@Inman thanks for a bit more information on the positions.
What data I attached was done quickly to pretty much back up your first points, I did caveat the data provided and suggest that you treat them with some caution.
Once you get into the real data you can play with the balance and actual Xover point and corner frequency to your hearts content. maybe making the tweeter a third order and taking it down to below 2 KHz whilst keeping a watchful eye on distortion.
When you have the drivers with you, and you can take real time measurement's the interactions and the iterations can be far quicker as you can measure what a component does to the response in a few minutes.
For @svp regarding your question about using the enclosure modelled impedance. Yes I could have used that data but I made the easy decision to skip that as the rapid changes of impedance around the reflex frequency would have a very small affect on the impedance around the crossover.
Back to the speakers, from your latest posts Inman it does seem like you will able to easily knock this into shape. I am intrigued to see how the Dayton driver performs and what effect all the felt will have on final measurements and values.
What data I attached was done quickly to pretty much back up your first points, I did caveat the data provided and suggest that you treat them with some caution.
Once you get into the real data you can play with the balance and actual Xover point and corner frequency to your hearts content. maybe making the tweeter a third order and taking it down to below 2 KHz whilst keeping a watchful eye on distortion.
When you have the drivers with you, and you can take real time measurement's the interactions and the iterations can be far quicker as you can measure what a component does to the response in a few minutes.
For @svp regarding your question about using the enclosure modelled impedance. Yes I could have used that data but I made the easy decision to skip that as the rapid changes of impedance around the reflex frequency would have a very small affect on the impedance around the crossover.
Back to the speakers, from your latest posts Inman it does seem like you will able to easily knock this into shape. I am intrigued to see how the Dayton driver performs and what effect all the felt will have on final measurements and values.
I just saw this, and think it is something to definitely consider.
see post 545.
Where Wolf informs us
"In my current process of implementing the SIG180-4 and trying to wring the best out of it, I find the 3kHz bump to still be breakup and require suppression. I think it would be lucky to get 2.5kHz at the top end of it. My usage is targeting 2.4kHz, and that blends to an LR6 relatively easily with a damped 2nd order and LCR across it for the breakup. For reference, the series notch across the driver is 1.2mH - 2.0uF - 1.2 ohms (coil DCR plus resistor) "
I think the tweeter you have chosen can get down low enough to help with this. Good choice.
see post 545.
Where Wolf informs us
"In my current process of implementing the SIG180-4 and trying to wring the best out of it, I find the 3kHz bump to still be breakup and require suppression. I think it would be lucky to get 2.5kHz at the top end of it. My usage is targeting 2.4kHz, and that blends to an LR6 relatively easily with a damped 2nd order and LCR across it for the breakup. For reference, the series notch across the driver is 1.2mH - 2.0uF - 1.2 ohms (coil DCR plus resistor) "
I think the tweeter you have chosen can get down low enough to help with this. Good choice.
You have to have measurements to use VituixCAD, without them you have to know what you are doing. Like literally know, in the @kimmosto level know (who is the author of VituixCAD), which is unattainable for mere mortals.
REW -> VituixCAD (with all the necessary tools) -> VituixCAD impulse response export -> EQ APO (or any other DSP) -> listen and make changes to your taste -> measure complete build -> hardware crossover
This work sequence works for me. Of course there is also physical build of box, maybe several boxes.
Now we are talking about nothing.
REW -> VituixCAD (with all the necessary tools) -> VituixCAD impulse response export -> EQ APO (or any other DSP) -> listen and make changes to your taste -> measure complete build -> hardware crossover
This work sequence works for me. Of course there is also physical build of box, maybe several boxes.
Now we are talking about nothing.
The baffle is stepped (I assumed the tw is 20mm beyond the wf)
Woofer
T (Sd for tw should be circa 6cm2)
Definitely use modeling for driver selection, but once you have selected the drivers, mount them and take measurements with them in the enclosure and on the baffle you will be using. If possible, flush mount the drivers to reduce diffraction issues. This may be a challenge, though, since you are using existing cabinets.
