Hi everyone! I am designing my first ever crossover for a 2.5way speaker using VituixCAD. I know these are all hypothetical and that real world response is affected by the room and baffle design, but I wanted to have a good starting point. I feel the frequency response looks reasonable as a good starting point, but wanted any feedback in a few areas that could improve its design:
Drivers being used: DA RST28F-4, and DA RS150-8 x2.
1. The response north of the crossover has multiple peaks and valleys with a spread of 3-4db. Is that a big deal? If so, is there a design that can flatten it out more?
2. The tweeter I am using seems to take a dive north of 17kHz. Is there a way to bump that up without simply raising the volume of the whole driver?
3. I integrated a highpass filter for the two woofers because I intend on integrating a subwoofer crossing over at around 80hz. Is this necessary? Or should I simply cross the sub and speaker using the avr/amp that will be used?
4. There is some sizable cone break up for the woofer past the crossover starting at 3300hz. The overall response is relatively flat however. Should I try a high order crossover to try and eliminate it more?
5. Lastly, I am over thinking this a bit too much? Is it reasonable to use this as a starting point and make adjustments later with further modifications/simply using a DSP to smooth things out?
Looking forward to everyone's thoughts!
Drivers being used: DA RST28F-4, and DA RS150-8 x2.
1. The response north of the crossover has multiple peaks and valleys with a spread of 3-4db. Is that a big deal? If so, is there a design that can flatten it out more?
2. The tweeter I am using seems to take a dive north of 17kHz. Is there a way to bump that up without simply raising the volume of the whole driver?
3. I integrated a highpass filter for the two woofers because I intend on integrating a subwoofer crossing over at around 80hz. Is this necessary? Or should I simply cross the sub and speaker using the avr/amp that will be used?
4. There is some sizable cone break up for the woofer past the crossover starting at 3300hz. The overall response is relatively flat however. Should I try a high order crossover to try and eliminate it more?
5. Lastly, I am over thinking this a bit too much? Is it reasonable to use this as a starting point and make adjustments later with further modifications/simply using a DSP to smooth things out?
Looking forward to everyone's thoughts!
It's either reflections from improperly gated measurements, as Allen suggested, or diffraction from multiple edges. In the first case it's merely an artifact and could be ignored, in the second peaks and dips are different for different points in space and cannot be corrected with a filter. You better off rounding the edges of the enclosure or putting tweeter in a waveguide. Or both.
Two woofers shring a single low-pass is a recipe for hideous vertical directivity. Better switch to 2.5-way. Not sure about integrated subwoofer crossover, as typical 80 Hz crossover may not be optimal in every situation. OTOH you get clean 80 Hz 24 db/oct high-pass, dispensing with the DSP channel for subwoofer.
It seems like every 'reference' driver must come in package with 10 dB breakup modes and surround resonance nowadays.
Ideally resonances should be 24 dB below the reference level, but two notches for 3.3 and 6.5 kHz peaks add cost and coplexity to the crossover. Broad hill at 1.3-2.3 kHz may be even more audible, as it sits in the most sensitive area. Small dip at 1.2 khz may be fixed by treating the surround with a 2-3 ml of vinyl acetate dissolved in acetone.
No, it better to have solid crossover before DSP optimization. Speaker should have good directivity, similar on- and off-axis response. Otherwise by improving on-axis you would hurt off-axis and acoustic power.
What's the purpose of 3,3uF cap across the tweeter?
That will probably give very low impedance at high frequencies. Be careful not to damage your amp. Add some resistor in series with that cap if you need to compensate something.
To answer your questions:
1. This might be reflections on the baffle edges. Rounded or chamfered edges should help.
2. I'am not sure I understand this, but if that's natural rollof of the tweeter than I think no.
You'll need active boost to fill it up.
3. I would avoid passive filters that low.
Use suitable closed box to achieve natural rollof. Or active cut.
4. What boders me are those peaks around 6,5 and 9,5 kHz. Graph shows only one midbass response. You have two in parallel, so they will be 6dB stronger. That gives around 11 and 15 dB attenuation of the peaks respectively.
I'am affraid that won't be enough to avoid HF smearing and its worth to try 3rd order filter. That also gives better protection of the tweeter, so you could also try a bit lower x-over point as well to further increase attenuation of those peaks.
And btw, x-over shown is not 2,5 way. With 2,5 way, one of midbasses usually have HF cut between 100-300 Hz.
5. Better think twice than sorry. Better starting point, better final result.
That will probably give very low impedance at high frequencies. Be careful not to damage your amp. Add some resistor in series with that cap if you need to compensate something.
To answer your questions:
1. This might be reflections on the baffle edges. Rounded or chamfered edges should help.
2. I'am not sure I understand this, but if that's natural rollof of the tweeter than I think no.
You'll need active boost to fill it up.
3. I would avoid passive filters that low.
Use suitable closed box to achieve natural rollof. Or active cut.
4. What boders me are those peaks around 6,5 and 9,5 kHz. Graph shows only one midbass response. You have two in parallel, so they will be 6dB stronger. That gives around 11 and 15 dB attenuation of the peaks respectively.
