Please critique this crossover

[DeZZar]

Also, not a good idea to add a series resistor to the woofer.

Not familiar with what software this is, but don't some of them show the inductors resistance as a separate resistor?

[rpb]

Not familiar with what software this is, but don't some of them show the inductors resistance as a separate resistor?

I must have misread something. I thought he said he added a resistor. There is a place for the coil resistance. I just read something wrong.

Edit... Mentioned the resistor in post #8.

[DeZZar]

Sorry missed that. Just not expecting to see people adding a 0.1ohm resistor on a woofer - seemed more like the resistance of the inductor...!

[Chris Roemer]

OP, in your 1st schemo., you SAID you had the 8n(ohm) woofers in "series", but you really had them in parallel (by your sim, as well as shown in your schemo.).
You can tell they were modeled in parallel, otherwise their (combined) sensitivity would be -6dB lower.

Also on your 1st schemo., the resistor (shown) next to each coil is where you put in the "DCR" of each coil. You show "0.1" ohms in each position. That's not an unreasonable value for the 0.18mH (woofer) coil (although 0.2n would be more realistic), but the (tweeter) 0.47mH shunt coil is more likely around 0.5n (although it won't make MUCH difference in the shunt position). PE lists the DCR value for all the coils they sell.

Also (on your 1st schemo) if you toggle your woofers' series coil between 0.18mH and 0.0mH (essentially = to NO coil at all) you'll see that even a value of 0.18mH has virtually NO effect on the woofers' output (not even a reduction of -2dB near the Fc around 3.5kHz). It's pretty common for a 1st order filter to be unworkable (in reality, for woofers or tweeters), unless the woofer has an extremely low "Le" value (meaning its impedance rises very little with an increase in freq.). Your paralleled pair of 8n woofers run around 4n near 250Hz, but they rise to nearly 30n @ 10kHz. To make something "happen" (to make them roll off at least somewhat) you can (in addition to making the coil significantly larger - as has been suggested) either go 2nd order (by adding a shunt cap - to groound) OR you can add a "Zobel" (like thingy - which is a series cap and resistor) to ground (just ahead of the woofer - pair, in this case).

If I can make a suggestion, go back to your 1st schemo and:
Add a (shunt) Zobel between the coil and the woofers, using a 5n(ohm) resistor and a 33uF (cheap "npe") cap. Also (more than) double your coil to 0.40mH (and use a DCR value of 0.4n).
On the tweeter, double the padding resistor from 4n to 8n (this will cut the tweeter's output by approx. -5dB - which will help give you BSC), and change the 0.47mH coil down to 0.30mH (w/a DCR of 0.3n).

This will give you a "system" that runs around 84dB w/nearly "full baffle-step" (about +5dB here) with the tweeter crossing near 3k. I think this would make for a pretty nice MTM (keep the drivers as close as you can (reasonably) get them) in your 0.22cf vented box - as you designed it. Another suggestion would be to use a single port. You COULD use a 1-1/2" id port that's about 5" long, but to help avoid chuffing (port noise), try to fit in a 2"id port - that's 10" long. You can use an elbow it you need to to fit it in there. If you DO end up w/a little port noise, you can reduce the odds of hearing it by exiting the port out the rear of the box (instead of the front). If the 8n version of these drivers is much like their 4n "brother" (and I think they are), you'll probably end up w/pretty good bass down to 40Hz (even though the box sim shows an F3 closer to 50). I don't think you'll need a sub w/these.

Your XO sim will show you your BSC, it'll look like about a +5dB rise in (woofer) output going from about 1200Hz down to about 200. I'd think you'd be quite blown away by this system, esp. given the small investment in drivers and XO parts. (post your reworked XO here and we'll double check it)

[billfitzmaurice]

don't some of them show the inductors resistance as a separate resistor?

Yes, some do. You'd insert the coil DCR there. Others even show the cap ESR as a separate resistor. You can do the same with XSim if you wish.

[DeZZar]

Yes, some do. You'd insert the coil DCR there. Others even show the cap ESR as a separate resistor. You can do the same with XSim if you wish.

Yeah thought so. xsim and vituix both accommodate them. I just keep seeing this style diagram around and was wondering what software it was.

[Jicafold]

Yes the woofers are parallel, not series. This will be the center channel for a 5.1 system. However, if things work well I might wish to do identical left and right front channels

I’m just getting started in this hobby and a lot of this information is way over my head, particularly response number 18. I’m not nearly at that level yet. I have only had a couple weeks of reading on this in my spare time. I don’t even know what all those acronyms mean let alone the theory behind it, so I don't have the ability to make any of those changes that were suggested.

I removed the series crossover from the woofer, understanding that that can cause a fire. I have a second-order crossover on the woofer and third order on the tweeter. However, my issues are still the large prominence in the lower frequencies and a dip in the mid range.

I could just copy what somebody else did or use a premade kit, but then I’m not really learning anything.

I need to go back and do more reading. I've gone to the Audioholics page to do some reading there. Thank you.

[billfitzmaurice]

However, my issues are still the large prominence in the lower frequencies and a dip in the mid range.

That's inherent to the driver native response. It's not sourced by the crossover. If you want flatter response start with a woofer that has flatter response. However, it's within the +/-3dB range that more or less defines a good result, so I wouldn't lose sleep over it. You can find more than a few $20k speakers that aren't as flat as that.

[Jicafold]

That's inherent to the driver native response. It's not sourced by the crossover. If you want flatter response start with a woofer that has flatter response. However, it's within the +/-3dB range that more or less defines a good result, so I wouldn't lose sleep over it. You can find more than a few $20k speakers that aren't as flat as that.

Excellent. Thank you.

[fpitas]

Bill's right; but if you keep heading down the DIY path, someday, on some track, you'll notice that dip..and it will bug you....

[Chris Roemer]

A good book is "SpeakerBuilding 201" by Ray Alden. Nice to have a paper "reference" around.

[rpb]

Yes the woofers are parallel, not series. This will be the center channel for a 5.1 system. However, if things work well I might wish to do identical left and right front channels

I’m just getting started in this hobby and a lot of this information is way over my head, particularly response number 18. I’m not nearly at that level yet. I have only had a couple weeks of reading on this in my spare time. I don’t even know what all those acronyms mean let alone the theory behind it, so I don't have the ability to make any of those changes that were suggested.

I removed the series crossover from the woofer, understanding that that can cause a fire. I have a second-order crossover on the woofer and third order on the tweeter. However, my issues are still the large prominence in the lower frequencies and a dip in the mid range.

I could just copy what somebody else did or use a premade kit, but then I’m not really learning anything.

I need to go back and do more reading. I've gone to the Audioholics page to do some reading there. Thank you.

When you actually measure the response in a box, the bass frequencies will drop by 6dB. This is due to what we call baffle step. When someone uses baffle step compensation, ie BSC, they are correcting for this bass drop. In a sim like yours, it might be desirable for the bass to show 5dB louder than the mid-range.

[rpb]

Try this. Put a cap in parallel with the .18mH coil, and see what it does to the woofer response at 10k. Adjust the cap value till it eliminates the response spike. Start with around 5uf, and adjust from there. a smaller cap raises the notch frequency. ( 5uf is a WAG, and might make a notch much lower than needed, but you should easily see where it is. )

[Jicafold]

When you actually measure the response in a box, the bass frequencies will drop by 6dB. This is due to what we call baffle step. When someone uses baffle step compensation, ie BSC, they are correcting for this bass drop. In a sim like yours, it might be desirable for the bass to show 5dB louder than the mid-range.

Oh.... So not just of prominence in the bass frequencies but a prominence that is higher than the midrange and treble. Thank you.

[Jicafold]

Thank you Chris. I see how your graph solves the problem of the baffle step. However, the resultant crossover is now at 3000 Hz. The spec sheet for that tweeter says there is a usable frequency response only down to 3500 Hz.

But I could use a different tweeter.

Additionally, I put these identical values in Passive Crossover Designer and came up with a similar sum response other than a dip at 4000 Hz. Now why would that be?

I see that book is on Amazon. I will have to order that.

[Chris Roemer]

Another aspect of the tweeter's power handling is how much its attenuated. In this case, its about -9dB down (at ALL freqs. - so its not being asked to handle rated power - or anything close to it). That puts it at about -20dB down at its Fs (resonant freq. - based off Dayton's .zma files). I think it'll be happy there, in this instance.

As far as your "dip" goes, when using mfr. data files, you really need to plug XYZ coordinates into your XO sim model. Also, I told you about DCR. If you showed me your latest XO schematic (where you tried to copy my suggestions) it might tell me more. I'm pretty sure you've got 0.1 ohms (now) for a 0.40mH coil, whereas its more like 0.4ohms. That added DCR (in series w/your woofer) will pull it down (at all freqs.) a bit. MY "phase" (model) is summing about +5dB (above where the 2 drivers cross). Your model is only half that. I'm not saying that MY phase (model) is right, but most likely yours isn't either. If it IS, one way to get better (higher) summing is to flip the tweeter polarity, OR add another order to the filter(s).