Mechanically passive cardioid doesn't use or require electronics. It's directivity cannot be tuned electrically.
Hmm..... thanks but not exactly what I was looking for.
By number of "bits" you mean number of samples?
I do see how the FIR has a limited length over a period of time... which means that with the proper number of samples over a given bit rate, and with a proper "decay" it is a temporal filter. Meaning it does go away, but if you have a low sampling rate and a long number of bit you'd run into a the issue that it could chase itself like an IIR....
In any event, i've been looking for a more mathematical description... but at least you did describe the "effects". For sure, we don't want to see pre ringing.
By number of "bits" you mean number of samples?
I do see how the FIR has a limited length over a period of time... which means that with the proper number of samples over a given bit rate, and with a proper "decay" it is a temporal filter. Meaning it does go away, but if you have a low sampling rate and a long number of bit you'd run into a the issue that it could chase itself like an IIR....
In any event, i've been looking for a more mathematical description... but at least you did describe the "effects". For sure, we don't want to see pre ringing.
It requires electronics because an 8 inch woofer with a pair of vents right behind it has a tapering frequency response below 200 Hz, and requires a bass boost to meet the subwoofers at 100 Hz.
Optimum is: "greatest degree attained or attainable under implied or specified conditions".
So, we are talking the same.
Cardioid speakers narrow the radiation pattern of the driver on the front baffle by sending the driver/s on the side/rear a delayed, reduced and inverted signal. The idea being to largely cancel the delayed and reduced sound from the front driver when it reaches the side /rear driver/s leaving mostly just the sound radiated forward. One can also create the delayed, reduced and inverted sound by sending the radiation from the rear of the front driver down a stuffed pipe but the cancellation only works well over a relatively narrow frequency band compared to using extra drivers.
The Beolab 90 has groups of similar drivers in triangles. The upper driver has a wide radiation pattern whereas the lower pair has a narrower radiation pattern. By changing the relative amplitude of the upper driver and the lower pair one can vary the width of the radiation pattern with frequency between that of the single driver and a pair of drivers.
There are options but, as you say, it involves playing with the relative phase and amplitude of multiple drivers. There is a price to pay in terms of multiple DSP controlled channels and needing lots of cone area to maintain clean SPL due to sound cancellation but it allows control of the dominant factor determining in room sound quality (high fidelity rather than audiophile sense) for speakers using drivers of adequate size and quality.
Hi. I'm a big advocate of open-architecture processors found in the prosound install world.
(I use q-sys and believe open-architecture DSP is DIY's gift from heaven 🙂)
I'm curious about the degree of open-architecture design available, with your processors and Audio Weaver. Thx.
Oversimplistic blanket statement.
All else being equal, active would be better than passive, and DSP active could make some things possible that would be impossible with analogue active.
BUT... all else is NEVER equal, and there can be many reasons why a superbly engineered passive system may sound better than a relatively cheap DSP solution.
I've personally tried several passive, analogue active and DSP solutions in my own system, and could clearly hear trade-offs... because in the real world, with limited budget, all systems will have compromises. To claim that one solution is always intrinsically better is miopic.
... and be aware that this nice (!) paper is from 2019: " ... Even the 3072 tap FIR filter can’t achieve the high-Q magnitude change desired at ~65 Hz. (It actually takes over 10000 taps at 48 kHz for the FIR filter to match the desired EQ.) ..." This reads as if FIR filters are inherently problematic because the need of long filters for a decent resolution expecially at lower frequencies. Today, it's no more probematic at all. A Rpi 4 may easily handle 8 FIR filters with 128k taps each. I consider a FIR filter length of 64k standard for a 44.1kHz/48kHz SR audio content. For SR of 88.2/96k I go with filter lenght of 128k as my personal compromize.
It's telling that you have to create this specific scenario of well engineered passive vs. poorly engineering active to support your argument.
Idc what any forum user has to say, active is objectively better if the goal is good summation of multiple driver sources. Frankly passive is just a waste of time to me. The pro world and commercial speaker market knows this already, lots of diy folk are playing catch up.
Just another dead horse being beat again and again. This place is just an echo chamber of noise isn't it?
You're missing (and/or misconstruing) my point.
I agree that active is theoretically better in principle.
I was just warning against misguided overconfidence in active always sounding better than passive, because, well, it depends! And DSP on its own is no guarantee of good sound, unless it's implemented correctly by someone who really understands what they're doing.
Also, the cheap AD/DA chips used in many DSP implementations aren't really transparent, as you quickly realise once the rest of the system is really, really good. Hence, what may look like the better option on paper (i.e., a well-designed DSP crossover), may not actually sound fully satisfactory when auditioned, even if compared to a theoretically inferior passive alternative.
Been there, done that myself.
I agree that active is theoretically better in principle.
I was just warning against misguided overconfidence in active always sounding better than passive, because, well, it depends! And DSP on its own is no guarantee of good sound, unless it's implemented correctly by someone who really understands what they're doing.
Also, the cheap AD/DA chips used in many DSP implementations aren't really transparent, as you quickly realise once the rest of the system is really, really good. Hence, what may look like the better option on paper (i.e., a well-designed DSP crossover), may not actually sound fully satisfactory when auditioned, even if compared to a theoretically inferior passive alternative.
Been there, done that myself.
Did anyone had the opportunity to listen to helios TX?
Any opinion on that?
It is free published at Loudspeaker pad page, and as far as I understand, not many speakers were DIY at axpona. At least free published DIY.
Any opinion on that?
It is free published at Loudspeaker pad page, and as far as I understand, not many speakers were DIY at axpona. At least free published DIY.
I think everyone agrees that if you define the goal to be something such that active is better, then active is better.
But you already made one of the best cases for passives earlier. I will have several expensive high quality stereo amplifiers. I also make a lot of speakers for family and friends. Active with NEVER EVER be the best alternative for me. TRADEOFFS.
Well, a really good active vs passive thread can be useful (like a couple on ASR). The main problem here is that this is the "my experience at Axpona" thread and the proponents of active trend to be very vocal and get pretty worked up when blanket statements about the superiority of active are questioned or dismissed.
I have only heard maybe 5 systems that were active that could hold a candle to a properly built and designed passive system, and I can usually tell which is which. Last year at AXPONA, I was really disappointed the active ATC 12" speakers did not sound better to me, just as an example.
I don't think it was Eclipse Audio's intention to create the impression FIR filters are inherently problematic for low frequency work.
Their market/audience is the prosound market, where latency is typically not tolerated...and then for sure, long FIR filters don't work well for low frequencies.
A Rpi or PC processor in general is very unlikely to be welcomed as a prosound processor.
Most prosound DSP amps and/or processors are very tap limited because who cares, if latency can't be tolerated.
It's worth noting that Fulcrum Acoustics includes FIR filter blocks in the multi-way presets they make for most all major processor brands, to be used with their speakers. The the FIR filters are only 384 taps !
Personally, the deeper I've delved into FIR, the more i've become convinced I'm getting better sound quality using as few taps as possible to get the degree of correction desired. Which begs for taps decreasing on a frequency dependent basis...which begs for each multi-way section having a different number of taps.
For the sub section ,the only task I give FIR is to achieve a complementary linear phase crossover with the low-mid section. All other work, EQ's and high-pass are IIR/min-phase. This gets taps down to about 6k at 48kHz, for any order xover at 100Hz. So ironically, after chasing all the taps I can find for many years, I now feel I don't need more than 6k per channel (and only a few hundred for VHF channel.)
I've just did a quick glance... it looks good. Thanks. I'll deep dive into it this afternoon.
Limitations based on processing do not affect the math behind the description.
Example, in '97 we were doing complex numerical analysis, it took us four Sun Ultras with an optimized interrupt handler running FORTRAN to get our daily data models done in 16 hours.
Today we can do that algorithm in a single multi core Xeon in just a couple of hours, max.
So, the algorithm implementation has to fit the processing available but that's just an implementation decision not the basics of the algorithm.
Thanks for the observation nonetheless.
Most important addendum:, I don't want to contest Eclipse Audio. Instead, I appreciate any background informations such as found in theirs very nice and educative paper.
Pro audio indeed has different primary requirements than home audio. I am no pro user, but a home user, audio only, no video, therefore not needing small or zero latencies. This is why I can afford use long LinPhase FIR filters, and I use them. For simplicity and pure comfort's sake. And I encourage anyone to do so if RT audio is not strictly required. It was my intent to point out not to read the Eclipse Audio paper in a discouraging way in terms of FIR filters.
This is a technically beautiful approach to mix IIR and FIR filters where they shine the best. This is art and craftmanship at the same time, while exactly knowing what you are doing. I like that kind of expertise, and can imagine that a good result after having done so gives a lot of satisfaction.
Just to mention another tweak, in order not to forget this alternative approach: One also could optionally resort to MixedPhase FIR filters (with their latencies between the MinPhase and LinPhase filters) in order to optimize a system for low latency. I admit never having done that one.
Maybe the RT/lowLatency problematic might be one reason why many commercial home-audio loudspeaker producers better stick to passive systems: Garanteed nobody then will complain about a delayed audio while watching a movie with a passive system.
Last edited:
Thank you for the kind words, Daihedz.
And please know I didn't mean to imply you were contesting Eclipse Audio. I thought you were simply helping folks put things into perspective ala long filters.
Just like i was trying to do the same, describing their target proaudio market.
I think once video delay in DSP gets as easy, precise, and affordable as audio delay is, we will see more actives, particularly lin-phase.
(JRiver might already have video delay..dunno.)
And please know I didn't mean to imply you were contesting Eclipse Audio. I thought you were simply helping folks put things into perspective ala long filters.
Just like i was trying to do the same, describing their target proaudio market.
Yes. I have a FIR generator program that does that, although I haven't played with it much, so i don't know how it works yet.
Good point.
I think once video delay in DSP gets as easy, precise, and affordable as audio delay is, we will see more actives, particularly lin-phase.
(JRiver might already have video delay..dunno.)
I apologize if I missed this by not reading every post in the thread yet, but can anyone tell me when and where the Iowa get together is? I live in the Des Moines area and would love to attend.