Depends on your budget for the project. But, designing a PCB and having it manufactured to your specifications is satisfying. Be sure to use a heavy copper layer. Rather than using relatively small traces, use large pads and just etch out the borders to minimize resistance, like in this picture:
It is a great tool, especially being for free. (I did make a donation to the developer since I often use it.) Ideally, you have a calibration file for your microphone that you can import into REW to maximize accuracy of your frequency response measurements. If not, but your microphone is accurate enough, that should be fine.
I'll throw out one more option:
Have you considered going all active? Good performing amplifiers are relatively inexpensive these days. You could use a miniDSP unit to implement your crossover and equalization using DSP. Or, you can use a Raspberry Pi (inexpensive) running CamillaDSP (free) and a DAC.
Another option is to use a Hypex FusionAmp, which provides the DSP, DAC and amplification. https://www.diyclassd.com/products/fusion-amplifiers/
The benefit of going all active is that you can experiment with all sorts of crossover topologies, crossover slopes, EQ, time delay, etc. with the fully assembled speakers, in your room, using your computer and REW. Also, going all active minimizes resistance between the woofer and the amplifier, providing a high damping factor, and thus making the most of the woofer's bass response.
I removed the passive crossovers from the speakers in my family room and went all active. It took those speakers to a whole new level. I still am impressed with the final results.
Have you considered going all active? Good performing amplifiers are relatively inexpensive these days. You could use a miniDSP unit to implement your crossover and equalization using DSP. Or, you can use a Raspberry Pi (inexpensive) running CamillaDSP (free) and a DAC.
Another option is to use a Hypex FusionAmp, which provides the DSP, DAC and amplification. https://www.diyclassd.com/products/fusion-amplifiers/
The benefit of going all active is that you can experiment with all sorts of crossover topologies, crossover slopes, EQ, time delay, etc. with the fully assembled speakers, in your room, using your computer and REW. Also, going all active minimizes resistance between the woofer and the amplifier, providing a high damping factor, and thus making the most of the woofer's bass response.
I removed the passive crossovers from the speakers in my family room and went all active. It took those speakers to a whole new level. I still am impressed with the final results.
@TerryForsythe I did consider going active and knew of some of the advantages while you mentioned some I did not know. I decided against it because it ultimately pushes my costs up too much. Loved the idea - my wife, not so much.
I have planned the board for 5mm minimum traces on 2oz board. Node spacing minimums are 5mm. Spacing minimum to air coils is 10mm for unrelated nodes. For 2nd order designs I tried, I was able to use just 1 layer. For 3rd and on I have very limited use of the 2nd layer. I'm not sure it matters but the point was to limit plate and fringing parasitics between nodes. I believe the boards should handle 7 Amps safely. I think that should be plenty with the drivers chosen. I set up the boards such that the high pass and low pass are independent. Should I bi-amp someday I'll remove jumpers merging the inputs.
Yep, I have the ability to see my mic is calibrated.
@shadowplay62 Wow! Thank you. I am learning this a bit at a time. So where did you get the inputs for the half space response? I'm familiar with the tracing tool and have used it quite a bit. Once done, you then used Export to fold into your simulation of the drivers? On the tweeter, I noticed you have the Full Space button selected but not on the woofer. Why?
@svp
@raymondj
I have planned the board for 5mm minimum traces on 2oz board. Node spacing minimums are 5mm. Spacing minimum to air coils is 10mm for unrelated nodes. For 2nd order designs I tried, I was able to use just 1 layer. For 3rd and on I have very limited use of the 2nd layer. I'm not sure it matters but the point was to limit plate and fringing parasitics between nodes. I believe the boards should handle 7 Amps safely. I think that should be plenty with the drivers chosen. I set up the boards such that the high pass and low pass are independent. Should I bi-amp someday I'll remove jumpers merging the inputs.
Yep, I have the ability to see my mic is calibrated.
@shadowplay62 Wow! Thank you. I am learning this a bit at a time. So where did you get the inputs for the half space response? I'm familiar with the tracing tool and have used it quite a bit. Once done, you then used Export to fold into your simulation of the drivers? On the tweeter, I noticed you have the Full Space button selected but not on the woofer. Why?
@svp
Perfect. Thank you for the steps.
@raymondj
I'm getting to understand the flow of the data in Vituixcad, so I plan to play around a bit more. I appreciate the confidence in driver selection. I really was not sure I could have these work out. I had information when selecting them, but not any experience to weight the information.
