-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Introducing Directiva - An ASR open source platform speaker project
- Thread starter Rick Sykora
- Start date
I am sure you guys will figure it out. Just to spitball guess...
It sorta reminds me of this:
https://www.diyaudio.com/forums/multi-way/374851-diffraction-effect.html
hifijim on diyaudio just traced some wobble back to diffraction off of the woofer. I know I have seen the same.
The purifi also has those humps around the cone. I wonder if everyone's woofer is oriented the same way?
I'm looking forward to seeing the result!
OP
- Thread Starter
- #623
So still have some anomaly below 1.5kHz, but is workable until I get the dual channel setup in place. The main issue indicates that my woofer and Amir’s woofer may not match well. From my measurements, it is not too bad, but @ctrl uses Amir’s Kilppel results to design new crossovers for me to try on the real speaker. To get the best results, need the best possible correlations between sims and real hardware.
We took another shot at alignment over the weekend and is still not as good as we both have experienced individually. Without this, cannot be sure the crossover designs being sent accurately represent @ctrl’s intent. Although I can say, they look and sound pretty good despite this annoying issue.
Am going to need to start from flat and add each filter and then compare the sim and real measurements. At this point, we have seen partial alignments, so hopefully we are close. We think my test rig is much cleaner now, but still need to do better before involving Amir!
We took another shot at alignment over the weekend and is still not as good as we both have experienced individually. Without this, cannot be sure the crossover designs being sent accurately represent @ctrl’s intent. Although I can say, they look and sound pretty good despite this annoying issue.
Am going to need to start from flat and add each filter and then compare the sim and real measurements. At this point, we have seen partial alignments, so hopefully we are close. We think my test rig is much cleaner now, but still need to do better before involving Amir!
I don't understand how is that possible. I'm reading how you're having problems with synchronizing measurements and simulation for quite some time. Crossover programs such as VituixCad and LspCad are laser precise.
If you did your measurements right, where lies the problem ?
When you have finished cabinets and mounted drivers, ports etc. you are literally hours away from good measuring crossover. It always needs some listening and light tweaking (attenuate tweeter 0.5dB for better subjective experience or slight change of midbass tuning to better fit the room).
If you did your measurements right, where lies the problem ?
When you have finished cabinets and mounted drivers, ports etc. you are literally hours away from good measuring crossover. It always needs some listening and light tweaking (attenuate tweeter 0.5dB for better subjective experience or slight change of midbass tuning to better fit the room).
Last edited:
OP
- Thread Starter
- #625
Most of the time has been unrelated to Directiva as nice weather and other work have been taking priority.
As mentioned earlier, the mic clip seemed the initial problem. Not sure I posted that diff, but was very measurable. In the past, was not worried as much about absolute correctness, so was not as disciplined as with test rig. A good example is the next post where I will show the change as I reoriented the speaker and mic in different positions throughout our family room.
The last issue is likely some diffs between Amir's woofer and mine. His is one of the original SPK5 woofers and mine is a newer one that I bought. Even the speaker terminals were different. His allow a smaller spades for the quick connects and mine only takes the 4.8 mm ones. Anyway, have more time this week and will get to the bottom of the remaining issue(s).
OP
- Thread Starter
- #627
So, just in case it helps others, here is a post of how much better my measurements have become. My initial gating was done by using a decent minidsp app note that helps determine the first major reflection and set the gate accordingly. It did not focus much on speaker height and so was using a nice 34" speaker stand and (due to space issues) my mic boom was more angled than perpendicular to the front baffle.
Here is my test rig measurement comparison...
Green trace is before (speaker closer to floor and mic boom parallel) and red one is after (speaker more than halfway to ceiling and mic boom perpendicular to baffle). At one point, I noticed I had not gotten as good a result and the difference was pushing the mic about a centimeter forward! These traces are separated to better see the differences. Notably above 800 Hz, you can see how much smoother the red trace is.
Here is my test rig measurement comparison...
Green trace is before (speaker closer to floor and mic boom parallel) and red one is after (speaker more than halfway to ceiling and mic boom perpendicular to baffle). At one point, I noticed I had not gotten as good a result and the difference was pushing the mic about a centimeter forward! These traces are separated to better see the differences. Notably above 800 Hz, you can see how much smoother the red trace is.
Both of those measurements are pretty lousy in terms of resolution. You need quasi-anechoic with at least 10ms time window to be able to see what you are doing and how the transducers behave in a cabinet.
I'm a little bit confused, why do you need to do any measurements at all?
I was under the impression that you build the cabinet based on simulations and then Amir measures each individual driver with the Klippel, you plug in the crossover and some PEQ filters to fix whatever non linearities there are in the response and voilla ?
Am i missing something?
I was under the impression that you build the cabinet based on simulations and then Amir measures each individual driver with the Klippel, you plug in the crossover and some PEQ filters to fix whatever non linearities there are in the response and voilla ?
Am i missing something?
OP
Yep, exactly the case, except for small details.
Amir has one speaker, Rick has the second speaker. Unfortunately, there are manufacturing tolerances in the production. Usually the first thing you do is measure the individual drivers in the speakers on axis as accurately as possible and then determine the differences. Unfortunately, Rick did not do that.
Differences of much more than 0.5dB and "small Q" between the same drivers of the two speakers, must be dealt with additionally in the crossover.
When there were significant differences between what the VCAD simulation with Amir's measurements predicted and Rick's actual measurements, we had a problem.
