Non linearity due to lower Gm than bipolars will be higher but
transition at the crossover point will be smoother as the crossconduction
region is wider.
Of course , this doesnt explicitly appear in doug self s curves
which show rather a bad behaviour, but this is due to bad use
of the 2SK135/2SJ50 devices , in that they are only 7A max Id
while Self test with +-40V input and loading with as low
as a 4R load, thus exceding the device max rated current,
not counting that this will exacerbate the Gm related THD,
as shown by his measurements...
Such laterals should always be used parraleled , with three
devices for a 100W/8R amp.
Hi,
Sorry, I would call these "Experiences" not prejudices. The post I made about my experiences in the western education system where I got my 2nd degree did not make it to the board somehow, so I have experienced both sides.
As for Class D amplifiers. I have wasted probably a year worth of man-hours evaluating solutions and to make them somehow loose that particular quality inherent to the lot, which I (and our listening group describe as amusical or dissonant).
Without going into details on specific solutions (so as not to be seen to attack any competitor), they all (including many very much hyped up here and elsewhere) failed blind listening tests abysmally.
I have yet to find any switching Amplifier I would prefer over a TDA2003 Chipamp (no need to go as far as formerly Nat Semi's much better stuff.
Please understand that for "the dayjob" I have very severe pressure from marketing and accounting to get a "good class D amp", but so far the listening tests where so bad, even marketing and accounting is shutting up over these oustdated, hot running linear amplifiers (and power supplies incidentally).
You don't.
Okay, let me put it this way:
1) I learned that what matters is not some cute measurements or theories, but the practical application.
2) I learned to trust my own empirical evaluation more than authority.
That is the difference. Hence I have an ambivalent relationship with "measurements". I know how how to perform them, I can interpret them, but I find them in-congruent with practical requirements.
And this is at least in part due to a difference in education, which engendered a different mindset (or as I like to call it weltanschauung).
Especially you as fan of Jean Hiraga should appreciate this...
Now do not get me wrong, I am in the end an engineer/programmer/human interactions designer. I like to have clear-cut, verifiable and measurable goals as much as anyone who ever had to deal with politically charged and high cost software projects would (that is still what brings in the real cash, audio is just a diversion and while business was way up post 2008, it is now down a lot).
But I find that in audio such reliable goals (on spec, on time and on budget - would be the one for software projects and make sure you write the spec, agree the budget and outrageously pad the timetable) are hard to come by when we wish to produce something that can reliably charm at least a siginificant number of listeners into parting with their cash.
Some companies can rely on reputation and marketing, as "low available marketing budget" newcomer you have to do by nothing but subjective sound quality.
Ciao T
Sorry, I would call these "Experiences" not prejudices. The post I made about my experiences in the western education system where I got my 2nd degree did not make it to the board somehow, so I have experienced both sides.
As for Class D amplifiers. I have wasted probably a year worth of man-hours evaluating solutions and to make them somehow loose that particular quality inherent to the lot, which I (and our listening group describe as amusical or dissonant).
Without going into details on specific solutions (so as not to be seen to attack any competitor), they all (including many very much hyped up here and elsewhere) failed blind listening tests abysmally.
I have yet to find any switching Amplifier I would prefer over a TDA2003 Chipamp (no need to go as far as formerly Nat Semi's much better stuff.
Please understand that for "the dayjob" I have very severe pressure from marketing and accounting to get a "good class D amp", but so far the listening tests where so bad, even marketing and accounting is shutting up over these oustdated, hot running linear amplifiers (and power supplies incidentally).
Still don't see what this has to do with measuring sound quality.
You don't.
Okay, let me put it this way:
1) I learned that what matters is not some cute measurements or theories, but the practical application.
2) I learned to trust my own empirical evaluation more than authority.
That is the difference. Hence I have an ambivalent relationship with "measurements". I know how how to perform them, I can interpret them, but I find them in-congruent with practical requirements.
And this is at least in part due to a difference in education, which engendered a different mindset (or as I like to call it weltanschauung).
Especially you as fan of Jean Hiraga should appreciate this...
Now do not get me wrong, I am in the end an engineer/programmer/human interactions designer. I like to have clear-cut, verifiable and measurable goals as much as anyone who ever had to deal with politically charged and high cost software projects would (that is still what brings in the real cash, audio is just a diversion and while business was way up post 2008, it is now down a lot).
But I find that in audio such reliable goals (on spec, on time and on budget - would be the one for software projects and make sure you write the spec, agree the budget and outrageously pad the timetable) are hard to come by when we wish to produce something that can reliably charm at least a siginificant number of listeners into parting with their cash.
Some companies can rely on reputation and marketing, as "low available marketing budget" newcomer you have to do by nothing but subjective sound quality.
Ciao T
Hi,
Let us be clear, the Gm discontinuity (what one may call static crossover distortion) when transiting from both device (block) conducting to the situation where only one does is much worse that bipolar transistors. It is not only D. Self who shows that, Cordell and others do too - a bipolar output stage can be made to have lower static crossover distortion.
However, had transient measurements been applied it would have been found that bipolar output stages have severe transient crossover phenomena (which can be reduced but not eliminated by suitable design) which the static view does not show and which, due the way the silicon is made for lateral fets does not affect them (but affects V-Fets even worse than bipolars).
So again, we actually find a case where traditional measurements (which would favour BJT's) fail to tell the whole story and may lead to faulty conclusions, if relied upon sufficiently uncritically.
Ciao T
Let us be clear, the Gm discontinuity (what one may call static crossover distortion) when transiting from both device (block) conducting to the situation where only one does is much worse that bipolar transistors. It is not only D. Self who shows that, Cordell and others do too - a bipolar output stage can be made to have lower static crossover distortion.
However, had transient measurements been applied it would have been found that bipolar output stages have severe transient crossover phenomena (which can be reduced but not eliminated by suitable design) which the static view does not show and which, due the way the silicon is made for lateral fets does not affect them (but affects V-Fets even worse than bipolars).
So again, we actually find a case where traditional measurements (which would favour BJT's) fail to tell the whole story and may lead to faulty conclusions, if relied upon sufficiently uncritically.
Ciao T
Conceptually, I can see that perhaps THD and IMD don't tell the whole story, and I read with interest an earlier link with Geddes' proposal for a new metric.
At the same time, comparing what goes into an amp with what comes out seems a pretty nifty way of finding out in what way an amplifier makes changes to the signal. Diffmaker does this, and some while ago a WAV file was posted comparing in and out of an ICEpower D-class amp. For your convenience, I would repost it here, but .wav files cannot be downloaded, so you have go to Audio DiffMaker and put in some effort.
I may warn you, because it will be a lot of effort for litterally nothing. The difference signal between input and output of this wretched Class D amp is inaudible, zip, niente, nada, rien, nichts, nothing at all. Tell me, how can an amplifier that does not change the signal in any percievable way, land on the bottom of your heap?
If I were the accountant or the marketing manager of your day job, I would tell you to stop making excuses and start work on an audiophile Class D amp.
vac
This is exactly what I am going to try. With positive feedback by current they should work exactly as I want from them.
I will hold my view for in-home on arrays, but they are far superior for club PA's. The big JBL may be fine in a theater or arena, but get a bad room with low ceiling, and going form the typical powered 15 and horn to an array is a huge step.
I probably got the $1.79 speakers for a buck, you probably got the $2.79 ones. With enough of them, the excursion is so short the overcome a lot of their basic limitations. Someday I hope to give it a try again with the $5 ones. Or dreaming the $12 ones. SB, Tang they do some pretty decent stuff. There is a lot to be said for not driving a cone into breakup instead of all the things we do in crossovers and exotic cones for when we do. But then again, the Bose line driver PA system sounds like, well Bose.
Yup, this thread is getting long. I have gleaned a lot of ideas on what is good practice in amp design and what may not be. I thank you all for that. I also still understand that there things that effect the system output that are not easy to point to or a simple test for. Of course, it would be disappointing if amps were perfected and that was the end.
I'm waiting on several books and thinking about building a DIY something in the Aimor chassis I have. Something I can play with and learn some more. Mid side AB or small class A. I built a couple tube amps to learn tubes, so I guess I had better do the same for ss.
I got all the new caps and hexfred rectifier in the Hafler but am waiting on a couple of details before I put it on the spectrum analyzer and see if I have lowered the noise floor. I'll report if the expensive diodes did any better than the conventional one with a single .01 disk across it over in the power supply forum.
Expensive diodes allow you to do more of obvious errors in PS design and layout, nothing more. Vacuum tube rectifiers forgive even more obvious errors. But you may apply gray matter once, and it saves green matter that goes to each device in production. Think about each wire that has resistance and inductance, that are sources of voltage spikes caused by current spikes, to be happy. Less of measurements are needed when you know what you do...
4" drivers I use in my home theater are around $20 each. 6.5" I use for PA are around $50 each.
3"x5" I bought in Parts Express are $0.99 each, I bought 140 and got such price and free shipment. They are Panasonic branded, and not ROHS compliant (I am not going to throw line arrays to waste basket anyway, so ROHS does not matter).
Still don't see what this has to do with measuring sound quality.
Hey Pano,
You can't measure an opinion.
Best,
revb.
Same old same old.
Is there something new and interesting in this thread, or just the same old opinions that we all know so well? Is the point of the thread simply to say "Measurements don't jibe with perceived sound quality" or is it perhaps to talk about some that do, or could. General opinions on the subject are OK, but I thought we had gotten past those in the first few pages.
Maybe it's asking too much of a public discussion.
There's this old thread on biasing speaker crossover capacitors: http://www.diyaudio.com/forums/multi-way/13184-pre-bias-capacitors-crossover.html and I've seen it done in a schematic of active electronics with back-to-back electrolytics (the - ends connected together and the connection going through a high-value resistor to ground or V-), ISTR by Jung.