Assuming you already have one amplifier, you could do it for around $300:
https://www.minidsp.com/products/minidsp-in-a-box/minidsp-2x4-hd
https://a.co/d/dgKUBBX
By the time you buy the PC boards and crossover components, I'm not sure how much you will save if you go passive, especially if you end up needing to tweak the crossover, and thus buy more parts.
"DSP" part can be absolutely free for the time of development.
Here is one of mine "2-way crossover" simulations attached. Yep, you see it right - that is all it needs for one channel
The inputs for the Half space FR are the traced curves or the FRD you can download from the websites of the manufacturers (when available).
So basically, you have to
Design your box in Enclosure tool and save the 2Pi FR and ZR
Design your baffle in diffration tool and export the curve at 10m and 1m (no full space/directivity check). Then load the half space response and check the boxes Full space/Directivity (distance should be 1m) and export the FR. For the Tw, load the half space response and check the boxes Full space/Directivity (distance should be 1m) and export the FR.
For wf and mid, that equire a box, open the Merger tool, load the FR previously exported and merge the FR to input in the main program.
To add to shadowplay62's information.
I used the woofer data that was included in the Dayton driver for that driver. In the FRD data these are typically 4 files zero degrees , 30, 45 and 60. I used the the zero degree data. Dayton are to be commended for this, but at the same time their resolution maybe be less than we would personally measure.
The Peerless tweeter was done by using the SPL trace tool. I used the 0 degree axis for SPL/FRD data and the impedance trace export for the ZMA.
Shadow play you mention "Design your baffle in diffraction tool and export the curve at 10m and 1m (no full space/directivity check)"
I am curious when do you utilise this information. For room interaction modelling maybe ?.
I liked your Xover that you posted the filter on the tweeter works well when i implement it in the data I generated.
I personally do not worry to much about the Lower Frequency range in the early stages of a design , 5inch, 6 inch, and 8 inch can only give so much.
Differing rooms , furnishings and position will be different across users. Its difficult to know what the bass quality will be like for everyone.
I most certainly do use the enclosure tool to show the basics in terms of maximum volume achievable against the optimum enclosure , try to keep group delay below 5-10 mS and port velocity within bounds. I grew up with the sound of Wharfedale, Leak and Celestion, speaker nearly all of them sealed consequently I am biased.
Finally for Inman one thing for sure the use of a cheap improvised rotating base will be really useful for measurements at various angles. I used a 10 or 12 inch Lazy Susan type rotating bearing from Amazon.
I have used DSP and hypex modules, in the past. I do wonder about their longevity past the 10 year point. What happens if a SMD component dies or a piece of silicon expires will you be able to source its replacement. Additionally lots of the software runs via windows. The same can be said of through hole passive components going forward I imagine.
What will the windows subscription model look like in years to come.
Yes, my Fine X crossover software, Hypex software, Clio pocket software all run on the windows platform. I am not sure how many will run in a Linux environment. One for further research.
I used the woofer data that was included in the Dayton driver for that driver. In the FRD data these are typically 4 files zero degrees , 30, 45 and 60. I used the the zero degree data. Dayton are to be commended for this, but at the same time their resolution maybe be less than we would personally measure.
The Peerless tweeter was done by using the SPL trace tool. I used the 0 degree axis for SPL/FRD data and the impedance trace export for the ZMA.
Shadow play you mention "Design your baffle in diffraction tool and export the curve at 10m and 1m (no full space/directivity check)"
I am curious when do you utilise this information. For room interaction modelling maybe ?.
I liked your Xover that you posted the filter on the tweeter works well when i implement it in the data I generated.
I personally do not worry to much about the Lower Frequency range in the early stages of a design , 5inch, 6 inch, and 8 inch can only give so much.
Differing rooms , furnishings and position will be different across users. Its difficult to know what the bass quality will be like for everyone.
I most certainly do use the enclosure tool to show the basics in terms of maximum volume achievable against the optimum enclosure , try to keep group delay below 5-10 mS and port velocity within bounds. I grew up with the sound of Wharfedale, Leak and Celestion, speaker nearly all of them sealed consequently I am biased.