I'am affraid that won't be enough to avoid HF smearing and its worth to try 3rd order filter. That also gives better protection of the tweeter, so you could also try a bit lower x-over point as well to further increase attenuation of those peaks.
And btw, x-over shown is not 2,5 way. With 2,5 way, one of midbasses usually have HF cut between 100-300 Hz.
5. Better think twice than sorry. Better starting point, better final result.
You have chosen to work with a little more difficult cone material so naturally you have to invest more work on these to behave the way they should.
Ok - so you haven't measured any of this? just manufacturer files?
First thing to address is net system sensitivity. you have 2 88dB woofers in parallel netting you 94dB. Then you need to allow for baffle step. Assuming you target ~4dB losses, your response target should be ~ 90dB.
I'm not sure the source of your frequency response files. but it's hard to know with a 101.5dB system sensitivity. The lead 0.39mH inductor is rather small to account for bafflestep - even with a 4 ohm load in the woofer circuit.
The reason I start with system sensitivity, is your woofer circuit will change quite a bit when allowing for baffle step. this impacts the rest of the crossover. So I typically work "woofer up" in design.
First thing to address is net system sensitivity. you have 2 88dB woofers in parallel netting you 94dB. Then you need to allow for baffle step. Assuming you target ~4dB losses, your response target should be ~ 90dB.
I'm not sure the source of your frequency response files. but it's hard to know with a 101.5dB system sensitivity. The lead 0.39mH inductor is rather small to account for bafflestep - even with a 4 ohm load in the woofer circuit.
The reason I start with system sensitivity, is your woofer circuit will change quite a bit when allowing for baffle step. this impacts the rest of the crossover. So I typically work "woofer up" in design.
Definitaly planning on rounding off the edges and I have an offset designed into the baffle. I also am trying to make the baffle as narrow as possible, which to my understanding, reduces diffraction for the woofers (which I know is oversimplified), but I may be miss understanding it. Heres an image from sketch on what my plans look like for the baffle which is 7inches wide including the width of the round off.
I am not sure what the acronym OTOH means. Care to explain? But understood, having a low pass probably doesn't make sense in this scenario.
I did do an updated design, which I will include below, that has the first break up 26dB below the reference which I hope helps. I also was able to flatten that hill which hopefully will help.
I really appreciate your input!
I originally had it in there because I found it served the same purpose to attenuate the tweeter while making the crossover slightly better. However I tweeked my design a bit and eliminated it and used 3 resistors in parallel to attenuate instead. Ill post the redesign below.
The graph does include both woofers and they are connected and sum to the system SPL shown. They are identical drivers and crossover the same so they overlap exactly. I am sorry for the confusion. I guess this would not be considered a 2.5 way design. I did however try playing around with 3rd order for quite a while but never could get it to improve much over my 2nd order and don't believe it would be worth the extra cost for me.
Thank you for your input!
No, these are manufacturer FRD and ZMA which I know are not ideal since it is not in real world conditions. My goal to to get a solid starting point with a crossover and do measurements once I have everything together and make adjustments from there. Id rather get 90% there before I build it all so Idon;t have to much added expense with unused crossover components.
I am not sure I fully understand you here. With my current crossover I am seeing the summation of the less sensitive drivers and the mild attenuation of the tweeter to combine to similar output levels. Are you suggesting I need to attenuate the tweeter even more in anticipation of a potential 4db loss from the baffle?
Below is my updated design. I tried playing around with 3rd order and notch filters for a while but felt they simply didn't show significant improvement to justify the increased costs. However, after everyone's suggestions, I did make some changes and tried to flatten the midbass hill and added each woofer to its own Lowpass. I also attenuated the tweeters with 3 in series resistors which have reasonable dissipation if driven with 140watts into 8ohms. So with all that said I have a few questions if you are all willing to humor me further:
1. Am I crazy in thinking this will be a 2 ohm load for the amp??? The tweeter is 4ohms, and each woofer is 8ohms. I believe the two woofers combined will be 4ohm load, but then combined in my current configuration with the tweeter the load at the amp will be 2ohms... Which is not good... I would like to wire this to be 8ohms for the amp. But when I try to wire it the way I think would give my 8 ohms, the whole spl graph goes on the fritz... How should I be wiring this to make the woofer system and tweeter into series to make it 8 ohms?
2. Any other adjustments or suggestions on this design to improve it further?
1. Am I crazy in thinking this will be a 2 ohm load for the amp??? The tweeter is 4ohms, and each woofer is 8ohms. I believe the two woofers combined will be 4ohm load, but then combined in my current configuration with the tweeter the load at the amp will be 2ohms... Which is not good... I would like to wire this to be 8ohms for the amp. But when I try to wire it the way I think would give my 8 ohms, the whole spl graph goes on the fritz... How should I be wiring this to make the woofer system and tweeter into series to make it 8 ohms?
