CHANGSTAR: Audiophile Headphone Reviews and Early 90s Style BBS
Lobby => Amp and DAC Measurements => Topic started by: Solderdude on July 30, 2015, 07:47:56 AM
-
Even though FFT has a purpose I think it doesn't say that much except for constant signals.
Why not 40Hz + 60 Hz to test for bass distortion ?
Agree with D North on most of his 'questions' .
I would suggest to use OTHER types of test signals.
Like real music under real loads, or tests that can show 'dynamics' or 'linearity' in an easy way or think of other artificial signals.
If you test for dynamics for instance then.... how much dynamics do you want tested ?
For instance if you measure with about 100dB peak SPL (and regard that as 0dB) why would you need to measure below -70dB and call it relevant as it is inaudible?
Even at 120dB ... why would one call -90dB from 120dB relevant ?
The dynamic range of hearing may well be 100dB but NOT at the same moment for SURE.
Can someone explain the correlation between 'known' measurements and sonic qualities/attributes and defend those ?
Can you take 'brain issues' (See Anax's PRaT remark for instance) and correlate that to measurements ?
Can you take mood, tiredness and time of day into consideration when evaluating static signals ?
Can you take cognitive (brain) aspects of different people in consideration when evaluating technical signals ?
Does one NEED to do that ?
Can you correlate with the standard set of measurements ?
Do squarewaves or Dirac pulses say ANYTHING at all about sonic attributes (in amps and DACs) as they don't exist in music at all.
Can we agree on audibility thresholds and get them to hold up in a court of audiophiles ?
What's the point of using -3dB / -6dB other than for obvious power/voltage/current levels that are obvious ?
What's the point to -10dB or -20dB for FR in a headphone or not mentioning it at all ?
What's a representative load ?
A resistor would be fine for mimicking orthos and very few dynamics.
What capacitance and inductance do you guys use (I test everything with 1 and 10nF in parallel to the load for instance as well as resistive only)
A headphone is always connected to a cord, how is your dummy load connected ?
Should we use inductors or even complex (LCR) loads to mimic an 'average ?' headphone and standardise this of sorts ?
You know what I don't like about standards ?
You can design something to work well during known tests but slum it on other specs and get out on top measurement wise ?
The question for me is WHAT aspects do we want to measure and what type of signals do you need.
Not a question of what 'current' methods should we use and what uniform settings .... though that may very well be a contributing part of a measurement set.
-
<<<stream of consciousness>>>
We have to start somewhere. Simple is a good place to start.
-
Ah too bad ...
I was hoping for more though facepalm.
As the measurement suite is now, the only thing that can be 'proven' is that the O2 is better than Magni, Vali and Studio in all 'electrical' departments that have been measured in the other thread.
One can also prove the Studio performs better in the bass area than a lot of other all tube transformer coupled amps.
Speculation doesn't cut it in the world of measurements... 8)
Where is the adventure in sticking to the 'old' ? :)p2
How to show in measurements that what cannot be shown in current measurements.
Sonic superiority like Studio > Vali > Magni > O2 as has been established subjectively ?
Surely there MUST be some proof somewhere ? :-Z
I mean after all.... stereo is just 2 voltages varying in amplitude over time or is there more to it ? :)p7
just teasing a bit ....
Following this with popcorn
-
No one has ever complained about the O2 having bad bass. People complain about it's Helen Keller level of resolution, lack of micro dynamics and tonally compressed and slightly foggy signature. Hell, you can hear that compared to an Ember. You helped to design that thing, you tell us.
-
I myself am fine with the fact that the Studio or any vacuum tube amp will never measure as well as a modern opamp design or any solid-state amp that uses plenty of negative feedback. Perhaps because I understand the mechanism behind the non-linearity in those amps and how the others "get rid" of it. As others have mentioned, the moderately higher THD won't keep me up at night when I consider the whole system and chain from a macro perspective.
It still doesn't explain the difference in sonics but fortunately we have this other great tool called the auris aurea to determine that for us.
The standard measurement suite as it is gives us a good idea of what's going on from a circuit design view and is still worth running through for any and all amps that can be measured in such a way. But if we want to a get a better gauge of "sonic superiority", best to look at amps with topology and component selection in mind. IMO and IME that is the best indicator of perceived sound quality we have at the moment.
-
But if we want to a get a better gauge of "sonic superiority", best to look at amps with topology and component selection in mind. IMO and IME that is the best indicator of perceived sound quality we have at the moment.
Heh
wire with tube and a mess of wires and a tube and another mess of wires
-or-
wire with tubes and cap and a tube and a mess of wires
-or-
wire with transistor and transistor and transistor and a loop going back to the first transistor.
-or-
wire with transistor and transistor and transistor (and a loop going back to the first transistor) and a cap and then a transistor and transistor and transistor with a loop going back to the first transistor.
The last one is Objective 2. Also, no such thing as wire with gain.
-
Oh I also understand the mechanisms behind negative, no, and even positive/forward feedback quite well, including the drawbacks and merits.
I also understand topologies and their up and downsides.
This does NOT explain the heard differences at all even though beautiful 'theories' go around by many designers and some even have valid points.
Indeed the standard measurement suite gives us (also me) a very good idea of what's going on.
Measurements as done by Atomic Bob for instance are outstanding examples of a nice set of measurements.
They are certainly worth running and say a LOT about amps and DACs for that matter.
I also understand that subjective detemined SQ and measurements do NOT correlate to measurements in a direct way and certainly not 100% which would have been nice.
Still, as you say, you can look at which components and topology are used and simply state that is the best indicator.
Nothing wrong with that and is what is usually done and subjective properties are coupled to components.
Unfortunately..... that does not reflect in any electrical measurements which would have been nice.
Why ?
That's what fascinates me till this day.
This is why I kind hoped there was an interest in alternative test methods or test signals that can ADD to the current measurement suite and could potentially reveal what current methods do not.
I would like more realistic 'test dummy loads' and sort of create a new standard and measurements that could be used to give more insight to perceived sound vs measured values.
For instance most amps measure well with just a resistive load but may reveal 'potential problems' when loading them with a capacitance and or inductance.
The same capacitance with different resistive loads yields different results.
THAT's what I was sortof hoping for, not the usual 'I feel attacked... you tell me' comments that are NEVER constructive.
I hoped the measurement suite attempt was an honest attempt to compile a measurement suite with the usual + 'other, not yet appllied' test signals as well that could perhaps shed some light in other ways than the already known ones.
I was already glad Marv measured in other ways than just 1kHz and the usual 'tests' that are simply made that way so others can get test in a standardised way.
Alas.... I misread it and all seems to have to remain the same perhaps using some different frequencies for FFT only that deviate from the 'standard' measurements.
It's what I liked about Donald remarks and hoped the gurus here would see 'beyond' standard measurements and come up with inovative test ideas.
Well maybe some day..
Count me in WHEN it gets constructive.
I have NO interest in playing the usual word games and childish remarks that are meant to be funny.
-
Well parts of your post seemed a bit more than cheeky. Your use of emoticons came off as a bit pretentious and condescending (a la Macedonian Hero) in the usual objective manner. So consider it lexical ambiguity. I'd like to think I know you better than that by now, but when one employs the usual flame bait posting tactics they usually get the same flame bait responses. That said, let's get back to it. As for lexical ambiguity, let's change the following absolutist remark to something more constructive:
"I also understand that subjective detemined SQ and measurements do NOT correlate to the usual current measurements in a direct way and certainly not 100% which would have been nice."
-
For instance most amps measure well with just a resistive load but may reveal 'potential problems' when loading them with a capacitance and or inductance. The same capacitance with different resistive loads yields different results.
Easy. Use a real load like a real HD600 or LCD2 like what Craig does. You can throw a cap and coil into the circuit to simulate the impedance, but that doesn't exactly simulate the non-linear behavior of drivers (if you've seen Klippel results, you would know). It's a good next step. Maybe you can put together a schematic to simulate HD800 (very popular headphone) impedance curve. As far as orthos, they are more similar to each other than not in terms of impedance. 32 ohms. Easy. Representative of Audeze, HFM, and even T50RP (close enough to 50).
This is why I kind hoped there was an interest in alternative test methods or test signals that can ADD to the current measurement suite and could potentially reveal what current methods do not.
No shit Sherlock. Unfortunately, I don't have all the fucking answers right now. Do you? Did you even bother what reading what I first wrote: "We have to start somewhere. Simple is a good place to start." Rome wasn't built in a day. Changstar headphone measurements have evolved over time. Let's take this step by step, unless you already have all the answers. I don't want ideas, I want methods and results.
And yes, I do have some ideas on more evolved measurement methodologies that I am keeping to myself. However the first step is to get everyone who is doing these organized and in-line to begin with. It's important to have a strong foundation, consistent results, no?
Count me in WHEN it gets constructive.
Post when you have something constructive. I've had enough of idiots at company meetings who propose pie-in-the-sky ideas with no idea how to get there.
-
This does NOT explain the heard differences at all even though beautiful 'theories' go around by many designers and some even have valid points.
I don't suppose you have any better theories. Let me guess, it starts with "expectation" and ends with "bias"
Still, as you say, you can look at which components and topology are used and simply state that is the best indicator.
Nothing wrong with that and is what is usually done and subjective properties are coupled to components.
Unfortunately..... that does not reflect in any electrical measurements which would have been nice.
No, I'd say that different components and topologies tend to result in very different electrical measurements. Whether these differences are positive, "below the threshold of audibility", or negative is a different story, but they certainly are quantifiably different.
I hoped the measurement suite attempt was an honest attempt to compile a measurement suite with the usual + 'other, not yet appllied' test signals as well that could perhaps shed some light in other ways than the already known ones.
I was already glad Marv measured in other ways than just 1kHz and the usual 'tests' that are simply made that way so others can get test in a standardised way.
I am very open to a new sort of test signal that could provide insight into the dynamic or "sonic" abilities of any given black box. If you have any sort of specific tone-burst or square wave or another sort of stress signal in mind that you think might be a valid stress test for amplifier performance, let us know and I for one would be happy to explore the idea. In terms of using some sort of musical signal...well that's certainly an interesting idea but I wouldn't know where to start
Count me in WHEN it gets constructive.
