-
X sim “as measured” vs “derived”
I use REW and Xsim. When importing FRD and ZMA files there is the option of “As measured” versus “derived.” I believe this is referring to phase. I export phase data with REW but I am not clear as which of the above options to use. Is REW exporting group delay??. ( I have asked this question on other forums and have not received an answer. Can’t believe Im the only DIYer non engineer struggling with this)
Another confusion on my part is the measurement of acoustic offset. I don’t understand why the individual driver plots do not fall exactly on the combined parallel drivers plot without any offset. Im obviously missing something essential . Is setting a timing reference essential?
If anyone can help.
PS A detailed video on “phase for idiots” would likely get a lot of views. I being the first !
-
16 Replies
-
I would implore you to ignore the measured versus derived for the time being. It does not sound as if you really need it. But what you do need to focus on is the z-offset. This is crucial to getting accurate crossover simulation. For those that are more visual learners, I did a video on why you need to account for Z-offset here:
From what I gather you are measuring your response of your tweeter and midrange separately, then measuring them together in parallel. You are importing your tweeter and mid measurements into X-sim, wiring them in parallel (in the program) and doing an overlay of the measurement you took of them in parallel. And for some reason these measurements are not lining up with each other? For anyone that is lost, check out this video, it should help. It also gives some really cool tips and tricks that will help you use X-sim more efficiently.
If that is the problem, let me try to explain it to you. If you look at the tweeter and midrange, you will notice the midrange magnet sticks back quite a bit further than that of the tweeter. That means the sound, although leaving at the same time, will arrive later on the midwoofer. Why? because it sits further back. Now Xsim doesn’t know this from the standard measurement. It just knows that whenever the sound hit the microphone, that is shape of the response. It has no idea there is a delay at all.
By wiring these together in parallel, we can now right click on the midwoofer and add a mod delay. This tells x-Sim that there was a time difference. We get this time difference by changing the mod delay until it matches the response that you took of them together. This is due to the fact that the response you took with them wired together, is now measuring the delay. So once the two graphs match up (by you changing your mod delay) you have now accounted for the time difference.
As far as being in phase at the crossover point. The easiest way to do that, is once you have your mod delay figured out, check the point at which your drivers cross. The final response should be 6db higher than that point. If they are, you should be in phase. You can easily double check by wiring up your crossover and moving the mic off axis. If you are in phase, you should not see any change in the crossover region as you go off axis. Off course, that assumes you haven’t crossed your mid over too high.
-
thank you. Very helpful. I will happily ignore the as measured/derived option. My initial attempts at z axis measurement gave me results that seem to be on track, if not exact, eyeballing the drivers. Here are two images of the project. In spite of the imperfect testing results ( a narrow null at about 1500 and question of z axis, they sound incredible. I plan to continue to refine the crossover but I wonder how much more there is to attain. Only one way to find out, I guess.
-
This reply was modified 1 month ago by
Toids_DIY_Audio. Reason: Adding as photos instead of files
-
This reply was modified 1 month ago by
-
Those look beautiful! I’m a sucker for open baffle as well. I’ve been wanting to design some on the channel but just haven’t yet. I especially like the baffle design? Did you do that fit looks or to help with diffraction? It should definitely help with diffraction.
As far as the null you’re getting at 1Khz, feel free to post some pictures. You are also able to upload your measurements along with what you believe the z-offset is and we can double check it for you.
We are happy to help
-
I wanted a stylish baffle and I used Basta to limit diffraction. Please understand this was a novice attempt, but I think I lowered diffraction with the lack of symmetry. I’ll include a screen shot of the crossover that resulted in the null, along with the in room scan in REW demonstrating the problem. ( A problem I don’t think I can hear.) I note the delays used in the screen shot. I found those using the technique as you described.
What is most troubling is the fact that there doesn’t seem to be any sign of the null in xsim. I am very confident of the REW scans as they were repeated numerous times, all in baffle, and outdoors showing excellent consistency. I tried simply flipping phase, and that was a complete mess.
I will say that the low distortion of the drivers and the use of the horn results in a very clear and realistic sound. Base is adequate, not earth shaking. I do have it crossed with a sub at 80 and that adds just a bit of depth when called for. I use solely for music, not theatre.
Im working on a complete revision of the crossover. I want to get their z offset right. I want the woofer and tweeter to cross lower and phase to be better matched at crossover points. Id like to cross the woofer around 400 and the horn at about 2000.
Im also considering , as a final step, adding room correction. Im hoping to do this through Roon, but that is for the future .
PS I tried so send the complete REW test group but I couldn’t load it up for some reason. It is all clearly labeled if I d get it uploaded
-
This reply was modified 1 month, 1 week ago by
rskmd.
-
This reply was modified 1 month, 1 week ago by
-
I apologize, the originally these were marked as spam. I fix that and I’ll keep an eye on it going forward.
-
I wrote a response that seems to be lost. In any event, the baffles were designed in Basta in attempt to minimize diffraction. Being a novice, not sure how well I did but overall, the asymmetry should lower diffraction,I believe.Ill include a few files demonstrating my problem . I also repeated my acoustic offset measurements, but still fail to get good alignment. Not sure if room reflections are the problem.
