Hi All,
Still experimenting with a pair of 2 ways and trying out some different things in DSP, in an active setup.
The creator of the below passive crossover for an Altec 14 has not been active online to respond and may have unfortunately passed or moved on to other things.
From chat forums, I do understand the following:
It crosses at 2Khz? (not sure if the woofer is running full range?)
There is a zobel network on the woofer audio path
The highs are attenuated xxx db vs the highs.
I'd be happy to put this in a simulator, but not sure where to start there as I'm a bit of a novice. Happy to provide the speaker specs, if that helps (ts parameters, impedance, etc.)
Was hoping to try to understand the slopes at the xover point, attenuation circuit and expected attenuation for the tweeter and what frequency ranges are getting to each driver. Also would be like to understand how the zobel network is impacting the frequency response and if that's needed in an active setup. Also assuming there's no EQ adjustment in the circuit, but not sure?
Still experimenting with a pair of 2 ways and trying out some different things in DSP, in an active setup.
The creator of the below passive crossover for an Altec 14 has not been active online to respond and may have unfortunately passed or moved on to other things.
From chat forums, I do understand the following:
It crosses at 2Khz? (not sure if the woofer is running full range?)
There is a zobel network on the woofer audio path
The highs are attenuated xxx db vs the highs.
I'd be happy to put this in a simulator, but not sure where to start there as I'm a bit of a novice. Happy to provide the speaker specs, if that helps (ts parameters, impedance, etc.)
Was hoping to try to understand the slopes at the xover point, attenuation circuit and expected attenuation for the tweeter and what frequency ranges are getting to each driver. Also would be like to understand how the zobel network is impacting the frequency response and if that's needed in an active setup. Also assuming there's no EQ adjustment in the circuit, but not sure?
Do you already using these drivers?
If not, you can simulate this easily enough. Using the factory impedance data gets you a more accurate simulation to show the filter action.
If not, you can simulate this easily enough. Using the factory impedance data gets you a more accurate simulation to show the filter action.
Hi AllenB, yep, I am currently using them now.
I do have the impedance information for the drivers from the factory data sheets.
Any suggestions on what software to use to simulate?
I do have the impedance information for the drivers from the factory data sheets.
Any suggestions on what software to use to simulate?
If you can measure the signal at the driver terminals using a standard sweep, you should have your answer. Is this an option?
Unfortunately, I don’t have the crossover itself, so cannot measure what it does.
I only have the schematic.
I only have the schematic.
Xsim and Vituixcad are commonly used crossover simulators. Some find Xsim easier to get in to. The result in this case will be the same for either.
If you can't trace the data sheet, post it and I'll do that.
If you can't trace the data sheet, post it and I'll do that.
Ps… here’s a version of the xover I found that was posted that appears to use some type of software simulator. The tweeter and woofer impedance/inductance are used in place of the speakers. Not sure if that helps.
Tweeter labeled 902-8a on the diagram and woofer Model 14 12” next to the resistor/ inductor series pairs representing them.
https://www.audiophilenirvana.com/wp-content/uploads/2016/05/model_14_final_w_zobel_hf.jpg
I’ll dig up and post what data sheets I can find.
Tweeter labeled 902-8a on the diagram and woofer Model 14 12” next to the resistor/ inductor series pairs representing them.
https://www.audiophilenirvana.com/wp-content/uploads/2016/05/model_14_final_w_zobel_hf.jpg
I’ll dig up and post what data sheets I can find.
Here's the impedance curve for the tweeter: https://www.diyaudio.com/community/attachments/mantaray-jpg.1229289/
I've not been able to locate the woofer info, but did find some info from the xover's author from woofer tester.
#34633 TS Parameters
Fs=22.4 Hz
Qes=.1569
Qms=6.8436
Qts=.1534
Zmax=261.65 ohms
Le=2.1394 mH
VAS=380 L (13.42 ft^3)
BL=15.8724 N/A
Sens=96.2 dB@1W/1M
Eff=2.63 %
Speaker brochure here
https://www.audiophilenirvana.com/wp-content/uploads/2016/05/Altec-Model-14-Full-Brochure.pdf
Tweeter brochure: https://www.audiophilenirvana.com/wp-content/uploads/2016/05/902-8B_hfdriver_spec_sheet.pdf
I've not been able to locate the woofer info, but did find some info from the xover's author from woofer tester.
#34633 TS Parameters
Fs=22.4 Hz
Qes=.1569
Qms=6.8436
Qts=.1534
Zmax=261.65 ohms
Le=2.1394 mH
VAS=380 L (13.42 ft^3)
BL=15.8724 N/A
Sens=96.2 dB@1W/1M
Eff=2.63 %
Speaker brochure here
https://www.audiophilenirvana.com/wp-content/uploads/2016/05/Altec-Model-14-Full-Brochure.pdf
Tweeter brochure: https://www.audiophilenirvana.com/wp-content/uploads/2016/05/902-8B_hfdriver_spec_sheet.pdf
Ah, a compression tweeter. The impedance varies with the horn you use which is stopping me from moving forward with producing the data files.
In your situation it might still be interesting to simulate using default drivers.. you have to start somewhere and I understand it's difficult to know how to move forward. DSP is an acceptable way to produce a no measure crossover.. otherwise you're likely to want measurements, either to simulate more accurately to reverse engineer, or to produce a crossover from scratch.
In your situation it might still be interesting to simulate using default drivers.. you have to start somewhere and I understand it's difficult to know how to move forward. DSP is an acceptable way to produce a no measure crossover.. otherwise you're likely to want measurements, either to simulate more accurately to reverse engineer, or to produce a crossover from scratch.
Thanks AllenB;
I've been playing with DSP to try things out and didn't like the 2k xo point at all with any levels of attenuation on the tweeter or EQ corrections.
I did finally try 6db slope crossovers and that's the ticket!
Landed on a 6db crossover with 15db of tweeter attenuation at 1.5Khz that sounds divine. (Also added a +9db low shelf up to ~165hz dropping off with a 6db slope to correct for the woofer dropoff, I'm assuming due to baffle step, driver fall off and room, etc)
Since they are first order slopes and my pre-amp has 2 sets of outputs (though one is cap coupled the other DC), I was thinking to try making these passive line level filters as follows:
For tweeter-- high pass 6db @ 1.5khz + 15 db attenuation/
For woofer -- low pass 6db @ 1.5hkz + 9db low shelf (Perhaps the 9db low shelf might be better in the pre-amps EQ loop to manage tr, but looking into that.)
Hoping to house them in the bodies of some XLR attenuators that I have to keep things simple.
I've been playing with DSP to try things out and didn't like the 2k xo point at all with any levels of attenuation on the tweeter or EQ corrections.
I did finally try 6db slope crossovers and that's the ticket!
Landed on a 6db crossover with 15db of tweeter attenuation at 1.5Khz that sounds divine. (Also added a +9db low shelf up to ~165hz dropping off with a 6db slope to correct for the woofer dropoff, I'm assuming due to baffle step, driver fall off and room, etc)
Since they are first order slopes and my pre-amp has 2 sets of outputs (though one is cap coupled the other DC), I was thinking to try making these passive line level filters as follows:
For tweeter-- high pass 6db @ 1.5khz + 15 db attenuation/
For woofer -- low pass 6db @ 1.5hkz + 9db low shelf (Perhaps the 9db low shelf might be better in the pre-amps EQ loop to manage tr, but looking into that.)
Hoping to house them in the bodies of some XLR attenuators that I have to keep things simple.
That could be done.
Second order filters may sound worse than first when they're wrong, but better than first when they're right. It may be that you need different filters for high and low. Different frequency, different Q factor, different order.. who knows.
Second order filters may sound worse than first when they're wrong, but better than first when they're right. It may be that you need different filters for high and low. Different frequency, different Q factor, different order.. who knows.
Thanks AllenB,
I do have to say, being able to adjust xo point & type, attenuation and EQ on the fly from my sitting position with an iPad makes it really easy with DSP. I did play with 12db and mixed 12db and 6db slopes, but I think my horn and woofer are happiest with 1st order. (12 or 24db on either side gives a significant suckout 1-3K). The DSP has a mic & embedded RTA tools, so I can measure & see results in real time, but I've been finding that tuning by ear/pink noise to be much faster than starting with a target FR curve for the EQ & level setting.
While the DSP sounds fantastic and I don't really' need' a passive analog xo, I want one, lol. Will do some research on the circuits and post a new thread on the endeavor. Plan is to start with/try a passive/single ended then balanced or skip ahead to active se/balanced if the desired results aren't achievable.
I do have to say, being able to adjust xo point & type, attenuation and EQ on the fly from my sitting position with an iPad makes it really easy with DSP. I did play with 12db and mixed 12db and 6db slopes, but I think my horn and woofer are happiest with 1st order. (12 or 24db on either side gives a significant suckout 1-3K). The DSP has a mic & embedded RTA tools, so I can measure & see results in real time, but I've been finding that tuning by ear/pink noise to be much faster than starting with a target FR curve for the EQ & level setting.
While the DSP sounds fantastic and I don't really' need' a passive analog xo, I want one, lol. Will do some research on the circuits and post a new thread on the endeavor. Plan is to start with/try a passive/single ended then balanced or skip ahead to active se/balanced if the desired results aren't achievable.
Last edited:
I can and reversing one side does improve the FR measurements slightly, but the sound gets a bit thin and hollow.
This is probably not what you should be listening for when determining a crossover and I'd advise against assuming the polarity reversal is the direct reason for it sounding that way.