The fundamental problem with your original post is that it's not particularly constructive. It's really just obfuscating and running circles around itself. Perhaps it's a language barrier thing or maybe just an internet communication misunderstanding, but you do speak like that a lot. I think we've all got accustomed to how we communicate with each other.... so I have no idea why you're surprised to see The Belligerent One or Chairman Marv having fun at the expense of your ideas. Best to take everything in stride rather than whining and abandoning ship.
-
Maybe you can put together a schematic to simulate HD800 (very popular headphone) impedance curve.
Was just about to propose this, I'll do it later tonight. I'm really not expecting any massive differences with most amps given low Z-out, sufficient output current and speed.
-
I was being somewhat cheeky ...
I should nuance the remark.
It goes without reason that a measurement that shows very high amounts of distortion is audible and correlates in a direct way.
THIS I do not doubt nor is it what I meant.
Sure if there is a clear roll-off in the extremes on FR in the considered audible range or if there is a substantial phase shift, ringing, oscilations or what not that can directly relate to what's perceived.
NOT talking about that either AND I am quite sure you know that pretty well.
Nope... What I am talking about is say comparing O2 with Magni (lets leave the obviously different Vali out of it.
Most like the Magni(2) over the O2.
I too like Jason more than Harold btw. 8) and tend to agree with Jason more than Harold for that matter.
But that's not to the point.
When you take the current 'usual' meaurement suite.
FR, distortion, under load or not, FFT perhaps squarewave, take clipping properties into account and drive both within their limits the O2 outclasses the Magni in the electrical measurements.
Yet..... when we look at subjective reports most feel the Magni sounds better/is preferred.
Likewise ... take the EC and put the measurements next to the O2 than it is obvious MEASUREMENT the O2 is the better one.
Regardless if the measurements of the EC amp being excellent for this type of topology.
Of course I am NOT saying the O2 is the better amp, just that it MEASURES better.
Obviously the EC amps get's lots of appraisal for its SQ where the O2 gets less (well maybe from fanboys or tube haters).
NOT talking about EC amps being 'worse' than Harolds 'standard application' with a twist at all.
Just saying that with the formentioned measurements you cannot conclude that the EC is the better sounding one.
There is NO correlation between the measurements and the perceived SQ.
Likewise take any 'C'Moy' and the UHA6S..
All here agree the Leck is excellent where most C'Moys are said to sound poor.
It's the exact same circuit ?
When you measure them they will measure the same yet they are not considered to sound the same.... why ?
What differs that does NOT show up in the usual measurements ?
Can other 'test signals' be constructed that are closer to music instead of standard sweeps, test tones, noise or whatever.
Signals that can give better insight in linearity or impulse behaviour using sine waves instead of dirac or squarewaves ?
With small different frequencies riding on bigger ones ?
Monotonicity tests but with larger scales ?
A more realsitic dummy load that challenges amps ?
That's what I meant.
-
Studio vs O2...
There is NO correlation between the measurements and the perceived SQ.
Wrong. There absolutely is. If you know how certain patterns of distortion sound like and if you understand that (harmonic) distortion <> resolution <> dynamics <> volume contrasts <> veil <> MOSFET mist <> etc.
What I've learned from the tests I've already taken is that my personal limit for hearing bass distortion (usually comes out as roundness or blurriness) is about 60db-65db when correlating to the specific measurements taken. I've already said, the Studio has the best bass performance I've ever heard in terms of distortion for a tube amp. It's almost 100%, but not quite, more like 90% there. Another person commented that the higher 3rd order for Studio 20Hz and 40Hz might give it an interesting character. I didn't make a follow up comment on that, but the Studio does actually have an interesting bass characteristic (very hard to use words).
Still, the bass is very close to the cleanest SS amps I have heard though (and overall bass in music is superior, but this would be including other factors). What this tells me is that it probably doesn't fucking matter if D2 or D3 are less than 60-65db. That itself is good information, unless you are dimwit like one of nwaguy minions or self-appointed engineers who insist that 0.001 is audibly superior to 0.05%
The problem starts when people equate distortion with everything else, every other sonic characteristic. Distortion has a very specific sound. Let me say this again: Harmonic distortion has a very specific kind of sound.
Also, let me say this in another way: harmonic distortion <> resolution <> dynamics <> volume contrasts <> veil <> MOSFET mist <> etc.
Can other 'test signals' be constructed that are closer to music instead of standard sweeps, test tones, noise or whatever.
Signals that can give better insight in linearity or impulse behaviour using sine waves instead of dirac or squarewaves ?
With small different frequencies riding on bigger ones ?
Monotonicity tests but with larger scales ?
A more realsitic dummy load that challenges amps ?
That's what I meant.
A lot of those are harder to do and require better measurement gear or measurement gear / methodologies that have not been invented. I've thought a lot already of what you have suggested. Again for the fifth time, my current challenge is to get every other fucker who is doing 284 different permutations of measurements, to do so in a few consistent ways, while also taking into account that not everybody has an AP or Stanford analyzer or dScope.
-
Easy. Use a real load like a real HD600 or LCD2 like what Craig does. You can throw a cap and coil into the circuit to simulate the impedance, but that doesn't exactly simulate the non-linear behavior of drivers (if you've seen Klippel results, you would know). It's a good next step. Maybe you can put together a schematic to simulate HD800 (very popular headphone) impedance curve.
The counter EMF is a valid point...
Howz about using the testsignal and also use that as a counter EMF + LCR circuit ?
We don't all have HD800's nor expensive test equipment (an HD600 is no problem though and indeed comparable)
Using real headphones is a good way but with amps in development stage a risky business.
No shit Sherlock. Unfortunately, I don't have all the fucking answers right now. Do you?
If I did ... would I ask... Watson ?
Did you even bother what reading what I first wrote: "We have to start somewhere. Simple is a good place to start."
Yes,
You have been around with measurements and so have others around here.
You say measurements have evolved here.. and that's certainly true so you (all) have been measuring for a while already.
Why 'start' with simple and basics then... we've all been there eons ago, why not raise the bar ?
And yes, I do have some ideas on more evolved measurement methodologies that I am keeping to myself.
However the first step is to get everyone who is doing these organized and in-line to begin with.
It's important to have a strong foundation, consistent results, no?
If you want to make a thread for newbies ... yes.
I thought you were planning to go beyond that.
My mistake.
-
Why 'start' with simple and basics then... we've all been there eons ago, why not raise the bar ?
Because it takes time and research to raise the bar. For the 6th time, this is simply a start.
If you want to make a thread for newbies ... yes.
I thought you were planning to go beyond that.
My mistake.
That's totally uncalled for and a passive aggressive slap in my face on your part.
I wouldn't call Donald, ultrabike, OJ or a few of the other guys noobs. It's not noob move to make sure we can first collaborate effectively and get consistent results with an expansion of the basics. I see this as a group project where people will contribute over time, like in the next few years.
EDIT: again, it's pointless to discuss advanced measurement techniques when none of us, including yourself or myself, have any actionable ideas or methods.
-
I didn't call Donald, UB, OJ nor others noobs.
In fact the opposite.
Raising the bar is what you and Craigg as well as others do on a daily basis.
So do countless others.
I figured you guys all knew how to measure and for the 7th time thought you wanted to ge beyond.
Go with the many ideas you already had.... FUTURE measurements.
Exchange ideas on those.
Progress ... and who says I don't have any ideas ?
Didn't think about 'just getting some basics' at all.
Sorry to have angered you.
Wasn't my intention.
-
Raising the bar is what you and Craigg as well as others do on a daily basis.
I figured you guys all knew how to measure and for the 7th time thought you wanted to ge beyond.
It's not about how to measure at this point. It's about data. To move forward we need data. At this point, we don't even have basic data and baselines using simple methodologies. It could very well be that this basic data is useless, or no more useful than 1kHz tests. But I at least want to see some of it.
The evolution of the headphone measurements could never have happened without data.
-
I feel we are not trying to discourage the development of new measurements that correlate better with perceived performance (i.e. rising the bar).
It seems to me we just want to have a basic set of measurements across the board, as a starting point, to allow comparative performance evaluations.
If some new measurement comes to mind that shines some light, then by all means it will be commented on and considered for addition. But we probably want to avoid 1,000,000 mind numbing tests that would yield the equivalent of 1 test of information. We probably want to also avoid (for example) having measurements at 10 ohms/1k ohm and then a new set of measurements at 50 ohms and 200 ohms on a different set of equipment and have no certain idea how one compares to the other. Which is the point of this discussion me thinks.
BTW, Harold seems to have done his earnest at this, and seemed to try to standarize 0.4Vrms, 15/150 ohms, 1kHz/20kHz THD, SMPTE, CCIF, Noise, Crosstalk at 15 ohms, and FR (into some load). But he seemed somewhat alone at this, and IMO made some mistakes by introducing 1 Vrms ocationally and sometimes failed on comparatives (too much drama and subjective bias agains this or that due to seeming hard feelings and misunderstandings). His list is also not very complete, and had perhaps way to much faith on his dScope. I feel the idea here is to expand on what its available. But not all in an all-over-the-map manner. Therefore the existence of this discussions.
-
I feel we are not trying to discourage the development of new measurements that correlate better with perceived performance (i.e. rising the bar).
Of course not. The reason we are not discussing these "advanced techniques" is because we do not know how or are not at the point where they are worth discussing. I've already gone down a few dead-ends already including measuring amp linearity, etc.
It's like the alternate Yggy output stage. Have several ideas how to tackle this problem... but not gonna talk about it. You know why? Because it would be pointless to discuss until I knew there was a viable solution that was 90% there.
May as well talk about how to construct a TARDIS.
-
My mistake Marv,
Seeing that you guys meet and talk about it a lot and you use similar plots, programs, ways of presenting and doing similar measurements just at 'non standardised' levels and loads I assumed you guys already had your measurements ways sorted.
For me doing measurements is a daily thing as well as working with standards.
As well in audio (hobby and used to be work) as well as far outside audio with measuring and interpreting data.
Judging from the measurements I assumed you already had most of it down.
Now I understand why my remarks were interpreted the wrong way and were seen as an attack instead of the (wished for) expansion of tests.
-
I think it would be very interesting if we could develop an input-output comparator and watch real-time the behavior of an amplifier, both in the time and frequency domain, into a real pair of headphones.
OJ: you're a hardware guy... :)p1
-
Built one 30 years ago ...
(http://i837.photobucket.com/albums/zz296/solderdude/distanalyzer_zps5501becb.jpg) (http://s837.photobucket.com/user/solderdude/media/distanalyzer_zps5501becb.jpg.html)
In those days (audio days) I had to use a B77 to 'record' the diff files.
Now that could be done digitally and would be easy to analyse supersonic frequencies (I could not hear above 18.5kHz in those days)
There are also much better devices to make something like this.
Schematics everywhere...
Just google 'differential amplifier' and the rest is peanuts to add.
Unfortunately such a device mixes FR (amplitude), phase and time and translates them all to amplitude.
With modern ways of analysing it should be possible to 'extract' aspects by viewing more signals at the same time.
-
That's cool - were you able to attain any meaningful insight from this differential data?
I wonder what those tapes sounded like...
-
Now I understand why my remarks were interpreted the wrong way and were seen as an attack instead of the (wished for) expansion of tests.
Basically a lot of us who were earnestly trying to mesh out the standards saw it as you thread-crapping.
--
On the comparator:
If the argument that tube amps do certain things better than solid-state amps is true, than it seems that a comparator (using solid-state parts or opamps) might not yield any meaningful results.
What I'd like to see is a method that can capture and present dynamic / transient behavior differences of amps - namely in amount of feedback. Feedback does wonderful things like tighten up the bass and extend FR, but it closes the soundstage and makes everything flatter and duller sounding.
-
I think I did get valuable info from it.
Of course I can spill the beans but won't.
Everyone that takes measurements seriously and uses their hearing and test equipment SHOULD find out for themselves.
They are easy to build and operate and very educative.
Don't use it any more ... it did open my eyes in how to look at measurements and which type of measurements are valid and telling.
What I liked about it .....
Real music, real loads.
Works for amps connected to speakers, headphones only though, [edit] and cables.
Possible caveat ?
How do you hear 'soundstage' or 'air around instuments' in a diff file ?
You don't ....
BUT with the current state of tech, PC's sampling with 20 bits resolution at high bitrates and analysing software you might come a long way by now.
Real loads .... real music ... repeatability .... analyse.
Of course you have to deal with ground loops, cable arrangements, carefull nulling, phase and time aspects (a long wire or electronics has a relevant delay) would have to be tackled.
By using longer pieces of wire to counter time delays.
-
I would use transformer summing instead of opamps ;) At least if this were going to be a real time sort of difference test
-
I hope those transformers will have silver windings and mumetal/nickel cores :)p7
-
Basically a lot of us who were earnestly trying to mesh out the standards saw it as you thread-crapping.
Get it now.... shouldn't have used the cheeky emoticons....
On the comparator:
If the argument that tube amps do certain things better than solid-state amps is true, than it seems that a comparator (using solid-state parts or opamps) might not yield any meaningful results.
Why ?
rethorical question as I understand where it comes from.
It comes from the notion/conviction/idea that transistors/opamps can't do what tubes can.
Let's just say I don't share your P.O.V. and see why you will probably not test this way, assuming an opamp cannot capture the effect.
Here is an idea for you to experiment with.
Play music on the tube amp and load with a headphone (or speaker)
record it using a PC (24bit many kHz) and play it back on an amp like the Rag.
Listen if 'the sound' is recorded and captured.
IF it has ... opamps can accurately follow it.
Can't do those tests myself. No access ... no time.
Yes... I also tested a (crappy, not high end) tube amp.
Was interesting to do.
Would like to do this again once with real tube amps.
Not in the business anymore nor access to high-end gear.
What I'd like to see is a method that can capture and present dynamic / transient behavior differences of amps - namely in amount of feedback. Feedback does wonderful things like tighten up the bass and extend FR, but it closes the soundstage and makes everything flatter and duller sounding.
You will need to create a signal that jumps from 0dB to say -70dB (which appears to me my limit when listening to loud extremely resolving speakers)
The signal needs to jump in level exactly around the '0'.
Reproduce it and look with a scope (not a digital recording) and see IF there is overshoot or the signal behaves other than the source.
Easy to see with 2 channnel scope and overlay.
That'll tell you which amp has problems.
Sort of a monotonicity test but with a much larger dynamic difference.
DC shifts will play a role as well, perhaps use a suitable high pass.
Certainly with non-balanced and coupled stages.
-
FWIW, from years ago:
http://www.head-fi.org/t/620770/waveform-fun-amp-headphone-output-comparison
-
I would use transformer summing instead of opamps ;) At least if this were going to be a real time sort of difference test
Trafo's may introduce higher distortions (hysteresis of the core), and won't do as well in frequency and phase response but the idea is intriguing.
