I don't think this is all that complicated. In fact I think it is easier than passive designs, and easier to do right without compromises (such as having to resort to asymmetrical slopes to compensate for delay differences).
I think it is also better to keep it simple when it comes to digital input included within the speaker: having spdif or AES/EBU inputs will let you plug a chromcast today, and maybe chromast 2 tomorrow, and whatever the day after. These things evolve too fast to be included within the very design of a loudspeaker that you want to be able to use 10 or 20 years from now.
Why not start with a IIR plate amp?
Thanks sharing details really helps to better understanding.
There is also this new thread if any interested http://www.diyaudio.com/forums/full...we-do-dsp-power-now-availabl.html#post4573181.
+1. Less hassle and better results.
Percussion is easiest, presumably because the most energy falls into the phase warp of the XO's transition band. For instruments with fundamentals and harmonics rather than something resembling an impulse it depends. In treble clef it's not too hard to hear the timbre's off compared to a real instrument, bass clef can get tricky for a ~2kHz XO; more a three way thing there.
It's my experience folks with limited live sound experience, especially unplugged, have a harder time with detection. Presumably because LR2 two ways are ubiquitous and the timbral shifts they introduce become the norm. It's also my experience the difference between warped and linear phase becomes less audible at LR4 and again at LR6, presumably due to decreased width of the transition band. (LR8+ is subjectively worse to me than LR4 or 6 in the configurations I've tested. I suspect mostly due to the more rapid change in acoustic center with frequency damaging the illusion of time coherence. Measured preringing increases too but it's difficult to implement proper controls to disambiguate between effects.)
Any particular song title you can share?
What about testing with only an all pass simulating the phase shift, using headphones for example?
That seems counter intuitive. Might be that in this case what you are hearing is not the phase shift per se, but summation issues from an non perfect acoustial crossover.
What about testing with only an all pass simulating the phase shift, using headphones for example?
In all systems I've measured the summation limit on typical two way crosses (and reasonably optimal TM XO in three ways) tends to be directivity matching between drivers. On axis is not difficult---particularly when working TM XO with mids that hold up to 10kHz---but first bounce is rather wigglier. Especially off the ceiling in most homes.
I've not ABXed allpass in cans but up round 2-3kHz my limit of discrimination measures out round 15 degrees. Significantly smaller than the 180 off an allpass or the 90 degrees of an LR2 at the XO point.
If you're trying to ask which tracks I use for testing it's usually whatever I've been listening to lately. Mostly jazz to metal. Not much for classical and taiko tends to gives me a headache after a while. Off the top of my head I've used stuff from Alvvays, Mark Knopfler (both solo and Dire Straits), Joe Satriani, Nanci Griffith, Swing Out Sister, Lacuna Coil, older Metallica, Tangerine Dream, Sibelius, Celine Dion, U2, Ladytron, some Spanish guitar guy whose name I can't remember, Mannheim Steamroller, various parts I've recorded myself, Bonrud, live music with the usual hodgepodge of instruments and vocals and mics and pedals (everything from solo celtic harp to eight piece bluegrass to didgeridoo; the violinist running a fuzz was fun), YouTube videos, and a gawdawful MP3 out of EZ Drummer I'd mercifully managed to forget until typing up this list. Plus quite a few which remain forgotten.
I've not ABXed allpass in cans but up round 2-3kHz my limit of discrimination measures out round 15 degrees. Significantly smaller than the 180 off an allpass or the 90 degrees of an LR2 at the XO point.
If you're trying to ask which tracks I use for testing it's usually whatever I've been listening to lately. Mostly jazz to metal. Not much for classical and taiko tends to gives me a headache after a while. Off the top of my head I've used stuff from Alvvays, Mark Knopfler (both solo and Dire Straits), Joe Satriani, Nanci Griffith, Swing Out Sister, Lacuna Coil, older Metallica, Tangerine Dream, Sibelius, Celine Dion, U2, Ladytron, some Spanish guitar guy whose name I can't remember, Mannheim Steamroller, various parts I've recorded myself, Bonrud, live music with the usual hodgepodge of instruments and vocals and mics and pedals (everything from solo celtic harp to eight piece bluegrass to didgeridoo; the violinist running a fuzz was fun), YouTube videos, and a gawdawful MP3 out of EZ Drummer I'd mercifully managed to forget until typing up this list. Plus quite a few which remain forgotten.
