Advanced passive crossover design
Printed From: Speakerplans.com
Category: General
Forum Name: Advanced Discussion
Forum Description: Advanced discussion area for higher lifeforms
URL: https://forum.speakerplans.com/forum_posts.asp?TID=108372
Printed Date: 26 March 2026 at 10:04pm
Software Version: Web Wiz Forums 12.08 - https://www.webwizforums.com
Topic: Advanced passive crossover design
Posted By: kevinmcdonough
Subject: Advanced passive crossover design
Date Posted: 23 February 2024 at 12:48pm
|
hey all Although I've designed and built quite a few speakers over the years, I've never really spent much time learning about passive crossover design. I've just run active, or made up a circuit from a pre-existing spec. While I know the basics in a vague sort of way and am watching lots of tutorials and things to learn more, I'd like to get some input and start some discussion on how to take it to the next level. What makes a good crossover a great one? People like Tom Danley and Ed Kinsella for example are known to design exceptionally high quality crossovers that are a huge step above the norm. What do they do different, what other components are they including, what is it that makes theirs stand out? I think it would be a great learning opportunity for everyone to have some discussion on this, and link to any resources anyone may have that could give us some insight. Annnnndddd, go...  K
|
Replies:
Posted By: VECTORDJ
Date Posted: 23 February 2024 at 12:57pm
| I use Xsim....Free and works good...VECTORSONICS |
Posted By: kevinmcdonough
Date Posted: 23 February 2024 at 2:14pm
I've saw that one, and I've also saw https://kimmosaunisto.net/" rel="nofollow - VituixCAD which looks really useful. My question is more about the theory really, rather than trying to calculate specific crossover values. Some crossovers have just a few basic components and do the job fine, but some are very big and complicated. The people I mentioned are renowned for making very complex crossovers that have lots of additional components and are seen as little works of art. I'm more trying to understand (and start a conversation to help others understand and learn) how/why they are doing this. What do they do differently that makes their crossovers stand out, what extra things are they adding and why? How do they elevate their crossovers to the next level. Kev
|
VECTORDJ wrote:Posted By: MarjanM
Date Posted: 23 February 2024 at 2:50pm
|
There are two types of crossovers. Simple one that pretty much only divides the bands and a complex one that actually shape the frequency response. Both types works well you just need to decide about you approach to the things. You want to correct everything by the dsp (which is still needed) or you want to "fix" as much as possible with the crossover it self. The complex crossovers do tend to create certain amount of loses because every single element adds a bit resistance in the chain. ------------- Marjan Milosevic MM-Acoustics www.mm-acoustics.com https://www.facebook.com/pages/MM-Acoustics/608901282527713 |
Posted By: Conanski
Date Posted: 24 February 2024 at 6:06am
A basic crossover simply creates separate frequency bands that are safe for the transducers being used, to get really good results with a simple crossover like this requires really really well behaved transducers so in general this just isn't enough to get even halfway decent results with pro audio speakers. To go to the next level we get into phase alignment, impedance flattening, response shaping, and level matching. To accomplish those things techniques like baffle step compensation, zobel networks, notch and shelf filters, and L-pads are added to flatten overall response and extend the extreme high and lowend response. Baffle step compensation isn't used as much in pro audio because it throws away so much raw sensitivity if used heavily, and because the same results can be achieved with a bit of EQ boost that is redily available in most cases. A really good passive crossover design can get into the ballpark of what is possible with well executed DSP processing but it will never match it, and it will take more time and effort to get those results.
|
Posted By: Conanski
Date Posted: 24 February 2024 at 6:09am
| duplicate |
Posted By: Earplug
Date Posted: 24 February 2024 at 6:41am
|
I've gotten good results using a "simple" passive xover plus 31 band eq to smooth out the response. Trying to do everything with a passive becomes a PITA. At best you can hope for 1 or 2 notches, or some attenuation. All that is far more easily obtained with a graphic, or parametric eq. Or DSP board. They are so cheap these days it doesn't really make sense to sweat over designing a (complex) passive anymore.  ------------- Earplugs Are For Wimps! |
Posted By: fudge22
Date Posted: 24 February 2024 at 5:58pm
|
I agree with this. Simple crossover circuits assume that the
load, i.e. the drive unit, has a constant resistive load. It doesn’t. Imagine building a loudspeaker cabinet for a 12” speaker,
and then picking the drive unit at random and expecting it to fit. If your main aim is learning how to design them, go for it,
but for practical systems putting the crossover before the amplifiers is a much
better option. The link below to Elliot Sound has some useful information. https://sound-au.com/lr-passive.htm" rel="nofollow - https://sound-au.com/lr-passive.htm
It can be done. Just as rowing across the Atlantic is
possible, it is far easier, quicker and cheaper to hop on an Airbus; or if you
are feeling a little more adventurous a Boeing. Plus, coming back to
crossovers, no matter how good the passive crossover is, you could improve the
performance by removing it and going active. |
Posted By: kevinmcdonough
Date Posted: 25 February 2024 at 12:20am
|
Thank you for your replies everyone As I said at the start, all of the speakers I've designed and built over the years have been run active, something I'm very comfortable doing. (Hard to believe it's been almost 20 years since I built my first subs and joined this site! )I've only ever built one passive crossover, for a double 8" + comp top I made, and it was just the very basic kind, everything else was done in DSP. My question was really just to satisfy my own curiosity, as I'd like to run the 12" wedges I'm currently working on ( https://forum.speakerplans.com/new-stage-wedges_topic108375.html" rel="nofollow - here ) passive so I can run 4 of a single 4 channel amp. I knew I could just run them the same way as the 8s, with a simple crossover and then do some EQ correction etc in DSP, but wondered if going down the rabbit hole of really learning about these far more complex crossovers would be worth it, if the results would justify the time and effort and research etc.
While at some point I might still do it, I don't have a lot of free time just now and it seems that no, sticking to simple is the way to go In this case I was wondering if really |
Posted By: fudge22
Date Posted: 26 February 2024 at 1:23pm
|
If you do get into deigning passive crossovers, a knowledge
of AC circuit theory, and inductor design would be very useful. Even sticking
to basic circuits, rolling your own can be cheaper than using bought ones. During the experimental/testing phase, leave the crossover
outside the cabinet. |
Posted By: snowflake
Date Posted: 26 February 2024 at 3:01pm
| try and work with the drivers and consider what their physical roll-offs are. If it looks like you need to do huge amounts of correction in the crossover then you probably need to reconsider the drivers and enclosure. Don't obsess about flat frequency response - you will end up using eq as soon as the speaker is in a real-world room. transient response and lobing are important and can't be fixed with DSP. |
Posted By: slaz
Date Posted: 26 February 2024 at 11:38pm
|
Agreed with the "just go active" posts - well most of the time .... but I'd say there's an exception - battery-powered systems like mine, consisting of 2 x FBT Maxx4 - very high quality passive 12 + 1 boxes + 2 options for bass augmentation. These speakers also get used for mains-powered setup (with different amplifiers obviously). I never took the FBT to bits to check the passive Xover, but I'd be willing to bet its a fairly fancy one. The 12" and the 1.4" are both B&C with FBTdesigned passive Xovers - so top notch - and they sound magnificient with superb vocal clarity. I wouldn't fancy my chances of matching the integration of those passives with DSP settings ! Well at least not without resort to fancy calibrated mics and software ..... Aside from that, if using class-D amp modules powered from (say) e-bike batteries or similar - to go bi-amping/tri-amping - you'd want to use lower-power modules for HF, but that would mean sourcing a different supply voltage - which is a right royal PITA :-) ------------- REMEMBER....POLITICIANS AND DIAPERS SHOULD BE CHANGED OFTEN AND FOR THE SAME REASON |
Posted By: fudge22
Date Posted: 28 February 2024 at 11:29am
|
I’m guessing that you have not read the paper “Improving
Loudspeaker Transient Response with Digital Signal Processing” by Dave Gunness.
You don’t need to take them apart. You can find pictures
online. It is a fairly basic second order filter with what looks like PPTCs to
pad the signal when the operator gets over enthusiastic with the volume
control.
I’ll take your word that you don’t have the skills to set up
a DSP properly, but even cheap measurement mics will get you within one or two
dB, and there are a number of free analysis software options available. It is a
skill well worth aquiring. |
Posted By: snowflake
Date Posted: 28 February 2024 at 5:05pm
|
the Gunnes paper is about systems with active filters. It says that: "the measured response includes two kinds of anomalous behaviors. The first are linear, time invariant, and spatially consistent anomalies, meaning behaviors that don’t vary with the loudspeaker’s operating conditions or the ambient environment. These are correctable behaviors. The second are nonlinear, time variant, and spatially variant anomalies, meaning behaviors that vary with the loudspeaker’s operating conditions or the ambient environment. These are uncorrectable behaviors." I'm not sure which category ringing in high-order passive filters falls into but I would guess it's the second 'uncorrectable' category. similarly I'm not sure DSP will solve lobing at crossover frequency of odd-order passive filters.
the paper says that " |