Including rerunning the FIR filter creation process? As above, even if nothing moved it still doesn't gaurantee that the exact same conditions were held.
For example, if there is an anomaly in the noise floor of your unfiltered measurement, then the FIR filter created will try to compensate for it. E.g. if there is some environmental 10KHz noise that occurs 0.9ms into the measurement, then the created FIR filter is going to make the system play the anti-phase impulse of that noise - i.e. you see a glitch 0.9ms into your measurement because the noise didn't exist when the system was remeasured.
Does that make sense?
So when exactly are you going to explain the same phenomenon in Halair's measurement without FIR or EQ? I already said it is obvious there was a difference in output at HF.
That was obvious in his measurement as well. Forget about the FIR part. It is not needed to show a difference here. The FIR correction isn't even that long at high frequencies anyway. It is a variable length correction but it isn't what is explaining this behavior.
So what else can you come up with. Seriously. I have my own theories and those did revolve around measurable differences between cables. I also believe that having an array here plays a big part in that.
Now if I only said I heard an improvement, would that have been any better?
I didn't answer the question. The microphone was at the listening position, I was not. I was looking for this so I knew not to touch that mike! The rest is run from the keyboard. Nowhere near the speaker or microphone.
That was obvious in his measurement as well. Forget about the FIR part. It is not needed to show a difference here. The FIR correction isn't even that long at high frequencies anyway. It is a variable length correction but it isn't what is explaining this behavior.
So what else can you come up with. Seriously. I have my own theories and those did revolve around measurable differences between cables. I also believe that having an array here plays a big part in that.
Now if I only said I heard an improvement, would that have been any better?
I didn't answer the question. The microphone was at the listening position, I was not. I was looking for this so I knew not to touch that mike! The rest is run from the keyboard. Nowhere near the speaker or microphone.
Last edited:
I'm not sure which measurement you're reffering to, can you link me please?
I'm not trying to spite you, i'm trying to help you interpret the results. The results don't line up with engineering theory that dictates the kind of phase shift you expect from the parasitic RCL of a cable. Any phase shift introduced by an RCL filter will exist equally at every time slice, not just pop up at 0.9ms or 5ms and disappear at every other time. I'm fairly confident that what you are observing from your measurement isn't caused by the cable. It is more likely an artifact of the measurement or FIR filter creation process.
If you change more than one test variable between measurements you can't be sure which one caused the difference in the result. Changing the FIR filter after changing the cable is changing more than one variable.
Like I said before, there isn't a need to apply the FIR to see this behavior.
To judge the cables I didn't change the processing. It is obvious enough in the measurement by seeing differences in the extension of the high frequencies.
I believe you are not here to be on my case. I wasn't sure at first but I see you're thinking. As am I. It wasn't until I saw measurements taken in a very damped room that I saw what is possible with a good speaker. Not my room and not my speaker. But it made me wonder about a few things. If you look at a non smoothed plot of an array you're going to notice the combing. No way around it with arrays.
But when I looked at Halair's measurement I saw an immediate difference compared to my own plots, but just on one side.
Just found the link again but it is dead. I saved it to be able to see differences to my own arrays. At the time I downloaded it I was helping him get useful measurements to base EQ on. I didn't have my speakers ready at that time. His REW measurement used to be here: http://www.diyaudio.com/forums/full-range/261157-room-treatment-full-range-line-array-rew-measurement-5.html#post4046219, it's not my place to put it up again and I generally don't share my measurements (yet). I might at a later time but first I need to know I've done all the little things I can to get them to perform the best they can.
Basically, the differences is in the dips. With the better performing wire the dips aren't as deep as with the old wires in the unsmoothed measurement. Had I used a single tweeter you would never have seen that much difference. It is because there is more than one driver playing the same frequency that this surfaced. And it's only a tiny difference for each speaker creating a bigger end result. Measurable even.
To judge the cables I didn't change the processing. It is obvious enough in the measurement by seeing differences in the extension of the high frequencies.
I believe you are not here to be on my case. I wasn't sure at first but I see you're thinking. As am I. It wasn't until I saw measurements taken in a very damped room that I saw what is possible with a good speaker. Not my room and not my speaker. But it made me wonder about a few things. If you look at a non smoothed plot of an array you're going to notice the combing. No way around it with arrays.
But when I looked at Halair's measurement I saw an immediate difference compared to my own plots, but just on one side.
Just found the link again but it is dead. I saved it to be able to see differences to my own arrays. At the time I downloaded it I was helping him get useful measurements to base EQ on. I didn't have my speakers ready at that time. His REW measurement used to be here: http://www.diyaudio.com/forums/full-range/261157-room-treatment-full-range-line-array-rew-measurement-5.html#post4046219, it's not my place to put it up again and I generally don't share my measurements (yet). I might at a later time but first I need to know I've done all the little things I can to get them to perform the best they can.
Basically, the differences is in the dips. With the better performing wire the dips aren't as deep as with the old wires in the unsmoothed measurement. Had I used a single tweeter you would never have seen that much difference. It is because there is more than one driver playing the same frequency that this surfaced. And it's only a tiny difference for each speaker creating a bigger end result. Measurable even.
Last edited:
What gets me is that in the 'good' cable measurement the phase response is pretty much perfectly flat out to 20KHz and in the 'bad' cable measurement there is more than 1080° of phase shift (three full 360º rotations) between 10KHz and 20KHz.
The parasitic components of an electrical cable form a 2nd order lowpass filter, which is only capable of introducing 180º of phase shift and where the phase is close to 180º the amplitude response would be down by at least 20dB. 1080º in one octave requires a >24th order filter (!). It's possible for the parasitic components of a cable to create a bumpy response instead of a simple lowpass if the speaker has a bumpy impedance, but then as the amplitude dips the phase would increase and as the amplitude rose back up the phase would decrease back to where it was. This is not the case in your measurement, the phase just keeps increasing.
The final indication that it is probably not a cable effect is that the effect changes with time. Electrical filters are time invariant.
The parasitic components of an electrical cable form a 2nd order lowpass filter, which is only capable of introducing 180º of phase shift and where the phase is close to 180º the amplitude response would be down by at least 20dB. 1080º in one octave requires a >24th order filter (!). It's possible for the parasitic components of a cable to create a bumpy response instead of a simple lowpass if the speaker has a bumpy impedance, but then as the amplitude dips the phase would increase and as the amplitude rose back up the phase would decrease back to where it was. This is not the case in your measurement, the phase just keeps increasing.
The final indication that it is probably not a cable effect is that the effect changes with time. Electrical filters are time invariant.
Last edited:
Don't think that way... it's not that. The phase shift happens due to the bigger dips in between the peaks. It's only the phase that drops due to the interpretation of the measurement in REW. No roll off orders here. No noticeable change in output. Difference in cable resistance perhaps? I couldn't measure it though with my low grade ohmmeter.
Before this little test I was sure it was something like resonances in the baffle etc.
Look at this post:
http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array-116.html#post4323842
Tiny changes in time, by using different length of wire. Multiply that with 25 drivers creating their own time smear on top of that at the mic position.
If that isn't it, it will still be the cable that has changed, nothing else. The experiment was repeatable. So no other influences play a part in this. Not even FIR filtering.
Halair, do you want to swap cables left and right and re-measure? That would do it. At least for me it would.
Before this little test I was sure it was something like resonances in the baffle etc.
Look at this post:
http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array-116.html#post4323842
Tiny changes in time, by using different length of wire. Multiply that with 25 drivers creating their own time smear on top of that at the mic position.
If that isn't it, it will still be the cable that has changed, nothing else. The experiment was repeatable. So no other influences play a part in this. Not even FIR filtering.
Halair, do you want to swap cables left and right and re-measure? That would do it. At least for me it would
.
Last edited:
A <1dB change is one thing and perfectly describable by electrical physics. 1080º phase shift which can only be observed when the gate size is greater than some amount which is a significant number of wavelengths at the frequency of interest is another thing altogether...
I agree that cables make a difference. Just not the difference that you want to attribute them to. This all falls back to experimenters bias. The test performed doesn't eliminate all but one variable. What you need to do is measure the system without a filter under quiet conditions, then very very carefully swap the cables making sure not to bump anything and measure it again with the different cable. Only then can you be fairly confident that the test conditions were the same and the only variable that changed significantly was the cable. If you can then swap back and forth between the cables and get the same results over and over again then you can be fairly confident.
Observing anything after more than a few wavelengths is sketchy at best anyway. If you even stand in a different position when you run the measurement you'll change the late reflections and things will change. It's pretty much the same for the FIR filter. What happens at 0ms is most important and easily repeatable.
I agree that cables make a difference. Just not the difference that you want to attribute them to. This all falls back to experimenters bias. The test performed doesn't eliminate all but one variable. What you need to do is measure the system without a filter under quiet conditions, then very very carefully swap the cables making sure not to bump anything and measure it again with the different cable. Only then can you be fairly confident that the test conditions were the same and the only variable that changed significantly was the cable. If you can then swap back and forth between the cables and get the same results over and over again then you can be fairly confident.
Observing anything after more than a few wavelengths is sketchy at best anyway. If you even stand in a different position when you run the measurement you'll change the late reflections and things will change. It's pretty much the same for the FIR filter. What happens at 0ms is most important and easily repeatable.
Last edited:
I can't help you as I do not possess a quiet room. I told you many times that I did do more than one measurement and that I didn't use any other processing in between.