When you have dual input windings you can easily change the phase if needed.
That would be a neat feature.
perhaps you could test it with the same signal on both input windings and see what it does.
You need to attenuate the input signal sligthly and match the other signal in amplitude to see how it nulls.
Would be elegant and fiully passive though.
Make sure there is NO DC present.
-
What I'd like to see is a method that can capture and present dynamic / transient behavior differences of amps - namely in amount of feedback. Feedback does wonderful things like tighten up the bass and extend FR, but it closes the soundstage and makes everything flatter and duller sounding.
I think that breaking the FB loop and looking at open-loop behavior would be very insightful when attempting to correlate amplifier performance to perceived SQ. I know that Jason has said that part of his design process includes carefully examining and optimizing open-loop performance. There's no doubt in my mind that that's a critical step for any analog design engineer.
If we operate on the premise that the transistors/tubes/components can add/subtract certain qualities based on their "intrinsic" sound, then it makes sense to try to examine the given device in all facets of its operation. It's worth asking exactly how linear a BJT is before you wrap it up in feedback. Same with a triode or a MOSFET. For amps that rely upon their global NFB loop to stay stable then forget about it...and same with opamps since they can have +100dB worth of open-loop gain. But interesting to think about nonetheless and something I'd like to examine myself.
-
Trafo's may introduce higher distortions (hysteresis of the core), and won't do as well in frequency and phase response but the idea is intriguing.
Yes, but we've already established that such distortion is less intrusive than the sound of an O2.
-
Bedtime here ... no more nonsense from me for a while. ::)
And ... that may depend on the used trafo AND how it is driven (which may be equally important)
besides ... I have not established anything yet before I see test results .
I have made 'educated assumptions' in the past which later on proved inaccurate so try to not assume or find things established, regardless of the confidence level.
And yes, all amplification devices are non linear, they do behave differently under different circumstances as well (voltage swing vs bias/power supply voltage ratio etc.)
Interesting thoughts though... I hope they keep coming.
-
Mumetal/nickel cores can have very low distortion. For example look at the distortion measurements for line-level balanced input transformers.
We will need a way to level match the output versus input.
I can imagine listening real-time and listening for the one with the least artifacts. For those artifacts which do exist, I wonder if those which sound least bad correlate with the better sounding amplifier....
-
Could use transformers with nano crystalline core to address hysteresis issue.
-
I suspected it may come to this: nickel versus nano crystalline :)
-
Huh, just tossing this here for reading later:
https://www.google.ca/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&ved=0CCwQFjABahUKEwjDq8GJr4THAhVPG5IKHaUyBtI&url=http%3A%2F%2Fwww.researchgate.net%2Fpublictopics.PublicPostFileLoader.html%3Fid%3D5410197ad11b8b096d8b461f%26key%3D88e37e18-2dd7-4188-a7c6-495387235c64&ei=dOO6VcOhHs-2yASl5ZiQDQ&usg=AFQjCNG6dgegg9Zg5lkmlrPM-oVpzUOYBA&bvm=bv.99261572,d.aWw&cad=rja
http://www.scielo.br/scielo.php?script=sci_arttext&pid=S1516-14392012000500020
The metallurgist in me is fascinated, but all my book smarts are on microstructural formations and mechanical behaviour, not electric/magnetic properties. I wonder if anyone's tried single crystal yet, or if that's viable. (wondering not for buzzwords, but because it's relevant to my area of research)
-
What's funny is that while various cables are marketed with certain unique features offering unique benefits with lots of associated debate and banter on both sides, some of the same attributes have a very real and quantifiable effects on transformers.
-
Pfft, if someone can market liquid cables, I'm just waiting for them to make a liquid transformer...
-
Lol, well I did say 'some'.
-
I just googled it so I'm an expert now... http://physics.aps.org/story/v27/st17
Now I just need a spinning toroidal chamber filled with molten sodium. Done. I've just created a newfangled power supply conditioner regenerator. Easy peasy. Now to print up a note telling people about the very long warmup times for optimal performance...
-
We will need a way to level match the output versus input.
I can imagine listening real-time and listening for the one with the least artifacts. For those artifacts which do exist, I wonder if those which sound least bad correlate with the better sounding amplifier....
Yes, that's the idea.
You can record the dif + source file using a soundcard and analyse afterwards as well.
If it needs to be done all tube you will need to find tubes + low distortion caps that buffer the inputs and an attenuator that can be set extremely accurately.
It all hangs on how balanced you can get it (probably around 1kHz using a sinewave).
I used a 20 rotation potmeter but I suggest to use 2 potmeters instead.
One for 'course' tuning and a second, small value one, in series with the input for fine adjust.
Then comes the difficult part.
The output signal represents the difference between the 2 inputs so when you have a difference signal it could be that what you hear is something that has been added but also something that may have been 'removed' from the original signal.
That will be an anlysing thing.
A difference in FR will show in an amplitude difference, so will a phase difference.
My questionmarks with traffos vs opamps lie in the fact that opamps can go from DC to MHz, transformers well...
With transformers a question will be how well both inputs balance.