2. Any other adjustments or suggestions on this design to improve it further?
You can't get a solid starting point, if your response files don't accurately reflect measurement conditions.
Credit goes to Scott Hinson for this graph but it illustrates the point with an unnammed Dayton driver, but the point is still valid
Scott has lots of very helpful papers here on a variety of topics - I recommend you read them:
https://docs.google.com/spreadsheet...-1cl8lSW_qwzWUA-XJyaNdMTg2xq-b0d4Czrwiw#gid=0
The Dayton spec sheet shows a nice flat response for this driver. it looks around 88dB sensitive (red curve) and had more bass than you'll get in a real cabinet (red infinite baffle bass response vs. green real world in cabinet bass response).
After you allow for baffle step rise (100Hz to 1KHz) you end up with an 84dB sensitive speaker (blue/purple curve) - not the inflated dayton infinite baffle sensitivity of 88dB.
In your design - I don't know what you did to get a 101.5dB FR curves... if you used Dayton FRD files - I would have expected both woofers to sum to 94dB.
If we translate the red curve to blue/purple below... to your design. Then your 101.5dB should be about 97dB across the board.
So yes - your woofers don't have enough baffle step compensation applied and your tweeter is too hot.
Credit goes to Scott Hinson for this graph but it illustrates the point with an unnammed Dayton driver, but the point is still valid
Scott has lots of very helpful papers here on a variety of topics - I recommend you read them:
https://docs.google.com/spreadsheet...-1cl8lSW_qwzWUA-XJyaNdMTg2xq-b0d4Czrwiw#gid=0
The Dayton spec sheet shows a nice flat response for this driver. it looks around 88dB sensitive (red curve) and had more bass than you'll get in a real cabinet (red infinite baffle bass response vs. green real world in cabinet bass response).
After you allow for baffle step rise (100Hz to 1KHz) you end up with an 84dB sensitive speaker (blue/purple curve) - not the inflated dayton infinite baffle sensitivity of 88dB.
In your design - I don't know what you did to get a 101.5dB FR curves... if you used Dayton FRD files - I would have expected both woofers to sum to 94dB.
If we translate the red curve to blue/purple below... to your design. Then your 101.5dB should be about 97dB across the board.
So yes - your woofers don't have enough baffle step compensation applied and your tweeter is too hot.
your 90% there crossover, I'd wager might be as little as "10% there" with bad data (manufacturer curves without any diffraction ripple, baffle step and LFE losses applied). You can simulate all of these to get closer... but I still wouldn't order parts without measuring to confirm.
Getting the bass and midrange response working well together and in balance with the treble is the hard part. anyone can "pad down a hot tweeter" but it won't do you any favours in the part that really makes or breaks a great speaker IMHO. #1 the midrange and #2 its integration / balance with the bass
Getting the bass and midrange response working well together and in balance with the treble is the hard part. anyone can "pad down a hot tweeter" but it won't do you any favours in the part that really makes or breaks a great speaker IMHO. #1 the midrange and #2 its integration / balance with the bass
Thanks! Dude's been keeping busy since his college? days of running the late, lamented BassList.
I did a calculation of the 4pi sensitivity of the woofer here. It’s a little below 80 dB. Two woofers in parallel gets you +6 dB. I’d suggest you aim for a system sensitivity around 86 dB as a starting point…..
Hi Dave, I really appreciate your input here. I have been learning a lot, but still have a ways to go. As far as the 101.5db FR curve, I had forgotten I had raised the voltage on the amp in the program a while ago and paid little to no attention to the height of the spl and was only paying attention to the "flattness" of the curve. After lowering the amp to the standard 2.83v you are correct and it levels off close to 95db for the woofers.
I am beginning to see that now... I guess it was a fools errand to attempt using factory measurements to get a starting point. The graph you shared is pretty eye opening. I of course knew baffle step, diffraction etc made a difference, but seeing that, shows me I can't predict it correctly. I will definitely have to order the drivers and at least make the baffle and take some measurements first before I spend the money on crossovers.
Anyway - you're not getting anywhere near 95 dB from the woofers in the bass/midbass. It will be more like 86-88 dB, because you will have no baffle loading like in the midrange - aka "baffle step".
Final impedance will be 4ohm.
Two woofers in parallel gives 4ohm, but x-over separates them from tweeter so they won't combine into 2ohm due different working ranges. Assuming, of course x-over is done right.
Your updated design (blue line) shows smoother response/rolloff of the midbass and much better attenuation of cone resonance peaks Good job.
Alot of talk about baffle step compensation here. But, no one tells that proximity of room boundaries plus room modal resonances can give significant boost to bass and low-midrange, all the way to few-hundred hertz.
Full BSC gives flat response in free-space/anechoic conditions.
Put such speakers in average listening room and you'll very likely get too much bass/low-midrange. Especially in smaller rooms with speakers close to hard (brick or concrete) walls.
So its a kind of double edged sword.
Since you going to use DSP for final response pollishing, I suggest you to use DSP for BSC as well. You'll have simpler x-over and won't loose speakers overal efficiency.