Build therad: OLA-WG 2.5 - a 2.5way with SB midwoofers and a Bliesma tweeter

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • ergo
    Senior Member
    • Mar 2005
    • 675

    #46
    I'll generally do the xover desing of 2way or 2.5way in this case so that the microphone stays at 1m on tweeter axis for all measurements. So the distance/phase difference of the mid is already accounted for and I do not need to change the coordinates.
    I did not have time today, but I might try to create calculated minimum phase copies of the measurements and then try with adding the coordinates too. I always do also a measurement with drivers in parallel, so I would have a reference to adjust to for the Z coordinate.

    Impedance is a bit more challenging than I'd like touching the 3ohm in lower mids . Maybe tunable still... My friend is leaning towards going to class-D amps as well (Hypes or ICEpower) from NAKSA. Hes been listening to mine and likes both and for class-D amps the 3 ohm would be fine.

    Click image for larger version

Name:	impedance_v1.jpg
Views:	1
Size:	28.0 KB
ID:	864323

    Comment

    • draki
      Member
      • Oct 2012
      • 37

      #47
      Thanks for the explanation.

      Comment

      • kimmosto
        Moderator
        • Dec 2006
        • 584

        #48
        Originally posted by ergo
        I'll generally do the xover desing of 2way or 2.5way in this case so that the microphone stays at 1m on tweeter axis for all measurements. So the distance/phase difference of the mid is already accounted for and I do not need to change the coordinates.
        I did not have time today, but I might try to create calculated minimum phase copies of the measurements and then try with adding the coordinates too.
        I suppose you're aware that simulated power response and DI are not correct if position of all drivers is X,Y,Z=0,0,0? Typical error is about 2 dB at crossover frequency with acoustical L-R i.e. phase matching.
        Minimum phase extraction produces wrong result to high off-axis angles due to missing path length around cone/horn/wave guide and box edges, and fully or partly dipole drivers due to wrong polarity to rear sector. So MP extraction is fundamentally bad method which is banned for good.

        Proper method to convert measurement data captured at constant mic position and the same time window and reference time settings so that you could enter actual driver locations to simulation is to subtract delay difference to axis origin (on the baffle surface) from all measurements of each driver. For example if measurement distance is 1000 mm and mic elevation is constant center of tweeter and mid-range driver is located 150 mm below tweeter (all drivers in straight/non-stepped baffle), Delay [us] parameter for mid-range driver is negative (1000-sqrt(1000^2+150^2))/344*1000 = -32.5 us. Own delay values for woofers of 2.5-way, which should have individual measurement data too. After this correction you can enter actual coordinates to all driver instances in XO. Verify that phase of axial response stays quite close to unchanged compared to original situation with X,Y,Z=0,0,0 when listening distance is set temporarily to 1000 mm in Options window. If okay, restore 2500...3000 mm to listening distance and check that vertical off-axis works as it should be giving more correct power & DI readings.
        VituixCAD, Features, User manual, Measurements with CLIO, ARTA, REW, SoundEasy, Download

        Comment

        • ergo
          Senior Member
          • Mar 2005
          • 675

          #49
          Thanks Kimmo! Reading the logic it makes total sense indeed and easy to implement. I'll try that out tomorrow and post a side to side comparison for learning reference too.

          Comment

          • draki
            Member
            • Oct 2012
            • 37

            #50
            Originally posted by kimmosto

            Proper method to convert measurement data captured at constant mic position and the same time window and reference time settings so that you could enter actual driver locations to simulation is to subtract delay difference to axis origin (on the baffle surface) from all measurements of each driver. For example if measurement distance is 1000 mm and mic elevation is constant center of tweeter and mid-range driver is located 150 mm below tweeter (all drivers in straight/non-stepped baffle), Delay [us] parameter for mid-range driver is negative (1000-sqrt(1000^2+150^2))/344*1000 = -32.5 us. Own delay values for woofers of 2.5-way, which should have individual measurement data too. After this correction you can enter actual coordinates to all driver instances in XO. Verify that phase of axial response stays quite close to unchanged compared to original situation with X,Y,Z=0,0,0 when listening distance is set temporarily to 1000 mm in Options window. If okay, restore 2500...3000 mm to listening distance and check that vertical off-axis works as it should be giving more correct power & DI readings.
            Is this a substitution to a multiple (i.e. each individual driver's vertical axis) measurement approach? Or just "good enough"?

            Comment

            • 5th element
              Supreme Being Moderator
              • Sep 2009
              • 1671

              #51
              Off topic aye but the first rise in distortion, below 300Hz, is due to the additional boost that the mid range sees to compensate for the cardioid losses via the sides of the box. This quickly drops below 100Hz as it's crossed to the subs on the back.

              The rise in distortion that the subs appear to show is quite alarming and I'm surprised at it. Two 8" drivers should not be showing distortion that high at 50Hz re 96dB 1 meter, even in a small cabinet with EQ. The drivers should be well within xmax and well under 10% distortion but here we are at 30% distortion... something doesn't add up. Maybe they need to switch to Dayton's ultimax 8" drivers instead.

              It's possible that this is entirely 2nd order distortion from suspension nonlinearities. It's one thing to have a motor that's got plenty of linear travel but if your soft parts lose their composure much sooner the 2nd order shoots up prematurely. Flipping one of the sub drivers so that it's magnet out would fix a lot of this. The subs suspensions have enough travel it's just the inward motion is quite different to the outward.
              What you screamin' for, every five minutes there's a bomb or something. I'm leavin' Bzzzzzzz!
              5th Element, otherwise known as Matt.
              Now with website. www.5een.co.uk Still under construction.