Finally for Inman one thing for sure the use of a cheap improvised rotating base will be really useful for measurements at various angles. I used a 10 or 12 inch Lazy Susan type rotating bearing from Amazon.
I have used DSP and hypex modules, in the past. I do wonder about their longevity past the 10 year point. What happens if a SMD component dies or a piece of silicon expires will you be able to source its replacement. Additionally lots of the software runs via windows. The same can be said of through hole passive components going forward I imagine.
What will the windows subscription model look like in years to come.
Yes, my Fine X crossover software, Hypex software, Clio pocket software all run on the windows platform. I am not sure how many will run in a Linux environment. One for further research.
In VituixCad you can import all of the frequency respones files. That way you can evaluate the dispersion characteristics of the design. I pay special attention to the directivity index, and try to keep is as linear as I can.
To all: Drivers were purchased this afternoon. Seems like I cannot procede without knowing more of the driver/enclosure interaction.
@TerryForsythe I didn't know of such devices. Very cool. I'm using a Sonos Amp for streaming, and phono. I'm not sure I'd trust biamp-ing with another. If there were timing issues or anything unusual, I doubt I could get Sonos to help. I've toyed with replacing the Amp with Sonos' streaming front end w/out amplification, then going with a conventional pair of amps. This will happen with a concurrent divorce😉
@svp interesting... so you have DSP while testing your speaker. What are the big picture steps? Are you expecting to finish and be using DSP for both channels using other methods or are you testing with DSP to end up with a passive components?
@shadowplay62 Thanks for the explanation. I have not even tried the Merger tool. It's on my list now. I put your crossover parameters in and made a few minor tweaks to see what moves what. Thank you. That bit of bad behavior between 420 and 780 Hz is annoying - nothing I did moved the plot. Looks like it will be audible but perhaps not. I know that I have tried several SPL and impedance files and not all display this bump.
@raymondj
@TerryForsythe I didn't know of such devices. Very cool. I'm using a Sonos Amp for streaming, and phono. I'm not sure I'd trust biamp-ing with another. If there were timing issues or anything unusual, I doubt I could get Sonos to help. I've toyed with replacing the Amp with Sonos' streaming front end w/out amplification, then going with a conventional pair of amps. This will happen with a concurrent divorce😉
@svp interesting... so you have DSP while testing your speaker. What are the big picture steps? Are you expecting to finish and be using DSP for both channels using other methods or are you testing with DSP to end up with a passive components?
@shadowplay62 Thanks for the explanation. I have not even tried the Merger tool. It's on my list now. I put your crossover parameters in and made a few minor tweaks to see what moves what. Thank you. That bit of bad behavior between 420 and 780 Hz is annoying - nothing I did moved the plot. Looks like it will be audible but perhaps not. I know that I have tried several SPL and impedance files and not all display this bump.
@raymondj
I wanted to know how sensitive the frequency response and DI are to cabinet and room changes. I realize speaker makers design for general rooms as a matter of course. Without experience, I wanted to know if I was creating something that would work in only a narrow set of cabinet and room variations. I learned that adding room reflections is like introducing a cat fight to a meditation.
Correction for some of the stuff in post 31. I didn't re read after I had mad some cut and paste.
I have successfully used Hypex DSP and hypex Ncore amplifiers in the past. I do wonder about their longevity past the 10 year point.
What happens if a SMD component dies or a piece of silicon expires will you be able to source its replacement.. The same can be said of through the larger wattage and voltage rating through hole passive components, will they be difficult to source because they are not utilised in computers, cars, Phones or televisions.?
Additionally, what will the windows subscription model look like in years to come.
Yes, VituixCAD plus Fine X crossover software, Hypex software, Clio pocket software etc all run on the windows platform. I am not sure how many will run in a Linux environment. One for further research by me
I have successfully used Hypex DSP and hypex Ncore amplifiers in the past. I do wonder about their longevity past the 10 year point.
What happens if a SMD component dies or a piece of silicon expires will you be able to source its replacement.. The same can be said of through the larger wattage and voltage rating through hole passive components, will they be difficult to source because they are not utilised in computers, cars, Phones or televisions.?
Additionally, what will the windows subscription model look like in years to come.