You will have to get a baseline by connecting the same signal and see how that nulls.
Still a nice idea and as Donald mentioned you could aim for the least amount of diff (if you are looking for hifi) or you can listen out for the most pleasant sounding diff file.
Another test that may be interesting to see is the dynamics test I proposed.
Chances are the -70dB signal may not look like the original.
You may have to clip the 0dB signal (diodes in antiparallel) to avoid problems with some scope inputs.
-
So that no one can say that nothing useful came out of this thread:
(http://i.imgur.com/EwcJoNx.png)
Check against Tyll's datasheet for reference. For actual testing, component values wouldn't have to be tight to get the gist of it
http://www.innerfidelity.com/images/SennheiserHD800.pdf
-
That's a good place to start.
For a passive emulation that is.
However, finding a 1.2H inductor may be a tall order.
the 2.2uF cap is no problem of course.
BUT... does a real HD800 really have an equivalent 1.2H and 2.2uF cap 'hidden' inside the voicecoil ?
It does have something like that of course .... the parts that make up the membrane + voicecoil + suspencion.
Here's the thing... the rising impedance at the top is determined by the inductance of the voicecoil (mainly)
The 100Hz hump is NOT a passive something.
It's caused by the back EMF of the driver.
The resonance frequency of the driver itself is the cause for the wide band.
The wide 'band' of course is better than a narrow one for obvious reasons.
This needs to be emulated by creating back EMF.
As Marv rightfully remarked is not done by a passive load but IS present using real headphones.
This would mean apply noise with a determined SPL TO the driver and measure the open voltage on the driver.
Next load the driver with a known resistance (say 330 Ohm in this case) and determine the source resistance and calculate the source impedance and EMF.
Relate that to the applied SPL and what voltage was needed to create the same SPL by the driver.
A much simpler but potentially less accurate way would be to make an assumption.
This could be that the internal resistance of the 'microphone' portion is around the DC resistance of the driver.
Given the fact that at 100Hz the peak is about twice the resistance (so half the current) I would suggest a simpler yet more complicated load to emulate high impedance headphones with back EMF.
I suggest the 2.5mH inductor + 560 Ohm to ground and connect a small amplifier via a 680 Ohm to the inductor as well to emulate the voicecoil.
Then use the test signal applied to the amp to be tested, split that into a small amp (use your obsolete O2 or a simple opamp for this) and bandfilter the input to 'approximate' the 100Hz hump.
Missing the cable capacitance of course, which will be around 150pF, and could well matter for some designs.
This means connect that load via a 1.5 (or other length) HP cable.
Emulating a low impedance HP could be done simliarly but with other resistor values to keep things simple so you just have to switch little components..
Finally a resistive load for testing 'orthos' could be an idea.
There you have 3 loads that could become a standard SHOULD that be warranted.
Another question about the Ortho load...
Does an ortho generate any back EMF ?
It should ... how low will that be ?
Could be determined with the above described test.
As you mentioned earlier, most amplifiers won't have problems with an inductive load of 2.5mH + 300 Ohm at all.
I fully agree with that BUT the capacitance of the HP cable + rising impedance of the voicecoil could potentially spell trouble with some designs (mostly those with overall feedback from the output) .
Is it worth going through all the trouble ?
Is it cheaper to just buy an HD800, or (used) HD580 or similar ?
Is it cheaper to buy a cheap driver instead that is close ?
Yes, it probably is + you have a 'monitor'
Damn ... back to where we started.
Simply use dummy headphone instead (for 300 Ohm)
Oh wait ... if we want to standardise everyone should have the same driver .... hmmm
For 32 Ohm this may also be a problem given how much power is generated by some amps and during testing something may go wrong and burn out a driver.
The horrors of testing.
But could it still be wiser to build the 'emulator' with standard parts anyway ?
What could be the potential benefits over a real driver (aside from the obious power rating) ?
Well for one you could 'switch off' the back EMF to see how this would affect the output signal.
You could also insert the 'back EMF' only into the amps output and see how it fares ?
You could alter the frequency response of the back EMF.
You could even insert a testtone to see how feedback in an amp handles stuff into its output.
Just some musings though ... and I want to add more than just a simulation could come from this thread.
Perhaps some novel ideas or test signals that aren't applied.
Perhaps someone dares to actually make the fransformer based or (heaven forbid) a simple opamp differential amp and starts educationg themselves further in ways they are unfamiliar with till now.
Get people out of comfort zones ?
Maybe someone will be brave enough to 'record' the output signal of a high end tube amp and play that back on the Rag (or other decent SS amp) to see if the tubeness is really gone ?
-
Given the fact that at 100Hz the peak is about twice the resistance (so half the current) I would suggest a simpler yet more complicated load to emulate high impedance headphones with back EMF.
No comment.
If anyone wants to push .5Vrms through their $1500 headphones they're welcome to do so. Not doing that with mine though
-
0.5Vrms should not damage the HD800. Their sensitivity is rated at 102dB with 1V input.
-
For low ohm testing everyone could just use a standard Knowles/Sonion integrated crossover drivers(TWFK or Acupass). If you are crazy enough you could still buy the HD800 drivers for around $500 for the stereo pair.