This file was truncated for preview. Please download to view the full file.This file was truncated for preview. Please download to view the full file.This file was truncated for preview. Please download to view the full file.This file was truncated for preview. Please download to view the full file.This file was truncated for preview. Please download to view the full file. -
I will take a look tomorrow. But typically when you have a null like that it is due to the fact you have reversed the polarity of one of your drivers. In this case it looks like the tweeter at the crossover point. Can you double check the polarity on the tweeter? Or even reverse it to see what happens?
-
This reply was modified 1 month ago by Toids_DIY_Audio.
-
This reply was modified 1 month ago by
-
@rskm a couple of issues. The files you uploaded are text files and not FRD files. Can you upload the FRD and ZMA files you have? It would actually be better if you can just send over the Xsim save data that has your saved offsets that way I can see where you went wrong and how you can fix it next time.
-
Interesting that the null doesn’t show up on Xsim. Makes me think there is a basic error in data. I thought likely in phase and a source of that error could be poor acoustic alignment on the z axis. Simply inverting phase causes a mess. So I thought to start from scratch and remeasure z offset. Note that I am using a compression driver for tweeter so in relation to the mid, its quite a bit back . I make adjustments relative to the mid and keep it as fixed. ( I had to Zip compress to upload. ) You have my last trial in Xsim and the windowed fed files. I windowed as I thought some reflections could be confusing the phase >
PS. I get an error regarding the types of files one can upload. Not sue if the files you requested are compatible. I hope I’m not causing confusion. My apologies if I am.
-
That all worked, but I don’t see any ZMA files in the folders. Do you have those? Without them you can’t properly design a crossover. I can take a look at the offsets, but that is about it.
As far as the types of files, you are correct, the site only like certain ones, luckily zip is one of them. You can always jsut zip them and upload it.
-
Just going off your frd with no ZMA data, it looks like 4.25 on the compression driver and 4.5 on the woofer. However, it isn’t lining up as well as I would like. Can you take me through how you measured them? Where the mic was, how far away from the speaker, etc.. Also I noticed you put windowed on your response. There is nothing wrong with that per se, but can you let me know what you did for the window?
-
seems too much. Eyeballing the drivers, should land somewhere between 1-2.5 inches. I used the standard approach. A sweep for each driver with mic positioned 1 meter from 0 degree axis of mid. Then horn plus mid in parallel, and woofer plus mid in parallel. Once again, mic not moved from original position where individual driver sweeps were done. Exposed from REW as text files. Name changed to 3rd files. All imported in Xsim. I used windows to avoid reflections altering phase measurement. ZMA files all done in DATs . All frd for crossover done outside in baffle. Repeated numerous times to confirm repeatability. I sent an image which gives an idea of the relative position of the drivers. 4 inches is too much. I was wondering given a wavelength of 1 meter about at the woofer/mid crossover of 300htz, an error on 1 inch represents about a 2% error in phase o or about 7 degrees. Is this at all relevant?? Id add that in reviewing this null problem, it seems it is falling just where the woofer and tweeter cross. Perhaps the problem in simply that they cross too high up. Ive worked on a complete redo of the crossover with steeper curves and the woofer/tweeter cross occurring down at least 25 db. Maybe that will resolve the problem
-
This reply was modified 1 month ago by rskmd.
-
This reply was modified 1 month ago by
-
A couple notes, first if doing indoor measurements there is no reason to be 3 feet away. This is designed for anechoic measurements. I would move the mic considerable closer to remove as much of the room as possible. Depending on the height of the speaker somewhere between 16-20 inches to the baffle would be good.
Second, I know you mentioned you measured from the midrange. Is that where you believe ear level will be? Typically it is measured from tweeter level. Measuring from the midrange isn’t necessarily wrong, but it will affect the crossover design and the final response if your listening windows is at tweeter level. IN general, it is usually recommended to take all measurements from tweeter level.
Keep in mind when we talk about the z-offset we are talking about the acoustic center and not the center of the magnet, etc. The acoustic center might be different that you think of a variety of reasons including the placement on the baffle, baffle angle, etc. If might be further back than you think.
Finally, I would take multiple measurements next time, one windowed and one not windowed. I am not saying you are windowing them incorrectly, but it would be nice to see the difference. This is one reason I do really appreciate the Omnimic, since it has a blended mode that give you really great in room measurements without the need to do much. Of course, currently you can’t even get an omnimic, but I am excited for V3.
-
OK. All points taken. Thank you once again for your time and effort. It is greatly appreciated.
-
Ok. Repeated 16 inches from baffle centered on tweeter 0 degree axis. No windowing. Got very nice alignment at 2.3 inches. Finally. And the result agrees with the range of values I thought most likely. I am surprised at the sign. I previously was at negative 1.6. I had read somewhere that the z axis turns negative further from the mic. Since this is a three way, I used the mid as a reference point. The tweeter is a compression driver in a horn, and therefore sticking a bit out more than the other drivers. In any event, the overlay of the curves is compelling. The alignment of woofer to mid remains a mess. I will continue working on I, but I hav e to admit, I am a bit skeptical of the importance of a few degrees of misalignment at those lower frequencies.
-
Sweet! Glad it worked out for you! As far as the lower frequencies go, I understand exactly what you are saying. However, you will still want to double check your off axis later, as it could negatively affect them.
-