I've been lurking around on DIY audio forums for the last year or so and built a dual TangBand 5.25" sub that I've very pleased with, but I now want to build a set of my own 2-way bookshelf speakers as my first 'real' speaker build.
My goals are as follows:
- Under $100 for all of the drivers
- Dig down to an F3 in the 30's without sacrificing high end 'sparkle' and mid range clarity
- Properly good SQ ("as good as, or better than" my Infinity Primus 6.5" bookshelves), but I know it won't rival $1,000 builds
- Excuse to buy more tools 😀
I had spec'ed out 2 drivers that looked like their responses would play nice together:
Dayton DC160 6.5" 4-ohm
Dayton DC28F 1 1/8" 8-ohm
Obviously, I've noticed that I'm mixing up my ohms with a 4 ohm on the bottom and an 8 ohm on the top. Dayton does make a 6.5" 8-ohm version of the DC160, but the params are completely different and, according to my WinISD modeling, I'm losing a solid 10Hz on the bottom end with the 8 ohm version (f3 around 45Hz where the 4 ohm has an f3 around 35Hz), which is very significant to me. After specing it out, I just happened to find a kit from PE that uses these drivers, but with the 8 ohm versions. I also found a crossover mod for the PE kit, but I'd really like to use the 4 ohm version if possible because of its extension and sensitivity benefits.
So far, my weak point is definitely the crossover and I know switching woofer ohms generally requires a full rework of the crossover. I downloaded Passive XO Designer just yesterday and am trying my hand at it, but I haven't made any substantial progress just yet. I don't mind my speakers being 6 ohm if that's what it comes out to, but if anyone has some pro tips, I thought I'd post up here and take them into consideration.
Perhaps this is a silly question, but is there any chance I could "simply" throw in a resistor or a bit of circuitry on the woofer to boost its ohms up to 8? Perhaps that would let me use the modified crossover design with the 4 ohm driver and not sacrifice the 10 Hz down low? I know I 'could' rig up 2 4 0hm drivers in series to get 8, but that messes with the sensitivity of the lows, which would require an XO rework as well, so that's no bueno.
My goals are as follows:
- Under $100 for all of the drivers
- Dig down to an F3 in the 30's without sacrificing high end 'sparkle' and mid range clarity
- Properly good SQ ("as good as, or better than" my Infinity Primus 6.5" bookshelves), but I know it won't rival $1,000 builds
- Excuse to buy more tools 😀
I had spec'ed out 2 drivers that looked like their responses would play nice together:
Dayton DC160 6.5" 4-ohm
Dayton DC28F 1 1/8" 8-ohm
Obviously, I've noticed that I'm mixing up my ohms with a 4 ohm on the bottom and an 8 ohm on the top. Dayton does make a 6.5" 8-ohm version of the DC160, but the params are completely different and, according to my WinISD modeling, I'm losing a solid 10Hz on the bottom end with the 8 ohm version (f3 around 45Hz where the 4 ohm has an f3 around 35Hz), which is very significant to me. After specing it out, I just happened to find a kit from PE that uses these drivers, but with the 8 ohm versions. I also found a crossover mod for the PE kit, but I'd really like to use the 4 ohm version if possible because of its extension and sensitivity benefits.
So far, my weak point is definitely the crossover and I know switching woofer ohms generally requires a full rework of the crossover. I downloaded Passive XO Designer just yesterday and am trying my hand at it, but I haven't made any substantial progress just yet. I don't mind my speakers being 6 ohm if that's what it comes out to, but if anyone has some pro tips, I thought I'd post up here and take them into consideration.
Perhaps this is a silly question, but is there any chance I could "simply" throw in a resistor or a bit of circuitry on the woofer to boost its ohms up to 8? Perhaps that would let me use the modified crossover design with the 4 ohm driver and not sacrifice the 10 Hz down low? I know I 'could' rig up 2 4 0hm drivers in series to get 8, but that messes with the sensitivity of the lows, which would require an XO rework as well, so that's no bueno.