Yes, VituixCAD plus Fine X crossover software, Hypex software, Clio pocket software etc all run on the windows platform. I am not sure how many will run in a Linux environment. One for further research by me
They will run in a Linux environment. Ideally in a virtual machine, but I've run VC and Xsim in WINE.
In said virtual machine, they're not especially fussy about which version of windows is used. The hypervisor can keep the windows machine isolated from the network so it's less critical.
In said virtual machine, they're not especially fussy about which version of windows is used. The hypervisor can keep the windows machine isolated from the network so it's less critical.
Well... This still applies:
REW -> VituixCAD (with all the necessary tools) -> VituixCAD impulse response export -> EQ APO (or any other DSP) -> listen and make changes to your taste -> measure complete build -> hardware crossover
Every step of this is in many threads here. It will be a very steep learning curve, but at the end you will understand a lot how sound waves work and how to build not just good, but VERY good speaker. Literally like or better than commercial 500-10k USD/EUR while using reasonable amounts of money.
In the screenshot is just the part of the wav files for each driver exported from VituixCAD impulse response and used in EQ APO. It mimics real crossover which you can listen before building physical one. There will be some differences of course, but simulations saves a lot of time and money for guesswork and prototypes.
Btw, for simulation you need any high quality soundcard and neutral amp. As someone already recommended TPA325x based 2 channel amp is a great choice, for soundcard I recommend Apple USB-C to 3,5mm dongle. It is the best souncard/DAC in the 1-100USD range by far. If you want to simulate/play 2 speakers at once, or more of them, with more separate drivers (4 channels total if stereo 2-ways, 6 channels total if stereo 3-ways and so on) - you need soundcard or audio interface with more channels, such as Xonar 7.1 and more amps.
That is provided data of the 15-30-45-60-etc of the flat big standard baffle, which you can see as close as in-wall results.
You still need full defraction-reflection data of the actual speaker box. VituixCAD does not sim full box, just front AND with some round over approximations. For full sims you need Comsol, Akabak or other similar soft.
Or just measure it.
That range is around the merging frequency and It is a bit critical in a two way. You can use a notch to reduce the peak but I would do It in second phase.
It is a pleasure to receive help from all around the world and it keeps me optimistic.
@raymondj Your points about reliability and repair-ability seem on point. I recently did a surface mount project with IC pins on a .5mm pitch. It was difficult and I think that's my limit. If the pitch gets smaller, the equipment needed to fix things goes out of my reach price-wise. The software will always be a pain e.g. fixing a bug in a program forces one to an OS upgrade, leads to other software no longer running, and a bunch of related scenarios.
I'm re-reading post #31. Thanks.
@svp thanks for the clarification. I'm checking out the TPA325x - not very expensive. The fairly small pipe going into my brain is a bottleneck for some of this stuff. For now my plan is "REW -> VituixCAD->..." and to measure actual values when the drivers arrive later this week.
@AllenB interesting. I used wine years ago - it was painful then but maybe not now.
@shadowplay62 thank you.
@raymondj Your points about reliability and repair-ability seem on point. I recently did a surface mount project with IC pins on a .5mm pitch. It was difficult and I think that's my limit. If the pitch gets smaller, the equipment needed to fix things goes out of my reach price-wise. The software will always be a pain e.g. fixing a bug in a program forces one to an OS upgrade, leads to other software no longer running, and a bunch of related scenarios.
I'm re-reading post #31. Thanks.
@svp thanks for the clarification. I'm checking out the TPA325x - not very expensive. The fairly small pipe going into my brain is a bottleneck for some of this stuff. For now my plan is "REW -> VituixCAD->..." and to measure actual values when the drivers arrive later this week.
@AllenB interesting. I used wine years ago - it was painful then but maybe not now.
@shadowplay62 thank you.
As promised, here is a library for making crossovers in LibrePCB. Unzip the file into LibrePCB's directory containing your libraries. On my MacOS the path is "~/LibrePCB-Workspace/data/libraries/local". Once you've done this it should be visible to LibrePCB's Workspace Library Manager. The usual disclaimer: it probably has mistakes, may not work, etc. The specifications of objects in the library were taken from online sources. I know for certain that some of the information was wrong and have tried to correct those known problems. That said, entering this information is time consuming, so I hope it benefits others.
Seet attached.
Seet attached.