|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
173.52.246.134
In Reply to: RE: Recent and future amplifier measurements posted by Frihed89 on November 06, 2015 at 11:35:37
>Judging from the latest Stereophile that reached my doorstep in Denmark,
>every "modern" solid state amplifier that JA measures from now on will be
>"well engineered", no matter what the price.
We are now into the third generation of solid-state amplifier designers
and each generation has learned from the previous one. It is rare, these
days, to find a solid-state amplifier that isn't well-engineered, unless
it has been compromised for reasons of cost.
>Isn't it time, now, John to measure some of the newest Class D amplifiers
>to see how far they have come (or not),
See the linked review below for my recent measurements of a modern,
no-compromise class-D design. (Scroll down past the digital section
measurements.)
John Atkinson
Editor, Stereophile
Follow Ups:
The method of testing audio power amplifiers is invalid because it does not simulate the way amplifiers are used in the real world Connecting the output to a 8, 4, or 2 ohm resistor is a walk in the park compared to what amplifiers really face. For example, to get speaker FR close to flat, most loudspeakers use very complex crossover networks that are highly reactive, not purely resistive. Look at the impedance measurements of YG Acoustics Sonja 1.1 and 1.3. The 1.1 has a reactive load that varies from 90 degrees capacitive in the bass to 33 degrees inductive in the treble, a very difficult load for any amplifier. To get that capacitive, the woofer must have had one humongous capacitor across its voice coil to filter out high frequencies. The 1.3 is better but even there you reported that of the three very expensive amplifiers you had, the best of them costing $44,000 had problems with it. The other factor is reverse EMF especially from woofers. Energy stored in the mechanical suspension as potential energy at its greatest point of excursion on each half cycle is converted back to electrical energy as it travels back through its zero crossing point. This generates a voltage that bucks the amplifier. Depending on the amplifier's power supply this can alter the bias voltage applied to the active devices.
What do these factors alter? Potentially everything including FR and non linear distortion, even short term stability. Exotic wires with highly reactive characteristics only make matters worse. This can easily be seen on an oscilloscope by looking at the output with a square wave input and observing the change in the waveform shape when a speaker is connected. This is why amplifiers that seem to measure about the same on a test bench sound different from each other.
Why are these measurements used? They are a legacy handed down from early times like the 1930s and 1940s when differences between amplifiers were so huge and the loads so simple that the tests showed meaningful differences. But those tests lost their value with time and tell us very little about real worth anymore. Unfortunately no one has bothered to invent better tests that are sensitive to meaningful differences in actual use. Therefore the measurements we do get are of little value.
Mark, at 90 degrees NO POWER is being delivered to the load.Can a wacky speaker crossover actually get to that point? Probably, but what would be the point?
As for a BETTER system?
Of course there are. A quick Google will show you several proposed 'real' speaker loads of known characteristics.
A 'PowerCube' device is availble to load an amp with inductive and capactive loads to +-60degrees at 2 to 8 ohms.
The resulting graphic tells a very revealing story.SEE LINK::
Too much is never enough
Edits: 11/10/15
The goal is to get the speaker system to sound the way the designer wants it to. To do that he invariably must include a filter network, an equalizer. He can do it at the low signal level cheaply, efficiently, precisely where it has no adverse affect on other equipment or he can do it at the amplifier output stage which is the worst possible way. But since audiophiles expect speakers to sound the way they want them to without any equalizers, tone controls, or other active filters in the low level signal path, we get these awful crossover networks that require super stable amplifiers to cope with them. That makes amplifier manufacturers very happy. It justifies a $40,000 amplifier instead of a $400 amplifier.
Really now , so active xovers and EQ have no sonic impact on the signal , must look into that...
:)
Go Rossi ......
Edits: 11/12/15
!
The Mind has No Firewall~ U.S. Army War College.
Simple loads indicate whether or not an amp is performing as designed and give some indication of the performance envelope. If there is a problem with simple loads, an amp certainly won't perform better with more difficult loads. When I see audible harmonics in the distortion spectra into a simple load, the amp will behave even worse into a more complex load and it's time to move on...
try it! you know you want to!
iF the distortion spectrum is 110 db below the principal tone, does it matter what shape it has?
110 fucking dB?
On the other hand humans are pretty insensitive to low order distortions far above measurabl thresholds so plenty of poorly measuring amps can sound not too bad at all. otoh, it sure seems that modern class D has perfectly good sonic performance, and unbeatable power efficiency, which actually matters to me.
iF the distortion spectrum is 110 db below the principal tone, does it matter what shape it has?
Your assumption is that with dynamic musical content - quite different from a sine wave - is that the performance will necessarily be the same. Engineers like Nelson Pass have proven otherwise.
" The resulting complexity creates distortion which is unlike the simple harmonics associated with musical instruments, and we see that these complex waves can gather to create the occasional tsunami of distortion, peaking at values far above those imagined by the distortion specifications."
Remember the old Crown gear like I used in high school with 0.05% THD/IMD? That should have been inaudible, right? Surely you don't believe that!
but some can't grasp the concept that what you can't hear doesn't matter...
try it! you know you want to!
...yes, they do seem to work great on Maggies.
The flaw is particularly obvious with class D amplifiers which are actually Pulse Width Modulation amplifiers, not digital amplifiers. The output has a series inductor to suppress high frequency noise. How that operates depends entirely on the load especially if it is reactive. It is therefore impossible to generalize how such an amplifier will sound with a given speaker system unless you know the X and R value of the series inductor, the complex impedance of the speaker system, the complex impedance of the speaker wire, and how to perform the network calculations. The YG Sonja 1.1 shows how drastically wide the range could be. Some speaker wires are so reactive they've been known to blow up amplifiers by simply creating a tank circuit at ultrasonic frequencies, a load amplifiers are not designed to handle.
performance is load independent?
try it! you know you want to!
Old news to anyone who has read ANY lab test data on these amps.
Check out the datasheet for the B&O ASP modules. While not 'latest', do show this effect in spades.
Too much is never enough
Oh, so now an audiophile is extolling the virtues of negative feedback. What a laugh. Negative feedback is supposed to be audiophile poison. Now suddenly it's the medicine that cures all ills. Well, which is it?
then have a listen. Then get back to use with an educated opinion...
try it! you know you want to!
Someone who can QUOTE Bruno about NFB should chime in. I think it boils down to a 'little' being no good, but a LOT being OK. I heard the amount of 60db bandied about but could be mistook.
Too much is never enough
Well .... listen some more and give us your "feedback " ......, :)
Go Rossi ......
I owned a 'd' amp for several years. The PSAudio GCC250 'integrated'. It was a fine piece and needed ONLY to have a stuck switch (button) freed and deoxit'ed.
But, it also did something to the HIGHS with my panels. I don't know the nature of the reaction but it was LONG TERM unsatisfying. This after MUCH adjustment to panel location. And I MEAN much. toe? Tweeter I/O? Flip 'em front to back? Spacing? All messed with and setups lasted anywhere from 2 minutes to 3 or 4 months.
I believe I gave 'd' a fair shot.
Where I LEFT them, which sounded best, also seemed pretty close (within cm) of the best place when I went to a Pair-O-Parasound.
Thus the need to move toward a line level crossover!
Too much is never enough
I don't know the nature of the reaction but it was LONG TERM unsatisfying.
Agreed. Clear and bright, but unnatural.
nt
try it! you know you want to!
Please, Kur, go find a review with Measured Data.
Frequency Response FALLS more quickly with lower impedance loads on amps with the output Zobel.
This falling response ALSO induces Phase Problems which MAY be audible.
Too much is never enough
Where? Oh, you mean out past 30KHz?
try it! you know you want to!
To also cherry pick the data, the ASP 500 module, the 'big guy' in the series was down 1/2db or so at 20khz. This produces notable phase shift in the audio band. And, if it matters, really odd looking 10khz square wave response.
I'm not going to try to find the measured data. You can find it if you are really interested.
Too much is never enough
hard to prove it's audibility at high frequencies where moving your head a few centimeters produced much more phase shift.
"The hardest thing of all is to find a black cat in a dark room, especially if there is no cat" - Confucius
I stated that the ncore uses post filter feedback and thus the frequency response was load independent. The graph I posted shows that....Half dB down at 20KHz might bother the dog but I can't hear it and I doubt anyone reading this can either...
try it! you know you want to!
If you bothered to READ my post, I stipulated the B&O ASP module.
Sure, other 'd' designs, notably the ncore have skipped the output filtration.
And it's not the amount 'down' that counts, but rather the phase shift this induces. I believe this was what caused problems with my panels. There was something just not right about the highs.
Too much is never enough
If you had bothered to read my post, you might have noticed it was a reply concerning class d amps in general that contain the output filter in the feedback loop. Ncore doesn't "skip" output filtration, it's in the feedback loop, same as with Icepower, by the way....though they take feedback both pre and post output filter.
try it! you know you want to!
- The industry is **not** listening, if you will pardon the expression :)
John,
Have you had a go with the black in your setup ? Also its pretty obvious even with filter, class-D has serious ussues with Square and sine Waves , what are your thoughts on this ..
Regards
Go Rossi ......
> Have you had a go with the [Bel Canto] Black in your setup?
See my comments at the link below.
> Also it's pretty obvious even with filter, class-D has serious issues with
> square and sine waves, what are your thoughts on this...
Not with sinewaves in the audioband. And the squarewaves are dominated by
ultrasonic noise which, being random, should not have any audible
consequences (unless a source component or preamplifier is not adequately
shielded against such noise when radiated by the speaker leads.)
John Atkinson
Editor, Stereophile
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: