Listening impressions
I spent some time critically listening to this “Terra Incognita” speaker over the past week. I also got input from my wife and the guy who will be the eventual owner of these speakers. I unpacked my old Class AB amp to drive them (BK ST-140).
I compared them to the two other systems I currently have set up, the LCCAM-10.3 and the Purifi-Waveguide. Since I aim for a particular tonal balance on all of my speakers, there is little difference in overall frequency response among them, so any differences I perceive are due to the small spectral differences that remain, as well as other factors such as distortions and resonances.
https://www.diyaudio.com/community/threads/compact-low-cost-active-3-way-speaker.402812/post-7558072
https://www.diyaudio.com/community/threads/purifi-waveguide-project.394174/post-7363646
The Terra Incognita speakers have a warm, relaxed, and smooth, quality. From the upper bass through the lower midrange, through the upper midrange and treble, it all seems very well integrated. There is no sense of strain or stridency at normal SPL. It is an enjoyable speaker to listen to for long periods. Compared to the LCCAM, it has a bit less detail and clarity. That is not necessarily a bad thing, depending on the recording.
The high end is smooth, and sibilance is handled well. It actually sounds very similar to the SB26STAC tweeter in the LCCAM speaker, which is not surprising since the two tweeters are closely related. Some people say that good silk dome tweeters have a forgiving, relaxed character which is non-fatiguing, and that fairly describes both the LCCAM and this speaker.
The spatial presentation of the two speakers is similar. Both do a good job of creating a wide soundstage with localization in 3 dimensions.
The midrange of the Terra Incognita is detailed with a natural presentation of overtones. Here I found a difference between this speaker and the LCCAM. The LCCAM had better clarity and detail. The LCCAM presented the silence between notes in a better way. Sometimes the LCCAM can sound overly analytical, and it can be unforgiving with poor recordings. The Terra Incognita did a better job of smoothing over the errors in bad recordings. The CRC midrange cone is a carbon fiber layer over a rohacell foam, and it reminds me of a good paper driver. It seems punchier than the NBAC aluminum cone in the LCCAM. It may seem strange to say it, but the CRC driver seems to handle the attack of each note better, while the NBAC driver handles the decay better. The CRC driver seems more harmonically rich, but this richness tends to smear details and, on some recordings, results in a loss of clarity. Overall I prefer the midrange presentation of the LCCAM speaker, but I imagine some would prefer the Terra Incognita, just as some people prefer paper cone drivers.
The bass of the Terra Incognita is warm and relaxed. Up until now, I have never been happy with any bass reflex speaker I have built. I like this one. Pitch definition is good, and the bass is articulate. Comparing it to the LCCAM, which has a sealed 10” woofer with DSP EQ, I found the LCCAM to have superior articulation, but that is to be expected. Nothing sounds quite like a sealed box woofer. But I was surprised and impressed at how well the Terra compared to the LCCAM.
According to simulation, the RS270 10” woofer in a sealed box should have a similar max SPL compared to the RS225 8” woofer in a reflex box. Subjectively however, the LCCAM could play bass louder than the Terra Incognita. The 8” reflex woofer started to sound hard and compressed before the 10” sealed woofer. It is not a big difference, about 2 dB difference, and it happened at a higher SPL than I normally listen to. Given this is a subjective assessment, there may be many factors that influence this perception, including the cabinets, the passive radiator, and the midrange drivers. In other words, it may not be the woofer.
When comparing to the Purifi Waveguide system, there is a clear difference. The Purifi system has a stepwise increase in every aspect of performance. The Purifi system had far better bass extension, articulation, bass power and dynamics. The Purifi system had much higher max spl. The Purifi system had much better midrange detail, clarity, and fine resolution of harmonics. The Purifi was better at resolving small changes in dynamic contrast, and seems effortless even at unwise SPLs. The Purifi system was much better at presenting a 3D image and sense of space, with excellent localizations, including depth. All of this is not surprising, considering the cost of the drivers and electronics in the Purifi system cost 4.5 times that of the Terra system.
So overall, I am happy with this design. It is smooth, easy-going, easy to listen to. Good imaging and spatial presentation, and enough detail and clarity to make music enjoyable, without making it tiresome.
I spent some time critically listening to this “Terra Incognita” speaker over the past week. I also got input from my wife and the guy who will be the eventual owner of these speakers. I unpacked my old Class AB amp to drive them (BK ST-140).
I compared them to the two other systems I currently have set up, the LCCAM-10.3 and the Purifi-Waveguide. Since I aim for a particular tonal balance on all of my speakers, there is little difference in overall frequency response among them, so any differences I perceive are due to the small spectral differences that remain, as well as other factors such as distortions and resonances.
https://www.diyaudio.com/community/threads/compact-low-cost-active-3-way-speaker.402812/post-7558072
https://www.diyaudio.com/community/threads/purifi-waveguide-project.394174/post-7363646
The Terra Incognita speakers have a warm, relaxed, and smooth, quality. From the upper bass through the lower midrange, through the upper midrange and treble, it all seems very well integrated. There is no sense of strain or stridency at normal SPL. It is an enjoyable speaker to listen to for long periods. Compared to the LCCAM, it has a bit less detail and clarity. That is not necessarily a bad thing, depending on the recording.
The high end is smooth, and sibilance is handled well. It actually sounds very similar to the SB26STAC tweeter in the LCCAM speaker, which is not surprising since the two tweeters are closely related. Some people say that good silk dome tweeters have a forgiving, relaxed character which is non-fatiguing, and that fairly describes both the LCCAM and this speaker.
The spatial presentation of the two speakers is similar. Both do a good job of creating a wide soundstage with localization in 3 dimensions.
The midrange of the Terra Incognita is detailed with a natural presentation of overtones. Here I found a difference between this speaker and the LCCAM. The LCCAM had better clarity and detail. The LCCAM presented the silence between notes in a better way. Sometimes the LCCAM can sound overly analytical, and it can be unforgiving with poor recordings. The Terra Incognita did a better job of smoothing over the errors in bad recordings. The CRC midrange cone is a carbon fiber layer over a rohacell foam, and it reminds me of a good paper driver. It seems punchier than the NBAC aluminum cone in the LCCAM. It may seem strange to say it, but the CRC driver seems to handle the attack of each note better, while the NBAC driver handles the decay better. The CRC driver seems more harmonically rich, but this richness tends to smear details and, on some recordings, results in a loss of clarity. Overall I prefer the midrange presentation of the LCCAM speaker, but I imagine some would prefer the Terra Incognita, just as some people prefer paper cone drivers.
The bass of the Terra Incognita is warm and relaxed. Up until now, I have never been happy with any bass reflex speaker I have built. I like this one. Pitch definition is good, and the bass is articulate. Comparing it to the LCCAM, which has a sealed 10” woofer with DSP EQ, I found the LCCAM to have superior articulation, but that is to be expected. Nothing sounds quite like a sealed box woofer. But I was surprised and impressed at how well the Terra compared to the LCCAM.
According to simulation, the RS270 10” woofer in a sealed box should have a similar max SPL compared to the RS225 8” woofer in a reflex box. Subjectively however, the LCCAM could play bass louder than the Terra Incognita. The 8” reflex woofer started to sound hard and compressed before the 10” sealed woofer. It is not a big difference, about 2 dB difference, and it happened at a higher SPL than I normally listen to. Given this is a subjective assessment, there may be many factors that influence this perception, including the cabinets, the passive radiator, and the midrange drivers. In other words, it may not be the woofer.
When comparing to the Purifi Waveguide system, there is a clear difference. The Purifi system has a stepwise increase in every aspect of performance. The Purifi system had far better bass extension, articulation, bass power and dynamics. The Purifi system had much higher max spl. The Purifi system had much better midrange detail, clarity, and fine resolution of harmonics. The Purifi was better at resolving small changes in dynamic contrast, and seems effortless even at unwise SPLs. The Purifi system was much better at presenting a 3D image and sense of space, with excellent localizations, including depth. All of this is not surprising, considering the cost of the drivers and electronics in the Purifi system cost 4.5 times that of the Terra system.
So overall, I am happy with this design. It is smooth, easy-going, easy to listen to. Good imaging and spatial presentation, and enough detail and clarity to make music enjoyable, without making it tiresome.
i had to look back at those threads, i am very impressed, and also somewhat jealuous of the space in your home you have to make these comparisons.Wrt to the purifi waveguide, i can only rely on memories, as my enclosure is now equipped with purifi bass and midrange, and as " upgrade path" with wwaveguided sb26adc tweeters. I have the same conclusions as you and also from your purfi thread that the critical listening distance is somewhat increased.
What i also noticed that the filtering and voicing is more sensitive , thus smaller corrections/changes can become audible. And bad recordings are just bad. But good recordings become better.
Your thorougness is remarkable.
When I built my full range line arrays I spent a significant amount of time changing the frequency balance through EQ. Because there is no crossover there is no affect to directivity from EQ. They defied all of my attempts to equalize to more anechoic response and I had to rely on in room measurements and tuning by ear in combination. This is very different than the more regular process of tuning a multiway speaker.
What I found was that the frequency balance is very important to how universally good a speaker can sound on a wide range of material. If I relied on audiophile music to tune the speaker I could get some real magic on those tracks, but it tended to make everything else sound awful. Of course a recording can be less than stellar due to over compression, clipping, strange mic choices etc. but once I had settled on a more universal balance there really were no "bad" recordings. Sometimes the differences to go from good to great were really quite small but had a disproportionate improvement to the sound.
I do not think there is one EQ that will work as well as possible for all music (setting aside gross differences in bass and treble). If there is only to be one tune a decision has to be made whether to swing in one direction or for a happy medium.
Some of the changes that made worthwhile improvements would be difficult or impractical in a passive crossover but if an overall active EQ can be used and you have the time to listen and listen while tweaking you might be surprised what happens.
For me the spatial image of the music is very important. Next is the rest, and indeed a small change in the filtering can make a remarkable difference. And for some bad recordings to become acceptable. It is my perception of getting engaged into the music, iow can i hear why the musicians is playing as he/she does. When i was as a business making fine art prints working mostly in close cooperation with the maker of the image, i learned as a rule that if a small change f.i. in exposure means from good to great print(reproduction), this means the image(recording) is spot on perception wise. This regardless of color or b/w or high key or low key , amount of pixels, all typical image classifications. Really bad elements like color diffractions ( green/purple fringing in high contrast edges) would also stand out more, and can destroy the perception of the image. As a side note those errors were most of time easily corrected, and more a case of "oops" overlooked. Great means the intent of the maker gets across, as a basic rule is that the maker is present in the image. For me this also applies for music.
This is my experience as well. I found that in a system with a very consistent directivity, a single voicing (EQ) works well with most (but not all) recordings. With my system with degraded directivity control (the TXT system), I found I needed three different presets on the Hypex amp to adjust the treble balance. The poor directivity control from 1k - 4k necessitated that I adjust the treble balance based on the recording. Today I have a new preamp with a nicely designed treble control, so I use that instead of a hypex preset, but the net result is the same... when I listen to the TXT system, I adjust for each recording. With the Purifi Waveguide system, I rarely need to adjust.... It sounds right on almost everything.
Your point about very small differences from good to great is very true. Accurate measurements and analysis can get us to 90% (good) fairly quickly. To go from 90% to 99% (good to great) requires intensive subjective assessment, and it can take a long time.
j.
Oh yes, the last 9% is costing 99% of the total time ( more or less ;-))
All joking aside it costs time indeed.
But i find intriguing that your experience with the purifi waveguide appears to be an outlier. Any idea why?
All joking aside it costs time indeed.
But i find intriguing that your experience with the purifi waveguide appears to be an outlier. Any idea why?
Back when I was the leader of an engineering team, I used to say that the first half of the job would consume 90% of the budget. The second half of the job would consume the other 90% of the budget.