-
0.5Vrms should not damage the HD800. Their sensitivity is rated at 102dB with 1V input.
Yeah but I'd rather not push things hard especially if we're going to put constant tones through it. Paranoid perhaps
-
It's good to be cautious, but unless you smell the voice coil heating up, I wouldn't worry. Plus they're replaceable. I would be more concerned about tests on a R10 or K1000. I have done a lot of high power sweep and Klippel tests over the years at Aura and now with ADX. A memorable one was when we were developing the NS15-992-4A 15" woofer at Aurasound, I did a 600W power test in a 2 cubic foot sealed box. At the end the frame was too hot to touch. Once I was able to lift it from the box, this plume of hot, smelly vapor came rushing out, caused by the hot MDF wood box.
-
If anyone wants to push .5Vrms through their $1500 headphones they're welcome to do so. Not doing that with mine though
0.5V in a HD800 = 0.8mW
The HD800 is rated at 500mW (12V)
Ratings are usually continous unless it specifically mentions otherwise.
Even a Sansa clip can reach 0.5V ... you are a very cautious man indeed.
With the simpler, yet more complicated load I mean 2 resistors + 1 inductor (which is simpler) + a small amplifier that generates the back EMF (making it more complicated)
As you don't like to use a HD800 as a load (I wouldn't either so you are not alone) the suggested 'replacement' with back EMF should be interesting.
-
Click quote linky
Just use a standard linear equivalent circuit and call it a day like OJ said. Sure, headphone non-linearities will not be incorporated, but it's better than a resistor if standardized (not saying it should though, it's a pain). But I fail to see the foundations for this "your amp will blow up" scare if using such an equivalent circuit.
Then test with a few headphones and disclose the model. Not afraid of using 0.5Vrms on HD600 or KSC-75s.
Don't subscribe to this back EMF voodoo notion. Better to lump it to the linear characteristics of the system, like most EE textbooks and applications do.
-
Don't subscribe to this back EMF voodoo notion. Better to lump it to the linear characteristics of the system, like most EE textbooks and applications do.
Sure, you can use either a simple resistive load or the suggested one by OJ or a real driver (can be a cheapy or a crappy one), test FR and distortion and call it a day.
If you don't test it, you won't know for sure and can only assume... is it not ?
How does the output stage react when current is being returned into the amp ?
Do all amps work like a perfect true 0 Ohm 'short' or does it introduce distortions, which ones and how much ?
Has anyone ever tested for this ?
It may not reveal anything at all as the currents are very small.
In case of the HD600/650/800 the back EMF current around 100Hz is half that of the supplied current around 1kHz so substantial, but still they are very small currents.
It is not a textbook test, just an idea to ADD to a test suite and may (or may not) give additional information about the amps performance.
Was just a musing anyway, nobody has to (nor probably will) ever build it.
-
How does the output stage react when current is being returned into the amp ?
From what I remember, this is due to the storage of charge (voltage) of a capacitor and magnetic field (current) of an inductor. All behaviors characterized by a set of linear differential equations that result from Maxwell's equations if I remember correctly. Laplace transforms are used to solve the resulting polynomials instead of the differential equations directly. If the excitation is sinusoidal, the phasors can be used. Transients can be found if initial conditions (voltage and currents) are known about the circuit prior to excitation.
Now how exactly, to the mosquito fart detail. I dunno. It probably depends on the amp. Use Spice or something. If the models don't fit, specially the non linear ones that are not covered by Maxwell's and so, I guess refine them.
Do all amps work like a perfect true 0 Ohm 'short' or does it introduce distortions, which ones and how much ?
No. Amps don't necessarily behave the same IMO. Non-linear distortion is likely due to non-linear interactions which are likely poorly described by back EMF and unique to a particular amp and load combination.
Has anyone ever tested for this ?
In many designs (not necessarily here), I believe so.
It may not reveal anything at all as the currents are very small.
Distortion is supposed to be small yet problematic.
In case of the HD600/650/800 the back EMF current around 100Hz is half that of the supplied current around 1kHz so substantial, but still they are very small currents.
You can call the hump around 100Hz on those headphones back EMF. I rather see it as a characteristic of the headphone reactive load.
It is not a textbook test, just an idea to ADD to a test suite and may (or may not) give additional information about the amps performance. Was just a musing anyway, nobody has to (nor probably will) ever build it.
This is IMO not new. I think such model loads have been built in the past for tuning the cross over of speakers. If it strikes our fancy to do it, we'll do it.
-
Yes, upon second thought, the reactive component of a real headphone load is adequately modeled with the passive equivalent circuit. We can't model the non-linear component but the passive load put forth should still reflect energy back to the amp. The definition of reactance...
-
Yes, upon second thought, the reactive component of a real headphone load is adequately modeled with the passive equivalent circuit. We can't model the non-linear component but the passive load put forth should still reflect energy back to the amp. The definition of reactance...
Yep. That's how I think things work.
-
the passive load put forth should still reflect energy back to the amp.
Yup, you may well have a point there.
Will do some tests someday and quantify some numbers.