              Comment

              • kimmosto
                Moderator
                • Dec 2006
                • 584

                #52
                Originally posted by draki
                Is this a substitution to a multiple (i.e. each individual driver's vertical axis) measurement approach? Or just "good enough"?
                Measurement instructions says that compact T+M pair - for example drivers of small 2-way can be measured at average elevation of tweeter and mid-woofer without moving mic or speaker vertically. That is more valid for small domes than long ribbons/planars of course. Measuring both at tweeter's elevation with delay compensation can be substitution for small 2-way (especially with ribbon/planar HF) if we know that mid-woofer's cone break-up resonance does not exist or will be properly captured at tweeter's elevation or will be measured separately for possible notch filter design.

                Small 2.5-way is equal to small 2-way by measuring far field of upper mid-woofer alone and loading the same data also for lower mid-woofer. That is natural with VituixCAD because both woofer instances in XO can and should use common measurement data in Drivers tab.

                But generally, measuring all drivers at the same elevation is bad thing and not recommended especially with big constructions because:
                - If woofer is closer to floor (or ceiling) than tweeter, time window for woofer should be shorter to avoid reflections. More LF and diffraction data will flow out with too short time window corrupting FR result. Woofers prefer longer time window than tweeters due to lower frequencies. At least speaker should be lifted so that delay of earliest reflections is longest with woofer or floor reflections properly damped.
                - Woofer rotates on circular path if baffle is stepped and woofer is installed on step closer to mic and not at Z of rotation center. That increases data loss (overflow from time window) to rear sector because distance to woofer increases more than without stepped baffle.
                - If woofer is much lower than tweeter, responses to small angles around driver's main axis will not be measured at all while off-axis sequence. There will be some variable offset in angle values from actual to file naming. For example when simulator selects 0 deg response, it will get actual 20 deg response. Error is highest at 0 and 180 deg. 90 deg is correct. Plane is not exactly hor or ver either, but that could be minor advantage if responses are measured horizotal alone and mirrored to vertical.
                - User needs to calculate required delay compensation for measurement data with previous formula in order to locate drivers to actual mechanical positions in the simulation.
                - My selfish point of view: If someone publicly violates recommendations without written warning, there could be few more e-mails in my inbox or messages on discussion forums. I have to explain what is wrong with that method, and when and how it can be used without significant negative consequences in simulation result. That could take few hours of my time while measuring person could save less than half an hour his time. This happens here and now, because I responded.

                So never too late to forget and leave few bad habits you've learned from measurements with constant mic position for simulation of axial response only "How to" easily measure and adjust e.g. 1000 mm distance from mic to rotation center i.e. center of driver cone on baffle surface or horn mouth was explained on diyaudio few days ago. There should not be valid reason to keep both mic and speaker at constant elevation while off-axis sequences, though lifting of heavy speaker on turning table is a problem without an assistant or a jack. I have used several large pillows for damping floor reflections while measuring woofer if weight of speaker has exceeded my "Force", but mic has been at woofer's elevation to capture cone resonances properly.
                Last edited by kimmosto; 04 June 2019, 02:03 Tuesday.
                VituixCAD, Features, User manual, Measurements with CLIO, ARTA, REW, SoundEasy, Download

                Comment

                • kimmosto
                  Moderator
                  • Dec 2006
                  • 584

                  #53
                  Originally posted by kimmosto
                  - User needs to calculate required delay compensation for measurement data with previous formula in order to locate drivers to actual mechanical positions in the simulation.
                  Best place to add delay correction for M and W is Reference time in Convert IR to FR. That would eliminate extra data loss because time window is shifted forward before FFT.
                  VituixCAD, Features, User manual, Measurements with CLIO, ARTA, REW, SoundEasy, Download

                  Comment

                  • ergo
                    Senior Member
                    • Mar 2005
                    • 675

                    #54
                    Thanks Kimmo, I have a terrace roof to wash with a power washer but after that this evening should allow few hours of time to take a new stab at it. I'll try both - using the existing data and adding the coordinates as well as gathering new set of data and trying to follow your best practice. Hopefully will teach me (and perhaps other too) on the difference.

                    Comment

                    • cochinada
                      Senior Member
                      • May 2014
                      • 658

                      #55
                      That will be a most interesting exercise. :T
                      Looking forward to check your results.
                      Joaquim

                      DIY 4 way speakers.
                      DIY subwoofers.
                      Zaph ZD3C.

                      Comment

                      • ergo
                        Senior Member
                        • Mar 2005
                        • 675

                        #56
                        First off - the new analysis of prior data

                        * Mic always at tweeter height for all drive units
                        * The layout measurement used as guideline for adding reference delay to the measured IR
                        Click image for larger version

Name:	triangles.jpg
Views:	1
Size:	26.1 KB
ID:	864330

                        16mm added for upper midwoofer - orange curve is the reference curve of tweeter and upper midwoofer measured in parallel. There is a good match
                        Click image for larger version

Name:	Correct_delay_MidWupper.jpg
Views:	1
Size:	125.5 KB
ID:	864331


                        62mm added to lower midwoofer- orange curve is the reference curve of tweeter , upper midwoofer and lower measured in parallel. There is a again a good matchmidwoofer
                        Click image for larger version

Name:	Correct_delay_MidWupper+lower.jpg
Views:	1
Size:	124.0 KB
ID:	864332

                        This is the prior simulation where all drivers had 0,0,0 coordinates
                        Click image for larger version

Name:	XoverV3_no_offset.jpg
Views:	1
Size:	121.6 KB
ID:	864333

                        And this is the result where Xover values have no been touched, just the FRD data is re-imported with the above extra reference time added and the driver co-ordinates correct in Y axis...