I have tens of saved configs/presets of my dsp-active speakers. I don't remember the differences between them exactly, but the latest always sounds better than the previous, until... 😎
And this has lasted for 10+ years. Now I understand why they are called "active" !
And this has lasted for 10+ years. Now I understand why they are called "active" !
Last edited:
😀 It's still a cool learning opportunity. Remember to take notes for each preset, so you could go back and kinda see the line of thought that made the sound turn south
On the subject of active systems... If I were going to apply active DSP 3-channel amplification to this project, this is the filter I would start with.
Comparing this to the as-built passive crossover is interesting, and it shows how the increase in control can lead to a higher performance.
Dashed line is the passive
Notice how the passive low pass filter on the woofer boosts the signal from 80 - 200 Hz. I think this is responsible for the warm character of the speaker. An active filter does not interact with the woofer impedance, so the filter acts in a more idealized way without the extra gain.
An active filter allows me to tame the 2k peak in a very precise way. This allows the whole balance of the midrange driver to be different, and I am able to optimize the PIR response a bit better.
As I said, if this were an active speaker, this is the filter I would start with to do the subjective voicing process. ....... j.
Comparing this to the as-built passive crossover is interesting, and it shows how the increase in control can lead to a higher performance.
Dashed line is the passive
Notice how the passive low pass filter on the woofer boosts the signal from 80 - 200 Hz. I think this is responsible for the warm character of the speaker. An active filter does not interact with the woofer impedance, so the filter acts in a more idealized way without the extra gain.
An active filter allows me to tame the 2k peak in a very precise way. This allows the whole balance of the midrange driver to be different, and I am able to optimize the PIR response a bit better.
As I said, if this were an active speaker, this is the filter I would start with to do the subjective voicing process. ....... j.
@hifijim As you mentioned, you used some PEQ on the midrange to attenuate its peak at 2kHz by 2.1dB. Did you consider boosting the frequency response at about 1.5kHz by about 1.5–2.0dB instead? Looking at the response curve of the midrange, this would have smoothed out that region quite considerably. I guess some attenuation at 2kHz might still have been needed, but it might have been expected to be less than 2.1dB.
The rise in the passive system's woofer output in the 80–200 Hz region seems to be a very common occurrence when using a passive filter on the woofer with a relatively low cut-off frequency, something that is typical of a 3-way loudspeaker system. It looks like it is some sort of series resonance circuit created by the capacitive behaviour associated with the reducing woofer impedance with increasing frequency in that region, interacting with the combination of the series filter inductor and the voice-coil resistance.
The rise in the passive system's woofer output in the 80–200 Hz region seems to be a very common occurrence when using a passive filter on the woofer with a relatively low cut-off frequency, something that is typical of a 3-way loudspeaker system. It looks like it is some sort of series resonance circuit created by the capacitive behaviour associated with the reducing woofer impedance with increasing frequency in that region, interacting with the combination of the series filter inductor and the voice-coil resistance.
Last edited:
Hi, this has been a thoroughly interesting project, thanks for sharing. Can you explain for me why you have cascaded the two high pass filters on the tweeter circuit?
Thanks
The acoustic slopes on the midrange and tweeter drivers through the crossover region are quite shallow. The resulting midrange/tweeter crossover region seems to be quite wide as a result. I'd be interested to know more about why such a design choice was made. Was it to try and get a more constant directivity through that frequency region?
I tried a lot of different EQ strategies, including that one. The 2k peak is caused by a well damped resonance, so it seemed more intuitively correct to cut the peak rather than raise everything else up to the level of the peak.
Yes, I talked about this in post #422 https://www.diyaudio.com/community/...ivers-a-new-3-way-project.413182/post-7775930
In post #428 I showed this plot. This is the measured near field response of the raw driver (light blue dashed line) and the measured near field response of the woofer with filter. You can see the 1.6 dB gain from 60 -120 Hz.
As you say, almost all 3-way filters will have this sort of issue. The only way around it is an impedance compensation network, and any such network in the 50 Hz range is going to require massive inductors and caps. It is just not practical.
I was trying to mimic the transfer function of the passive filter, but moving the crossover frequency up closer to 3k. Someone on this forum (I think it was @fluid) advised me that waveguide tweeters often respond well with a first order filter at a high Fc, and I have found this to be generally true. The first order filter counteracts the natural 6 dB/octave rise in output as the frequency drops, and then combines with the natural rolloff of the tweeter to give a combined 2nd order acoustic rolloff. The second lower 1780 Hz first order simply maintains the desired response down to 500 Hz. With real world drivers in a real world baffle, I don't expect the filter network to be idealized or mathematically derived. It is simply a collection of compromises which work well enough.
I think we talked about this in post #484, yes? https://www.diyaudio.com/community/...ivers-a-new-3-way-project.413182/post-7790092
The midrange and tweeter slopes look shallow, I agree. They follow the 2nd order Linkwitz-Riley target curves reasonably well.
j.
"The first order filter counteracts the natural 6 dB/octave rise in output as the frequency drops, and then combines with the natural rolloff of the tweeter to give a combined 2nd order acoustic rolloff. The second lower 1780 Hz first order simply maintains the desired response down to 500 Hz. With real world drivers in a real world baffle, I don't expect the filter network to be idealized or mathematically derived. "
Yes, this is the nicest thing with dsp. Loudspeaker's output is acoustical and the acoustic response is what we are modifying to match, with far more liberties than passive cirquits can. Of course we can make hybrids with coils and caps to eg. provide protection for the tweeter (pops occur at startup). Low order xo requires good slopes far beyond nominal Fx
I have noticed that with 3-4 way speakers slopes come rather naturally to 2nd order Duelund type (elliptic slopes), like this case of hifijim's.
https://musicanddesign.speakerdesign.net/Duelund_and_Beyond.html
2nd order Duelund
Modified 4th order
Yes, this is the nicest thing with dsp. Loudspeaker's output is acoustical and the acoustic response is what we are modifying to match, with far more liberties than passive cirquits can. Of course we can make hybrids with coils and caps to eg. provide protection for the tweeter (pops occur at startup). Low order xo requires good slopes far beyond nominal Fx
I have noticed that with 3-4 way speakers slopes come rather naturally to 2nd order Duelund type (elliptic slopes), like this case of hifijim's.
https://musicanddesign.speakerdesign.net/Duelund_and_Beyond.html
2nd order Duelund
Modified 4th order
That's an interesting approach to provide a bit of EQ to a waveguide tweeter's response characteristics. 👍
We did, and at some length; thanks for reminding me!
Regarding the SB15CRC30-8 midwoofer.
Several people cautioned me about using this driver as a midrange, particularly @wolf_teeth . Now that this project is nearing completion, I can say that this driver is a challenging driver to use effectively, especially in a passive network.
For comparison, I will show the 0-90 degree polar response of the SB15CRC, the SB15NBAC, and the SB15CAC. These are all gated far field measurements merged with near field scans. Each driver was measured in a cabinet, and all three cabinets are different, so the baffle step will be different and the higher frequency diffraction effects will be different. Regardless, I think it will be obvious why the SB15CRC is a more challenging driver than the other two. The first breakup mode on the SB15CRC is a half octave closer to the typical crossover frequency of 2.5k – 3.5k compared to the other two.
I have no regrets about using this driver. Part of the goal of this project was to learn about and use new drivers. And the fact that it was challenging means I learned something.
Several people cautioned me about using this driver as a midrange, particularly @wolf_teeth . Now that this project is nearing completion, I can say that this driver is a challenging driver to use effectively, especially in a passive network.
For comparison, I will show the 0-90 degree polar response of the SB15CRC, the SB15NBAC, and the SB15CAC. These are all gated far field measurements merged with near field scans. Each driver was measured in a cabinet, and all three cabinets are different, so the baffle step will be different and the higher frequency diffraction effects will be different. Regardless, I think it will be obvious why the SB15CRC is a more challenging driver than the other two. The first breakup mode on the SB15CRC is a half octave closer to the typical crossover frequency of 2.5k – 3.5k compared to the other two.
I have no regrets about using this driver. Part of the goal of this project was to learn about and use new drivers. And the fact that it was challenging means I learned something.
Some people assert that measurements won't tell you the difference between cone materials.
You can only listen with your ears... etc
Clearly @hifijim can.
Thank you for unwinding, even if just a little bit, the mysteries of different cone materials...
Next up for me- the SB17MFC. Allegedly the smoothest FR of all the SB15/SB17, not only the front side, but also the rear of the cone, as @CharlieLaub will tell you...
You can only listen with your ears... etc
Clearly @hifijim can.
Thank you for unwinding, even if just a little bit, the mysteries of different cone materials...
Next up for me- the SB17MFC. Allegedly the smoothest FR of all the SB15/SB17, not only the front side, but also the rear of the cone, as @CharlieLaub will tell you...
My projects aim to have a dipole radiation characteristic, so I look for drivers with as similar a front and back radiation pattern as possible. There was a "quick" measurement I would perform, where I would use a panel of that 1x1 vinyl coated steel wire material that was formed into a shape looking like this: "∩" that was about 1.5m tall and 50cm wide. The curve of the ∩ was gentle enough so that I could just rest a driver on it and it would stay in place, and as a driver support it is acoustically transparent. I would take an on-axis measurement at around 16-20 inches away and then rotate the driver by 180 deg and do the same from the rear. I could then immediately compare the two responses in my measurement software.
Here is that measurement for a couple of the SB17 family drivers, the CRC and MFC version. I also threw in another standout, the Audax HM130C0. Oops and here is the Textreme cone MW19TX that has very good matching of front vs rear output. It's a little on the large size for use up to 2kHz due to beaming, but you will get a very nice dipole pattern from it. I use it in a project and hope to again. It's much more expensive than the other drivers shown.
I was searching for a nude midrange for my project and could not cross over below 2kHz. A little higher would be even better and would put less strain on the tweeter. The MFC version was best in this regard and of the tens of drivers I tested was a great combination of smooth and extended response as a "nude" driver. It's quite affordable and has decent performance all around (e.g. distortion). The Audax has a smoother FR but the distortion, especially IM distortion, was definitely worse. It's another good option and a cool looking and well built driver, just with an older motor design. I was very happy with the MW19TX and being a bit larger it could also be used to a lower frequency compared to smaller drivers (that start to have too much dipole cancellation as high as 600-700Hz for the 130mm Audax).
As soon as you put the driver into any sort of baffle, the response around 2kHz and other places will change. If you are making a PAP-style or any large planar OB speaker, then your needs for a driver are different. This is really only important when you are using the midrange nude, or in a very minimal baffle. It's my particular niche interest in DIY loudspeakers I guess...
Here is that measurement for a couple of the SB17 family drivers, the CRC and MFC version. I also threw in another standout, the Audax HM130C0. Oops and here is the Textreme cone MW19TX that has very good matching of front vs rear output. It's a little on the large size for use up to 2kHz due to beaming, but you will get a very nice dipole pattern from it. I use it in a project and hope to again. It's much more expensive than the other drivers shown.
I was searching for a nude midrange for my project and could not cross over below 2kHz. A little higher would be even better and would put less strain on the tweeter. The MFC version was best in this regard and of the tens of drivers I tested was a great combination of smooth and extended response as a "nude" driver. It's quite affordable and has decent performance all around (e.g. distortion). The Audax has a smoother FR but the distortion, especially IM distortion, was definitely worse. It's another good option and a cool looking and well built driver, just with an older motor design. I was very happy with the MW19TX and being a bit larger it could also be used to a lower frequency compared to smaller drivers (that start to have too much dipole cancellation as high as 600-700Hz for the 130mm Audax).
As soon as you put the driver into any sort of baffle, the response around 2kHz and other places will change. If you are making a PAP-style or any large planar OB speaker, then your needs for a driver are different. This is really only important when you are using the midrange nude, or in a very minimal baffle. It's my particular niche interest in DIY loudspeakers I guess...