Seas ER18RNX and 27TDFC 2-Way
Sept 19, 2007 - Initial posting
Nov 4, 2007 - Crossover revised
Nov 30, 2007 - Crossover revised; Comparison to Zaph's design; Baffle diffraction sim changed to use the driver positions of Madisound's precut baffle
Dec 2, 2007 - Options added; in-wall mounting crossover included
Dec 3, 2007 - A note added in the "Comparison to Zaph's design" section
The new Seas ER18RNX (H1456) 7" midbass driver seems interesting. It's basically a paper cone driver like its sibling CA18. But according to Zaph's nonlinear distortion test, thanks to the new motor design its harmonic distortions, especially odd order harmonics, are significantly lower than those of the CA18. Considering its price, it is a very attractive choice as a midwoofer in a small 2-way or as a midrange in a large 3-way. Actually, I know that Zaph has been working on a 2-way that uses this driver and the 27TDFC/27TBFCG. But I just wanted to design my own version of a crossover for these drivers before he posts his. Because I already designed two other 2-way designs using his infinite baffle data and voice them to my ear, a similar 2-way design using his data will be meaningfully compared to them. I wanted to first design a crossover whose predicted voicing matches the voicing of my already built designs and then to compare it to what Zaph will post at his website. This will give me an interesting chance to compare my voicing preference to his. Once he posts his design, I'll apply his XO to my simulated in-box responses and then compare the result to mine.
If John sees this page by any chance, I hope he won't be offended. I have no intention to offend him. In fact I respect him as an excellent speaker designer and I learned a lot from his website. I simply wanted to do this for fun. People may not want to build my design which has not been built and tested, unless they really believe that my voicing preference is exactly what they've been looking for (if mine and John's turn out to be different).
The ER18RNX is appropriate for a small vented 2-way. For instance, it will fit nicely in the Parts Express .5 cu ft or the Madisound MD14B cabinet---Madisound also sells a precut baffle for Seas standard 176 mm woofer and 104 mm tweeter. I used the MD14 cabinet and its precut baffle for box and baffle diffraction modeling. In the precut baffle of this cabinet, the tweeter is placed with no offset from its vertical center line. Offsetting the tweeter will affect the frequency response only in a minor way, the effect of which can be accommodated by adjusting the tweeter's lower rolloff. Tuning with a 7" long, 2" diameter port gives a tuning frequency of 44 Hz and an F3 point of 48 Hz with no midbass peak (see the predicted bass response). If deeper bass is desired, a larger cabinet can also be used. For example, Madisound MD20B (w/ precut baffle available) can be used. If you want to build one yourself, use the same vented box dimensions as shown in my Dayton 2-way design page (you can use either a stand-mounting or a floor-standing cabinet). For these cabinets, tuning with a 6" long, 2" diameter port will give a tuning frequency of 37 Hz and an F3 of 42 Hz with a little lean bass, which will provide a flat response in a small to moderate size room with expected room gain (see the predicted bass response).
Here are my results of modeling:
The crossover point is about 1.55 kHz and the designed listening axis is tweeter axis. In case you're further interested:
System impedance and phase
Filter transfer functions
Speaker Workshop file for the design (zipped 1.4 MB)
Comparison to Zaph's design
Important Note: In the following sections I DO NOT mean that my suggested crossovers are better than Zaph's. Read carefully. In fact, my suggestions are nothing special. They can be viewed as sort of extension of Zaph's "reduced baffle step compensation" option with more specific implementations. This is for people who want more tuning options in order to customize their speakers to their own listening environments. Zaph put a lot of effort to bring this great design to the DIY community for free. I have no intention to undermine this fact.
Zaph finally posted his ER18 design (AKA Madisound ZA-SR71 kit). Here's the result from applying his crossover to my modeled in-box responses, and Zaph's simulation using his measured in-box responses:
First, notice that modeled in-box responses obtained from baffle diffraction simulations applied to infinite-baffle measurements are very close to measured ones. Using my method of baffle cavity effect approximation, I was able to closely replicate the real measurement of the tweeter response. Only differences between modeled and measured responses are the woofer's sensitivity and its high-frequency rolloff. I don't know why the former occurred. My guessed reason is that above 500 Hz I used Zaph's infinite baffle data, and his measured ER18RNX somehow generated a bit lower sensitivity, and that below 500 Hz, I did not use real measurement but simply spliced a modeled response. The steeper higher rolloff of the woofer is due to its vertical offset from the microphone axis (in this case, the tweeter axis), which barely affects the crossover simulation with a relatively low xover frequency.
Looking at his final design choices, a couple of things became clear to me. The ER18's characteristic of rising nonlinear distortions over its frequency band made him choose to use full baffle step compensation. In fact, it is a little more than full 6 dB compensation if we closely examine the woofer's transfer function, acoustic rolloff, and the system frequency response. Even if the woofer's early rolloff due to the low crossover point is taken into account, the amount of compensation should be about 6.5 dB. I guess this choice gave him his preferred tonal balance and desired system nonlinear distortions. One possible disadvantage of this choice is that these speakers will be picky in their room placement and may produce a bit too warm and muddy tonality of midrange in a situation where they need to be used in a small room or a little close to walls. Of course, he provides a reduced BSC option, and also suggests trying the default design and then unwinding the primary coil in the woofer net if the user wants a different tonal balance. But this method will move the crossover frequency a little to the higher side, and combined with the tilted up woofer response, will result in higher system nonlinear distortions in the 800 Hz to 1.5 kHz range. For this reason, if you want to use reduced BSC for this design, I recommend trying my crossover. It uses a lower 1.55 kHz crossover point and thus the tweeter's lower nonlinear distortions will dominate the woofer's higher distortions a little earlier to result in better system harmonic distortions in the above range. For those who may be concerned about the tweeter's power handling with a 1.55 kHz crossover frequency, I can say that the 27TDFC is capable of handling this demand. I used this tweeter in my RS180 2-way design with the same 1.55 kHz LR4 slope. I haven't heard any hint of distortions from the tweeter even at an ear hurting volume level. Out of curiosity, I compared the tweeter's filter transfer function of my RS180 design to that of this ER18 design:
Comparison of tweeter filter transfer functions between two designs
For easier comparison, two functions' levels were matched at 6 kHz. As you can see, the ER18 design's tweeter load at its low end is a bit lower than the RS180's. So I don't think we have to worry about the tweeter's power handling issue.
The tweeter level difference between Zaph's and my designs above 2 kHz is minor because this can be easily adjusted by changing a padding resistor. But according to my experience of replicating nearly all of Zaph's published designs using the same IB measurements of his, I noticed that he tends to use a higher tweeter level that makes the measured system frequency response flatter than I use for voicing my designs. In my experience, this voicing for a flat tweeter level gives most of pop music CDs more "air" and makes them more lively. But in the case of most of classical music with string instruments, which I listen to most of the time, it makes strings' "sibilant" (i.e., sound of strings drawn by bows) somewhat obtrusive. Anyway, this is more of a matter of personal taste.
In sum, I recommend trying my crossover if you think you're in one of the following situations:
1) Your listening room is relatively small and acoustically reflective Þ Use my default crossover;
2) You want to build the design in a floorstanding cabinet, which means a little more baffle gain due to the extended baffle height (see the effect of using a tall baffle on frequency response) Þ Use my default crossover;
3) You need much reduced BSC in order to use the speakers for nearfield listening/monitoring, or to place them close to walls, on a desk, in an entertainment rack, on a bookshelf, etc, which results in a significant boundary gain Þ Use the reduced BSC option I provide below.
Using the 27TBFC/G tweeter
If you want to use the Seas 27TBFC/G tweeter (a metal dome version of the 27TDFC), use the following crossover. This crossover takes into account the frequency response of the 27TBFC/G that is different from that of the 27TDFC (i.e., different sensitivity and rolloffs at both frequency ends). The crossover frequency is the same as the 27TDFC version's 1.55 kHz.
Crossover for 27TBFC/G
Frequency response comparison of 27TBFC/G and 27TDFC versions (on magnified SPL scale)
Tweeter network fine-tuning option
Don't be afraid to fine-tune the tweeter network by your ear. It can be easily done and you will be satisfied with speakers' sound customized to your taste. I suggest trying two different kinds of adjustment. Tweeter level adjustment will mainly affect the speaker's treble above 2.5 kHz, and tweeter lower rolloff adjustment will affect the response in the upper midrange (1.2 k to 2.5 kHz). In my experience of voicing speakers, a change in the upper midrange, even if the adjustment is small, has more audible effect on a speaker's perceived "brightness." To get a better idea about these adjustments, read the "Voicing your speakers" section in my note Designing Crossovers with Software Only. This option will also come in handy when you use different driver positions from those of the default design (e.g., tweeter offsetting). Some possible effects of using different driver positions, such as different diffraction ripples and lower rolloff of tweeter response, can be taken care of sometimes by listening off-axis as well as by fine tuning the tweeter network. Shown below are the effects of adjustment when the 27TDFC crossover is used---you can do a similar thing with the 27TBFC/G crossover.
Effect of tweeter level adjustment (w/ 27TDFC)
Effect of tweeter's lower rolloff adjustment (w/ 27TDFC)
Impedance flattening circuit
In case you want to use these speakers with a tube amp, shown below is an impedance flattening circuit (R20, C21, and L22). Using an electrolytic capacitor for C21 is perfectly okay and has nothing to do with the speaker's performance. Inductor L22 is 19/20 gauge. Note: this option can be used with the tweeter fine-tuning option suggested above as long as the change of component values is minimal (e.g., values suggested in the figures), but cannot be used with reduced BSC and no-BSC options below.
Crossover with impedance flattening circuit w/ 27TDFC
Crossover with impedance flattening circuit w/ 27TBFC/G
Reduced baffle step compensation
If you have to place the speakers close to walls, on a desk, in an entertainment rack, on a bookshelf, etc, which results in significant boundary reinforcement, or if you want to use them for nearfield listening/monitoring, the following version of crossover will be useful with reduced, about 3.5 dB baffle step compensation. The reduction of BSC still keeps the design's crossover frequency as low as 1.65 kHz:
Reduced BSC crossover for the 27TDFC version
Reduced BSC crossover for the 27TBFC/G version
No-BSC crossover for in-wall mounting
This 2-way design can be an excellent alternative to low-quality in-wall speakers for those who are looking for a great sounding DIY solution. Below is a crossover version with no BSC for in-wall mounting.
Crossover for in-wall mounting
Modeled frequency responses on an infinite baffle
The crossover frequency is about 1.65 kHz. Unlike the in-room version with BSC, the design axis of this no-BSC crossover is not tweeter axis but in the middle of tweeter and woofer axes. Its vertical off-axis performance (up to +/- 15 degrees) will be very good thanks to excellent acoustic phase alignment over a wide range below and above the crossover frequency. Particularly, vertical downward off-axis listening (assuming tweeter is placed above woofer) will be exceptionally stable. This is on purpose in the design because in many cases in-wall speakers tend to be installed a little above ear level. Even so, placing the speaker more than 15 degrees above ear level is not recommended. The center-to-center distance between tweeter and woofer should be less than 6". If you want to use this speaker without a subwoofer, I recommend building it in a ported box. If an in-wall cabinet of 17 liters is used with a 6" long, 2" diameter port, the port tuning frequency and -3 dB point will be 43 Hz and 44 Hz, respectively. When this is installed in a sealed box (at least 7 liters), a subwoofer should be used with at least 80 Hz crossover point. If you want to use a subwoofer with a crossover frequency below 80 Hz, a vented design will be required because the ER18's -3 dB point is as high as 86 Hz in a closed box. The speakers won't sound bad even without fine-tuning, but nonetheless I recommend experimenting with R9 and C6 values to find ones that suit your taste. Try values between 3 and 4 ohms for R9, and 10, 11, and 12 uF for C6---you can do this by purchasing a 10 uF and two 1 uF capacitors for each speaker. To understand how to adjust these values, read the "Voicing your speakers" section of my page Designing Crossovers with Software Only.
Feel free to email me with any questions or comments.