure that people realize how expensive this tool is. :o
-
How expensive is it?
-
How expensive is it?
It was $10,000 four years ago:
http://www.tmworld.com/article/317063-Prism_Sound_slashes_price_of_audio_analyzer.php
-
Mike,
Never meant to insinuate that you might be goosing anything. Just wanted to suggest why people might not trust manufacturer measurements, since they have a very vested interest in presenting their products in the best possible light.
The measurements in this thread are very much appreciated.
Best,
Roscoe
-
It was $10,000 four years ago:
http://www.tmworld.com/article/317063-Prism_Sound_slashes_price_of_audio_analyzer.php
Makes SpectraPlus-DT (http://www.telebyte.com/pioneer/FeaturesAndSpecifications_DT.htm) look like a genuine bargain. :wink:
-
Makes SpectraPlus-DT (http://www.telebyte.com/pioneer/FeaturesAndSpecifications_DT.htm) look like a genuine bargain. :wink:
A very different instrument though there is some overlap. One can purchase the dScope without the digital analysis options and save either $2K or $4K depending on other options.
-
Mike, the FFT looks pretty good, the higher harmonics should be attenuated by the output filter so they will look increasingly better as the frequency goes up.
That should only relevant above 20k. Anything lower than is in the audio band and so unattenuated
It's certainly hard to find fault with how this amp measures.
Thanks for the new data, this is the first FFT I have seen of Brunos's new baby.
Scotty
agreed
-
Here is a link to a Harmonic overtone calculator. http://www.sengpielaudio.com/calculator-harmonics.htm
As you can see from the calculator about the time you reach 5Khz to 6kHz the higher harmonics above the 3rd overtone fall at or above 20kHz.
The output filter will quickly become a factor in your measurements as you look at the THD the amplifier might have at frequencies above 5Khz.
Of course the bottom line is, the amplifier sounds good in addition to doing well on the test bench.
Scotty
-
Here is a link to a Harmonic overtone calculator. http://www.sengpielaudio.com/calculator-harmonics.htm
As you can see from the calculator about the time you reach 5Khz to 6kHz the higher harmonics above the 3rd overtone fall at or above 20kHz.
The output filter will quickly become a factor in your measurements as you look at the THD the amplifier might have at frequencies above 5Khz.
Of course the bottom line is, the amplifier sounds good in addition to doing well on the test bench.
Scotty
Okay, to be pedantic. Mike's original plot was for a 1kHz fundamental. Harmonics will occur at every 1 khz. The output filter won't have any attenuation until (approximately) the 20th harmonic (20 khz)
For higher fundamentals (eg 6 khz) the output filter won't have an effect until the frequency of the harmonic exceeds the corner frequency of the filter, or. 20 khz, which for 6 khz will be 4th harmonic (24 khz) upwards
So the output filter won't have any practical effect on the harmonics in Mike original plot
-
Correctamundo, Mikes 1kHz FFT will not show the effects of the filter. The filters effect however can be seen on the THD measurements presented in the Ncore data sheets.
For higher fundamentals (eg 6 khz) the output filter won't have an effect until the frequency of the harmonic exceeds the corner frequency of the filter, or. 20 khz, which for 6 khz will be 4th harmonic (24 khz) upwards
From the the online calculator the 3rd harmonic overtone falls at 24kHz.
From the THD graph in the Ncore's datasheet, the output filter's effect on the amplifiers high frequency response can partially explain the reduction of THD seen in the graph as the frequency increases.
Scotty
-
Thanks for posting this! Those are very good measurements. Is the noise just over -115 dB part of the noise floor of the measuring device, or is that the noise floor of the amp?
-
Thanks for posting this! Those are very good measurements. Is the noise just over -115 dB part of the noise floor of the measuring device, or is that the noise floor of the amp?
The dScope III has a rated noise floor of -115dB. Not quite as good as the Audio Precision 2700 series but awfully close, the AP is -117.8dB but is about 4x the cost. :o
If I measured at a higher power level it would likely show a greater S/N ratio since the signal would be higher. I'm still learning how to use it but getting there. :)
-
Correctamundo, Mikes 1kHz FFT will not show the effects of the filter. The filters effect however can be seen on the THD measurements presented in the Ncore data sheets. From the the online calculator the 3rd harmonic overtone falls at 24kHz.
The fundamental, the base frequency, is the '1st harmonic' eg. 6kHz
The 2nd harmonic is 2x base freq ie 12kHz
so 4th harmonic is 4 x 6 = 24kHz
That's why measurements of harmonic distortion always start with '2nd harmonic distortion' and go up
The 'harmonic' is different to the 'overtone'
The 'overtone' is always 1 number lower than the harmonic, ie. the 3rd overtone is the 4th harmonic
So, for example, at 6kHz there will be no attenuation of any 3rd harmonic distortion
From the THD graph in the Ncore's datasheet, the output filter's effect on the amplifiers high frequency response can partially explain the reduction of THD seen in the graph as the frequency increases.
Scotty
That will also apply to any amp with hi-freq roll-off (including most/any class D?)
However I believe the good hf distortion figures of the ncore (good distortion figures across the whole audio band) really are because it's highly linear, ie. it really has low distortion. It has low open loop distortion; and it has even lower closed loop distortion, because it has lots of feedback. The hf filter will help but isn't the underlying reason
This is why Bruno/hypex publish the 18.5k + 19.5k Intermodulation distortion plots - they let you see the hf distortion reflected back down into the audio band
PS. the freq response is less than 1dB down at 24kHz and less than 2dB down at 48kHz, so it won't be having that great an effect even on the 6kHz 8th harmoinc distortion
The freq response is still less than 3dB down at 60kHz, the 10th harmonic of 6kHz
-
Hi guys,
on this link there is a small signal frequency response for all loads (8, 4, and 2 oHms), but I`m curious what the large signal response would tell for the same loads, specifically, what would be the electrical power into 8, 4,and 2 oHms for 20Hz, 1kHz and 20kHz?
http://www.hypex.nl/docs/nc1200%20folder%20web.pdf
The reason I`m asking is I have noticed non-linear behavior at small and large listening levels with some class-D amps. Some of them sound detailed at low listening levels, but much less at higher ones.
I hope I can get a straight answer.
Thanks!
-
Hi guys,
on this link there is a small signal frequency response for all loads (8, 4, and 2 oHms), but I`m curious what the large signal response would tell for the same loads, specifically, what would be the electrical power into 8, 4,and 2 oHms for 20Hz, 1kHz and 20kHz?
http://www.hypex.nl/docs/nc1200%20folder%20web.pdf
The reason I`m asking is I have noticed non-linear behavior at small and large listening levels with some class-D amps. Some of them sound detailed at low listening levels, but much less at higher ones.
I hope I can get a straight answer.
Thanks!
I'm not sure why you wouldn't get a straight answer.
Have you looked at this link, and in particular the PDF file, section 9 (Figure 9.2?):
http://www.audiocircle.com/index.php?topic=108251.0
Best,
Anand.
-
That`s THD versus frequency and it tells nothing about the frequency linearity into various loads at full power.
I would like to see the same frequency graph as in in 9.1, but measured at full power.
-
Got it. Sorry, can't find an equivalent with just FR measured...might want to email Bruno directly...
Anand.
-
That`s THD versus frequency and it tells nothing about the frequency linearity into various loads at full power.
I would like to see the same frequency graph as in in 9.1, but measured at full power.
Maybe, if I get some time. I already owe a friend some measurements that I haven't gotten done yet. :duh:
So you want to see THD vs Freq at 200W into 8R? By definition that is the clipping point, so it would be more relevant at something less than full power, especially, IMO, as 99.999% of all listening is done at low to medium power, at least in my world. :)
-
Got it. Sorry, can't find an equivalent with just FR measured...might want to email Bruno directly...
Anand.
Maybe the measurements show the same what I did noticed.
:roll:
-
Maybe, if I get some time. I already owe a friend some measurements that I haven't gotten done yet. :duh:
So you want to see THD vs Freq at 200W into 8R? By definition that is the clipping point, so it would be more relevant at something less than full power, especially, IMO, as 99.999% of all listening is done at low to medium power, at least in my world. :)
no THD, just like I said above.
Freq response at 8,4 and 2 Ohm at full power
The frequency response seems to change with different amount of power. As it does with different impendance.
Thanks!
-
Are you assuming there's a correlation between subjective "sound detail" at low or high listening levels and actual objective non-linear behavior? I'm not sure a straight answer is attainable when the premise is rather gray. :)
Anyways, it would be interesting to see the type of test results you're mentioning. However, it's going to require someone to fabricate a hefty dummy load and possibly reconfigure their NC400 mounting scheme since this testing scenario would be completely different from normal usage. Type and capability of the power supply could become a big factor in this type of testing also.
Cheers,
Dave.
-
I have a load capable of absorbing over 4kW, so that part isn't a problem but I may not run it at full power. While I doubt the amp would fail, I am not willing to sacrifice it for something that is not a real world scenario, IE: full power listening. Per the data sheet, the average level may reach 1/3rd of full power when peak clipping is occurring, so in the real world 1/3 rated power for a FR graph would be more useful. However, the datasheet also specifies the FR from 0 to 50kHz as +0/-3dB, all loads.