It's difficult when you have multiple problems at once to deal with and I can't give quick answers on this one. The sound distribution throughout the room will be a large factor in contrast to the crossover blending itself, since reversing polarity can affect the direct/reflected ratio. The speaker needs to be designed for favourable acoustic characteristics to be able to be crossed properly, and it also needs to be adjusted to be balanced across all frequencies.
It's difficult when you have multiple problems at once to deal with and I can't give quick answers on this one. The sound distribution throughout the room will be a large factor in contrast to the crossover blending itself, since reversing polarity can affect the direct/reflected ratio. The speaker needs to be designed for favourable acoustic characteristics to be able to be crossed properly, and it also needs to be adjusted to be balanced across all frequencies.
The paper schematic of a passive crossover does not always represent what is really happening.
David Rich wrote:
“In the old days, there was a lot of trial and error in designing a crossover. So, adding more than a few components was futile, since it wasn’t practically possible to optimize them. Simplicity yielded better optimizations”.
“The individual crossover components usually don’t have functions in a way that can be isolated. A crossover is a filter network that implements a transfer function as a result of all of its parts. Sometimes it is possible to generalize and say, Oh, this resistor adds damping. Or, This network compensates for a resonance. But, really, the better a crossover design is, the more the parts work together, symbiotically”.
Dave Rich
David Rich wrote:
“In the old days, there was a lot of trial and error in designing a crossover. So, adding more than a few components was futile, since it wasn’t practically possible to optimize them. Simplicity yielded better optimizations”.
“The individual crossover components usually don’t have functions in a way that can be isolated. A crossover is a filter network that implements a transfer function as a result of all of its parts. Sometimes it is possible to generalize and say, Oh, this resistor adds damping. Or, This network compensates for a resonance. But, really, the better a crossover design is, the more the parts work together, symbiotically”.
Dave Rich
Thanks AllenB, certainly very unscientific on my part. I'm less looking to design a xo as I feel like I've landed on parameters I would like to try out in a purely analog method. I'm certain you are correct about the crossover being improved in 2nd order (and I'm sure active vs. pll), but I'm wanting to try the very simplest circuits for my 1st try at a pllxo but certainly willing to try other things in parallel.
However, certainly open to suggestions. I don't have DATS or anything like that to properly measure the drivers, but do have REW setup to confirm the DSPs RTA and get other information/graphs not available on the DSPs UI and used this extensively when trying to get the analog active 24db/4th order balanced boards I purchased from Xkitz dialed in. While I did mess with phase (at the speaker terminals and subwoofer I had allowed adjustment) I didn't try overlapping the xo point but perhaps that's something I can try. (It's a bit more involved as this prebuilt xover has pcb slots for xo points, level pots for outputs and a bsc pot only)
With 2nd order it's not the filters but the needs of the speakers. Some situations are more tight than others.
I'd advise you to get yourself an equaliser (even a software one). This way you can begin to address the balance and voicing needs by using it in the signal path before the cross. It will let you see whether you can fix the crossover region without addressing the crossover itself. When it isn't possible to get it to sound right, it shows you that the cross isn't right.
On the other hand it balances the rest of the spectrum for when you do work on the cross.
Primarily, the best you can do will be the same. Some methods make it easier to land on a good result when using a trial and error method.. it's the result that counts.
You can do perfect impedance measurements without it using REW.
I'd advise you to get yourself an equaliser (even a software one). This way you can begin to address the balance and voicing needs by using it in the signal path before the cross. It will let you see whether you can fix the crossover region without addressing the crossover itself. When it isn't possible to get it to sound right, it shows you that the cross isn't right.
On the other hand it balances the rest of the spectrum for when you do work on the cross.
Thanks again, great info.
I do have an EQ (Kenwood GE-7030 14 band or 3 band quasi parametric - frequency is one of the 14 bands and is either 'wide' or 'narrow') that I put in my pre-amps processor loop when trying to dial in the 4th order xkitz bouards and got to a happy place with the xo, until I decided to try the 6db slopes + Low boost shelf in DSP with the sub removed.
I got close to what I'm hearing now using the Kenwood EQ and the 4th order active analog xover boards, but doesn't sound as smooth as the 6db slopes. Did play with phase at the terminals which did help flattening things out, but didn't try overlap on the xo point.
I do have an EQ (Kenwood GE-7030 14 band or 3 band quasi parametric - frequency is one of the 14 bands and is either 'wide' or 'narrow') that I put in my pre-amps processor loop when trying to dial in the 4th order xkitz bouards and got to a happy place with the xo, until I decided to try the 6db slopes + Low boost shelf in DSP with the sub removed.
I got close to what I'm hearing now using the Kenwood EQ and the 4th order active analog xover boards, but doesn't sound as smooth as the 6db slopes. Did play with phase at the terminals which did help flattening things out, but didn't try overlap on the xo point.