Were the differences
- due to manufacturing tolerances of the drivers
- by non-optimal measurement conditions in Rick's measurement room
- by the measurement equipment
- by errors in the implementation of the VCAD parameters in the miniDSP (or should I say vice versa
)or by a combination of all.
Then best bet would be raw inbox measurements from Amir sent to you and then modeling the crossover based on that. Then send the finished loudspeaker to Amir to measure it. Amir doesn't lack the expertise to add some correction himself if need be. Exchanging the midwoofers to see if the design holds is a must.
If there is still some problem, then change midwoofer to any Seas transducer. I've compared their midwoofers (alu coned) that are fabricated 10 years apart and still measure the same.
If there is still some problem, then change midwoofer to any Seas transducer. I've compared their midwoofers (alu coned) that are fabricated 10 years apart and still measure the same.
OP
OP
- Thread Starter
- #635
So, using a nearfield woofer measurement was able to clearly establish that the sim is working correctly again. Have been prepping to have @amirm measure and encountered a major hurdle. His Klippel NFS is only wired for a single channel! 
So, the bad news is no Directiva spins until we get an alternative plan determined. However, due to @ctrl's good work, the sims are good and so expect to post them soon.
So, the bad news is no Directiva spins until we get an alternative plan determined. However, due to @ctrl's good work, the sims are good and so expect to post them soon.
Last edited:
Wouldn't it be easier to send him a MiniDSP with the build?
OP
- Thread Starter
- #637
I might if it fixed the issue. The issue rn is that he can only run the bass OR high freq signal - there is only one speaker wire. Otherwise, the electronics have to sit on top of cabinet. We do not think this will yield worthwhile results.
how does he test active monitors? Maybe send an Aiyima A07 or something along the MiniDSP?
OP
- Thread Starter
- #639
An active monitor has amp built into speaker, so in that case, he just needs a single line in.
We will figure it out. Amir needs to sleep a little and I am doing some impedance measures rn. We have lots of ideas, just will take more time before spin can be done unfortunately.
OP
- Thread Starter
- #640
Since it is summer and am taking a long weekend, this post is not really overdue.
But here it is the first of many project posts, but decided to share the latest from @ctrl. Since Amir cannot test as designed, we will likely take this version and make a passive crossover so, the can test. Until that day, if you should choose to build this speaker beforehand, please be aware that you will need to measure and possibly adjust the filters to get a flatter response than the sim allows. We have not been able to fully explain some of the differences between the Klippel measurements and my own (crude but improving ones). The most major diff is that the sims have an upward tilt above about 10 kHz.
After we do some explaining of this design, I will do a simple walkthrough of doing your own with VituixCAD in the build thread. The attached file is compressed and contains the project file and 2 folders with Amir's measurements of the drivers. EDIT: See following post for files. When you open the project, you are likely to get a couple of warning like this one...
This is due to the paths changing between the project and drivers measurements. To fix, you need to do the following for both drivers:
On the drivers tab, click on the driver to select it..
Click on the folder icon (next to Frequency responses) and navigate to the folder where you extracted the driver measurements and select all the files in the folder and click Open on the dialog box. You will see another error about duplicate response, you can ignore and click OK. Repeat for each driver. If you did this correctly, the frequency response, the frequency response plot should flatten and excess phase traces will be gone. It should look like this...
Note that below the Frequency response information there is a comparable one for Impedance response. This project does not need as it uses an active crossover, but if you want to do a passive design, you will need to associate the impedance files for each driver (we will show how once Amir supplies the files).
Click on the crossover tab and this will display the crossover diagram and a full set of graphs. it should look like this...
You can double click on any the graphs and they will be show in a larger view. At this point, expect there will be questions and I will assist as I can. As this design was done by @ctrl, he will take over from here.
Enjoy!
But here it is the first of many project posts, but decided to share the latest from @ctrl. Since Amir cannot test as designed, we will likely take this version and make a passive crossover so, the can test. Until that day, if you should choose to build this speaker beforehand, please be aware that you will need to measure and possibly adjust the filters to get a flatter response than the sim allows. We have not been able to fully explain some of the differences between the Klippel measurements and my own (crude but improving ones). The most major diff is that the sims have an upward tilt above about 10 kHz.
After we do some explaining of this design, I will do a simple walkthrough of doing your own with VituixCAD in the build thread. The attached file is compressed and contains the project file and 2 folders with Amir's measurements of the drivers. EDIT: See following post for files. When you open the project, you are likely to get a couple of warning like this one...
This is due to the paths changing between the project and drivers measurements. To fix, you need to do the following for both drivers:
On the drivers tab, click on the driver to select it..
Click on the folder icon (next to Frequency responses) and navigate to the folder where you extracted the driver measurements and select all the files in the folder and click Open on the dialog box. You will see another error about duplicate response, you can ignore and click OK. Repeat for each driver. If you did this correctly, the frequency response, the frequency response plot should flatten and excess phase traces will be gone. It should look like this...
Note that below the Frequency response information there is a comparable one for Impedance response. This project does not need as it uses an active crossover, but if you want to do a passive design, you will need to associate the impedance files for each driver (we will show how once Amir supplies the files).
Click on the crossover tab and this will display the crossover diagram and a full set of graphs. it should look like this...
You can double click on any the graphs and they will be show in a larger view. At this point, expect there will be questions and I will assist as I can. As this design was done by @ctrl, he will take over from here.
Enjoy!
Last edited:
