This network I want to try isn't a zobel, it's like a zobel, posted a few pages back by BYRTT. It's a conjugation network for the speakers plus enclosure. So it has a lot more components besides the cap and resistor.
The resistor I posted a picture of from ATE is a military spec that's at least wired to have low induction. I guess I just have to try...
I also want to try a resistor in series, kind of like what Nelson Pass suggests in his paper:
http://www.customanalogue.com/various/Current-Amplification.pdf
Just want to learn a bit more here, I've seen and read claims... I want to know if I can spot a difference.
Different wires made a real difference, at least looking at my phase plots. So why shouldn't this do something as well.
If it does do something besides burn power I might have learned something. BYRTT got the idea from a thread on here that got cut short.
Basically because it always leads to controversial discussions. An array like this, acting as a single big full range driver might reveal something.
It's worth the lesson I hope. I learned a lot by just playing with these arrays. Trying different things. I love the way they sound right now.
Measurements do show delay at or close to the box resonance, but that could also be caused by another problem (room or array size). If there is a chance this will do something beneficial it's worth the try. Look at this post: http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array-116.html#post4323229
Sims are included that show tiny effects in timing when an amplifier is used that doesn't have a real zero impedance. The effects get larger with more resistance in series. Based on those sims and the background from Nelson Pass it's worth further investigation I'd say.
Best or worst case (depending how you look at it) it will lead to me needing to save up for one of those big Pass Labs amps...
The resistor I posted a picture of from ATE is a military spec that's at least wired to have low induction. I guess I just have to try...
I also want to try a resistor in series, kind of like what Nelson Pass suggests in his paper:
http://www.customanalogue.com/various/Current-Amplification.pdf
Just want to learn a bit more here, I've seen and read claims... I want to know if I can spot a difference.
Different wires made a real difference, at least looking at my phase plots. So why shouldn't this do something as well.
If it does do something besides burn power I might have learned something. BYRTT got the idea from a thread on here that got cut short.
Basically because it always leads to controversial discussions. An array like this, acting as a single big full range driver might reveal something.
It's worth the lesson I hope. I learned a lot by just playing with these arrays. Trying different things. I love the way they sound right now.
Measurements do show delay at or close to the box resonance, but that could also be caused by another problem (room or array size). If there is a chance this will do something beneficial it's worth the try. Look at this post: http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array-116.html#post4323229
Sims are included that show tiny effects in timing when an amplifier is used that doesn't have a real zero impedance. The effects get larger with more resistance in series. Based on those sims and the background from Nelson Pass it's worth further investigation I'd say.
Best or worst case (depending how you look at it) it will lead to me needing to save up for one of those big Pass Labs amps...
Last edited:
Not there yet... First I want to establish if there is a reason to think about them.
I am intrigued by them though. Enough to try and emulate it with a series resistor.
If it turns out to be an improvement I'll give it a more serious thought.
I'm a poor man so it may take me a while to get there. Ideally I'd need plenty of power.
At it's current stage, my arrays sound much sweeter with the high frequencies clear of phase rotation like I showed a few pages back.
This is a big part of the direct sound hitting the listening position. I'll always have to deal with the room, but that was known from the start. Kind of like a goal really. To get a speaker perform that well in my room. But I'll continue to work on little improvements, if my time and budget allows me to do so.
I am intrigued by them though. Enough to try and emulate it with a series resistor.
If it turns out to be an improvement I'll give it a more serious thought.
I'm a poor man so it may take me a while to get there. Ideally I'd need plenty of power.
At it's current stage, my arrays sound much sweeter with the high frequencies clear of phase rotation like I showed a few pages back.
This is a big part of the direct sound hitting the listening position. I'll always have to deal with the room, but that was known from the start. Kind of like a goal really. To get a speaker perform that well in my room. But I'll continue to work on little improvements, if my time and budget allows me to do so.
wesayso,
I have a question regarding your measurements showing severe changes in phase by introducing a different speaker cable type, which, according to transmission line theory, should not be happening at the frequency range involved here.
That is:
What with and how exactly are your arrays wired internally? Connections soldered, crimped, hookup wire in one contiguous piece? Just curious.
Best listening, Marin
I have a question regarding your measurements showing severe changes in phase by introducing a different speaker cable type, which, according to transmission line theory, should not be happening at the frequency range involved here.
That is:
What with and how exactly are your arrays wired internally? Connections soldered, crimped, hookup wire in one contiguous piece? Just curious.
Best listening, Marin
Hello Marin,
My speakers are wired with 2 long wire runs from top to bottom. Short leads in between the speakers to keep wiring length to a minimum. Wires used are multi strand type speaker wire. I used crimped connectors on the speakers. I mounted the speakers with wiring tabs facing each other from 2 adjacent drivers.
The whole setup is intended to be serviceable. After the wire test results I might even rewire with solid core wire at some point.
It's not the end of the world as it isn't soldered, except at the binding posts.
A clear raise in high frequency output was evident after the wire change without FIR correction. To keep output the same for comparison I ran FIR correction on impulses after the wire swap. I didn't expect to see it. Makes me glad I tried.
Edit: without seeing this in Halair's measurements I wouldn't have tried it. It makes me want to shorten the leads to both ends (*), placing my amp in between or maybe use 2 separate amps in the future, placed close to the speakers.
(*) = All wiring is same length in both speakers and from amp to speaker.
My speakers are wired with 2 long wire runs from top to bottom. Short leads in between the speakers to keep wiring length to a minimum. Wires used are multi strand type speaker wire. I used crimped connectors on the speakers. I mounted the speakers with wiring tabs facing each other from 2 adjacent drivers.
The whole setup is intended to be serviceable. After the wire test results I might even rewire with solid core wire at some point.
It's not the end of the world as it isn't soldered, except at the binding posts.
A clear raise in high frequency output was evident after the wire change without FIR correction. To keep output the same for comparison I ran FIR correction on impulses after the wire swap. I didn't expect to see it. Makes me glad I tried.
Edit: without seeing this in Halair's measurements I wouldn't have tried it. It makes me want to shorten the leads to both ends (*), placing my amp in between or maybe use 2 separate amps in the future, placed close to the speakers.
(*) = All wiring is same length in both speakers and from amp to speaker.
Last edited:
I looked into those graphite resistors. It's easy enough to try. But power handling quotes of the Duelund resistors are quite low. Might be worth trying though if it turns out the proposed idea of a conjugate network combined with series resistor to tune actually works. I'll definitely keep it in mind.
After playing around with my DRC templates a bit more I discovered another cute song to try, probably more widely accepted than anything by Zappa 😛,
Steely Dan's Hey Nineteen is awesome when the backing vocals come in. All in all most Steely Dan recordings are definitely reference type recordings, very well done.
Edit: DRC template can be found here: http://www.diyaudio.com/forums/full-range/275730-convolution-based-alternative-electrical-loudspeaker-correction-networks-4.html#post4360956
Last edited:
That phase wrapping is just strange. You shouldn't get anything like that due to cable RCL parameters. If the cable caused it then your frequency response would drop like a rock as soon as the phase got to 180deg, which it doesn't seem to and it wouldn't wrap because a cable can't cause anything more than a 2nd order filter.
Have you tried doing back to back measurements altering nothing except the cables? Surely that measurement is not repeatable.
Edit: just saw you are using FIR correction. Are you reoptimising the system between changing cables or something? High Q FIR filtering can cause something like you see.
Have you tried doing back to back measurements altering nothing except the cables? Surely that measurement is not repeatable.
Edit: just saw you are using FIR correction. Are you reoptimising the system between changing cables or something? High Q FIR filtering can cause something like you see.
Last edited:
Sorry to say but it was repeatable, and no, it didn't drop like a rock.
The phase wrap with original used cable originally occurred after about 0.9 ms. With the new cable it happens past 5 ms. I tried some other cables than the two mentioned here. Without correction it was clear the newer cable had more extended HF output. As did the solid copper electrical wiring I tried. Had I only looked at the REW results of the end frequency response I wouldn't have seen it.
I did back to back measurements to settle on my new cables. I even tried paralleling cables to see what that did for the output. And yes, I used FIR correction based on the newly inserted cable measured impulses between swaps. But first I looked at HF output and I didn't correct with every wire I tried. The settings(*) of my correction didn't change. What did change is the needed correction due to different output levels of HF of the different wires.
To see it for yourself look in Halair's thread for his (latest) REW data (if it's still up) that triggered me to do this test. The only thing I saw that was different was the wires so I just had to try and I indeed could repeat those results. Did I expect it? No, not after believing everything I had read about cables and the effects it has on high frequency output. Am I glad I tried? I sure am! Will I try more tests after this: Absolutely!
(*) corrected to the same target using the same template in DRC.
Last edited:
You're right, somehow mentioning the word Zobel triggers different ideas for different people. Most think about the HF impedance correction.
What are those timings referring to exactly? Impulse response?
Because the effects of a cable shouldn't cause an effect after 0.9ms or 5ms. That's far too long for an effect that shows up >10KHz. You'd expect to see significant effects in the impulse response at 0.05ms to 0.1ms.
edit: Sorry not familiar with REW, are you looking at a frequency response for a time slice 0.9ms after the initial impulse? Because any cable effects should affect all time slices, not just after some specific time. Most likely what you're observing is a late reflection in the system causing cancellation at the microphone. 0.9ms aligns with the observation that something was going on at 1KHz (1sec/1khz = 1ms). It doesn't take much for a delayed reflection to invert the phase response at high frequency because the wave lengths are so short (3.4cm for 10KHz). You have to be very careful that absolutely nothing in the room moves even 1cm between measurements if you want to compare time slices >0.1ms from two measurements like that.
Yeah, that's pretty much what you expect for a cable with lower capacitance and inductance - less roll off at high frequency. If you try to compensate high frequency dips with a FIR filter, or compensate for one mic position and then move the mic (edit: or speaker) slightly (even 1cm!) you get something like what you observed - the phase rotations. Hence why I suspected there was something more going on than just swapping a cable and hitting remeasure.
Last edited:
I'll try that next time I measure. I'll move the mic around. Can't hurt right?
I don't feel like I left anything out here though. It's all there. So you didn't need to suspect anything, just read it 🙂. I stated that I corrected the impulse with FIR filters etc. The amount of correction needed did change. Just to be clear, the mic didn't move during this excersize, nor did the speaker.
By the way the timing references are explained in my post: http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array-118.html#post4352467
One more note to add, the REW data from Halair, that triggered this experiment did not use any form of EQ and showed similar results between two cables that peeked my interest. So to blame it all on FIR correction is going a bit to far for me. Maybe Halair's right speaker just got lucky and be placed at exactly the right spot for the microphone to pick up a signal without early phase wraps 🙂.
Last edited:
Ah right, that explains it. If you correct the impulse response with a FIR filter then the correction is only valid for that exact mic/equipment position. If you move your microphone towards or away from the speaker by half a wavelength your phase response will rotate by 180degrees. Also if any equipment moves, any reflections off walls and objects will arrive at different times so they will be under/over compensated so you will observe strange effects at different time slices.
Extreme care must be taken to do a valid comparison at high frequencies because the measurement will be extremely sensitive to changes in the positions of all the equipment.
You can observe this by measuring and compensating the impulse response of the system, then moving your microphone about 17cm closer to the speaker (approximately half a wavelength at 1KHz). Then without recompensating the system, measure the response again. You should see about a 180degree phase shift around 1KHz and lots and lots of phase rotations above 1KHz.
Extreme care must be taken to do a valid comparison at high frequencies because the measurement will be extremely sensitive to changes in the positions of all the equipment.
You can observe this by measuring and compensating the impulse response of the system, then moving your microphone about 17cm closer to the speaker (approximately half a wavelength at 1KHz). Then without recompensating the system, measure the response again. You should see about a 180degree phase shift around 1KHz and lots and lots of phase rotations above 1KHz.
But you haven't explained anything about the un-EQ-ed results from Halair. No FIR was used and no EQ. Just a difference in how and when the phase wraps show up. Just lucky placement?
A small FYI - my measurements were done with none of this in mind, and with very limited experience with REW 🙄 Still havent changed the cable 🙄
Phase wrap in +/-180 degree view of phase plot is just the view. That little dotted vertical line is not a spontaneous 180 degree phase flip.
You tell me what to call it and I'll do that for you.
To me the phase wraps with my old wire, not with my new wire at the ~5 ms gated view.
I'm happy with the new wire, out with the old. Measurement at 3 meters in a live living room full of reflections.
These are smoothed, the old ones were too, but not as much.
To me the phase wraps with my old wire, not with my new wire at the ~5 ms gated view.
I'm happy with the new wire, out with the old. Measurement at 3 meters in a live living room full of reflections.
These are smoothed, the old ones were too, but not as much.
Last edited:
Phase shift is inevitable in the unfiltered measurement, especially for line array systems. Every time you move the mic you get a new phase response because you've changed the path length from the mic to each driver, and the arrival time of all the reflections.
The point here is that time alignment and phase are intrinsically related. A small time alignment difference can cause a drastic change in phase at higher frequencies.
Creating a FIR filter from a measurement further introduces variations because of any differences in the configuration of the system before and after filtering. If anything changes between measuring the unfiltered response and measuring the filtered response, there will be a glitch in the response of the system.
This also includes that the noise floor of the measurement must be the same at both times.
You are right it is not, but the plot shows more than 180degree of total phase shift, which is impossible with a second-order filter introduced by the parasitic RCL of a cable.
The point here is that time alignment and phase are intrinsically related. A small time alignment difference can cause a drastic change in phase at higher frequencies.
Creating a FIR filter from a measurement further introduces variations because of any differences in the configuration of the system before and after filtering. If anything changes between measuring the unfiltered response and measuring the filtered response, there will be a glitch in the response of the system.
This also includes that the noise floor of the measurement must be the same at both times.
You did read the point where I said the mic and speaker didn't move, yet the result was repeatable didn't you?
The FIR correction isn't needed to see this behavior. And yes, line arrays act differently due to time smear created by the differences in distance to the microphone.
I agree about timing differences at play here, but not due to a change in position of the speaker and the microphone in this case.
I did not touch the microphone and the speaker didn't move (it weighs close to 70 Kg I guess).
The FIR correction isn't needed to see this behavior. And yes, line arrays act differently due to time smear created by the differences in distance to the microphone.
I agree about timing differences at play here, but not due to a change in position of the speaker and the microphone in this case.
I did not touch the microphone and the speaker didn't move (it weighs close to 70 Kg I guess).
Last edited: