Allen, when I say "Let's try to nail down the basics of my measurement scheme and driver positions in Vituixcad", this is my way of saying I need help to understand this. I get that it's important to understand this, this is why I asked the very specific questions earlier about measurement scheme.
I usually take everything people write here literally. I'm not sure @DcibeL meant something else, I can't speculate about this.
Is there a measurement scheme that would allow all driver positions in Vituixcad to be fixed at 0?
I must admit it can be hard to follow you Allen, you are cryptic at times 🙂
Member
Joined 2003
Not sure why I'm being quoted here replying to someone else in another thread, but since you are bringing up using a constant mic location for all individual driver measurements..
With the use of VituixCAD, CTA-2034-A and reliable results regardless of speaker type, size, etc, yes, measuring drivers from a single mic location is not good practice. With a dual or semi-dual measurement configuration in mind, there is no situation where measuring individual drivers off axis (at the tweeter axis) makes sense. The "problem" is not only one of delay between drivers, but also SPL over distance, and accurate results for power/DI chart and directivity plots, which is why I say the simulation is locked in to a response that may not be an accurate representation of the response at listening distance. DI/power spinorama will also be incorrect unless the mic location is 2m+ which is impractical for most indoor measurements.
To answer your question, just follow the measurement instructions for VituixCAD, it doesn't require very expensive gear and is a simple method that overcomes the limitations of leaving the mic in a single location. Following this process, and assuming your speaker has a flat baffle, z axis offset remains at 0 for all drivers, tweeter x,y coordinates at 0, and other driver x,y coordinates are entered relative to the tweeter location.
With full set of on and off-axis measurements completed at driver axis, listening distance can be easily adjusted in VituixCAD with accuracy, simply done by interpolating off-axis measurement data and calculation of SPL adjustment over distance to listener. The main benefit here is accurate response result at normal listening distance of 2m+ using measurement data taken at a shorter ~1m.
To be honest I'm finding this thread quite difficult to follow, can you elaborate a bit on the specific problems / questions you have in the measurement / design process?
Attachments
Hi DcibeL,
I referenced your post in the Vituixcad thread because you provided some excellent feedback that cleared up some questions for me. I got very confused when Allen then disputed this.
Thank you, this is exactly how I have done my measurements and set driver positions in Vituixcad.
Allen has been very kind and patient with his contributions here, but I think we may have been talking past eachother.
So I fully understand that you find the thread hard to follow 🙂
Initially I wanted some response to my preliminary crossover simulation that I attached in my first post. I also attached FRD and ZMA files.
Do you think this looks okay DcibeL, or do you see any room for improvement (I would be very surprised if not)?
Member
Joined 2003
Ok, now I am up to speed 🙂
Given the posted data, first suggestion is that you need response to 180 degrees in order to fill in the power and DI data. Measurements are incomplete, so only on-axis response is useful. If you use D2608, with the amount of off-axis data available, only listening window will be useful.
Your woofer response files are showing a major problem between 100-200Hz, and are also not showing typical baffle step response for drivers measured on a speaker baffle, so I don't have much confidence in your splicing of near/far measurements.
Second suggestion is to provide a higher order filter on the tweeter than just a single cap. But first you need to provide the software with good data, so focus on the measurement and response processing procedure first.
Given the posted data, first suggestion is that you need response to 180 degrees in order to fill in the power and DI data. Measurements are incomplete, so only on-axis response is useful. If you use D2608, with the amount of off-axis data available, only listening window will be useful.
Your woofer response files are showing a major problem between 100-200Hz, and are also not showing typical baffle step response for drivers measured on a speaker baffle, so I don't have much confidence in your splicing of near/far measurements.
Second suggestion is to provide a higher order filter on the tweeter than just a single cap. But first you need to provide the software with good data, so focus on the measurement and response processing procedure first.
Last edited:
Hi DcibeL,
this response is exactly what I needed, thank you!
