Sounds good, but there's nothing to simulate with a DSP because it delivers the signal to the drivers exactly right, at least electrically speaking.
OK, there's actually a lot to simulate as far as your room, walls, floor, carpeting, furnishings, polar plot of drivers, "house curve" (representing how YOU want your FR to be), etc.
But that's nothing you can do for now. What you can do, is buy a DSP and fiddle till you are happy with the sound (aided by interative REW measurements step by step).
BTW, has anybody mentioned the obvious half-way smart move: passive crossover and DSP to access some DSP powers like fabulous multi-band parametric EQ, effective sub-sonic filtering (see Danley RTA thread in sub forum), etc.?
B.
Last edited:
In today's world, I think passive vs active needs to be expanded to...
passive....., vs active (IIR only)....., vs active with FIR.
IMHO, the benefits of active with FIR are so compelling, there is little reason to consider anything else,
unless maybe you simply enjoy building passives, or you truly need to drive the speaker with only one amp channel.
passive....., vs active (IIR only)....., vs active with FIR.
IMHO, the benefits of active with FIR are so compelling, there is little reason to consider anything else,
unless maybe you simply enjoy building passives, or you truly need to drive the speaker with only one amp channel.
Indeed, passive performance is dreadful and active is soooo much easier to modify, especially the digital versions.
What I actually meant was to stay out of the debate, but to merely use active bi-amp as a design tool for passive crossovers, for the simple reason that the world is not going to rid all its passive crossovers overnight.
There is no other way I can evaluate a tweeter and see how steep a slope would prevent it from audibly distorting, or how deep I need to bury the cone break-up of the woofer underneath to become unnoticeable.
But what I said scored a point for active: any transfer function which can be realized by passive can be done by active, but the reverse is not true. FIR cannot be done by passives, and also one cannot realistically expect to implement 8th-order LR passively.
Good crossover parts are not cheap - at some point extra channels of active amplification are going to be more cost effective.
But I found the hiss level of miniDSP to be quite high. (It is inherent noise shaping dither required by Delta-Sigma converters, not anything that can be fixed by merely upgrading components.) It isn't meant to be audiophile grade equipment and there are better sounding DSPs out there for sure, But the hiss added by a passive is zero.
Only if you're doing it wrong.
I don't know of a simple software route, but this is how I design my passive crossovers and it makes the job a lot faster. I use active with slopes and EQ that can be achieved without too much trouble in a passive crossover, working and changing until I get the summation I want. Then I record or export those active slopes and use PCD or XSim to find the same filter functions with passive parts - and the driver impedance curves. That usually goes pretty fast and gets me very near the target.
Maybe I can add something here.
Passive crossovers are very difficult to get right. Active (DSP) is so easy to get right and can do a little more accurate a job than passive. However ...
Since I have software that can do both and optimize both, I have done many speakers both ways. My latest speakers - my own set of NS15 - are active. But what a pain implementation was. It's not so bad if you only do stereo, but if you use the same speakers for a home theater then implementing a bi-amped system with a Home theater receiver is anything but trivial.
After having my bi-amped system for about a year, I am canning the active system and going back to passive. The differences are just not that great that it is worth all the trouble.
So, if you are a newbie who cannot make a complex passive crossover, then DO go with DSP (MiniDSP is ideal). But if you can do both just about as well, then the ease of passive is hard to pass up.
Passive crossovers are very difficult to get right. Active (DSP) is so easy to get right and can do a little more accurate a job than passive. However ...
Since I have software that can do both and optimize both, I have done many speakers both ways. My latest speakers - my own set of NS15 - are active. But what a pain implementation was. It's not so bad if you only do stereo, but if you use the same speakers for a home theater then implementing a bi-amped system with a Home theater receiver is anything but trivial.
After having my bi-amped system for about a year, I am canning the active system and going back to passive. The differences are just not that great that it is worth all the trouble.
So, if you are a newbie who cannot make a complex passive crossover, then DO go with DSP (MiniDSP is ideal). But if you can do both just about as well, then the ease of passive is hard to pass up.
You are absolutely right and even wise in what you say. That's the problem.