I’ve been listening rather critically to various things, and playing with different XO’s and Time Delays with an active crossover. I’ll share some of my favorite music, and what I notice (apologies in advance for the long post
). I’m not sure if what I notice may be a result of proper or improper frequency summation between the drivers, and/or proper or improper time alignment between the drivers. Once I began to fiddle with time delay between drivers to try to correct it, I feel like several things “fell into place”. Also, make note that I am listening to L+R Channels summed to Mono through One 3-way loudspeaker. I’m building the other one very soon.
As mentioned earlier, percussion seems to be the place where I really notice differences with the phase coherency (or time alignment,… whatever you want to call it).
Tsuyoshi Yamamoto, on an album called Autumn in Seattle has a song called “No Problem”.This entire album is in my opinion the best reference recording I have ever heard. This particular No Problem song has lots of Piano, sometimes using the sustain pedal, and other times quick little bursts, sometimes hitting the keys soft, and sometimes really banging a chord with authority. Also in the middle of the track, there’s a bit of a stand-up string bass solo, followed by a drum solo. This drum solo really shows the “air” in the soundstage when the timing is correct, I think I can “feel it”, even at fairly low volume levels.
My second favorite is Tony Overwater, Over the Rainbow, two songs “When I fall in love”, and “Pour un Elefant Terrible”. These two songs are an upright string bass, and again –this is a super good recording. As he plays up and down the keys, he hits several octaves, covering all three drivers. I can hear through the fine details of the strings slapping the neck as his left hand changes position.
There’s a James Taylor album One Man Band (live), the song “My Traveling Star” has James singing lead, and there’s a Chorus behind him. I notice better separation of his voice from the Chorus with in one XO setting vs. another. To me this is very interesting that I notice so much image and placement difference because I am listening monophonic.
I notice in Pink Martini “U Plavu Zoro” when the Violin in the beginning of the track seems to come not from any one driver, but seems to hover somewhere between the drivers.
I listen for the same thing with Dire Straits “Where do you think you’re going.”, the guitar in the beginning sounds very deep mid-bass to me (and we know there are harmonics that are much higher)
Thanks for the titles guys.
I found that listening through only allpass (simulating typical crossover phase shifts using rephase in time=inv mode and using the resulting FIR in foobar's convolver) with headphones these things are really subtle.
It is easy to detect very high order phase shifts, and it is a good method to put your finger (or ear) on the effect, but then slowly lowering the order it quickly becomes less and less audible to the point of not being detectable for low orders (for me).
If we are talking about phase coherency then yes this is audible, but this is not directly related to phase linearity.
I found that listening through only allpass (simulating typical crossover phase shifts using rephase in time=inv mode and using the resulting FIR in foobar's convolver) with headphones these things are really subtle.
It is easy to detect very high order phase shifts, and it is a good method to put your finger (or ear) on the effect, but then slowly lowering the order it quickly becomes less and less audible to the point of not being detectable for low orders (for me).
If we are talking about phase coherency then yes this is audible, but this is not directly related to phase linearity.
Hello,
I'm new to rephase, I read many, many articles all over the places on how to use it. My first question is not linked directly to rephase. How do you know or get the information on your speakers to enter values in filter linearization like "ventex low Q" and the frequencies ?
Thanks
David
I'm new to rephase, I read many, many articles all over the places on how to use it. My first question is not linked directly to rephase. How do you know or get the information on your speakers to enter values in filter linearization like "ventex low Q" and the frequencies ?
Thanks
David
If you cannot measure them, or get reliable measurements from the manufacturer or a review, or even reliable specs, then this is going to be difficult.
One can still guess based on previous models and/or driver models and/or simply driver diameter for the frequency and then bet for 24dB/oct acoustical slopes, but that is a bit like shooting in the dark.
Bass reflex tuning frequency can still be calculated based on port dimensions and enclosure internal volume, and any Q would do really...
What speaker is this?
One can still guess based on previous models and/or driver models and/or simply driver diameter for the frequency and then bet for 24dB/oct acoustical slopes, but that is a bit like shooting in the dark.
Bass reflex tuning frequency can still be calculated based on port dimensions and enclosure internal volume, and any Q would do really...
What speaker is this?
Yes, I used REW to measure them a 70cm distance, height of ears at listening position. Using the linearization filters and paragraphic EQ I managed to adjust a phase. I've not had time yet to listen to them.
One additional question. Once I get the right phase, I'd like to adjust amplitude at the average of various points around the listening point. How can I sum the phase correction at 70cm and the amplitude correction at listening point ?
Thanks
David
One additional question. Once I get the right phase, I'd like to adjust amplitude at the average of various points around the listening point. How can I sum the phase correction at 70cm and the amplitude correction at listening point ?
Thanks
David
Did you get the polarity and t=0 of the measurement right?
This is important to get the proper visualization of phase shifts and correct... correctly. They have been some advice in the previous posts that you should follow.
What phase corrections have you been using?
Regarding amplitude correction, you should only use minimum-phase ones, and as for the phase correction try to keep things simple. You don't want to correct measurement artifacts (position related, or windowing related...).
Phase and amplitude corrections done (properly) at different position can be "mixed" without problem.
Minimum-phase EQ should also help solving some phase variation, and eliminate the need for some phase EQs.
Not not hesitate to share you corrections and/or measurements.
If you end up with complementary acoustical slopes (same crossover type for low and high pass) then it is only a matter of getting the levels and delays rights (as well as polarity if you inverted it in some measurements!) and you should get a flat summation.
This is easy to check, for example using the reversed polarity null method.
Having two IR-wav files imported into REW there is some tools available in there, it can make the delays by offsetting IR and offsetting spl level is also easy, then on "All SPL" tab there is average response button and other tools if choosing "Controls" when view is in "All SPL"-window buttons as make A+B plus more. But to blend more than two imported IR is over my head.
XSim seems a possibility connecting its build in textbook perfect wideband driver direct parallel to power amp and get real world measured drivers response blended with correction filter linked to driver in XSim. XSim can't read wav-files but REW can read and export to frd-file that then is linked to driver in XSim.
Control settings in Power amp in XSim is watt/impedance/delay/polarity and control settings for each driver is polarity/mute/spl/delay, and add to those settings that one can pull each driver in and out of network on the fly, and if that is not enough control then open XSim as many times as computer can pull with same template circuit and make different settings within each template so as to compare results when hovering mouse over MS taskbar.
Below is a 4-way speaker example probably close to joip targets, first is pure IRR setup and second is IRR system pass band with FIR XO slopes seen on steps and square wave plot. Drivers pass band for examples is textbook slopes made in Rephase imported to REW and exported as frd-file for use in XSim. For IRR example had to offset IR to zero for them to sum right but could have slided delay up or down for each driver to get to same result.
So think above request can be summed and steered with controls for each driver for polarity/mute/spl/delay inside Xsim. To create one of four band pass for XSim then open that drivers measurement in REW plus import its Rephase correction and then in "All SPL" tab under "Controls" chose "A:" to be driver measurement then "B:" to be Rephase correction and set command "A * B" and hit "Generate", export this new "A times B" as frd-file to link in XSim.
Attachments
Thanks POS/BYRTT.
BYRTT, much appreciate all the details you provide. I will go thru them. Unfortunately day job is quite demanding at this time, so unable to do a lot of justice to the excellent inputs/help provided on this thread. But soon will post something for review.
You both had suggested amplifier as the preferred point for driver level matching. I am using Hypex modules based power amplifier so cant do level adjustments there. ESS9018 DAC I am using doesn’t have analog volume control on the 8CH of analog output. However the chip does support good digital volume control on the channels (Currently not supported in the firmware but can be added fairly easily)
So at this point, level adjustments to match the drivers will be done using
1. Volume control plugin added to each channel in Jriver
2. In future will try to achieve the same using per channel digital level control on the ESS DAC.
BYRTT, much appreciate all the details you provide. I will go thru them. Unfortunately day job is quite demanding at this time, so unable to do a lot of justice to the excellent inputs/help provided on this thread. But soon will post something for review.
You both had suggested amplifier as the preferred point for driver level matching. I am using Hypex modules based power amplifier so cant do level adjustments there. ESS9018 DAC I am using doesn’t have analog volume control on the 8CH of analog output. However the chip does support good digital volume control on the channels (Currently not supported in the firmware but can be added fairly easily)
So at this point, level adjustments to match the drivers will be done using
1. Volume control plugin added to each channel in Jriver
2. In future will try to achieve the same using per channel digital level control on the ESS DAC.