While it is relevant for Fourier math, it is irrelevant for the topic: it is enough that it was infinite long and stable before we started our observations/measurements. And it does not matter if it ends, or not after we measured.
I myself use a pair of Wavetek functional generators, one to VCA of another, to observe transients.
They can't kick me physically through the Net. It is a good side effect of computer usage.
You are missing the point. I am looking at what the amp/driver SYSTEM is doing. By changing only one component of the system, the amp, I see the system behave differently. It is not phase as it changes. It is a representation of the driver's ability to get into motion from a stop. BTW, I did these at 4KHz, not 10 Hz.
No, I didn't miss the point, don't worry. I did misunderstand about the 10 cycles, tho.
Measuring at the speaker terminals may give you a better look at what's going on, which is why I suggested it. Whatever the driver cone is doing, you should see it there electrically. I can even see cabinet resonances in my impedance sweeps. If, however, you are seeing something in the acoustic measurements that is not at the driver terminals, that would be good to know, and surprising. If the amps are doing different things, it ought to be showing up at the speaker terminals. You will see the amp/driver system that way - without the uncertainties of air and microphones.
And if you do see these differences as measured at the speaker terminals, then substitute a resistor as load and the difference go away, you've found something important. I.E., the amps measure the same on a resistive load, but not on a real speaker load.
Measuring at the speaker terminals may give you a better look at what's going on, which is why I suggested it. Whatever the driver cone is doing, you should see it there electrically. I can even see cabinet resonances in my impedance sweeps. If, however, you are seeing something in the acoustic measurements that is not at the driver terminals, that would be good to know, and surprising. If the amps are doing different things, it ought to be showing up at the speaker terminals. You will see the amp/driver system that way - without the uncertainties of air and microphones.
And if you do see these differences as measured at the speaker terminals, then substitute a resistor as load and the difference go away, you've found something important. I.E., the amps measure the same on a resistive load, but not on a real speaker load.
All very true. Starting at 0 volts AND 0 current as it is just turning on could have quite an interesting spike in current, or not. That would be very interesting to measure. I could see where the difference to supply near instant maximum current would be a reasonable cause. Thinking on the keyboard out loud, output zobel differences? Location the feedback is picked up from? I don't have a good current probe, just my buggered amp-clamp with a voltage output. Guess I should troll e-bay for a real one.
The particular tweeter is the Seas 27TBFC/G, a ferofluid filled metal dome. I would not suspect it to be terribly difficult to deal with. It's one of their better ones before moving to the Excel line where they get really serious. I have about 6 other tweeters in the ~ $100 range on the shelf. It would be good to test them.
No, I didn't miss the point, don't worry. I did misunderstand about the 10 cycles, tho.
Measuring at the speaker terminals may give you a better look at what's going on, which is why I suggested it. Whatever the driver cone is doing, you should see it there electrically. I can even see cabinet resonances in my impedance sweeps. If, however, you are seeing something in the acoustic measurements that is not at the driver terminals, that would be good to know, and surprising. If the amps are doing different things, it ought to be showing up at the speaker terminals. You will see the amp/driver system that way - without the uncertainties of air and microphones.
And if you do see these differences as measured at the speaker terminals, then substitute a resistor as load and the difference go away, you've found something important. I.E., the amps measure the same on a resistive load, but not on a real speaker load.
I agree with these comments too. Note I am comparing the amp output with the mic output. I was not looking at the difference in amp input to output. I should do this too.
What you're describing is how uncorrelated the "distortion" is with the signal. Things such as harmonic distortion and baffle edge diffraction are highly correlated to the signal a hifi is trying to reproduce. That sort of thing makes it easier to tell if some "distraction" such as someone who coughs is in the room with you, or if the coughing is on the recording and being reproduced.
The sudden (instantaneous!) startup and stop of the waveform adds a "click" to the signal, and this shows up as wideband noise in the spectrum. This is why windowing was developed for practical FFT measurements.
If you ramp up the level during the first six cycles then ramp down the level for the last six cycles, it will substantially reduce the generation of signals other than your 4k burst.
This article should demonstrate. The spectrum of what you're generating is shown for the rectangular window here.
Window function - Wikipedia, the free encyclopedia
Someone should find out what they are and how important each one is. At first glance it doesn't seem possible, but I think with measurements AND listening tests of lots of speakers (and adequate questionnaires for the listeners, this may be the hardest part), it should be possible to find out by taking all this data and doing a regression analysis.Pano, with his "Sound Quality Index" ?
Toole, with his speaker ranking studies ?
Note: not a single measurement, but a single aggregate figure of merit derived from a group of measurements. That's what I'm saying is futile, because I think you simply can't boil down so many competing and largely orthogonal factors into a simple scalar figure of merit, especially if personal preference is indeed a factor.
And there's not much reason to, unless your goal is to replace controlled listening tests with measurements to see how speakers rank against each other in listener preference. A single boiled down figure of merit gives no guidance into what needs improving in each speaker, so doesn't help the designer.
I'm 100% with you there, you're arguing the same side of the argument as I did in my longer post...?
...
Formal listening tests are a lot more expensive and time-consuming than taking measurements, and if a few semi-reliable "figures of merit" can be correlated between the two, I'd think a designer could make more progress by using such "figures" calculated from repeated measurements as design changes are made, and then find out how these correlate with listening tests.
"Data mining" is a Big Thing for businesses thesedays (for things such as finding out what customers Really Want so they can sell more of it), and it looks like an interesting topic:
Data mining - Wikipedia, the free encyclopedia
Yes, a nice envelope would probably provide a nice pretty test. I am looking for something that causes a difference and to see if it correlates to the subjective test. Something extraordinary, a stress test if you will. All the nice "reasonable" tests I know of don't show anything different. Hey, I could be so far off in left field I am totally lost. I did see a difference. Relevant? Don't know. Just a difference.
If I can reliably reproduce the difference, than I can hypothize what parameters could be driving it and change that in an amp to see if it correlates to the subjective test. I suspect making a "good" amp bad is easier than making a bad amp good. Only then can I draw a conclusion with enough substance to have others reproduce the experiment and draw their own conclusions.
Tangent thought. You may have noticed I have them. We all know the original Fletcher-Munson experiments and how those evolved into the ISO standard for relative loudness. What I have never seen is the deviation across the tested population. We "know" the fair gender hears higher frequencies and likely the critical 4K range with greater sensitivity that we do. By how much? All I can find is the normalized curves. Has anyone seen the data correlated to differences in the population? Has anyone done any measurements on brain activity to determine if there is a difference in what we detect, and what we process as sound? I my not be using the best search terms.
If I can reliably reproduce the difference, than I can hypothize what parameters could be driving it and change that in an amp to see if it correlates to the subjective test. I suspect making a "good" amp bad is easier than making a bad amp good. Only then can I draw a conclusion with enough substance to have others reproduce the experiment and draw their own conclusions.
Tangent thought. You may have noticed I have them. We all know the original Fletcher-Munson experiments and how those evolved into the ISO standard for relative loudness. What I have never seen is the deviation across the tested population. We "know" the fair gender hears higher frequencies and likely the critical 4K range with greater sensitivity that we do. By how much? All I can find is the normalized curves. Has anyone seen the data correlated to differences in the population? Has anyone done any measurements on brain activity to determine if there is a difference in what we detect, and what we process as sound? I my not be using the best search terms.
benb,
Data mining is old school. We have been doing it with data warehouse products like IQ for a long time. Where it gets interesting is when we shift to big data. THAT is something I actually do know a bit about. Multi-petabyte is one thing, but doing sliding window analysis on data bigger than can fit in memory for near real time analysis gets fun.
Data mining is old school. We have been doing it with data warehouse products like IQ for a long time. Where it gets interesting is when we shift to big data. THAT is something I actually do know a bit about. Multi-petabyte is one thing, but doing sliding window analysis on data bigger than can fit in memory for near real time analysis gets fun.
P6302's seem out of my price range. Anyone have suggestions on current probes with a more limited bandwidth? I have been out of the lab so long I don't know what is out there. They were really expensive back then. I guess I could just rig a hall effect transistor to a op-amp or something to give relative difference measurements. Butcher an old Bosch ignition module maybe?
[venting]Why does this discussion remind me of the metaphor of 'the blind men and the elephant'
I'm so glad the micro analysis of topologies is over...whew. When I set up a system for live performances. You know, the ones you like to recreate in a bedroom then complain about the timber of the drivers being wrong. It takes a long time to learn how to tune a room. If the amps are good enough*, then I say "good, next problem". (problems mostly created by architects, grrr) btw, I hate large hanging panels and Jumbotrons. Thanks for the large sonically reflective surface,
.
*(btw, my definition of "good enough", is stunningly transparent, crowd pleasing and very very clean)
can you tell what I did today...
Tuning the room takes much more effort and makes much more of a difference than tweaking the output of the HF drivers or adding a soft filter and a notch filter to kill a standing resonance.
I wonder why designers get sucked into the minuta trap. ie. it could be .00000002% better if I just changed this resistor... why, it's good enough. I guess it would take a renaissance man to stop designing the amp and move on to design the drivers, cabinets and room. Well that and you can't sell the whole system just most of it.
I need a beer.
[/venting]
Sorry guys, I needed to get that out.
cheers,
revb.
I'm so glad the micro analysis of topologies is over...whew. When I set up a system for live performances. You know, the ones you like to recreate in a bedroom then complain about the timber of the drivers being wrong. It takes a long time to learn how to tune a room. If the amps are good enough*, then I say "good, next problem". (problems mostly created by architects, grrr) btw, I hate large hanging panels and Jumbotrons. Thanks for the large sonically reflective surface,
.*(btw, my definition of "good enough", is stunningly transparent, crowd pleasing and very very clean)
can you tell what I did today...
Tuning the room takes much more effort and makes much more of a difference than tweaking the output of the HF drivers or adding a soft filter and a notch filter to kill a standing resonance.
I wonder why designers get sucked into the minuta trap. ie. it could be .00000002% better if I just changed this resistor... why, it's good enough. I guess it would take a renaissance man to stop designing the amp and move on to design the drivers, cabinets and room. Well that and you can't sell the whole system just most of it.
I need a beer.
[/venting]
Sorry guys, I needed to get that out.
cheers,
revb.
Tvrgeek,
just a thought, have you tried taking one of the amp speaker combinations that your wife likes and moving it somewhere else. Say, outside, or in a bathroom or unfinished basement?
I wonder if it's the amp-driver or the amp-driver-room-number of people in the room- humidity level-time of day-etc etc.. you get my drift.
just a thought, have you tried taking one of the amp speaker combinations that your wife likes and moving it somewhere else. Say, outside, or in a bathroom or unfinished basement?
I wonder if it's the amp-driver or the amp-driver-room-number of people in the room- humidity level-time of day-etc etc.. you get my drift.
Yup. It seems independent of the room. She does appreciate the tuning I have done in some of our rooms. I can't say enough about what large absorbers can do for imaging when behind the speakers, all of what I have done for the bass and so on. That seems independent of the harshness she hears. I would agree with the comment room tuning is a lot harder than system tuning. Doing it in a room and making it look not dominated by the sound system is even more of a trick. We try and keep our living room looking like a living room.
- Status
- Not open for further replies.