Last edited:
After a bit of research, it looks like even if you add a 4-ohm resistor to the woofer, it messes with the SQ and I'd be better off making a new XO for it.
just re-calculate the woofer crossover component values using the woofer's impedance at the crossover frequency. Despite what some say, an on-line calculator is OK, if you know how to use it: Crossover Calculators : Custom Car Stereo - Complete Car Audio Building Guide
You may also need to tweak the tweeter attenuating resistors
You may also need to tweak the tweeter attenuating resistors
Oh, that's super nifty for a n00b although it seems more of a boxed XO rather than what some would consider a 'proper' XO. That's actually round about what I already got piddling around with Passive XO today as well, so I must be on the right track. Although I'm still trying to figure out how to pull up a total response graph in Passive XO. I'm also still trying to figure out if the sensitivities are taken into account properly.
Last edited:
Anyone else have some pro tips? If nothing else, I'll just do a 'basic' crossover, measure the speaker response and tweak the XO again later on...
Start playing with PCD.
Use REW and calibrated mic. Set up speaker towards middle of room so tweeter is about between floor and ceiling. Aim speaker 45 deg from walls (or along diagonal of room). Get an 8ohm resistor and drive it with sine gen through your amp from REW. Use voltmeter to find what setting on amp makes (at say 1kHz) 2.83v AC. Record that number. Reduce REW drive signal by -6dB. Measure again and you should have 0.71v if all is working well. Now remember that dBFS setting in REW.
Set up mic exactly 50.0 cm from tweeter bezel to mic tip and at same height.
Hook up the DC28 tweeter straight to amp and do a sweep in REW full range 10Hz to 20kHZ at a setting that was 0.71v above. Then without touching the mic or speaker position, switch amp wires to woofer only. Take another full range sweep. Don't worry, this low signal won't hurt your tweeter. Then, also very important without touching mic or speaker placement - connect woofer and tweeter in parallel. Then take final sweep.
Now in REW you have the three most critical measurments needed to fully simulate your xo and apeaker in PCD or Xsim.
Assuming Dayton had the ZMA files you are basically set to start modeling.
You need to convert your REW measurement to minimum phase before it can be used in PCD. Go to export impulse as wav (select minimum phase). Do this for each file.
Now close all files in REW and import impulses that you just saved. They are now minimum phase data and export this as a txt file .FRD.
Load all 3 FRD's info PCD. Set the parallel response as the target curve so you can turn it on and off. That curve is an acoustical interferogram with dips and peaks that you can now try to simulate using the tweeter and woofer only responses. Do this by adjusting the driver Z offset on tweeter until the simulated summed response matches exactly the target that you measured. You have now confirmed the acoustic centers offset (typically around 25mm). And you have confirmed that PCD can simulate these drivers in this baffle perfectly.
Next step is start on your xo design. Set your target response (LR2, BW2 etc) the level and frequency of the XO.
The SPL levels in the sim have been scaled by -6dB to match 2.83v at 1m and measured at 0.5m to reduce floor bounce cancellation.
That should get you started.
Use REW and calibrated mic. Set up speaker towards middle of room so tweeter is about between floor and ceiling. Aim speaker 45 deg from walls (or along diagonal of room). Get an 8ohm resistor and drive it with sine gen through your amp from REW. Use voltmeter to find what setting on amp makes (at say 1kHz) 2.83v AC. Record that number. Reduce REW drive signal by -6dB. Measure again and you should have 0.71v if all is working well. Now remember that dBFS setting in REW.
Set up mic exactly 50.0 cm from tweeter bezel to mic tip and at same height.
Hook up the DC28 tweeter straight to amp and do a sweep in REW full range 10Hz to 20kHZ at a setting that was 0.71v above. Then without touching the mic or speaker position, switch amp wires to woofer only. Take another full range sweep. Don't worry, this low signal won't hurt your tweeter. Then, also very important without touching mic or speaker placement - connect woofer and tweeter in parallel. Then take final sweep.
Now in REW you have the three most critical measurments needed to fully simulate your xo and apeaker in PCD or Xsim.
Assuming Dayton had the ZMA files you are basically set to start modeling.
You need to convert your REW measurement to minimum phase before it can be used in PCD. Go to export impulse as wav (select minimum phase). Do this for each file.
Now close all files in REW and import impulses that you just saved. They are now minimum phase data and export this as a txt file .FRD.
Load all 3 FRD's info PCD. Set the parallel response as the target curve so you can turn it on and off. That curve is an acoustical interferogram with dips and peaks that you can now try to simulate using the tweeter and woofer only responses. Do this by adjusting the driver Z offset on tweeter until the simulated summed response matches exactly the target that you measured. You have now confirmed the acoustic centers offset (typically around 25mm). And you have confirmed that PCD can simulate these drivers in this baffle perfectly.
Next step is start on your xo design. Set your target response (LR2, BW2 etc) the level and frequency of the XO.
The SPL levels in the sim have been scaled by -6dB to match 2.83v at 1m and measured at 0.5m to reduce floor bounce cancellation.
That should get you started.
If these are all dayton's, you can even go high tech. PE has the FRD and impedance charts for all the Dayton drivers. Grab them and import to XSim to do pretty accurate crossover design. It's better to get the FRD directly from the cabinet you build though.
NO reason at all to add a 4 ohm resistor to match them! Plus you'd have to pad the tweeter down even further. Ugh.
Erik
Ooo!!! XSim is super nifty! I've just piddled around with it a little bit here at work, but it's looking like it'll be super valuable for XO stuff. Thanks!
Last edited:
Here's what I've got for an XO so far using XSim. I think it's looking rather decent, but I don't know if there are any 'cardinal rules' of XO design that I may be breaking. I'd like to remove that little dip in the 500Hz range, but I'm not sure how to do that just yet. Any useful critiques are welcome!
Drawing -----------------
Use the "Ground" component to tie all the ground going wires, this will clean up your schematic. Don't forget to put a ground at the (-) side of the amp as well as each driver.
Tweeter -------------------
Put the resistor on your tweeter right next to your tweeter. This will save you money in caps. 🙂 You'll need less uF to cross an 18 ohm tweeter than an 8 ohm tweeter, but also consider using an L-pad instead. It's easier to adjust, and you can make changes without re-calibrating all of your filter values. Often crossover developers will make the tweeters 8 Ohms (effectively), but 8 to 12 ohms effectively is a good range. The higher the effective range, the larger caps and more $$$ you'll have to spend. Doesn't matter so much with low end caps, but when you are spending $10/uF it adds up quickly. 🙂 Don't attempt to make your effective resistance smaller than your actual tweeter resistance with an L-PAD. You'll waste a lot of amp power that way.
XSim has circuit blocks for L Pad's and low and high pass filters, but for some reason they stop at 2nd order, which is kind of annoying.
Best,
Erik
Last edited:
Before simulating a XO have you adjusted the driver acoustic offsets to get the simulated woofer and tweeter to match the measured woofer tweeter parallel measurement?
You need to set that to get the phase right moving forward. Otherwise you can design a xo but it won't be right once you build it because the sim did not account for the offset - which is a phase shift.
You need to set that to get the phase right moving forward. Otherwise you can design a xo but it won't be right once you build it because the sim did not account for the offset - which is a phase shift.
Further notes:
For each filter section, from left to right, always start with the series components first. For the tweeter, that's a cap first, then the inductor to ground. For the woofer, it's the inductor first, then the cap to ground. In your case the woofer section was correct.
This improves the minimal impedance you'll see.
Use the built-in CircuitBlocks for notch filters. It let's you dynamically adjust the various parameters of frequency, depth and width. Again, I would put them closest to the driver, but before any L pad.
On the filter parallel components (inductors on HP, caps on LP) there is a 0.5 to 1 ohm resistor in series. This again prevents the minimum impedance from getting too low but also may help tune the overall phase response.
When you've got all of this down, you'll need to worry about phase matching. 🙂 That's a good time to consider adding a Zobel network to the bass driver. Not only may it help your FR, but it can really help fix up the phase matching angle.
Best,
Erik
XRK is right of course. 🙂
So usually the correct order is to put the drivers in the cabinets, measure and then design the crossover. This will also ensure any baffle step compensation needed is taken into account in your crossover design.
You can get a rough estimate of the woofer offset by subtracting the depth of the tweeter from the depth of the woofer. This is from the back of the face plate to the back of the magnet. This will be very rough, but plugging this number into your woofer you'll have some idea of what to expect and is a good starting point when attempting to set them correctly after measurement.
Best,
Erik
Thanks for all the pro tips! I'm re-working my schematic a little bit and I've got the tweeter section done (I think) and am moving on to the woofer.
I know to do it "properly", I'd need to build the box, drop them in the box, measure it, fiddle with stuff, etc, but unless it really makes a significant difference, I'd rather skip that bit because I don't have any proper measuring equipment and I'd rather keep it relatively simple being my first 'real' speaker and all. If I can get good, but perhaps not astounding results without doing all that, I'd be ok with that for my first build. I can always redo the XO later on if I decide go that route too.
Something I've not completely understood is the phase shift. I understand it's essentially just sliding the sine wave a bit, but I don't understand why it shifts, how to fix it, and what the phase target is unless it's just trying to match the tweet and woofer.
I know to do it "properly", I'd need to build the box, drop them in the box, measure it, fiddle with stuff, etc, but unless it really makes a significant difference, I'd rather skip that bit because I don't have any proper measuring equipment and I'd rather keep it relatively simple being my first 'real' speaker and all. If I can get good, but perhaps not astounding results without doing all that, I'd be ok with that for my first build. I can always redo the XO later on if I decide go that route too.
Something I've not completely understood is the phase shift. I understand it's essentially just sliding the sine wave a bit, but I don't understand why it shifts, how to fix it, and what the phase target is unless it's just trying to match the tweet and woofer.
You are welcome.
Gads! Noooooo... :0) If you have a cheap mic, and a sound card you can get the distance yourself. Any free measurement software like REW will help you get this to within a millimeter. 🙂 Otherwise you'll end up completely redesigning your crossover later.
At the very least, estimate your woofer offset from specs.
Try this. Look at the Dayton drivers specs and estimate the woofer offset. Plug that offset into XSim and look at your FR again. Did it get ugly?
Mismatched phase angles and distances can lead to serious comb filtering effects in your FR, very unpleasant and eliminate the hard work you put into your crossover design. If you are going that route, just buy any pre-built crossover. If you are going through the trouble of crossover design you'll want to keep this in mind.
The test for a good phase match is to invert your tweeter from normal. You should see a deep and symmetrical valley appear. You can do this in XSim fairly easily.
What causes phase shift? Differences in driver distances, even as you walk around the speaker, crossovers and the amplitude changes inherent in the drivers themselves.
Best,
Erik
Right on! Lots of good info in here that I will definitely reference!
What about mic coloring though? Different mics have different frequency responses, so wouldn't that affect the modeling?
So, basically, the only way to properly design a XO is to build the whole speaker except the XO, measure it all, then whip up the XO as the very last thing rather than spec it all out on paper/cad and slam it together.
What about mic coloring though? Different mics have different frequency responses, so wouldn't that affect the modeling?
So, basically, the only way to properly design a XO is to build the whole speaker except the XO, measure it all, then whip up the XO as the very last thing rather than spec it all out on paper/cad and slam it together.
Unnecessary for interferometry (measuring distance by how much one driver interfere's with another). So long as your three measurements:
- Tweeter
- Woofer
- Tweet + Woofer
are all done with the same microphone AND your simulation uses the same Tweeter and Woofer data then the frequency response of the measurement microphone is irrelevant.
Well, I wouldn't say the ONLY way, but it is my personal preference. Some do complete designs from cabinets to crossovers first, or use baffle step compensation calculators to guide them in addition to the other data.
I find this method most quick, accurate and with least amount of complications and errors compared to the fully simulated or paper-only designs.
Either way, knowing the tweeter to woofer offset is key no matter what method you use. Also, the FRD files from Dayton won't know how your cabinet behaves in terms of baffle step and exact Q. Measuring your woofer in place is key here.
Best,
Erik
Oh, I see what's going on with the REW stuff. It's basically just plotting how the drivers "mess with" each other compared to their normal response, then taking that into consideration for the XO design. Got it.
Well, thanks a bunch for all the tips and tricks! I've learned more in this thread than about 2 years over on the AVS forum and trying to figure out all that technical documentation.
I'm going to go ahead and order some drivers (although I'll probably spring for 4 DC160's as I may decide to go with a WTW setup to reinforce the lower end a bit). I'll post back as things evolve!
Well, thanks a bunch for all the tips and tricks! I've learned more in this thread than about 2 years over on the AVS forum and trying to figure out all that technical documentation.
I'm going to go ahead and order some drivers (although I'll probably spring for 4 DC160's as I may decide to go with a WTW setup to reinforce the lower end a bit). I'll post back as things evolve!
If you are going that route, do 4th order Linkwitz-Riley filters, they have the best off-vertical axis response. Remember though that the order is additive. The driver + the crossover should be 4th order.
One neat trick with XSim is to use an ideal driver, crossed over this way, and then match your actual driver to it.
There are other free tools out there which help you by designing a crossover to match an arbitrary shape, not sure what they are though.
Best,
Erik
- Status
- Not open for further replies.