Series Crossover for Zaph's L18/27TBFCG Design?

[ABS]

Having built a pair of speakers using Zaph's L18/27TBFCG design, I found that overall I was quite happy with them but I wanted to see if I could build a series crossover version which would still work well and possibly provide some additional benefits - quite a challenge given John's already very good design!

To begin, I spent quite a lot of time on the www.humblehomemadehifi.com web site familiarizing myself with the typical design elements of series crossovers. I also downloaded a .zip file posted here on DIYAudio by Zaph (John Krutke) which are his measurements for frequency response/phase and impedance of the drivers in the L18/27TBFCG system. The file may be downloaded here: .Zip File What I don't know is if these measurements were taken in the actual .5 cu ft PE boxes used by John in the final design, but these are the measurements I used in my simulations (perhaps John can let us know?). Using SpeakerWorkshop, I imported the files from John and simulated John's crossover original design. Then, using the frequency response from John's design as a target, I spent a lot of time simulating new designs to develop the new series crossover version.

I have been performing listening tests on the new design for the last couple of days and I feel confident that this design is very good! Truth be told the performance has exceeded my expectations. Some of the advantages of the new design include the following:

1. Both Tweeter and Woofer have somewhat steeper slopes (a side effect of the series crossover).
2. The crossover point is moved up to around 2300 hz.

These two changes mean that the woofer's upper frequency range (where the cone break-up occurs) is down an additional 6 dB (above 6000hz) and that the tweeter's output at 1K is down an additional 8 dB. These changes should improve power handling and reduce distortion.

3. The woofer output at around 2500 hz (frequency of 3rd harmonic resonance) is down an additional 1db. I do not know if this change is audible but I don't think it can't hurt given John's comments regarding the negative effects of the 3rd order harmonics.

6. Midrange phase shift slowly rotates 360 degrees over a wide bandwidth (250 hz to 7500 hz). This seems to provide improved driver integration.

7. Reduced sibilance on S's with the spoken word.

The negatives I have found with this new crossover:

1. No opportunity to bi-amp with series crossovers.
2. Increased parts count.
3. The impedance has a dip to approximately 3 ohms at 2000 hz.
4. Slightly worse off-axis imaging.

For those interested, I have attached some images. In all the images, blue lines are for the series crossover and white lines are for John's original crossover.

1. A chart showing the individual driver and combined frequency responses and total phase for John's original design and the series crossover design.

2. The impedance curves of both designs.

3. The crossover diagram for the series crossover.

I would like to emphasize that John's design already is quite excellent. The purpose of this post is simply to share with the community an option I developed which uses a different kind of crossover and to generally open the discussion regarding series crossovers for use with the same drivers, enclosure and layout as that used in John's original design.

Please note also that I have not performed any equipment based testing of this design - I have used my ears only. Subjectively, I feel this design has a very good sound. I believe the driver integration to be quite good and the new crossover retains the same detail and sparkle as the original. The hint of sibilance which I had continually experienced when listening to the news or other spoken material is all but gone with the series crossover. Imaging and stage depth are also both very good although I believe the imaging may have been slightly better with John's design (this may be a side effect of having moved the crossover frequency up a few hundred hertz).

Should you decide to try/test this crossover, please post your results.

Also, as a novice with series crossover design, I welcome comments and feedback.

Regards,

Andrew

[ABS]

Impedance Curve

[ABS]

Here is the design of the series crossover. A few things to note, electrically, this is a second order series crossover on both the woofer and the tweeter (for series crossovers the 2nd order electrical slopes are steeper than 12db). Of note, C10 and R10 alter the impedance near the crossover point and effectively tweak the woofer's output between 1500-2000 hz. Changing the value of R10 up or down will increase/decrease the woofer output in this range (higher value resistors will drop output). C1 and L1 are the same values for the notch filter as in John's original design and still work quite well to reduce the woofer's HF cone resonances. R1 and R2 make up the tweeter L-Pad and those values also remain unchanged. R2 may be slightly reduced to 3.0 (versus 3.3) to bring the frequency response even closer to John's original curve if so desired. It should also be possible to reduce this value further to obtain a bit more of a laid back tonal balance . . . L2, L3, C2 and C3 form the 2nd order crossovers and the values should not be changed. L3 uses the same low DCR steel core inductor used in John's original design. L2 uses a .3mH low DCR 14 gauge Erse PerfectLayer inductor. I use and like Clarity Caps as they seem to provide more detail and "air" than the similarly priced Solen units.


