CHANGSTAR: Audiophile Headphone Reviews and Early 90s Style BBS
Lobby => Amp and DAC Measurements => Topic started by: Marvey on July 23, 2015, 08:32:04 PM
-
Did these for fun. I think most people here have had a chance to hear these amps for themselves.
Two tone tests
1kHz and 123Hz @ ~2V
0dbFS = 20 volts (using 10x probe so QA400 doesn't explode)
100 ohm resistor load.
Note that output had to be reduced with Obj2 to prevent clipping because of its retarded architecture.
Vali is not as bad it it looks. 2nd harmonics are -45db (0.5%) down, and IMD bands are -40db (1%) down. Lots of smaller IM bands from the HD distortion too! I won't kid you as I do feel this distortion is audible. This distortion doesn't take away that the Vali subjectively more lively and resolving, albeit slightly warmer and thicker compared to the excellent measuring Obj2.
P.S.
Distortion measurements should always be on a log scale, never linear percentage. We hear logarithmically, not linearly. Distortion graphs in linear scale can be very misleading and make things look better than they really are. Here are some examples:
http://www.changstar.com/index.php/topic,2105.msg59347.html#msg59347 (http://www.changstar.com/index.php/topic,2105.msg59347.html#msg59347)
http://stereos.about.com/od/Measurements/ss/Oppo-Digital-PM-1-Headphone-Measurements.htm#step5 (http://stereos.about.com/od/Measurements/ss/Oppo-Digital-PM-1-Headphone-Measurements.htm#step5)
-
The distortion of the Vali might probably drop a lot when the output voltage is lowered by 10dB.
How does a TOTL EC amp measure comparatively ?
-
I'll measure the Studio when I get a chance. It's at the lab. We are still checking it over since it was a first build in a new chassis that we just finished the day before the meet.
In the meantime, here is Vali 10db lower @ ~0.6V peak. Both IMD and HD are down another 6db. Same 100ohm load. Not going to bother with Magni 2 or Obj 2 since measurements were already awesome and distortion was far far below what people can hear.
M0 and M2 are the tones. M1 and M5 and their second harmonics. M3 and M4 are the IMD bands.
-
Looking at the measurements: I kinda wish the second channel was connected, to look at the distortion of the crosstalk. John Atkinson has said that the distortion of the crosstalk can be pretty ugly, but he hasn't shown it. Seems possible that the distortion of the crosstalk in a channel could be worse than the inherent distortion of the channel itself, in some instances. Anyway, I should investigate this for myself.
-
Noise profile and distortion for the Vali seem in good agreement with what I had before. Though I used slightly different voltage settings. When comparing to other measurements note the noisefloor is less than -130 from 20Vrms.
It is also nice to see a zoom in plot aroudn the 1kHz frequency. Previously I was using RMAA which did not allow that AFAIK.
The Magni 2 and O2 measurements seem pretty close. O2 seems to measure a hair better.
-
Looking at the measurements: I kinda wish the second channel was connected, to look at the distortion of the crosstalk. John Atkinson has said that the distortion of the crosstalk can be pretty ugly, but he hasn't shown it. Seems possible that the distortion of the crosstalk in a channel could be worse than the inherent distortion of the channel itself, in some instances. Anyway, I should investigate this for myself.
Maybe later. I currently only have one set of differential /10 /100 probes.
-
I'll measure the Studio when I get a chance. It's at the lab. We are still checking it over since it was a first build in a few chassis that we just finished the day before the meet.
Thanks Marv, quite curious how TOTL measures.
-
I assume you don't find the QA190 probe to make much of a difference even with low signal levels?
-
I assume you don't find the QA190 probe to make much of a difference even with low signal levels?
Raises the noise floor about 5-10db on top of the QA400 (already crazy good) depending upon conditions.
-
Here is a follow up with the EC Studio (not yet in production). These measurements were taken at inside my home where the power is worse than where the prior measurements were taken. Same 100ohm load. Single-ended output. /10 probe.
