Well, I have both hybrid and fully active options for my NaO II system. I have yet to have anyone ably to identify which is which when playing the system with one speaker fully active and the other channel passive on the panel, let alone the fully active system vs the hybrid.
In theory you can list lots of things that could contribute to differences. Whether they make a practical different or not depends on implementation.
There are a lot more issues with analog active. What what topology to use, discrete transistors (or tubes) or opamps, what kind of caps, resistors, circuit layout, do you cascade text book filters like or combine stages were possible. Do you use VSVC, IG-MFB or biquad filters....
First I would like to thank 5th element for his extensive comments at # 23 on my earlier post. It is indeed not as black and white as I depicted. Horses for courses and all that. For simple situations, passive might indeed be the preferred choice. There is one thing however in 5th element's post that I don't quite agree with, and that is the influence of voice coil temperature on passive filter behaviour. This can be very significant, certainly with higher order filters. Le is a bitch too.
On AlanB's post, I'd like to share the approach I have been taking for a while.
The first step is to measure the drivers in their enclosure, and to fine tune with MiniDSP the curves that work best to get the desired combined response. Next step is to translate this into an analog xover that mimicks these curves. The top screen shows a simulation of one leg of the analog filter in LTSpice. The final step is to build the filter and measure it's behaviour.
This is what came out at the end of the process.
Now, why not stop after the DSP-stage? Some reasons for this (but it is highly dependent on the end use):
- I like to see the amplification including the xover as close as possible to the drivers. With DSP, you get gain structure issues, which you don't have with analog.
- analog sounds better to me (although not by a wide margin).
- analog is way cheaper, even using high quality components, than DSP.
vac
But didn't you buy a miniDSP in order to create said passive crossover. Doesn't seem that cheap to me once you factor that in
After seeing the arguments in this thread I don't think you can really compare active to passive. Its clear to me that they're both very different routes and folks have different priorities so picking the 'best' is never gonna happen. Also comparing DSP to passive is underselling DSP because you have to cull all the useful stuff in order for it to be a fair comparison.
I see it like this, if you look at some of the best from each breed like that Dolby Lake or the DEQX and pitch it against a really well sorted passive then generally active is going to have the advantage because room correction and driver linearisation can be game changers but as I've said that isn't a fair comparison and the two aren't directly comparable in most situations.
I think either aproach can work equally well. Passive is cheaper but there in no real flexability. Once you design and build a crossover for a driver set it's not like you can just substitute new drivers and expect the same results as with the original driver set. If you are doing a lot of driver swaps active DSP makes thing a bit easier and quicker.
Rob
Rob
Is what I said in post 23 that states that voice coil heating will affect a passive crossover.
Le can be a bitch, but it's as much of a bitch in a passive crossover as it is in an active one. Le is reflected directly in the loudspeakers unfiltered frequency response, if Le is higher then the bandwidth is reduced.
In an active crossover you can somewhat counteract the effects of Le by applying some boost. In a passive loudspeaker you are of course able to apply a zobel network, this will flatten the impedance and possibly make the drive unit easier to drive. You can't do this actively (yay for passive?), but then again, you are always free to add a passive zobel network to an active design.
This is simply not true. DSP chips aren't that expensive. The Sigma line from Analogue devices (as used in the minidsp) cost less then $10 when bought in bulk for the top line processors. This will handle 192/24 and will have more then enough power to easily implement a 4 way stereo system. They also come with a built in SPDIF receiver.
There are a plethora of different ways that one can implement the 8 channels of D/A conversion and at a wide range of different price points. The same can be said for the clocking scheme you wish to use and if you wish to use an A/D converter.
An analogue implementation of an active crossover for a 4 way system can get expensive pretty quickly. Currently my 4 way as done on the DSP requires 26 2nd order hi/low pass filter stages, 3 Linkwitz transform circuits, 3 shelving filters, 6 notch filters and 2 delay networks. That is a lot of opamps and filter caps and could reach as high as 50 opamps if you're going to cascade standard Sallen Key filters. This will also require some quite relatively large film capacitors which will cost quite a bit.
The D/A conversion (that I've chosen to use) on the other hand requires 24 opamps and only relatively small value film caps. With the digital system I could also double the number of filters and still only require the same D/A circuitry, whereas the analogue version would basically double in cost. D/A converters can be pretty pricey for the top models, but something like a PCM1796/8 are certainly no slouch and can be had for as little as $3 when bought in bulk. You'll want to incorporate some form of 8 channel volume control too if you want to maximise sound quality and a CS3318 will do that for you and do it quite cost effectively, interestingly though it will probably be the single most expensive IC in the entire system.
At the end of the day though, which is more expensive, for an equivalent level of performance, is entirely related to what the end system requirements are. Just as an analogue active crossover will be cheaper and outperform a passive version in certain situations, so will a digital active crossover vs an analogue active crossover.
Hi John, long time no see
I'm completely with you on this one and I did the exact same test myself with Summas. Side by side there was no detectable difference, either measurable or audible.
But I will certainly agree with the guy who is trying get a good response with his horns - go active, its a whole lot easier, (passive crossovers to horns are a real PITA) and the sound will be the same. Its just more expensive thats all.
Personally I would have thought that Dynaudio sells speakers with decent passive xovers.
They use 6dB slope ones for all the usual reasons and the same in the active variety. Frequency remained the same to the best of my knowledge (can't check as they revised the model since and deleted the passive from their line-up).
In the case of my Tannoys I changed a decent passive 12dB Butterworth to 24dB L-R active, frequency remained the same as it is nicely placed where the 12" woofer narrows to 90deg for the 90deg conical treble horn to take over.
As a nice bonus the removal of the series inductor increased the woofers output from 92dBspl to 95dBspl at 1W/1m. Interestingly Tannoy quotes the passive woofer section as 100w while the bare driver is clearly marked 50w.
Tannoys present a totally benign typical 8Ohm load, never dropping below 5.5.
JFYI
They use 6dB slope ones for all the usual reasons and the same in the active variety. Frequency remained the same to the best of my knowledge (can't check as they revised the model since and deleted the passive from their line-up).
In the case of my Tannoys I changed a decent passive 12dB Butterworth to 24dB L-R active, frequency remained the same as it is nicely placed where the 12" woofer narrows to 90deg for the 90deg conical treble horn to take over.
As a nice bonus the removal of the series inductor increased the woofers output from 92dBspl to 95dBspl at 1W/1m. Interestingly Tannoy quotes the passive woofer section as 100w while the bare driver is clearly marked 50w.
Tannoys present a totally benign typical 8Ohm load, never dropping below 5.5.
JFYI
If you want to cover the full range, and get smooth bass response, you pretty much have to go active - at least for the subs. My system is based around Unity horns (waveguide with mids combining near the throat of a compression driver). The compression to mid crossover is done passively (but could be active) and the waveguide to midbass is active (though it could be passive), as is midbass to subs. The response of the waveguide doesn't have to be perfect, because the overall response is corrected with a DEQX. I plan to use the Geddes approach for the subs, which will require an DSP additional modules for at least one of the subs.
I can mix and match amps, because the DEQX allows a lot of flexibility in adjusting the gain structure, while preserving digital dynamic range. And, because my amps are all DIY, I can easily adjust the gain structure at the input, if I need more range - as in combining low powered tube amps with higher power ss amps. I always set it up so that the amps never clip.
Sheldon
Hi Earl,
Yep, I don't post much anymore. Not much interested in discussing what color blue the sky is, if you know what I mean. I guess after all these years I'm getting a little jaded.
What color Jade?
JADE?
They look nice in black, and Peter Comeau has managed to make an allegedly "phase linear" 4 way for your passive aperiodic pleasure!:
http://www.wharfedale.co.uk/Products/Product/tabid/78/PID/244/CID/249/language/en-GB/Default.aspx#detail
Copy that for the asking price!
The fact they miss-spell Klippel in the techno section doesn't exactly fill me with confidence
In the PDF they get it right, but still, not good!Edit - I think if you wanted to clone this, to a certain extent you could use a vifa DQ25 a scan speak 10F, a SB acoustics SB17NRX35 and a pair of RS225s per side. And without knowing what Wharfedales xover does, those would certainly give a huge amount of driver overlap and smooth extended responses within probably what would be in their pass bands and well into their transition bands. This is kind of getting off topic though. Anyone want to hazard a guess at what their crossover looks like?
Last edited:
Passive crossovers can be simple and deliver excellent results, provided you use quality bass-medium speaker drivers exhibiting moderate cone resonance in the highs, and provided you use quality tweeters producing a low distorsion even when receiving some attenuated medium frequencies. Bass-medium speaker drivers and tweeters meeting those specifications are not the cheapest.
There are some high quality bass-medium speaker drivers on the market equipped with metallic or carbon-fibre membranes, delivering high power, low distorsion and a reduced high frequency directivity, unfortunately exhibiting one or two huge resonance peaks in the highs.
When using bass-medium speaker drivers exhibiting more cone resonance in the highs, the passive crossover needs to be high order : at least 4th-order, with an impedance correction network, and with a frequency notch. Such complicated passive crossover, built using long-lasting parts, could costs more than a comprehensive active crossover put after the volume control, feeding two power amplifiers built around a TDA7293 or LM3886, having a common power supply. Even is if this is not the optimal layout for an active filter, it can help you get fine results at a decent cost when not using expensive speaker drivers.
If you target a perfect acoustic wave reconstruction, you need a linear phase.
If you target a wide and smooth directivity, you need zero relative phase shifts in the transition band.
Analog crossovers (passive or active) can't deliver both. With passive crossovers, even if it is possible to get a linear phase, you still get a relative phase shift in the transition band, something like 90 degrees.
Digital crossovers, when running a Lipshitz-Vanderkooy delay-compensated crossover with a 4th-order lowpass Bessel function, can deliver a perfectly reconstructed wave (linear phase) and can reduce the relative phase shift in the transition band to something like 30 degree.
In crossovers, you need to a precise time alignment of all speaker drivers. This is usually not done using analog (passive or active) crossovers. This is sometimes done by mounting the woofer on a thick baseplate (see Cabasse), however this may induce diffraction in the highs. Time aligning all speaker drivers is easy in digital, using delay lines.
There are some high quality bass-medium speaker drivers on the market equipped with metallic or carbon-fibre membranes, delivering high power, low distorsion and a reduced high frequency directivity, unfortunately exhibiting one or two huge resonance peaks in the highs.
When using bass-medium speaker drivers exhibiting more cone resonance in the highs, the passive crossover needs to be high order : at least 4th-order, with an impedance correction network, and with a frequency notch. Such complicated passive crossover, built using long-lasting parts, could costs more than a comprehensive active crossover put after the volume control, feeding two power amplifiers built around a TDA7293 or LM3886, having a common power supply. Even is if this is not the optimal layout for an active filter, it can help you get fine results at a decent cost when not using expensive speaker drivers.
If you target a perfect acoustic wave reconstruction, you need a linear phase.
If you target a wide and smooth directivity, you need zero relative phase shifts in the transition band.
Analog crossovers (passive or active) can't deliver both. With passive crossovers, even if it is possible to get a linear phase, you still get a relative phase shift in the transition band, something like 90 degrees.
Digital crossovers, when running a Lipshitz-Vanderkooy delay-compensated crossover with a 4th-order lowpass Bessel function, can deliver a perfectly reconstructed wave (linear phase) and can reduce the relative phase shift in the transition band to something like 30 degree.
In crossovers, you need to a precise time alignment of all speaker drivers. This is usually not done using analog (passive or active) crossovers. This is sometimes done by mounting the woofer on a thick baseplate (see Cabasse), however this may induce diffraction in the highs. Time aligning all speaker drivers is easy in digital, using delay lines.
Last edited:
The fact they miss-spell Klippel in the techno section doesn't exactly fill me with confidence
Yes I agree that's bad, that will be the Chinese marketing side not the developer. Although smelling pistakes are surprisingly common in commercial sales literature.
I was going off topic, running with the Jade theme!
It's good to see closed box and aperiodic designs, but that's just personal preference.
on with the proper debate................
We don't need no activation, we don't need no driver control, hey designer, leave those cap's alone, all in all inductor, your just a filter that's all!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.