The discussion starts here:
http://www.diyaudio.com/forums/full-range/265915-hyperfast-hypercube-based-2-way-6.html#post4152525
Based on Gmad's speaker from here:
http://www.diyaudio.com/forums/full-range/265053-hypercube-loudspeakers.html
Measured IR (near field):
Here is square wave from same:
Btw, Wesayso is doing the same thing but x50 for his line arrays. Now that is an impressive speaker!
Last edited:
Probably nobody has ever compared similar 4" drivers in a blind test. That's exactly why doing that test makes sense. I certainly do have a certain expectation what the result is but that would only be conjecture. The idea of doing a blind test is to gain real knowledge and do away with conjecture.
This test would not prove that axial response is all that matters. It would prove that other things matter most.
Measured IR (near field):
Quite some pre-ringing. Do you also have data that shows how the impulse looks at other locations? It's pretty easy to create a near-perfect impulse response with EQ -- but only at a single point in space 😉
This is Gmad's work - you will have to ask him. There is pre-ringing which is the nature of FIR filters - but the amplitude is relatively small compared to the peak. Certainly less than the typical ringing we see on all the drivers I measured.
Tony, while it may be true that uniform polars matter, it is not true that their importance outweighs the on-axis response. On-axis response, the first thing your ears hear, is THE primary determinant of sound quality. Second comes the off-axis response, which with holes is ok, but peaks are not good. Distortion, unless it is very high, is usually not a concern in properly designed speakers.
I would highly doubt you would be able to hear the difference between two systems with a flat on-axis response, assuming the polars are not extremely different. Most likely you would conclude that they sound alike with some minor differences.
I'll relate a personal experience. I had a huge horn system with the large Altec VOTT bass horns combined with the SEOS-24 horns. Each speaker is about the size of a modern refrigerator. I also have a KEF Q100 5" dual concentric driver in an 8" x 11" cabinet. I can swear to you that with approximate level matching, I couldn't tell which system was playing in my room, once the frequency responses were alike. This is a sighted test. I mean yes, the VOTTs will go loud and the two systems have completely different polar patterns, but at comfortable levels, they sounded more alike than different. I would describe the character as neutral, listenable on all sorts of music, really clear, non-fatiguing.
At listening position see #437 http://www.diyaudio.com/forums/full-range/265053-hypercube-loudspeakers-44.html#post4215089.
hi ra7, I may have been misinterpreting but I got the impression that there were at least a few who were saying that it didn't matter at all, and that the *only* thing that mattered was on axis frequency response, and that if that were made the same then it would not be possible to discern a difference.
There was also the suggestion that the test should be done outside so that there could be no reflections. This to me sounds like rigging the test to show that there *is* no difference, when in a real world test (properly conducted and controlled) may well show that there are discrenable differences (due to polar response differences).
It just seemed to me that one of the things that is likely to be a differentiator was potentially being excluded from the test regime.
Now for my anecdote 🙂
I recently made some repairs to my parents 1970's pioneer three ways, and set them up in my room for a listen. I was very surprised at how little difference (if any) I could hear with them compared to my very different MTM's I don't have any measurements of the pioneers, but my MTM's are pretty flat. This was a bt of a surprise,, but not a shock 🙂
Tony.
There was also the suggestion that the test should be done outside so that there could be no reflections. This to me sounds like rigging the test to show that there *is* no difference, when in a real world test (properly conducted and controlled) may well show that there are discrenable differences (due to polar response differences).
It just seemed to me that one of the things that is likely to be a differentiator was potentially being excluded from the test regime.
Now for my anecdote 🙂
I recently made some repairs to my parents 1970's pioneer three ways, and set them up in my room for a listen. I was very surprised at how little difference (if any) I could hear with them compared to my very different MTM's I don't have any measurements of the pioneers, but my MTM's are pretty flat. This was a bt of a surprise,, but not a shock 🙂
Tony.
Ah, understood! Yes, if the test is done, it should be done indoors with location of the speakers and microphone fixed. I also suggested that rather than EQing the response flat, we should let the drivers be and let people pick their favorite sound clip (drivers should not be identified before hand). That way, we would know people's preferences.
Sure. This notion that detail resolution is caused by peaks in response is partially true, but misleading. Yeah, you'll get what might be perceived as enhanced details if, for example, you have a lift from 5-8kHz. However, I think it's basically a beginner's mistake. With some experience, the difference between resolution and non flat frequency response is obvious.
As I've generally EQ'd or otherwise balanced most of the systems I've made to have a flat FR, the fact that some drivers have better detail retrieval is abundantly clear. (pardon the pun) I have a few tracks I'll use to test this, which have subtle details that few drivers can pick up. These tend to be things like sax keys clacking, or someone walking in the back of a recording studio.
These sorts of details can be easily swamped by room noise, room reverberation, mediocre electronics, etc., so it's important to consider these things. That's true for listening tests and (presumably) testing as well.
I find that CSD is a pretty good indicator for this, but it's not often available.