So, you guys are probably thinking "Wow, horrible!". I bet some people are now thinking "These measurements totally explain why the Studio, a tube amp, sounds so great to these moronic audiophools. It's all that euphonic even order harmonic distortion! Yup! Yup! Distortion!"
Well, hold on a minute. Let's think about this. At 2V (probably really fricking loud), the 2nd harmonic (M1) of the 123Hz tone is -63db down. That's 0.07%. Maybe Anaxilus can hear this distortion, but I really doubt noobs like nwaguy has ears well trained enough to hear this. The 2nd harmonic (M3) for the 1Khz tone is even lower at -73db down. That's 0.02%. The IMD bands are -62 to -66db.
The only thing that looks funny compared to the other graphs, especially the SS amp measurements, is all the feathery crap that sticks out from the noise floor. Is this a big deal? Probably not unless your ear can make sound sounds 90db down from from the fundamental tones. The feathery stuff is basically the noise floor rising when there are signals present. It is true that tube amps are noisier, but you are not going to hear any noise from this amp with the volume at reasonable levels.
There is one thing I've learned. I have awesome test equipment that can measure stuff so far beyond what the human ear can detect.
-
Thanks for measuring this Marv.
Kuddos for posting them as most high-end amp manufacturers never post plots like these as they are hard to 'defend' given how many people adhere to nothing but 'superb' measurement values.
It's actually pretty good for this type of amp design.
Not worried about everything below -70dB myself.
When you look at the spectra of real instruments you often find the 2nd, 3rd harmonics are just 10 to 20dB below or even above the fundamentals so I don't worry about harmonics usually unless they are really high order ones that are high in levels as well.
I think one has to 'fear' more for IM.
What I find funny though is that most people worry (and/or claim) that everything above -100dB is potentially audible (especially in DACs) and subsequently amplify those signals with amps that 'add' god knows what to their signals.
But IMO if one likes what one hears who cares how it measures as it is all about personal enjoyment of music/sound and as this amp appears to deliver who cares.
Would be fun to have a 'full' set of measurements on some of the better sounding gear though.
Willing to bet that NO relation to subjective reports will ever be found though.
-
What I find funny though is that most people worry (and/or claim) that everything above -100dB is potentially audible (especially in DACs) and subsequently amplify those signals with amps that 'add' god knows what to their signals.
Pretty sure quite a few ppl on this forum have made that claim inc Marv, unless I misinterpreted his claim to be able to hear ~18 bits dynamic range? Not a jab, just sayin.....
I have no idea what my audible threshold is so can't say either way practically speaking , although the engineer in me is naturally bias towards less distortion...
-
Pretty sure quite a few ppl on this forum have made that claim inc Marv, unless I misinterpreted his claim to be able to hear ~18 bits dynamic range? Not a jab, just sayin.....
The bit-depth with DACs and associated noise floor with those bit-depths is a totally different story from distortion:
- Digital hard clips at 0db, so with good uncompressed masters, we are unlikely to see much or any of the codewords near 0db. This in essence reduces the effective bit-depth of a recording.
- In my own experience with true 24-bit recordings, I found that decimating down to 18-bits to be no different from original 24-bits, and 18-bits to be possibly better than 16-bits. It's possible that 18-bits sounded better because of less quantization error.
- Finally, we are talking about distortion signals. Distortion signals are likely to be masked by the fundamentals which are playing much louder. It's important to not confuse the ability to hear dynamic range with the ability to hear distortion.
-
What I find funny though is that most people worry (and/or claim) that everything above -100dB is potentially audible (especially in DACs) and subsequently amplify those signals with amps that 'add' god knows what to their signals.
This is probably one of the most harmful things that came from nwavguy. Misunderstandings or obsessive focus from laymen and sadly "engineers" (or other self-appointed experts who cite various unrelated credentials) on getting the best results from a suite of five or six measurements. Results that far far surpass what these very same people can actually hear.
Willing to bet that NO relation to subjective reports will ever be found though.
I would say there is quite a bit of correlation - if and only if you know what specifically to listen for*. For the EC Studio, Craig and I, and to a lesser extent Anaxilus, went through almost a dozen interstage transformers. We listened and measured (the transformers). It's no accident that the best of the bunch had the best distortion characteristics - mainly in the bass - where distortion tends to be highest and most audible.
However, the notion that distortion on a FFT spectrum plot is the end-all-be-all is completely ridiculous. We simply need to do a thought exercise:
The distortion of transducers is a magnitude or two (or three) higher than that of amplifiers. A headphone's distortion will completely dominate a amp's distortion on an FFT spectrum plot. If one wanted to be a hard-ass and "objective" about measurements, then amp distortion measurements (at least from most amps) really don't matter. If one wanted to be true to measurements, one cannot cherry pick ... it should be that the measurements that matter most are those coming out from the transducer as part of the total system.
*I've always maintained that measurements must be correlated to subjective experience and not be considered in a vacuum - measurements for measurement's sake. Jason @ Schiit said the same thing in his measurements post on HF.
-
More measurements. Again 100 ohm load at ~0.6V. /10 probes. Just getting some baselines. Ultimately, I'd like to take the measurements at 300ohm and 30 ohm loads which represent Sennheiser and typical orthos respectively. We at Changstar should try to develop some baseline measurements which are reflective of typical conditions experienced in the hobby.
The Studio at 20Hz and 40Hz. I feel that tube amps biggest disadvantage compared to solid-state is bass performance. Solid state typically has better bass definition and articulation, at least to my ears. Keep in mind that The Studio has uber (near $1000 / pr.) custom interstage transformers with nano-crystalline cores (which did measure better than all the other interstages we tested).
What's interesting is that we see more 3rd order harmonic distortion (-51db) for the 20Hz tone. Second order distortion is -58db down. For the 40Hz tone, the 2nd harmonic is -65db (0.06%) The 3rd order is a little bit higher -57db (0.14%). Wait a minute, aren't tube amps supposed to have a high level of "pleasant sounding" even order harmonics? The moral of the story is that we shouldn't overgeneralize. FWIW, all distortion sucks to my ears, including even order. Whoever said even order harmonic distortion is pleasing should be shot.
I really doubt most people will be able to hear these small levels of distortion in the bass. Those who have heard The Studio know how good the bass quality is on the amp. Finally, it should be noted that 99% of music has very little or rolled off energy below 40Hz.
-
Vali (hybrid amp) bass distortion. Same conditions. 20Hz and 40Hz for comparison.
For the 20Hz tone, Vali is slightly worse in 2nd order, but much better for 3rd order compared to The Studio. Same situation for the 40Hz tone, although 2nd order on the Studio is significantly better -65db vs -54db (Vali). Vali does have much lower 3rd order, about 33db better!
Succinctly, The Studio does have better 2nd order performance than the Vali, but is worse in the higher order distortions.
-
Magni 2 here.
As we can see Magni 2 destroys anything prior with a tube. The interesting thing is that even order distortion is non-existent.
Now the important questions are:
- What levels of distortion at 20Hz and 40Hz can we actually hear?
- Does bass distortion, even from the tube amps, even matter if the distortion from the transducers are going to dominate (HD800, HD600, etc.)
P.S. 100 ohms is a odd load. Going go run these tests again with ~300 ohm load which should cover a lot of the Sennheiser headphones.
-
Holy crap, that's brave, posting 20Hz measurements out of a transformer-coupled tube amp! Transformers are most disadvantaged in the bass (just check some bad transformer-coupled designs for bass distortion, holy moly.) The Studio is daaaaammmmnnnn impressive.
I expect midband measurements to be really amazing.
-
How relevant is tube rolling and tube condition to these? Would be cool for people to see how/if different tubes alter what you've shown and whether there would be a subjective correlation.
-
From what I remember on the Torpedo measurements (which is not the same as the Vali and the Studio), different tubes had indeed different characteristics. The 6J6Ws had less overall distortion than the E90CCs, but a little more higher order distortion in the upper frequencies.
I would expect similar in other systems. However, obviously is not always possible to tube roll.
Also, those are indeed fairly good measurements for the Studio given the Vali reference, which I don't think uses transformers in the topology.
-
Holy crap, that's brave, posting 20Hz measurements out of a transformer-coupled tube amp! Transformers are most disadvantaged in the bass (just check some bad transformer-coupled designs for bass distortion, holy moly.) The Studio is daaaaammmmnnnn impressive.
I expect midband measurements to be really amazing.
Have you had a chance to hear it or are you just looking at the measurements?
How relevant is tube rolling and tube condition to these? Would be cool for people to see how/if different tubes alter what you've shown and whether there would be a subjective correlation.
I'd also be curious to see the effects/their magnitude.
-
Have you had a chance to hear it or are you just looking at the measurements?
Don't know if Jason has heard it. Mike Moffat has heard it with the Yggy and pretty darn good (for the $$) turntable rig.
Casey at LHLabs and Jazmin at UE have also heard it. She's not an audio forum troll though, but her reaction was priceless. Wish I had that on video.
-
Don't know if Jason has heard it. Mike Moffat has heard it with the Yggy and pretty darn good (for the $$) turntable rig.
Casey at LHLabs and Jazmin at UE have also heard it. She's not an audio forum troll though, but her reaction was priceless. Wish I had that on video.
Hearing truly is believing. Those few listens at the SF Meet were as impressive as anything I've ever heard.
I don't know squat about tubes so I look forward to hitting you and Marv up for recommendations on a good set (or two) of 2a3s.
-
Have you had a chance to hear it or are you just looking at the measurements?
It's simply a matter of "holy cow" measurements for a transformer coupled amp. Both the Tribute interstages and Cinemag OPTs are mighty impressive as far as performance at the low extreme. Most transformers can't even hit 20Hz flat, much less have inaudible or near inaudible distortion. To be fair, there is 1.45db feedback from the tertiary windings of the Cinemags. That does tend to help extension and distortion.
Are these measurements the end all be all? Of course not. Do I personally hear perfection in the bass? Almost. 90% there. Nothing is perfect. Still the best bass quality from a tube amp I've ever heard - subjectively exceeding that of many solid-state amps.
See FR below. Using FFT averaged white noise, so a big jaggy. FFT size of 65536 pts might limit measurement of low bass, but we still got some idea. Graph goes down to 1Hz. 100 ohm load. Amps go in this order:
The Studio
Vali
Magni 2
Magni wins again. But The Studio reaches down to 20Hz clean with -1db at 16Hz.
-
One of the most interesting thread I ever read. Thks :)p1
-
It's simply a matter of "holy cow" measurements for a transformer coupled amp.
This. The Studio's distortion performance is *very* good. The slight odd-order distortion at the bottom of the response might add a small bit of character there, but it might help keep the bass from going limp as the volume goes up.
The only thing that looks funny compared to the other graphs...
The only thing that looks funny to me are the uneven sidebands in the Studio's IM plots (http://www.changstar.com/index.php/topic,2617.msg74824.html#msg74824), which imply a level of phase-intermodulation-distortion that is within spitting distance of the (amplitude)IM itself. I'm not suggesting it is audible, just notable.
-
Would be interesting to have some measurement of bad transformer coupled tube amp to compare. Is there any chance to get one Marvey ? No nedd to name it. The idea is to see a reference of bad measurements.
-
It's simply a matter of "holy cow" measurements for a transformer coupled amp. Both the Tribute interstages and Cinemag OPTs are mighty impressive as far as performance at the low extreme. Most transformers can't even hit 20Hz flat, much less have inaudible or near inaudible distortion.
Exactly. After testing some passive transformer-based summing solutions for transforming a circlotron's balanced output to single-ended, I have immense respect for any transformer that can (a) do anything at all at 20 Hz, and (b) meet or exceed its stated distortion performance at low frequencies. You're talking $$$$$$$ iron to do that.
And nope, I haven't heard the Studio yet. I'm a little scared to. I think, at heart, I really do enjoy doing the most we can at the lowest cost. Oddly enough (considering his history), so does Mike.
-
Any idea for test that can separate the men from the boys then? Would be interesting to see where tubes edge out, unless its unmeasurable with the current setup. The 20hz distortion for the Studio is pretty sick for a design of its kind.