Take from it what you want. Call it whatever you want but I know what I did do and did observe. It isn't a phase rotation the way you see it. Many will observe phase rotations at the listening position in their room and see them change in time. To me I am only chasing a clean result in the first few milliseconds, changing the wires has been part of that.
Do not think the "real acoustic phase" has been shifted that many degrees. That hasn't happened.
Take from it what you want. Call it whatever you want but I know what I did do and did observe. It isn't a phase rotation the way you see it. Many will observe phase rotations at the listening position in their room and see them change in time. To me I am only chasing a clean result in the first few milliseconds, changing the wires has been part of that.
Do not think the "real acoustic phase" has been shifted that many degrees. That hasn't happened.
No magic, just tiny differences... lets see if this clears things up.
Group Delay in REW of the different wires, you guess which is which...
Both smoothed 1/6 oct and gated 5 ms. Note the scale on the left!
I told you it was timing differences, didn't I. Too bad I didn't think looking at this before.
Here's the worst as you normally view it:
Keep in mind that you're looking at the highest frequencies of 25 3.5" full range drivers.
Group Delay in REW of the different wires, you guess which is which...
Both smoothed 1/6 oct and gated 5 ms. Note the scale on the left!
I told you it was timing differences, didn't I. Too bad I didn't think looking at this before.
Here's the worst as you normally view it:
Keep in mind that you're looking at the highest frequencies of 25 3.5" full range drivers.
Last edited:
Halair, do you want to swap cables left and right and re-measure? That would do it. At least for me it would
I'll see what I can do, in order to do that I need to move my stationary Mac downstairs and have the hous to myself for a couple of hours - which happens all to rarely
No hurry for me. But maybe we could prove something here. If it wasn't for your measurements I wouldn't have looked at it. Now I want to know more (lol).
If you could reproduce this effect and show it to follow the cable it does mean at least something. I already saw that at my home but that would or could be considered inconclusive.
If you could reproduce this effect and show it to follow the cable it does mean at least something. I already saw that at my home but that would or could be considered inconclusive.
Interesting topic. I've run into the same phase issue and the mic/speaker/in room position was quite critical. Once those things were considered many of my measurements still would show phase wrapping. Hmmm Then I considered inaudible low frequency noise as this can skew phase results. It's handy to have two cal mics so I put them to use measuring the speaker and the room. Bingo
Anymore the first thing I look at is the phase response, if its not clean, discard and repeat.
Then I considered inaudible low frequency noise as this can skew phase results. It's handy to have two cal mics so I put them to use measuring the speaker and the room. BingoAnymore the first thing I look at is the phase response, if its not clean, discard and repeat.
Well I can't say if this is comparable. Different wires gave different results. Swapping them in and out confirmed that difference. I kept the best performers in there. A repeatable concept really. Made sense to me to publish the result of it. I still think I wouldn't have gotten results like this with a single tweeter. That's not indicating that there wouldn't have been a difference in wires with that single tweeter but it would have been more difficult to spot it. 25x a tiny difference in behavior somehow stands out more I think.
No amount of processing was able to get the original wires to behave similarly.
If I only had said I heard a difference it would have been less controversial (lol).
The full rangers get less boost up top due to the wire change and have less big dips in HF leading to the phase and GD graphs posted. And no, the FIR processing didn't do that.
I only used it to see if a wire with less HF output could match up, it didn't. It seems you can't compensate for everything by boosting.
No amount of processing was able to get the original wires to behave similarly.
If I only had said I heard a difference it would have been less controversial (lol).
The full rangers get less boost up top due to the wire change and have less big dips in HF leading to the phase and GD graphs posted. And no, the FIR processing didn't do that.
I only used it to see if a wire with less HF output could match up, it didn't. It seems you can't compensate for everything by boosting.
I have no idea, I haven't tried. I have the first option but I know there are people with different wiring schemes. If I have to guess the final impedance is going to be close to the same or be the same. The one thing I did work on is to get a clean as possible impedance curve from the drivers in the enclosure. That took quite a bit of work.
I tested all drivers separately and noticed it was very important how I had the baffle supported because I had all drivers mounted on the baffle. Every tiny resonance is going to show up on the impedance curve.
I tested all drivers separately and noticed it was very important how I had the baffle supported because I had all drivers mounted on the baffle. Every tiny resonance is going to show up on the impedance curve.
I finally managed to find the feet of my amplifier and the rest of the equipment so I took them off the wall were it was hanging beside the line array. It did clean up the right side response but it's still a little rougher than the left side. It seems the damping on the left side-wall really helps there. I have received warning though that I may not place more damping panels in the room . So for now this seems the best I can do to get both sides as similar as possible:
Red is right, blue is left... 1/12 octave smoothing...
. So for now this seems the best I can do to get both sides as similar as possible:
Red is right, blue is left... 1/12 octave smoothing...
