Other than changing the amp, how would you increase the H2?
I've seen frequency response of many small speakers but hardly ever saw ones with bump at its lower octaves. Which particular speakers were you talking about, especially the majority of them?
My question was, where and when did you observe those people?
First order of (your) business is to get to know what speaker transient response is by using that access.
Speaking about self-test, I think that foobar ABX is perfect for this purpose. I think that the only reason why some are suggesting different methods is that it doesn't bring the expected result to them. They want to get a method telling them they hear a difference between DACs, wires, 0.5 % H2 in music. Not the method that tells them they do not hear it. So the debate would be endless, without any output, other than wasting time.
I can use tweeter with high H2 and design the crossover accordingly.
I can discuss deeply the effect of many engineering/design decisions if you genuinely want to know something new. I think I have seen every/all loudspeaker measurements in Stereophile site. Let me start with showing you an example of bloated response at the lower end of a (not even a narrow bandwidth) Mirrage speaker, attached. Please note that this is something considered as 'error' by people (like you?) but it doesn't mean that the designer didn't know how to make one that will get your approval (he knew about room response, baffle diffraction loss, etc.)
If you don't want to get involved with deep technical discussion, that's also fine with me, actually I love it because I actually don't like to share my knowledge that I got from making my hands dirty 😀
Attachments
If you read up on speaker design, you will find a lot of designers (Troels Gravensen, KEF etc) slope the response downward - ie the bottom few octaves are 2-3 dB higher than the mid and HF. The reason they do that is to ensure the speaker does no sound ‘forward’ and to extend the bass response in a typical living room.
No bs. That’s why a lot of speaker curves look like that - even my B&W’s
No bs. That’s why a lot of speaker curves look like that - even my B&W’s
No. VFA can have loop gain flat until 10kHz and higher. For example is amplifier that using OITPC compensation by Dadod.
Gladly PMA used "I think" to mark it as an opinion piece. 😉
I hope we can agree that currently, no data is available to decide whether he is correct or not?
The scientific evidence shows that the ABX protocol has certain disadvantages, compared to other test protocols, if the data is analyzed by the procedure Foobar ABX uses.
So it is harder for test participants to get a positive result.
PMA has to rate this disadvantage low (or dismiss/neglect it) to reach his opinion.
If I'd say:
"I think, PMA likes Foobar ABX because he gets the negative results that he likes to see, and therefore demands its use without warning readers that extensive training might be needed."
is it "exactly, well said" too? 😉
Basically, we all have our biases, but shouldn't the scientific approach help us to reduce the impact of those?
Originally Posted by Evenharmonics
Better in what way? Even cheap DACs these days perform better than what people can hear, including you. If you mean better in measurements, OK, that can be used as status symbol or bragging right but still well beyond our hearing limits.
Well, the point made by Evenharmonics makes a lot of sense to me. You may disagree, but the same as is true for amplifiers also goes here imo. The good ones all sound alike, nobody has proven to be able to distinguish by ear alone two properly implemented recent DAC's.
I think it has tremendous value to make this point.
Better in what way? Even cheap DACs these days perform better than what people can hear, including you. If you mean better in measurements, OK, that can be used as status symbol or bragging right but still well beyond our hearing limits.
Well, the point made by Evenharmonics makes a lot of sense to me. You may disagree, but the same as is true for amplifiers also goes here imo. The good ones all sound alike, nobody has proven to be able to distinguish by ear alone two properly implemented recent DAC's.
I think it has tremendous value to make this point.
Obviously. Have you any practical suggestions how a Foobar ABX test on the forum could be improved?
"Harder" might be questionable. I was expecting a null ABX result in case of my latest absolute phase test and obviously there are participants with repeated positive result. Which surprised me and I am happy. Because the reason of the test is not to get assured in some opinion or prejudice, but to find answers to questions.
I told Dan here days ago, that if only he choosed amplifier design instead of cable design, with his ears and experience he will be able to design better amplifiers (though may be harder to commercialize) and make far more improvement than a cable could make. His hearing is accurate (though there is more to it).
So GNFB is related in a unique way with OLG (and other variables). Usually, there is trade-off whether we want wide bandwidth or high loop gain. The idea to make the amp flat up to high frequency (10k/20k) before applying feedback is I think has been known since may be Williamson or Ottala, and this has been followed by many here including OStripper and Dadod.
My latest finding was I didn't need this wide bandwidth before FB. Because I found compromises with that. But I don't have many/sufficient experience with this as it was not easy to do it. But since Dadod shows his latest 100W schematic recently, I realized that with his compensation scheme wide bandwidth before FB is easy!
Add: please note that what I mean with 'wide bandwidth' is that it is FLAT up to high frequency. Ignore use of wrong terminologies.
Sure, as stated numerous times before, take time to get used to the specifics of this test protocol, use some training files and positive controls, select the number of trials in each experiment before starting and do this number of trials.
And, notice that you don´t have to use ABX if you dislike it (even after trying to get used to it); there are other protocols to choose from, and it depends on the task which one to use.
@ PMA,
No, "harder" is what the evidence shows. Depends on the statistical approach.
And of course, it is an empirical result, which means the risk is higher compared to other protocols.
As usual, it is always possible that some humans don´t have any difficulty with this protocol at all, ever find it easier to use and get good results.
As already written quite often (see for example posts from 2009) people like Bruno Putzeys and Paul Frindle did report very impressive results (means would be associated with high sensitivity under the specific test conditions), but, as said before, the risk is higher overall.
Last edited:
Bob Cordell picked that up in his book*. He convincingly showed that in a modern, reasonably linear amp to start with, feedback clearly reduced ALL harmonics, not creating new ones either.
I have posted the relevant graphs repeatedly at diyaudio. Unfortunately, those who's agenda is furthered by misleading the audience will not be interested at all in the real facts so they never get mentioned. And so the misconceptions are propagated.
* You DID order his 2nd Edition, didn't you??
Jan
Thanks. I don't know whether that is conducive to maximum audience participation though 😉 Does the person conducting the test have to decide on the most suitable protocol or is that something the individual can decide if they prefer one over another?
I showed this compensation OITPC OITPC - Output inclusive TPC (not TMC) but I used it in my 200W CFA and 100W VFA quite time ago.
Damir
PS. I hope Richard will show us soon some measuring and listening results as his amp use OITPC too.
That's I think, may be, 80% of your intentions with the tests. 20%, sometimes you do bad things too 😀
Discussions about this ABX thing is difficult here because there are too many conflicting agendas in this thread since it is first introduced. I can see people claimed to hear things that do not exist and some objectivists here so eager to challenge these people with ABX. But on the other hand, MMerill was correct about the 'weakness' of the test etc. To most of the objectivists this has been misinterpreted as 'defending' those who make false claims.
Yes, I know, but I didn't realized the benefit of the compensation scheme back then. That's why I asked the benefit of the complex compensation as I could do with one cap, -0dB up to 1kHz and the 20kH THD is even better. Now I realized that with using your compensation scheme I would have less compromises. I don't think this compensation has timing issues as it works on 'VHF' but I'm not decided yet whether it will present difficult load for the front end or not.
My agenda is democratisation of ABX testing in the context of a DIY audio forum, why should anyone else's agenda conflict with that? 🙂
Forgot to add above, use repetitions/replications of the test but report all results.
As usual, it depends.
PMA does not demand a "Foobar ABX" (at least it is my interpretation of his posts) because it is the best suited for all the tasks, but because he appreciates the additional "anti-fraud" features.
So, if you don´t like it, do not participate; conduct your own instead.... 😉
- Status
- Not open for further replies.