I am aware of this. I was a bit lazy, not having a turntable this was a tedious process. So I decided to take a shortcut this first time, and tweak and evolve this crossover over time when I get more experience. I just need something to get me started. But I suppose I can still simulate directivity to 60 degrees with woofers and D2608 tweeter.
For the Morel tweeter I only have on-axis measurments at the moment.
This is a good point. I have attached an image showing my nearfield measurements of woofers and port. Green = Woofers, Red = Port.
And also one of the Merger tool in Vituixcad, I may have messed up something here.
I will try to simulate a second order filter for the Morel.
Attachments
Member
Joined 2003
Neafield response should be the immediate area of focus. Area from 100-200Hz should be smooth, that may be cabinet reflection, is there no stuffing in your cabinet? Port response showing a big peak at 200Hz is also of some concern. If you load in just nearfield and port response to the merge tool, no diffraction, the result should closely resemble your enclosure model. For the sake of simplifying the merge process, you can always skip the port response, it's not going to have much influence on a crossover design >1kHz.
Beyond that, I would recommend processing each upper and lower woofer responses individually. Diffraction response is fine, but just for the sake of argument, since nearfield response is for a single driver, diffraction simulation should also only include single driver, and distance = typical listening distance of 2-3m. Not that it will make much difference either way, but 10m distance is a bit odd to see here. You will find that making this adjustment has little effect to the diffraction response of interest for this task, <1kHz.
Beyond that, I would recommend processing each upper and lower woofer responses individually. Diffraction response is fine, but just for the sake of argument, since nearfield response is for a single driver, diffraction simulation should also only include single driver, and distance = typical listening distance of 2-3m. Not that it will make much difference either way, but 10m distance is a bit odd to see here. You will find that making this adjustment has little effect to the diffraction response of interest for this task, <1kHz.
Last edited:
There is definitely stuffing in the cabinets, but I will see if I can rearrange it or experiment with different types of stuffing.
It looks like I have some more work to do with the nearfield measurements before moving on, I'm going to remeasure this.
And thanks for the tip about skipping port response, since it's also on the back it will have even less influence on the higher frequencies.
I just copied the example from kimmosto in his tutorial here, where he used 10m distance. I will try this using listening distance instead.
When you say processing each woofer individually, you mean the farfield responses? I did measure each on their own axis, and imported them individually to the merger tool.
Wrt the diffraction response, the nearfield response is for a single driver but bass response includes both drivers. Both drivers were connected, but one driver was isolated with a pillow in front of it. I'm a bit confused about only using one driver for diffraction response.
Member
Joined 2003
In retrospect I think what you've done for diffraction is fine. Both woofers included in diffraction model to create average diffraction in the far field, and 10m distance is used to avoid response null from differing driver distance to mic. Keep on as you are, near field response is the only real point of interest to address.
Hello.
Designing a commersial speaker is a something including the speaker elements parameters and the volume/size of the enclosure. That and meeting desired retail price still resulting in profit. Guess You all know this, not unique for speakers - applies basically anything on the shelf.
My question or thinking is regarding the crossovers, as a example the (famous) Rogers LS3a where a small near field monitor used by BBC (UK) and still can be bought as a clone. But the crossover in these are very complex and big, lots of coils, condensors and resistors and are, according to what i learned not the most efficient speakers around. Never heard them but have absolutely no doubt they are amazing.
Anyway, to make this long story shorter: "Less is more" (old saying) and if applied on crossovers AND speaker elements, spending more time on matching the elements to each other and the box/volume/type of enclosure (closed/vented/horn/open baffle/...) must (?) lead to less "manipulation" via added components, which inturn create problems just been added (phase shifts at x-over frq etc).
Summarizing since I seemed to lost myself, possibly You aswell; Crossovers, is less more?
Depending on the design-
It can take a simple nudge, or it can take a hammer. What the wanted end goal is, is up to the builder.
It can take a simple nudge, or it can take a hammer. What the wanted end goal is, is up to the builder.