Same old same old.
Is the point of the thread simply to say "Measurements don't jibe with perceived sound quality" or is it perhaps to talk about some that do, or could. General opinions on the subject are OK, but I thought we had gotten past those in the first few pages.
Maybe it's asking too much of a public discussion.
I've answered you already many times: when you design something you whould know what, when and how to measure, in terms of what you design. For example, in order to draw best pictures Leonardo was spending some time in morgues measuring body parts, bones, etc... He knew what to measure, and what for. But it does not mean that we can define some simple measurement standards to judge works of Leonardo using measurement standards.
The same goes to audio design. When you write a book about design of some selected topology like Douglas Self does, you can measure every detail to understand better which decision lead to better end result. And even in such case many factors are still outside of equations, since blameless amp does not mean blameless "amp + speaker + power supply + signal source" system.
The answer is: measurements are vital. But can't be found such set of simple standard measurements that reflect all possible flaws of all possible topologies and system components. Design of devices that fool imagination still is the art since different designers differently understand how to design optimal thingy that better fools imaginations. The majority of designers I am pretty sure even did not hear such quality that fooled their imagination as if sounds are real, so they even don't approach such a goal.
Member
Joined 2009
Paid Member
Wavebourne:
I understand and am not looking for a measurement that "tells all". But is there some set of measurements, perhaps a set that could be combined, that will be closer to a "Sound Quality Index"? We know that within reason FR and THD are not indicative. That's a dead horse.
But what might be? DF96 had a proposal. Are there not any measurements or measurement techniques that would relate to SQ for you? Nothing that would work across topologies?
I understand and am not looking for a measurement that "tells all". But is there some set of measurements, perhaps a set that could be combined, that will be closer to a "Sound Quality Index"? We know that within reason FR and THD are not indicative. That's a dead horse.
But what might be? DF96 had a proposal. Are there not any measurements or measurement techniques that would relate to SQ for you? Nothing that would work across topologies?
There were tons of proposals similar to what DF96 proposed, but no agreement about what and how to weight. People want something simple, like 2x2=4, so any marketing department guru can understand, any lawyer agrees.
I've found one measurement: if subconscious reaction on sounds happen before people realize that it's reproduction, it is good quality. My usual test: close eyes and try to imagine that the sound is real. The easier it is to imagine, the better is quality. However, such measurements require some calibration. Untrained people eager to judge experienced designers can't use them.
I've found one measurement: if subconscious reaction on sounds happen before people realize that it's reproduction, it is good quality. My usual test: close eyes and try to imagine that the sound is real. The easier it is to imagine, the better is quality. However, such measurements require some calibration. Untrained people eager to judge experienced designers can't use them.
Cool, thanks.
Over the years I've found that I can look at the harmonic structure of an audio circuit and have a good idea how it will sound (to me) subjectively. Not 100%, tho. But I have no idea what to measure or look at to determine if a circuit sounds subjectively muddy or dull. Wish I could figure that one out.
Over the years I've found that I can look at the harmonic structure of an audio circuit and have a good idea how it will sound (to me) subjectively. Not 100%, tho. But I have no idea what to measure or look at to determine if a circuit sounds subjectively muddy or dull. Wish I could figure that one out.
I have measurements, but they are hard to reproduce!
My wife's tolerance for horns. Second track on King James Version for example.
My hearing Julian Bream's strings sounding metallic
Applause on Clapton Unplugged sounding more horrid than others.
Now, how do I equate these with a repeatable scientific measurement, topology, components, mechanical construction etc? I can't even suggest what we hear as "wrong" are the same things that bother others.
The only thing I have between several amps into the same speaker is a differences picked up my a microphone the shows a bit more delay in the driver starting and catching up with the signal from a dead stop. Looking at 4k pulse my Rotel 840 driven system it "caught up" in 2 cycles. With my DH 120, it was closer to three. My own 6P1 PP 6W amp it was, well ugly. I have no idea if this has anything to do with what we hear, but at least it is a measured difference. The load was just a Seas tweeter through a 20uF cap.
My wife's tolerance for horns. Second track on King James Version for example.
My hearing Julian Bream's strings sounding metallic
Applause on Clapton Unplugged sounding more horrid than others.
Now, how do I equate these with a repeatable scientific measurement, topology, components, mechanical construction etc? I can't even suggest what we hear as "wrong" are the same things that bother others.
The only thing I have between several amps into the same speaker is a differences picked up my a microphone the shows a bit more delay in the driver starting and catching up with the signal from a dead stop. Looking at 4k pulse my Rotel 840 driven system it "caught up" in 2 cycles. With my DH 120, it was closer to three. My own 6P1 PP 6W amp it was, well ugly. I have no idea if this has anything to do with what we hear, but at least it is a measured difference. The load was just a Seas tweeter through a 20uF cap.
- Status
- Not open for further replies.