                        There is a clear difference. I need to get the emulation working in LspCAD to verify if this new attempt is closer than the older one.
                        Click image for larger version

Name:	XoverV3_pre-delay+driver-offset.jpg
Views:	1
Size:	126.0 KB
ID:	864334

                        Comment

                        • ergo
                          Senior Member
                          • Mar 2005
                          • 675

                          #57
                          With a bit of tweaking of the Xover values the result can look good again
                          Click image for larger version

Name:	XoverV3_pre-delay+driver-offset_tweaked.jpg
Views:	1
Size:	117.6 KB
ID:	864335

                          Comment

                          • ergo
                            Senior Member
                            • Mar 2005
                            • 675

                            #58
                            As a next exercise I did redo all of the measurements at 1m distance

                            * Mic at tweeter height for
                            tweeter on and off axis
                            tweeter and upper midwoofer in parallel
                            tweeter + uppar and lower midwoofer in parallel

                            * Upper midwoofer raised by 180mm for on and off axis measurements

                            * Mic lowered to be on axis with lower woofer for on and off axis measurements

                            Tried to follow Kimmos best practice guidelines and also used a pillow this time to cover the unused midwoofer as well as shorting the leads on the unused woofer

                            Click image for larger version

Name:	pillow.jpg
Views:	1
Size:	53.4 KB
ID:	864336

                            The similar procedure of adding a reference time offset goes ok, but I am struggling a bit to get similarly good match with the reference measurement made with the drive units playing in parallel... It's getting late, so seems that I have to leave that part of the fun for a later date.

                            Comment

                            • JonMarsh
                              Mad Max Moderator
                              • Aug 2000
                              • 15261

                              #59
                              You're doing this "by the numbers", which can certainly be time consuming and a lot of work... sometimes one gets to a certain point and you just have to call it a day and come back fresh to the situation and problems... BTDT many times.


                              With a situation as you're working with, sometimes I've literally made three baffles and measured everything with just one driver in position on the baffle... that ignores the possibility of acoustic interaction with diffraction off the adjacent driver, but one does get good driver data that way. PITA, as we say over here...

                              Is your listening distance relatively near field, (as your diagram at the top of #56 suggests?). Is this representation how you will be listening?

                              I think I do recall you mentioned earlier relatively near field use. What about boundaries? (walls).

                              Something I've been doing of late that is possibly questionable is how I locate test cabinets- basically, at golden means from the floor and wall boundaries, and the minimum boundary distance is over 1.3 meters. And very little smoothing. Setup for good measurements, and measure far field enough (as well as near field to check what the driver is doing) that summing is fairly straight forward.


                              Here is an in room polar measurement farfield of the Satori WO24P... sometimes it's a PITA to find a position that really gets you what you want, but in my condo in San Ramon (since February) I'm doing fairly decent... OTOH, it does have 9 ft ceilings, too.

                              Click image for larger version

Name:	WOP24P Polar.jpg
Views:	1
Size:	64.0 KB
ID:	864337

                              Something like this won't help you fix positioning issues with the speakers how they're finally placed, but at least the crossovers should be generating a total power response into the room that's in alignment with your goals...


                              The way you are doing it is somewhat like what Michael Bell has been trying to do with his new project (which I call the Osiris) and he even brought someone in to do measurements, but they setup badly in the room, ignored boundary issues, and the data he sent me just really isn't usable. I'm having to redo most of it here, and have new test cabinets for midwoofer and midrange in process for that...

                              Geez I wish I was retired!

                              BTW, VERY nice looking cabinets and waveguides!
                              the AudioWorx
                              Natalie P
                              M8ta
                              Modula Neo DCC
                              Modula MT XE
                              Modula Xtreme
                              Isiris
                              Wavecor Ardent

                              SMJ
                              Minerva Monitor
                              Calliope
                              Ardent D

                              In Development...
                              Isiris Mk II updates- in final test stage!
                              Obi-Wan
                              Saint-Saëns Symphonique/AKA SMJ-40
                              Modula PWB
                              Calliope CC Supreme
                              Natalie P Ultra
                              Natalie P Supreme
                              Janus BP1 Sub


                              Resistance is not futile, it is Volts divided by Amperes...
                              Just ask Mr. Ohm....

                              Comment

                              • ergo
                                Senior Member
                                • Mar 2005
                                • 675

                                #60
                                I did a my own project in VituixCAD 1 timeframe last time and then there was the Wavecor Ardent build in between. Now with VituixCAD 2 I've been dabbling with it but decided to use this opportunity to learn it better and follow Kimmo's process (which I see I failed at first go ... anyhow learning and trying again....

                                With the new turntable, even though it's manual, it does allow collecting the 3X 0..180 deg data pretty fast. So that's not really a problem. Just have to figure out the rest of the process and get better at that too.

                                The listening distance will the the usual 2.5...2.7m or thereabouts. My friend has a low long shelf for TV that will also have the speakers on it. The back wall is another 1.8m or so behind the speakers. So room mode wise these will be close to free standing and the only strong first reflection will be the floor/ceiling one. I have pictures of his room, but I'll ask him first if it's ok to post those.

                                Above diagram is just to help get the distances right for the measurement case and I'm trying to match up the simulation with the true "all parallel" measurement at 1m.

                                Comment

                                • kimmosto
                                  Moderator
                                  • Dec 2006
                                  • 584

                                  #61
                                  Originally posted by ergo
                                  also used a pillow this time to cover the unused midwoofer as well as shorting the leads on the unused woofer
                                  Pillow trick is for near field only. You (should) have both woofers connected in parallel to amp. to get correct LF response with possible HH-resonator, and mic is measuring single cone; lower or upper. Pillow should block mid-range sound from the other driver so that mid-range is mostly measured from driver close to mic. Pillow should not brake movement of cone to get proper LF result. This assuming that woofers have shared box volume.

                                  Main frequency range of far field measurement is above 300 Hz so there should not be extra reflecting objects close to DUT. Soft pillow is not bad but not mandatory either. Blind flange in place of lower woofer while far field measurement is one option to prevent passive radiator effects and get some - hopefully decent directivity information below merging frequency.
                                  VituixCAD, Features, User manual, Measurements with CLIO, ARTA, REW, SoundEasy, Download

                                  Comment

                                  • kimmosto
                                    Moderator
                                    • Dec 2006
                                    • 584

                                    #62
                                    Originally posted by ergo
                                    but I am struggling a bit to get similarly good match with the reference measurement made with the drive units playing in parallel.
                                    Comparing to parallel measurement will not usually give exact match. Measurements in parallel and sum are taken to exactly equal angles, but not necessarily when measuring at horizontal plane only e.g. with 10 deg angle step. VituixCAD does not interpolate responses within measured angles. It just selects the closest measured angle, so there will be some error due to angle stepping and probable difference.
                                    In your case, exit angle from upper woofer to mic at 1 m is atan(180/1000)=10.2 deg ver. You have measurement to hor 10 deg which is selected and mirrored to ver by the program. That small difference is okay for sure. Exit angle of lower woofer is atan(360/1000)=19.8 deg ver. You have measurement to hor 20 deg which is also really okay. But some other cases angle difference could be half of angle step e.g. 5 deg, so difference to parallel measurement will be self evident at frequency range where radiating surface of one (or both) DUT is very directive. But that range is usually irrelevant in practice because XO usually has (~should have) low-pass filter where radiator is so directive that response changes radically within 5 deg off-axis difference to angle <30deg. This is the reason why I would never compare to parallel measurement as long as measurement gear capable to capture sound flying time is available. I've had either dual channel gear or CLIO since 2k so parallel verification is something I have not done for speaker design. I've just tested difference between acoustic centers with some coaxial and noticed that method is totally useless without adequate response overlapping.
                                    VituixCAD, Features, User manual, Measurements with CLIO, ARTA, REW, SoundEasy, Download

                                    Comment

                                    • ergo
                                      Senior Member
                                      • Mar 2005
                                      • 675

                                      #63
                                      I figured the same that the most probable cause of this difference is the lack of actual vertical offaxis data for the angles needed. In theory if I would measure 0...180deg in 10deg steps for horizontal plane and then only 0...30deg in 5 or 10deg steps in vertical plane, then would the VituixCAD take those few vertical data points into account and use the horizontal data for the rest. Or will it need a full set for both hor and ver if done so?

                                      That comparison to summed response is one of the 'old habits' I used with LspCAD based process. I'm pretty sure the phase data should be ok as I too use the CLIO FW for this. Anyhow as mentioned this is a learning exercise too, so thats why I'm exploring these things perhaps 'too deep' at some points.

                                      I will try to get the Xover emulation going and in past I have successfully passed measurement signal through the emulation. Thats the best way in this case to get a true answer to how the two woofers sum and whats the LF balance etc.

                                      Comment

                                      • kimmosto
                                        Moderator
                                        • Dec 2006
                                        • 584

                                        #64
                                        Originally posted by ergo
                                        In theory if I would measure 0...180deg in 10deg steps for horizontal plane and then only 0...30deg in 5 or 10deg steps in vertical plane, then would the VituixCAD take those few vertical data points into account and use the horizontal data for the rest.
                                        If you measure 0-180 deg with 10 deg steps in horizontal and 0-30 deg with 5 deg steps in vertical:
                                        - vertical 5, 15 and 25 deg are mirrored to horizontal plane
                                        - horizontal 40-180 deg is mirrored to vertical plane.

                                        'Mirror missing' could be fatal setting with rectangular or otherwise very unsymmetrical radiators. Therefore it's recommended in 'Measurement prep' document and 'How to start' in www that the same angles should be measured in vertical and horizontal planes with unsymmetrical drivers.
                                        You could measure also 5, 15 and 25 deg in horizontal plane to avoid mirroring from vertical.
                                        All drivers in your project are circular so measurement in vertical plane is not mandatory. Full mirroring from horizontal to vertical causes very small (~0.5 dB) error to power response at baffle step hump frequency with slim "decoration speakers" Baffle step hump is a bit pronounced in vertical plane if baffle width << height. So you could measure 0,5,10,15,20,25,30,40,...,180 in horizontal and that's it.
                                        VituixCAD, Features, User manual, Measurements with CLIO, ARTA, REW, SoundEasy, Download

                                        Comment

                                        • cochinada
                                          Senior Member
                                          • May 2014
                                          • 658

                                          #65
                                          Why did you measure tweeter in parallel with upper woofer and then all 3 together? I'm completely lost...
                                          Joaquim

                                          DIY 4 way speakers.
                                          DIY subwoofers.
                                          Zaph ZD3C.

                                          Comment

                                          • 5th element
                                            Supreme Being Moderator
                                            • Sep 2009
                                            • 1671

                                            #66
                                            Originally posted by cochinada
                                            Why did you measure tweeter in parallel with upper woofer and then all 3 together? I'm completely lost...
                                            You do this to create a reference for how all the drivers sum together with no crossover.

                                            This sum is then compared to how the simulation program predicts they are going to sum. If the predicted sum matches reality then you know your software is doing the right thing.

                                            You usually alter the simulation parameters, such as driver offset in the Z axis, until the simulation matches reality.

                                            Matching three drivers at a time is complicated, two less so much. So here ergo matched the tweeter and the upper woofer first, then added in the third driver later.
                                            What you screamin' for, every five minutes there's a bomb or something. I'm leavin' Bzzzzzzz!
                                            5th Element, otherwise known as Matt.
                                            Now with website. www.5een.co.uk Still under construction.

                                            Comment

                                            • JonMarsh
                                              Mad Max Moderator
                                              • Aug 2000
                                              • 15261

                                              #67
                                              Exactly.

                                              This is the only reliable way I’ve found to be sure you’ve got the time offset values for the (minimum phase) driver data corrected for acoustic origin in your design tool. One of the most critical initial things to do....
                                              the AudioWorx
                                              Natalie P
                                              M8ta
                                              Modula Neo DCC
                                              Modula MT XE
                                              Modula Xtreme
                                              Isiris
                                              Wavecor Ardent

                                              SMJ
                                              Minerva Monitor
                                              Calliope
                                              Ardent D

                                              In Development...
                                              Isiris Mk II updates- in final test stage!
                                              Obi-Wan
                                              Saint-Saëns Symphonique/AKA SMJ-40
                                              Modula PWB
                                              Calliope CC Supreme
                                              Natalie P Ultra
                                              Natalie P Supreme
                                              Janus BP1 Sub


                                              Resistance is not futile, it is Volts divided by Amperes...
                                              Just ask Mr. Ohm....

                                              Comment

                                              • kimmosto
                                                Moderator
                                                • Dec 2006
                                                • 584

                                                #68
                                                Originally posted by 5th element
                                                You usually alter the simulation parameters, such as driver offset in the Z axis, until the simulation matches reality.
                                                If or when we're talking about simulation programs such as LspCAD or VituixCAD, driver's origin X,Y,Z is logically "timing origin of measurement data" - usually mechanical center point of speaker/mic rotation while off-axis measurement sequence.
                                                Z coordinate can be used for difference of acoustic centers if driver's radiating surface is visible for microphone while off-axis measurements or delay changes to different angles are equal to all drivers in construction. Simulators (at least LspCAD and VituixCAD) do not calculate actual path length of sound (or delay difference to minimum phase response) from acoustic center/throat around possible horn/cone surface and around box edges to microphone located on the back side of speaker. So possibilities to use Z value for acoustic centers and minimum phase response data are quite limited.

                                                ergo measures and simulates to 0-180 deg and acoustic center of tweeter is hidden from mic >~60 deg and not necessarily at the same level with woofer, so he should measure actual sound propagation delay (sound flying time + possible dsp time). Z should be at rotation center while off-axis sequence (not Z of acoustic center!) and minimum phase extraction is not allowed to capture variable differences from acoustic centers to mic. ergo has CLIO 12 fw-02 which measures propagation delay by default so minimum phase extraction and parallel measurement are not needed.

                                                concinada had dual channel measurement gear, connection and mode in ARTA while measuring his 4-way project so parallel measurement and minimum phase extraction was not needed or used. Exactly how author of simulator has recommended since day one of VituixCAD.

                                                There are easier and faster methods than parallel measurement to verify is measurement system capable and stable for measuring propagation delay, and how simulator sums two responses with different excess phase/GD. But parallel measurement with two individual measurements can be used for sanity check; to verify are some basic things in measurements and parameters okay at least in single spot in 3D space.
                                                VituixCAD, Features, User manual, Measurements with CLIO, ARTA, REW, SoundEasy, Download

                                                Comment

                                                • ergo
                                                  Senior Member
                                                  • Mar 2005
                                                  • 675

                                                  #69
                                                  Click image for larger version

Name:	VituixCAD_vs_LspCAD_vs_Real.jpg
Views:	1520
Size:	91.8 KB
ID:	864340 Lets see if this works.

                                                  What I've done is to take the first version of measurements where Mic was always at tweeter height. Then the respective delays were added in VituixCAD IR -> FRD tool and project created (same as the above sample)

                                                  Next I replicated same Xover with same 0deg measurement data in LspCAD and asked it to emulate the crossover.

                                                  The CLIO SPDIF out is then routed to RME UCX soundcard -> LspCAD Xover Emulation --> RME 6 channel analog out --> 6 channel Hypec amp --> Each drive unit in separate channel (only one speaker connected)

                                                  And then the measurements run in normal fashion.... A pretty good correlation of measured versus simulated.

                                                  I'll try the same exercise now for the version where I measured each driver on it's own axis.

                                                  PS. Note that VituixCAD range is 40dB, while other two are 50dB. VituixCAD has only 40 or 60 option but then CLIO does not use 5dB per division and it makes graph harder to read
                                                  Last edited by theSven; 07 April 2023, 16:16 Friday. Reason: Update image location

                                                  Comment

                                                  • ergo
                                                    Senior Member
                                                    • Mar 2005
                                                    • 675

                                                    #70
                                                    Here a tweeter reversed graph is added and a long gated version showing the low end with room modes.

                                                    Click image for larger version

Name:	VituixCAD_vs_LspCAD_vs_Real+LF+reverse.jpg
Views:	1
Size:	94.7 KB
ID:	864341

                                                    Comment

                                                    • draki
                                                      Member
                                                      • Oct 2012
                                                      • 37

                                                      #71
                                                      Originally posted by ergo
                                                      VituixCAD has only 40 or 60 option
                                                      The two sets of expanding/collapsing arrows right next to "reference angle / deg hor" at the top will allow a 5 dB increase/decrease of the graph dB range...just saying.

                                                      Comment

                                                      • ergo
                                                        Senior Member
                                                        • Mar 2005
                                                        • 675

                                                        #72
                                                        This is the same exercise but the measurement result files are for each drive unit measured at it's own axis. For purely on axis response tuning both methods give a close match with simulation. But as Kimmo explains the difference could come out in offaxis / power response and directivity.
                                                        I'm afraid the time is running out for today, but having the system working now for emulation I'll attempt to do a comparison on 0..180 deg in 10deg steps and then calculate the power response for each method.

                                                        Click image for larger version

Name:	y4mLabx8bDPG68Rb4t7ufhg6GJq4ArZxYnys0JPY33R3VGgE-cL98enlxhLNIuKqfggKvGYH5mo_oy0AcjjWe-Jkioqd8ig639_MkB0zVIg86uzl_goghkx3YWE6NilKV4I8zoDaMalY-Yo3yWMB0BXRMi63mdwLEyicY77pq0KC0_rp95DJDolgf_ixpmndTYFRfz95ZtwyyWmYpWJ112gHQ?width=2478&amp;height=1431&amp;cropmo
Views:	213
Size:	268.4 KB
ID:	932750
                                                        Last edited by theSven; 07 April 2023, 16:16 Friday. Reason: Update image location

                                                        Comment

                                                        • ergo
                                                          Senior Member
                                                          • Mar 2005
                                                          • 675

                                                          #73
                                                          Originally posted by draki
                                                          The two sets of expanding/collapsing arrows right next to "reference angle / deg hor" at the top will allow a 5 dB increase/decrease of the graph dB range...just saying.
                                                          Aargh - thanks! Indeed they do. I failed to try those and looked under options only.... but smarter now again

                                                          Comment

                                                          • ergo
                                                            Senior Member
                                                            • Mar 2005
                                                            • 675

                                                            #74
                                                            After all this theory I decided to listen to some music also... only one mono speaker mid room for now.

                                                            But there is potential here for sure. I do hear the transparency of the Beryllium I know from Ardents. Mids sound nice and revealing too and the bass is not shabby. A true subwoofer support will be a good addition I'm sure but these compare well to Modula MT mkII 2ways and probably beat those if I would compare.

                                                            Lets see when I get time to fine tune the Xover a bit more and we'll order the parts. There is quite an urge to do that fast though.

                                                            Comment

                                                            • draki
                                                              Member
                                                              • Oct 2012
                                                              • 37

                                                              #75
                                                              Ergo, whenever you have time: could you please post the tweeter response without the two EQ notches. Curious to see where and what are you EQ-ing.
                                                              Have a combination tweeter (not Bliesma!) with Visaton WG on a similar width baffle, and the response behaves (wavy peaks and dips) so I have to use two EQ notches (three even...). The rough original prototype baffle was almost 35 cm wide and the FR was very smooth, no dips.peaks, just the expected low-range boost (but looked ugly - the box I mean - so had to redesign). Couldn't believe the difference when the new family council approved slim-front enclosures were built (and finished, painted). Lesson learnt.

                                                              Comment

                                                              • ergo
                                                                Senior Member
                                                                • Mar 2005
                                                                • 675

                                                                #76
                                                                With notches
                                                                Click image for larger version

Name:	TW_with_notches.jpg
Views:	1
Size:	101.4 KB
ID:	864342

                                                                No notches
                                                                Click image for larger version

Name:	TW_no_notches.jpg
Views:	1
Size:	102.0 KB
ID:	864343

                                                                Comment

                                                                • draki
                                                                  Member
                                                                  • Oct 2012
                                                                  • 37

                                                                  #77
                                                                  Sorry if this is diverging OT, I will stop with this.

                                                                  Here is the same tweeter+WG on a wide (35 cm) and narrow (23 cm) baffle.
                                                                  Attached Files

                                                                  Comment

                                                                  • ergo
                                                                    Senior Member
                                                                    • Mar 2005
                                                                    • 675

                                                                    #78
                                                                    draki - are you sure you did not have some other measurement issue causing a bad test result?
                                                                    I assume the WG was equal distance from top edge in both tests and the WG was mounted in center horizontally in both cases?

                                                                    It's not quite correct to analyze the WG on baffle with the VituixCAD but it should give general idea what type of difference is to be expected. And the difference does not look nowhere as extreme as you see in your measurement. So that is why I'm asking how sure you are about the source data.
                                                                    Click image for larger version

Name:	WG_baffle.jpg
Views:	1
Size:	64.7 KB
ID:	864345

                                                                    Comment

                                                                    • kimmosto
                                                                      Moderator
                                                                      • Dec 2006
                                                                      • 584

                                                                      #79
                                                                      It's certain that baffle cannot cause that much difference. Should be quite close to equal. Something is wrong with measurement or driver itself or mechanical connection to WG.
                                                                      VituixCAD, Features, User manual, Measurements with CLIO, ARTA, REW, SoundEasy, Download

                                                                      Comment

                                                                      • Zvu
                                                                        Senior Member
                                                                        • Oct 2013
                                                                        • 434

                                                                        #80
                                                                        Dip at 6,5kHz and at 13kHz. Something happens at much shorter wavelength than baffle width.
                                                                        Tesla; George Carlin;

                                                                        Comment

                                                                        • draki
                                                                          Member
                                                                          • Oct 2012
                                                                          • 37

                                                                          #81
                                                                          I will dig out the original protobox and will repeat all measurements.
                                                                          As I said, I was surprised as well since both boxes were measured under same conditions (not at the same time though), same mic, sound card, with JustMLS (and later the same results with ARTA). Drivers OK.
                                                                          But really left it aside at the time and left it for to-do list.
                                                                          Now that you reinforced the suspicion, I am more determined to find out what is going on.
                                                                          Will report back - could be weeks though...

                                                                          Comment

                                                                          • JonMarsh
                                                                            Mad Max Moderator
                                                                            • Aug 2000
                                                                            • 15261

                                                                            #82
                                                                            Originally posted by draki
                                                                            Sorry if this is diverging OT, I will stop with this.

                                                                            Here is the same tweeter+WG on a wide (35 cm) and narrow (23 cm) baffle.

                                                                            I would be reluctant to trust this without a thorough controlled evaluation. Particularly, comparative data makes sense if it's the same exact waveguide and tweeter measured on each baffle, and the driver/waveguide combo is verified not to have any other issues, say, making impedance tests to confirm integrity.

                                                                            I had issues with my first set of BlieSMa tweeters, they had to be returned.

                                                                            OTOH, I've measured a variety of tweeters on Visaton waveguides, and mostly on somewhat narrow baffles (8-12" wide, but with full 3/4" roundover).

                                                                            The issues you're seeing in this measurement above 5kHz have almost always (in my experience) been due to the tweeter design and construction and interface to the WG throat. Little to no impact from external baffle. Below 3 kHz, it's another story. Someday I'd like to do some detailed studies to understand this better. In the meantime, though, we just try to focus on finding combo's that work well.
                                                                            the AudioWorx
                                                                            Natalie P
                                                                            M8ta
                                                                            Modula Neo DCC
                                                                            Modula MT XE
                                                                            Modula Xtreme
                                                                            Isiris
                                                                            Wavecor Ardent

                                                                            SMJ
                                                                            Minerva Monitor
                                                                            Calliope
                                                                            Ardent D

                                                                            In Development...
                                                                            Isiris Mk II updates- in final test stage!
                                                                            Obi-Wan
                                                                            Saint-Saëns Symphonique/AKA SMJ-40
                                                                            Modula PWB
                                                                            Calliope CC Supreme
                                                                            Natalie P Ultra
                                                                            Natalie P Supreme
                                                                            Janus BP1 Sub


                                                                            Resistance is not futile, it is Volts divided by Amperes...
                                                                            Just ask Mr. Ohm....

                                                                            Comment

                                                                            • draki
                                                                              Member
                                                                              • Oct 2012
                                                                              • 37

                                                                              #83
                                                                              Ergo and all, sorry for highjacking the thread.

                                                                              I did manage to find the original box on a first try, and have mounted the tweeter+WG and remeasured. The tweeter, btw, is a SEAS NoFerro650 on Visaton R146 WG. The mechanical fit between the two is pretty straightforward and solid, and looks gapless. Similar combo at HeissmanAcoustics.de (but the driver is SEAS NoFerro 950, very similar). The drivers were not tampered with, dropped or otherwise misused.

                                                                              While at it, might as well say that the BM is SB17NRCX-8, just for info. CB of 18 L. No complaints for what it is. The usual peak notched down. XO at 1700 LR4 acoustical. But I digress..

                                                                              Attached FRs are the original measurement in green (more than 2 years ago); the same driver in the same original box measured today (blue); and in the narrow baffle box, today (red). I can't repeat the first original measurement, even though I am using the same hardware and software, and doing it in the same room/spot/distance ... But 2 years have passed so maybe I am not replicating all or I did something that I don't recall. I do remember though that I had measured very carefully, several runs, and was very satisfied since the measurement were as per book.

                                                                              The impedance today (blue) does show a little difference, it measures a little lower in the absolute value that the original measurement (green), but the Fs is pretty much spot-on.

                                                                              I am as puzzled as anybody. Not that this is a problem, rather a puzzle in a need of solving..
                                                                              Attached Files

                                                                              Comment

                                                                              • ergo
                                                                                Senior Member
                                                                                • Mar 2005
                                                                                • 675

                                                                                #84
                                                                                draki, try also making a few more off axis measurements 0,10 ,20, 30 for example. It might give you and answer

                                                                                This is a project from years back where I tried to mate Seas XT25 with a waveguide bought from Finland. measurement is sample A and sampel B in 0deg on axis and 15deg off axis. As you see the 15deg offaxis smooths the 7kHz ripple considerably. So tweeter in waveguide can have abrupt issues right at 0deg, but smoother as you move a bit off axis... so maybe you happened to have the wide baffle turned (not perpendicular to mic) ?

                                                                                Click image for larger version

Name:	sample1 and sample2 - 0deg and 15deg.jpg
Views:	1
Size:	221.9 KB
ID:	864348

                                                                                Comment

                                                                                • draki
                                                                                  Member
                                                                                  • Oct 2012
                                                                                  • 37

                                                                                  #85
                                                                                  Originally posted by ergo
                                                                                  so maybe you happened to have the wide baffle turned (not perpendicular to mic) ?
                                                                                  No, the original measurement was 0 deg on axis. And I just can't get it again (as smooth as then) and that is irritating.
                                                                                  I made few quick 0,10,20,30 and 40 deg horizontal measurements in the original wide box, not enough for a definite design but still a good indicator for an initial xo (if I decide to reuse the box and put an effort to finish it cosmetically).
                                                                                  Here is it. Workable, IMO. Design target at 10 deg. No nearfield meas for the BM, only farfield at 90 cm. 4 us gate.
                                                                                  Attached Files

                                                                                  Comment

                                                                                  • draki
                                                                                    Member
                                                                                    • Oct 2012
                                                                                    • 37

                                                                                    #86
                                                                                    .... and the x/o schematic
                                                                                    Attached Files

                                                                                    Comment

                                                                                    • ergo
                                                                                      Senior Member
                                                                                      • Mar 2005
                                                                                      • 675

                                                                                      #87
                                                                                      Bit of a progress report. The physical Xover parts should arrive next week for an Xover that we have tuned some more from the last I posted. We've done few listening sessions with the LspCAD emulating the xover. The initial version did not sound quite right balance wise especially. Comparison to Minerva's in my room confirmed that. So I went 'back to drawing board' and redid the Xover. We listened to few slightly different variations and then settled on one that really seemed to 'click' for all aspects.

                                                                                      Now what is interesting is that with this speaker I did not use "Jon's voicing curve" as a design stage target, but when I did decide to put it on the version we settled on it is a very close match... so for me it's a proof again that at least for speakers with good uniform power response this is a good target that I like a lot. So thanks again to Jon for that one

                                                                                      (reference angle overlay is Jon's voicing curve)
                                                                                      Click image for larger version

Name:	OLA - Ver2 xover - Jons voicing curve.jpg
Views:	1
Size:	40.2 KB
ID:	864378

                                                                                      Comment

                                                                                      • ergo
                                                                                        Senior Member
                                                                                        • Mar 2005
                                                                                        • 675

                                                                                        #88
                                                                                        It's been a long time since I updated the progress on this build. There has been a lot going on actually and meanwhile I have measured and re-measured the speaker a number of times. Then we built 3 or 4 variations of the Xover while speakers were at my place. Eventually around October we moved them to their actual home at my friends place and he kept fine tuning the Xover while listening daily. Some two weeks ago he arrived to a "happy place" with the tuning. Xover is still in a 'ratnest' form but it does sound good. We are building the Purifi power amps (one for me and one for him) and then we'll see if the Xover will be made more permanent as it is.... as we do hear differences between the amps and DAC-s now with ease it is a good sign I guess. A good speaker also means there is no mercy on components upstream.

                                                                                        Right speaker in it home
                                                                                        Click image for larger version

Name:	OLA_Xover_readyproto_Nov2019.jpg
Views:	1
Size:	54.2 KB
ID:	864638

                                                                                        Measurement at 1m ini his room (same position as above pic with mic 1m in front about half way between WG and upper midwoofer)
                                                                                        Click image for larger version

Name:	FinalXover_Meas_Nov2019.jpg
Views:	1
Size:	127.6 KB
ID:	864639

                                                                                        Xover as is currently
                                                                                        Click image for larger version

Name:	FinalXover_sch_Nov2019.png
Views:	1
Size:	54.6 KB
ID:	864640

                                                                                        Comment

                                                                                        • Bear
                                                                                          Super Senior Member
                                                                                          • Dec 2008
                                                                                          • 1038

                                                                                          #89
                                                                                          Congrats! That looks like a relatively benign set of parts (number of parts and part values). How deep did your final waveguide end up? It looked like you tried a number of different profiles and depths.
                                                                                          Welcome to Rivendell, Mr. Anderson.

                                                                                          Comment

                                                                                          • ergo
                                                                                            Senior Member
                                                                                            • Mar 2005
                                                                                            • 675

                                                                                            #90
                                                                                            The one we chose in end was the 26mm deep version
                                                                                            Click image for larger version

Name:	All_WG-s_responses_selected.jpg
Views:	1
Size:	68.1 KB
ID:	864648

                                                                                            With the shown Xover it measure like this
                                                                                            Click image for larger version

Name:	FinalXover_TW.png
Views:	1
Size:	55.2 KB
ID:	864649

                                                                                            Comment

                                                                                            Working...
                                                                                            😀
                                                                                            😂
                                                                                            🥰
                                                                                            😘
                                                                                            🤢
                                                                                            😎
                                                                                            😞
                                                                                            😡
                                                                                            👍
                                                                                            👎
                                                                                            Searching...Please wait.
                                                                                            An unexpected error was returned: 'Your submission could not be processed because you have logged in since the previous page was loaded.

                                                                                            Please push the back button and reload the previous window.'
                                                                                            An unexpected error was returned: 'Your submission could not be processed because the token has expired.

                                                                                            Please push the back button and reload the previous window.'
                                                                                            An internal error has occurred and the module cannot be displayed.
                                                                                            There are no results that meet this criteria.
                                                                                            Search Result for "|||"