Nobody doubts you can emulate the basic crossover curves of a DSP with a multi-element passive crossover. The issue is the extra DSP powers that can't be emulated that way. In the case of great (or lucky) speakers, those extra elements - whether crossover, EQ, time travel, polarity, loudness of individual drivers, etc. - may not add so much and a passive crossover may be pretty much indistinguishable from an electronic one.
Likewise for the advantages of bi-amping which - in some lucky cases - may not add much even if it is a monumental benefit in other cases (or be simply non-emulatable in the case of subs).
But with a fully crafted passive crossover, as soon as you move the speakers to a new room or even change the location of one of them in your old room, you are back to basics and a real good reason to have a DSP in the system.
B,
Last edited:
Hi!
Yes, passives cannot do FIR, or handle steep x-overs well.
And IIR-only actives cannot do FIR,
and are comparatively limited in the number of IIR filters they can implement vs FIR.
I'm an unabashed FIR head
Sometimes i wonder which property about FIR is more valuable,
the ability to make as many minimum phase corrections on a driver-by-driver basis as needed, via IIR filters embeded into the FIR file,
....or linear phase x-overs.
Both of those properties make for ease of tuning, on and off axis consistency, and simply superior sound IME.
And another yes to your post, I found the DAC hiss high in the mini-DSP's too...but as you mention in fairness, it's not meant to be an audiophile solution.
I'm using miniDSP's openDRC-DI's, staying entirely digital, to feed cleaner DACs, and couldn't be happier.
I have no DSP experience. All my work is passive and I have had literally hundreds of "rats nests" on the floor for decades. Mostly two way systems with ribbons on top. Again easily a hundred or so various designs both OB and boxed, cones, domes, ribbons, and planer magnetics.
A few things that stand out from many long battles.
1- when the crossover is right the whole system can go from so so to surprisingly good. Often with only very small changes and this i suspect is why so many think that passive is just awful.
2- well behaved drivers make getting passive right an easy affair. I admit however that I mostly work with ribbons that can be taken lower than typical domes so I have a much wider window to work with. It makes it all much easier when you dont have to take cones above 1 khz and dont have to deal with limited response of domes into lower freqs.
3- after many years of playing with all sorts of crossover component qualities ( constantly swapping out various types during development work) I am humbled by the fact that the ultra expensive stuff makes little difference. So long as inductor DCR is low,cores far from saturation,resistors are not heating too much, caps are high enough voltage rating , and all components are separated enough , I hear no significant difference. This on quite revealing ribbons of all masses and constructions.
Yes there are the occasional situation where a type of cap is singing on its own or an inductor has loose winds that are rattling etc etc, but apart from the strange I have not experienced the "revelations" some wax on about over certain components.
A few things that stand out from many long battles.
1- when the crossover is right the whole system can go from so so to surprisingly good. Often with only very small changes and this i suspect is why so many think that passive is just awful.
2- well behaved drivers make getting passive right an easy affair. I admit however that I mostly work with ribbons that can be taken lower than typical domes so I have a much wider window to work with. It makes it all much easier when you dont have to take cones above 1 khz and dont have to deal with limited response of domes into lower freqs.
3- after many years of playing with all sorts of crossover component qualities ( constantly swapping out various types during development work) I am humbled by the fact that the ultra expensive stuff makes little difference. So long as inductor DCR is low,cores far from saturation,resistors are not heating too much, caps are high enough voltage rating , and all components are separated enough , I hear no significant difference. This on quite revealing ribbons of all masses and constructions.
Yes there are the occasional situation where a type of cap is singing on its own or an inductor has loose winds that are rattling etc etc, but apart from the strange I have not experienced the "revelations" some wax on about over certain components.
Hi!
I think speaker tuning, whether it be passive or active, is independent of room tuning....or should be for best results.
I agree that it's great to have DSP capability for room tuning.
It's just that IMO, room tuning needs to be accomplished in another layer from speaker tuning, so to speak.
As you say, the room tuning layer varies every time you change rooms, or even change speaker placement.
Speaker tuning doesn't change, which is why a good passive still works.
I have to agree.
Passive crossovers get a lot of stick they don't deserve, largely due to inexperience in passive crossover design on the part of the claimants.
Yes they are harder to get right than active systems, a lot harder in some cases, but the simulation software available today makes it much easier to achieve a really good passive crossover design and considerably narrows the difficulty gap between active and passive designs.
There are certainly some things that can be done with active designs but can't be achieved using a passive crossover, such as:
1) True time delays between drivers to allow for driver acoustic centre offsets.
You can use tricks like unequal filter slopes to bend the phase enough to compensate for small offsets, but that brings its own issues, and while widely used is not an approach I'm keen on.
Or you can use a passive all-pass filter to approximate the necessary delay over a few octaves around the crossover frequency - this can solve the phase tracking issue in the overlap region (at the expense of additional parts) but as it is not a true equal time delay at all frequencies you will still end up with a different overall group delay across the entire spectrum compared to a pure delay.
Or just mechanically design the speaker not to require delays in the crossover!
2) Any types of FIR filter responses of course cannot be implemented in passive crossovers, so you're limited to IIR filter topologies, although you could still in theory use some linear phase DSP correction prior to the crossover to some advantage.
3) Generally only attenuation is possible with a passive crossover, making any correction that requires a peak above the average infeasible without significant complication or loss in efficiency by trying to attenuate everything but the peak.
There are a couple of exceptions though - if you need a boost in response at the turnover point of a filter, for example to compensate for the sagging response of a driver you can often arrange that by deliberately tuning the filter (or one filter section) to be underdamped - by doing so you can actually get a higher output voltage than the input to the crossover, which I don't think many people realise is possible in a passive network.
You get voltage gain because the filter acts as an impedance transformer near its resonance. Thus a lower impedance is presented to the amplifier at that frequency to draw more current and therefore more power, which is transformed into a higher voltage at the output.
Another option would be a step up transformer or autotransformer to do the same thing over a wider frequency range, however I have not felt the need to try anything like this.
The biggest difficulty with a passive crossover though, is understanding the interaction between different components and different filter sections, and that's where the current generation of simulation software really excel in making the job so much easier.
An active DSP design has the nice advantage that each filter section is an isolated black box with no interaction with neighbouring boxes. You can adjust the baffle step correction without affecting anything else, the low pass/high pass section without affecting anything else, any PEQ peaks or notches individually etc... nothing interacts with anything else and it doesn't really even matter in which order you place those black boxes in the signal chain.
Not so with a passive crossover. Everything interacts with everything else due to the load impedance of one section propagating back to the section preceding it.
The order in which the filter sections are placed in a passive design is critically important as you have to think about the load that the following section will be placing on a given section at different frequencies, and that some filter sections may be series and some shunt to achieve the desired result.
For example consider the 2 way crossover I recently finished:
If you look at the order of the sections for the woofer there is baffle step correction (and a peaking filter) first, low pass filter second, shunt notches and zobel network last.
For the tweeter the high pass filter comes first, then the all pass filter, then the L-Pad. There is a good reason for the order chosen for each section.
Trying to design a network this complicated manually (on paper and physically without simulating) and then tune it to the exact response required by substituting actual components would be very tedious and time consuming to say the least, as every time you adjust one component several around it will need re-adjusting as well.
However with the right simulation software, an accurate measured raw driver impedance and frequency response(s) you can get feedback very quickly in real time as you develop the design.
I did the crossover design above from just an idea in my head to more or less completed in a single weekend using Vituixcad. I did further tinker with the values over the following few weeks while I waited for the opportunity to order the components, (pay day
) but the initial over the weekend design was 95% of the way there and only slight component value changes occurred afterwards.Furthermore, when I assembled the crossover and tested the filter electrical response with the actual speakers connected as loads, apart from discovering two out of tolerance components and replacing them, the actual measured filter response was so close to the predicted response that I was able to export the predicted filter response from Vituixcad, import it into ARTA as an overlay, overlay it on top of the measurement and the curves overlapped perfectly... (Within a tiny fraction of a dB)
With the two out of tolerance components I saw significant deviations between the two and that's what brought my attention to the component tolerance problems.
With such accurate simulation possible today and a bit of understanding of the limitations and techniques used in passive designs there is no reason why you can't get a really top notch result with a passive crossover.
Sometimes having a speaker be a single passive box with just two terminals on the back that can be driven by any amplifier is very advantageous. The inside of the speaker is more complicated but everything outside it becomes much less complicated!
Last edited by a moderator:
That's an aesthetic choice, not a technical judgment. Makes no sense to do layer upon layer when you can just do one layer, unless you are a manufacturer. shipping to stores.
We are talking about passive crossovers versus wide-ranging DSP adjustments. True, with a lot of care you can get the crossover middle-frequency band behaviour OK with passive. But that's all you've accomplished.
When you use REW, mics at your chair, listening at your chair, and DSP, you work on the whole intact package to make the whole experience right.
Personal note: with great all-but full-range electrostatic speakers, my interest is focussed on the challenging region where subwoofers come in, below say 130 Hz. While you can emulate good DSP using passive circuits north of say 500 Hz, it is not possible to avoid DSP and bi-amping south of 200 Hz or in ordinary rooms.
B.
Last edited:
It's nice to see some of the well reasoned arguments in favor of passive crossovers. 
Too many get seduced by the whiz-bang technology and sound of active crossovers. "It's active, it has to be better!" I was in that camp for years; until some superb passive crossovers changed my mind.
But hey, I've often used both active and passive in the same system, just like a lot of others have.
Too many get seduced by the whiz-bang technology and sound of active crossovers. "It's active, it has to be better!" I was in that camp for years; until some superb passive crossovers changed my mind.
But hey, I've often used both active and passive in the same system, just like a lot of others have.
I know we've discussed this point in another thread recently but I have to agree.
In particular one thing I've found is that anything that affects the frequency response over a large number of octaves, such as the relative attenuation of two drivers in a multiway system, becomes unduly critical.
I was literally noticing a 0.2dB change in the amount of baffle step correction in the midrange (relative to the tweeter) or in the tweeters attenuator (relative to the midrange) as an obvious change in the imaging and presentation that changed it from "great!
" to "doesn't sound right. ".I'm sure some will scoff and say that 0.2dB is impossible to hear, and I'm fooling myself, but that's what I've found time and time again. When you start to get close, the relative attenuation of the drivers is unduly critical to get that last mile from good to great, and while it might sound fine on a pink noise test when it is slightly out it won't image properly.
This is where a passive crossover starts to become a bit of a pain - adjusting the L-Pad for a tweeter by 0.2dB is a hassle, because to do it right you have to adjust both series and parallel resistors by very small amounts to maintain the same impedance. If you don't, the shape of the tweeters frequency response changes as well. (And even if you do, it may still change slightly since the tweeter's impedance isn't flat)
Or if you adjust the baffle step resistor on the woofer, if you change it more than a little, you'll also have an undesired effect on the roll off shape of the low pass filter that will confuse the issue.
So the problem of component interaction becomes quite problematic when tuning a passive network, and that's where some give up and declare passive as too hard or just no good.
But I find that a very accurate simulation based on very accurate measurements is a godsend here. I do my tweaking and component interaction correction in the simulator first and then make the physical component value changes after I know exactly what effect it is going to have. So I work out exactly what changes I need to shift the tweeter by 0.2dB, verify that it actually does so and without any undesired interaction, and then try it.
By the way, the extreme sensitivity to relative driver attenuation to get optimal results is one thing that turns me off bi-amping or active crossover designs.
With a bi-amped system, whether passive or active crossover, you have the added variable of non-matching amplifier gain, especially if the amplifiers themselves are not provided with the speakers, for example internal amplifiers.
If you just give a bi-amped speaker to someone to hook up to amplifiers they have provided themselves, who is to say that the gain matching of the two amplifiers is closer than 0.1dB ? Chances are, they are not matched that closely.
Also you are powering each driver using different connectors and cables, that may suffer from different connection quality. For example banana connectors are notorious for high and intermittent contact resistance - in a bi-amped speaker this can lead to a relative shift in attenuation between the two drivers.
With a passive design, although it's more difficult to adjust the attenuation and get it bang on during the design process, once it's done, it's set in stone, and provided that there are only good soldered connections between the crossover and driver units (oxidised spade connectors need not apply) it shouldn't ever change in the life of the speaker, as any slightly dodgy external connections will tend to affect both drivers equally.
Even small ribbons have a much flatter impedance curve than non-ferrofluid domes. In the crossover I posted above, there is no resonance compensator on the tweeter - it just doesn't need one as the impedance is very flat, with only a very minor bump a full octave below the crossover frequency.
Exactly the same opinion here. The super expensive golden ear crossover components that cost 10x as much as normal decent quality components are just snake oil like monster cables. There is no audible difference in quality.
For the crossover I just did I used relatively affordable Jantzen coils, ceramic resistors and their metalized polypropylene "cross cap" series. Not the cheapest you can buy, but not expensive either.
The main area of concern for a passive crossover and a point of weighing up balances is inductors of course - I prefer to use large air cored where I can as you can eliminate any concerns of saturation, and did for all the coils for the tweeter and for the low pass filter coils for the woofer, but the baffle step correction coil is laminated steel and the others are permite bobbins.
Interestingly I did originally have a permite core for the baffle step coil as well, and despite it having a power rating far in excess of the speakers I could hear some saturation effects on a few very specific songs (midrange breakthrough due to the inductance dropping during saturation) so I changed it to a laminated steel I core and the problem went away - no audible distortion I can hear even on heavy bass.
So coils certainly do vary in their effectiveness.
Agreed 100% (except possibly only for where you express your dislike for asymmetrical slopes: these can work very well in my experience).
Just for grins, here's the result of one of my own passive crossover labours
Marco
I would argue that a speaker crossover should not be trying to bake in room corrections, whether passive or active, and therefore doesn't and shouldn't need to change when moving the speaker from room to room.
If the crossover needs changing when moving the speaker from one room to another to sound right then there is probably a design problem with the speaker, especially if the crossover that you are changing is between midrange and treble frequencies, which shouldn't need any changes from room to room unless there are issues like poor polar response etc, which should be addressed directly in the design.
Bass correction up to about 200Hz is an entirely different thing, and can and should be applied actively.
One potential pitfall with an active crossover/EQ is that it's easy to muddle together the parts of the correction that are there to correct for actual speaker response errors, with the corrections made for the room.
My approach is that crossovers and room correction should be kept segregated. In my current system this is done by the crossover being passive, which contains both the high pass and low pass filters along with various corrections (notches etc) for the drivers to give an optimal anechoic response with a standard amount of baffle step correction.
They sound pretty good and reasonably balanced in room even with no additional room EQ or correction, although the lower room modes can be a bit boomy as I sit right at the back wall.
Then on top of that I use a DEQ2496 to provide active EQ for room problems <200Hz, in the form of 3 PEQ notches and a slight shelf increase in bass below 100Hz to compensate for the overall loss of bass level introduced by the notches.
This room correction transforms the bass response, practically eliminating any room mode effects at least at the primary listening position.
If I was to move these speakers to a new room I would start over with the EQ for below 200Hz to match the new room, however I would not change anything about the crossover response in the speaker, nor would I try to equalise anything above about 200Hz with the DEQ to try to solve a room problem.
Doing that just doesn't work.
Last edited:
I ran a system that was a mixture of both passive crossovers and active EQ for years and it worked very well. One thing the active EQ was responsible for was all the baffle step correction!But hey, I've often used both active and passive in the same system, just like a lot of others have.
A lot more efficient than doing it in a passive crossover.
Simon, I followed a similar path to yours. Get the speaker right with the crossover (active or passive) then do room EQ with the DEQ2496. I probably cheated some with speaker correction in the DEQ.
I understand Ben's approach to build it all into the crossover, but I find that a little clumsy for me. I have heard it done to good effect purely passive, so many approaches work.
I understand Ben's approach to build it all into the crossover, but I find that a little clumsy for me. I have heard it done to good effect purely passive, so many approaches work.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.