Series Crossover

[kyselym]

Hi ABS

What enclosure are you using? same as John? or more like Tony?

[ABS]

I am using the exact enclosure design provided by John. It is the Parts Express .5 cu ft enclosure, lined with Sonic Barrier 3 layer 3/4 throughout and the same baffle layout as provided by John. I did select a different port - I am using the 2" Precision Port. I hope this helps. You may already know this, but Tony's design uses a different woofer . . .

[kyselym]

I know but it can be used with both versions of L18. On the other hand tony's design works better with older H1142 I think. I have both versions and now can try both for zaphs

How much will affect performace if I use same enclosures I have for tonys DD8?I mean in normal room. I think it will not be very noticeable but haven't tried yet.

[ABS]

Kyselym:

This is very interesting - you have both Zaph's design and also Tony's? That is really great! I think I agree with you that both woofers should work with in either cabinet from a bass tuning standpoint. However, I know that Tony's crossover will not work properly with the new L18 woofer used in Zaph's design. The break up modes are different as is the frequency response. Now, if you took the new L18 and installed it in Tony's enclosure and used either the series crossover I just presented or Zaph's original crossover, I think you would get a reasonable result but with a few caveats: 1. Zaph's design has the tweeters mounted off-center and Tony's design does not - I think this will definitely impact the off-axis HF response; 2. The size of Tony's front baffle is different than that in the Zaph design - I am sure this will impact (to a greater or lesser degree) the baffle step correction built into Zaph's crossover (and my series crossover); 3. It appears that the driver to driver spacing is quite a bit greater on Tony's DD8 design than on Zaph's design and I feel this may have the biggest impact of all since it will alter the relative phase of the drivers and therefore will likely impact the combined output of the drivers.

However, having said all that, if you have the parts at hand, I would say you should give it a try and let us know how it turns out. I bet there are other folks who previously built the DD8 and would be interested in an upgrade. I don't have any of Tony's designs otherwise I would consider trying it myself. Let us know how it turns out!

Andrew

[Zaph]

Interesting work, thanks for sharing.

One thing I want to note for those using that data I posted ages ago - the response and phase curves are on each driver's axis. I always do it that way so I can use soundeasy's virtual front panel to sim multiple mic locations. That's normally a different procedure for people who don't use soundeasy. Response curves are normally taken on the tweeter axis, putting the woofer off axis.

So, when you work with this data you are essentially designing for an infinite mic distance. My designs normally simulate a mic distance of 1 to 2 meters, on or slightly above tweeter axis. People who model my crossover using LSPcad or SW with that data posted will get different results than I do. Soundeasy's virtual front panel simulates off axis rolloff based on radiating diameter. (and does it very accurately too) It's a powerful tool and something no other package does as far as I know.

Just wanted to make sure you were aware of that. For this reason, I don't post response data anymore. It's slightly less useful and occasionally confusing to people not using soundeasy.

If I were to improve my old L18 design today, I'd move the crossover point down a little bit, though I'd have to add components to do it.

[ABS]

Thanks for the response John. I was not aware of this difference with SoundEasy vs. SW and I know that the question of phase as determined by mic position is very critical to developing a good crossover. SW recommends that the mic stay in one position while both the tweeter and woofer are tested. I can easily see the benefit of testing each driver on axis and then simulating various mic positions in the software since one would get a more accurate freq. response curve from each driver but still would account for the differences in driver centers. I think I can try something similar in SW using the driver offset adjustment - I will have to look into that.

John, could you let me know if the response curves you posted were taken in the PE .5 cu. ft box with the final baffle layout and with the mic the same distance from the front baffle for each driver?

Thanks!

[Zaph]

Well, I can't remember what I did yesterday, much less 2004, but I think they were taken in the PE box with the final driver location. The mic was the same distance from each driver.

Note that even if you were to add time to the woofer to account for a mic position on the tweeter axis, the woofer will still not have the correct response curve for that off axis angle. It makes quite a difference above the crossover point. Hmmm, you may be able to sim it and sum it into the response somehow.