I'm working with 17NBAC 4ohm version, and yeah while nice there's hint of 'metal cone' coloration. Slapped RLC trap to reduce 4-9kHz area, capacitor across coil to remove the cone breakup hash, but while quite clean and detailed, there's something there, a glare as you described it. I guess it's just the way they behave, sort of like alu done tweeters vs fabric domes.Getting back on the M105 copies;
After living with the M105 'copies' there is something that bugs me - it's like a sheen/glare over the entire midrange. I kept trying to ignore it but eventually I couldn't and it bugged me. Perhaps I failed to supress the woofer breakup sufficiently. I suspected the woofer and have since swapped it for a paper cone scan speak - if the problem goes away I will provide updates with the scan speak unit. Here's preliminary quick data
View attachment 200672
View attachment 200675
Very interesting building and this 17NBAC youre mentioning seems to be very good . Maybe you have to do a crossover with at least 24 dB/oct slopes because there IS a nasty breakup higher in freq and a notchfilter in my opinion doesnt do any good for the sound .I'm working with 17NBAC 4ohm version, and yeah while nice there's hint of 'metal cone' coloration. Slapped RLC trap to reduce 4-9kHz area, capacitor across coil to remove the cone breakup hash, but while quite clean and detailed, there's something there, a glare as you described it. I guess it's just the way they behave, sort of like alu done tweeters vs fabric domes.
This paper Scan looks well behaved, without the 'hash' of alu SB midwoofers.
It is 24dB/oct. IMO most 6" drivers need this, I'm actually a bit sceptical how J.Bagby used 12dB in his Satori builds. Seas ER18 maybe isn't as bad as it looks. The SB17NRXC isn't easy to hammer into LR4 either, shooting into Himalayas past 4kHz but I hear it. So definitely paper must dump the resonances better. You might be right about something steeper than 24dB/oct for some alu cones, 36 might work nicely, just the parts count will high.Very interesting building and this 17NBAC youre mentioning seems to be very good . Maybe you have to do a crossover with at least 24 dB/oct slopes because there IS a nasty breakup higher in freq and a notchfilter in my opinion doesnt do any good for the sound.
Lesson learned for me - If the driver has a nasty breakup such as in seas er18rnx or 17NBAC its probably much better to do steep crossovers than start to implement notch filtering .
17NBAC has no more than about 3 mm linear extension so it should also be crossed at about 80 Hz or higher.
Maybe metal coned midranges are better suited for steeper active crossovers , even if they look quite good on raw measurements ?
Apparently that the only way to reduce the non-linear distortion components: https://www.diyaudio.com/community/...er-speaker-builds.352063/page-70#post-6983512Thanks Auger. Hmm, I looked at Lars network for his Purifi 6" paper cone and couldn't figure out why he used series RLC this way, when there are easier ways to hammer the freq resp into specific transfer function. Back to drawing board and some head scratching.
thx again
It seems like this is an inherent advantage to using passive crossovers over active crossovers. I've never considered this.Apparently that the only way to reduce the non-linear distortion components: https://www.diyaudio.com/community/...er-speaker-builds.352063/page-70#post-6983512
Active is certainly better overall. I was just noting that this is one particular advantage that passive has over active. The optimal setup might be an active setup that also has passive series notch filters to tame breakup-induced distortion.With active, especially DSP, there are more possibilities than passive. With passive we are married to driver's freq response characteristics, impedance and the way crossover parts interact with them. Also with large excursion and higher than average signal level certain parameters change their nominal value and might affect the crossover network (slope). The amplifier sees the combined impedance of the crossover and transducers, with all the impedance dips and peaks, plus phase changes. DSP can add specific EQ, notch filters, time delay, it's mostly immune to parameter change of the drivers with signal level, and drivers are directly coupled to amplifiers. So, I would say that active crossovers should actually have better distortion performance and better control over the driver(s) transfer function. Most studio monitors are active in design for a reason.
Just read the paper and I don't see any advantage for a passive crossover. All filters can be implemented much more precisely in DSP, and since the driver is connected directly to the power amp its very low output impedance can control the driver better than through a passive network.It seems like this is an inherent advantage to using passive crossovers over active crossovers. I've never considered this.
Will take a look for you soon.McFly,
it would awesome if can you make frd zma files available. I would like to check what are the possibilities other than x-over you presented (more filtering above 4k, where Alu cones show their character). And of course I'm interested what the sb26adc in augerpro vaweguide can do.