I suggested the test with equalised frequency response for one reason: to get it out of the way (as far as that is possible). What should remain are the other factors at play.
We have endless discussions on here already that don't solve anything.
Preferably I would suggest to exclude just one factor (a time) and see what that gets you. That way we would maybe learn something. To those who did not do a test like this, why bother thinking about what it would sound like? Do it or don't I say. Speculating about it doesn't move us forward, right? 😉
There is a lot of talk of what we can't measure. But I can think of a lot more ways/things what we could (or maybe should) do with measurements to actually learn something.
What you might be calling resolution could really be how linear the driver is over it's operating range. When pushed, drivers distort, including changing the frequency response, i.e., linear distortion. The better quality drivers with linear suspensions, linear motors and copper shorting rings (found in the TC9, btw), help keep the driver's linearity over a larger range of operation. I'm sure you'll agree that these are secondary to the on-axis frequency response, which must be linear in the first place. The kind of distortion that a +9db shelf in the midrange can produce is far too large compared to the secondary parameters, such as, how loud does it go.
Interesting as this could be it still won't result in "better" loudspeakers. The enjoyment of music is a subjective experience, much as the enjoyment of art, food, or wine. Personal preference will always come into it, and one person's taste will differ from anothers. Transducer design can never be perfect.
Measuring and building a dataset of driver performance though is a useful exercise as it can help diyers decide before purchase which drivers might meet their personal taste.
I'm missing why it wouldn't lead to better speakers. In todays world we have a lot of relatively new tools available to make them better even with the drivers we have now.
Sure, some would never accept these new tools but does that matter? We would be wiser from a couple of tests, and if you're willing you could take advantage of the things we learn. I'm all for it! Wait, I suggested it (lol).
Sure, some would never accept these new tools but does that matter? We would be wiser from a couple of tests, and if you're willing you could take advantage of the things we learn. I'm all for it! Wait, I suggested it (lol).
In the imprecise world of transducer design how would you define better? Unless you can do perfect the resulting variations will result in disagreement amongst different people when listening to music. Don't lose site of the end goal.
Let's focus on recording the differences and note what characteristics may lead to which particular sonic results.
Thanks for the links.
Thanks, I'll have a read!
By the way I totally agree with your objective philosophy and your test methods are more than accurate enough for any open minded DIY'er to learn a lot from all your hard work.
All the best
Derek.
PS Someone made a comment about you only having 2 years experience Vs one of the old timers 30 years experience......I would say that many of the old guard dont have 30 years experience....They have one....They have just repeated their first year 30 times.....!!!!
Thanks, I'll have a read!
By the way I totally agree with your objective philosophy and your test methods are more than accurate enough for any open minded DIY'er to learn a lot from all your hard work.
All the best
Derek.
PS Someone made a comment about you only having 2 years experience Vs one of the old timers 30 years experience......I would say that many of the old guard dont have 30 years experience....They have one....They have just repeated their first year 30 times.....!!!!
The discussion starts here:
http://www.diyaudio.com/forums/full-range/265915-hyperfast-hypercube-based-2-way-6.html#post4152525
Based on Gmad's speaker from here:
http://www.diyaudio.com/forums/full-range/265053-hypercube-loudspeakers.html
Measured IR (near field):
Here is square wave from same:
Btw, Wesayso is doing the same thing but x50 for his line arrays. Now that is an impressive speaker!
No problems with that at all! Changing only one thing at a time is the best way to identify what has an impact and what doesn't. I didn't have any issue with the flattening, just what I perceived as an assertion (by some) that once it was done then there would be no discernible difference.
Also there was a suggestion to do the listening test outside to eliminate reflections. This could be done I guess as a separate exercise, and would be interesting to compare the results to an in room listening test, where all other factors were equal.
Tony.
That's exactly what a 'properly conducted and controlled' is about, isolate potential differentiators as much as possible.
If no difference can be heard between two drivers equalized to have the same frequency response then we know that other factors like polar pattern and therefore interaction with the room has the most important effect on what we hear. At the same time we would have learned that subjective audiophile concepts like 'DDR' have no real meaning.
DDR is not an audiophile concept. As Barleywater pointed out, it is dynamic range -- a well established engineering term. But the conotation ofdynamic range most often used is the ability to deal with a large increase in output without compression or other problems. Adding the prefix Downward just focuses the attention to the other end of the scale, the small stuff, detail, low level information. So it is just saying don't forget the small stuff. Thanx Barleywater fro that.
Now how does one measure the lower limit of a speaker's dynamic range? I'd be tempted to call itthenoise floor, but it is known that we can hear into the noise.
dave
PS: how is your morning coffee?
Could also come from narrow peaks or a difference in how the speaker interacts with the room. You can't be sure what exactly it is as long as you're doing sighted listening tests that do not isolate potential differentiators.
Sure it is an audiophile concept and Barleywater did a great job in identifying it as such. Please go back an re-read what he said. There's not much I could add.
- Status
- Not open for further replies.