Speakerplans.com Homepage
Forum Home Forum Home > General > General Forum
  New Posts New Posts RSS Feed - Passive crossover advice
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

Passive crossover advice

 Post Reply Post Reply Page  <12
Author
Message
RobinMatrix View Drop Down
Young Croc
Young Croc


Joined: 23 August 2013
Location: Kidderminster
Status: Offline
Points: 567
Post Options Post Options   Thanks (0) Thanks(0)   Quote RobinMatrix Quote  Post ReplyReply Direct Link To This Post Posted: 15 February 2020 at 12:07pm
I use somethign very similar to Xsim,  with response and impedance sweeps captured out of REW and the results are incredibly accurate.

Being able to adjust component values and see the actual reponse curves and impedance curves vary is incredibly useful.  I don't think it is possible to relaibly produce a crossover design without software like this ... unless you have eithre lots of time, or lots of luck.

What I always find fascinating is how well they work ... if you take a good impedance sweep and a good reposnse measurement and leave the measuring mic setup and then make a crossover up and re-sweep the speaker ... if you don't change the mic position and re-sweep it, the response curve will *exactly* overlay the prediction ... right down to the smallest details and bumps.  I supposes its not that surprising, it's just electronics, but it is amazing to see.

The thing I use is no longer avaialable, but it looks very similar in functionality to Xsim,  I'd be surprised if it wasnt the same author.
Matrix Pro Audio ::http://matrixproaudio.com/"
Back to Top
ceharden View Drop Down
The 10,000 Points Club
The 10,000 Points Club
Avatar

Joined: 05 June 2005
Location: Southampton
Status: Offline
Points: 11776
Post Options Post Options   Thanks (0) Thanks(0)   Quote ceharden Quote  Post ReplyReply Direct Link To This Post Posted: 15 February 2020 at 1:05pm
You have to be careful not to fall into the trap of trying to EQ or Align out a feature of your measurement environment.  Not many people have access to chambers to do their measurements in so you have to make do.

I think one of the most difficult issues is that you can't time align with a passive crossover.  On the one I've just done I was fairly lucky that I could put the crossover point at the equivalent of a half wavelength of the offset, flip the polarity and get an approximation of 'in-phase'.  However crossovers being a long way from brick walls you still get some interaction and octave or so either side.

I could also have made my life easier by doing a proper impedance compensation network on both drivers but I didn't want to end up with a design the size of an A4 sheet that wouldn't fit in the cab!  Managed to adjust the values in the filter sufficiently to get a good approximation.

Back to Top
odc04r View Drop Down
Old Croc
Old Croc
Avatar

Joined: 12 July 2006
Location: Sarfampton
Status: Offline
Points: 5483
Post Options Post Options   Thanks (0) Thanks(0)   Quote odc04r Quote  Post ReplyReply Direct Link To This Post Posted: 15 February 2020 at 7:48pm
Originally posted by RobinMatrix RobinMatrix wrote:

I use somethign very similar to Xsim,  with response and impedance sweeps captured out of REW and the results are incredibly accurate.

Being able to adjust component values and see the actual reponse curves and impedance curves vary is incredibly useful.  I don't think it is possible to relaibly produce a crossover design without software like this ... unless you have eithre lots of time, or lots of luck.

What I always find fascinating is how well they work ... if you take a good impedance sweep and a good reposnse measurement and leave the measuring mic setup and then make a crossover up and re-sweep the speaker ... if you don't change the mic position and re-sweep it, the response curve will *exactly* overlay the prediction ... right down to the smallest details and bumps.  I supposes its not that surprising, it's just electronics, but it is amazing to see.

The thing I use is no longer avaialable, but it looks very similar in functionality to Xsim,  I'd be surprised if it wasnt the same author.


On the few times I've made a passive xover, LTSpice and REW did me well. LTSpice takes a bit of getting used to of course but once you're in parametric component sweeps are possible.

Next level would be doing it in something like Scilab/Octave?Matlab where you could use regression to compute a nominally flat bandwidth whilst varying components. But I suspect this would fall foul of real life not being as well behaved as measurements. Still, be interesting to try on day.

For full nerd points, simulate your entire enclosure after the xover in spice by converting the enclosure model into its electrical analog, and congratulations you have basically re-programmed an Akbak script. Maybe when I am retired.
Back to Top
Jack1991 View Drop Down
Registered User
Registered User


Joined: 05 February 2020
Location: Bristol
Status: Offline
Points: 491
Post Options Post Options   Thanks (0) Thanks(0)   Quote Jack1991 Quote  Post ReplyReply Direct Link To This Post Posted: 27 February 2020 at 8:49pm
Decided i will be going with pre made crossover for my first time.

Am i right in thinking this though: 

My mid drivers SPL is 94db and my Tweeters SPL is 104db. So i know i need to lower the tweeter by 10db with a L-Pad. However i just noticed the crossover i want to use has a "HF level -3.5db", so i assume that now makes the tweeter 100.5db. So now i just need to L-pad down by 6.5db correct?

Mid driver

Tweeter

Crossover
Back to Top
studio45 View Drop Down
Old Croc
Old Croc
Avatar

Joined: 16 October 2007
Location: United Kingdom
Status: Offline
Points: 3863
Post Options Post Options   Thanks (0) Thanks(0)   Quote studio45 Quote  Post ReplyReply Direct Link To This Post Posted: 28 February 2020 at 2:50pm
Right on, that's a good starting point. Your maths are correct, on paper you need to add a 6.5dB L-Pad after the tweeter output on the crossover.
However, that crossover lacks a couple of features that you may want to explore for higher sound quality. Specifically, it doesn't include a Zobel network for the woofer, or an EQ network for the tweeter. 
The lack of these, is likely to result in excess output in the high mids, compared to the bass and treble. So it will work, and nothing will break - but the sound is likely to be quite bright and forward, maybe too harsh at high levels.
The Zobel network rolls off some extra hi-mids from the woofer, in addition to what the main crossover filter does. It also linearises the woofer's impedance, which otherwise is a rising curve that may reach 20-30 Ohms at the desired crossover frequency of 2500Hz. This means the crossover filter is not really correctly loaded, if it is designed for 8 ohms. The Zobel sorts that right out. It consists of one resistor and one cap in series, connected in parallel with the woofer, after the crossover. Here is a handy calculator https://www.diyaudioandvideo.com/Calculator/ImpedanceEqualizationCircuit/
The tweeter's response is not flat to 20kHz, in fact most cheap 1" drivers like that have a large amount of output from 2-8kHz and start to roll off thereafter. The EQ network works in conjunction with the L-Pad, and allows extra high-frequency signal to bypass it and go directly to the tweeter. This flattens the response and stops the hi-mids being too prominent. It consists of an inductor and capacitor in parallel, placed in parallel with the series resistor of the L-Pad. There are not really any handy calculators online that I have found, but you can experiment with small values of C and L around the 1uF/0.1mH range and see if you can get an improvement. You want to end up with an LC pair that has its lowest impedance at about 10-15kHz with quite a wide bandwidth, so it is "helping" from around 8kHz up.
To make this whole process easier, it is handy to have impedance measurements (graphs) for your drivers. There is free software available called ARTA Suite, which includes a program called LIMP (Loudspeaker impedance measurement program). Then, you need what is known as an "ARTA box" which is a simple passive interconnection and signal level management circuit that allows you to use your computer sound card to sample signals being sent to the driver from an amp, works out the AC current flowing at various frequencies, and from that derives a graph of impedance vs frequency. 
Google all that, and you should make some good progress as a crossover designer :) 
Studio45 - Repairs & Building Commotion Soundsystem -Mobile PA
Back to Top
Jack1991 View Drop Down
Registered User
Registered User


Joined: 05 February 2020
Location: Bristol
Status: Offline
Points: 491
Post Options Post Options   Thanks (0) Thanks(0)   Quote Jack1991 Quote  Post ReplyReply Direct Link To This Post Posted: 28 February 2020 at 4:08pm
Originally posted by studio45 studio45 wrote:

Right on, that's a good starting point. Your maths are correct, on paper you need to add a 6.5dB L-Pad after the tweeter output on the crossover.
However, that crossover lacks a couple of features that you may want to explore for higher sound quality. Specifically, it doesn't include a Zobel network for the woofer, or an EQ network for the tweeter. 
The lack of these, is likely to result in excess output in the high mids, compared to the bass and treble. So it will work, and nothing will break - but the sound is likely to be quite bright and forward, maybe too harsh at high levels.
The Zobel network rolls off some extra hi-mids from the woofer, in addition to what the main crossover filter does. It also linearises the woofer's impedance, which otherwise is a rising curve that may reach 20-30 Ohms at the desired crossover frequency of 2500Hz. This means the crossover filter is not really correctly loaded, if it is designed for 8 ohms. The Zobel sorts that right out. It consists of one resistor and one cap in series, connected in parallel with the woofer, after the crossover. Here is a handy calculator https://www.diyaudioandvideo.com/Calculator/ImpedanceEqualizationCircuit/
The tweeter's response is not flat to 20kHz, in fact most cheap 1" drivers like that have a large amount of output from 2-8kHz and start to roll off thereafter. The EQ network works in conjunction with the L-Pad, and allows extra high-frequency signal to bypass it and go directly to the tweeter. This flattens the response and stops the hi-mids being too prominent. It consists of an inductor and capacitor in parallel, placed in parallel with the series resistor of the L-Pad. There are not really any handy calculators online that I have found, but you can experiment with small values of C and L around the 1uF/0.1mH range and see if you can get an improvement. You want to end up with an LC pair that has its lowest impedance at about 10-15kHz with quite a wide bandwidth, so it is "helping" from around 8kHz up.
To make this whole process easier, it is handy to have impedance measurements (graphs) for your drivers. There is free software available called ARTA Suite, which includes a program called LIMP (Loudspeaker impedance measurement program). Then, you need what is known as an "ARTA box" which is a simple passive interconnection and signal level management circuit that allows you to use your computer sound card to sample signals being sent to the driver from an amp, works out the AC current flowing at various frequencies, and from that derives a graph of impedance vs frequency. 
Google all that, and you should make some good progress as a crossover designer :) 

Really appreciate your long and detailed answer but making a crossover is out of the question. I literally dont understand it whatsoever. This is my first build too so i want to make it fairly simple so i can start to begin how it all works. (even the simple way is advanced to me as i havent done anything like this before)

So my main concern was that because the crossover i suggested has the "HF Level -3.5db" i just need to make a LPad consisting of 2x resistors to lower it the rest of the way. 

Or would i be better off finding a different pre build passive crossover that doesnt have a HF Level reducer and just use a LPad. Or does it not make a difference?
Back to Top
Jack1991 View Drop Down
Registered User
Registered User


Joined: 05 February 2020
Location: Bristol
Status: Offline
Points: 491
Post Options Post Options   Thanks (0) Thanks(0)   Quote Jack1991 Quote  Post ReplyReply Direct Link To This Post Posted: 28 February 2020 at 4:57pm
Or just thought maybe id be better off just getting a Eminence Mid Driver so then it will be eminence tweeter, eminence crossover and eminence mid driver. Surely they will be made more to suit?
Back to Top
odc04r View Drop Down
Old Croc
Old Croc
Avatar

Joined: 12 July 2006
Location: Sarfampton
Status: Offline
Points: 5483
Post Options Post Options   Thanks (0) Thanks(0)   Quote odc04r Quote  Post ReplyReply Direct Link To This Post Posted: 28 February 2020 at 11:03pm
It really won;t make a lot of difference if you buy all from one manufacturer or not.

There is no escaping that if you want to design your own passives there will be some maths. The only other alternatives are to buy a pre-made board, or to simulate and randomly change values until you possibly like what you see.

I would suggest you read https://sound-au.com/lr-passive.htm in its entirety, and then try to follow the maths in sections 5 and 6. It isn't that difficult, and if you can follow it then you just copy it with your preferred values of gain and frequency and you will get the right numbers. You can then re-simulate those in XSim or LTSpice to make sure you are getting the right crossover points.

Specifically section 6 deals with attentuating levels, whilst ensuring your xover points don't change depending on your driver DC resistance.

I'm afraid it will never get easier than that for your first go.


Edited by odc04r - 28 February 2020 at 11:05pm
Back to Top
studio45 View Drop Down
Old Croc
Old Croc
Avatar

Joined: 16 October 2007
Location: United Kingdom
Status: Offline
Points: 3863
Post Options Post Options   Thanks (0) Thanks(0)   Quote studio45 Quote  Post ReplyReply Direct Link To This Post Posted: 29 February 2020 at 3:26pm
Originally posted by Jack1991 Jack1991 wrote:


Really appreciate your long and detailed answer but making a crossover is out of the question. I literally dont understand it whatsoever. This is my first build too so i want to make it fairly simple so i can start to begin how it all works. (even the simple way is advanced to me as i havent done anything like this before)
You don't understand it *yet* ;) It's worth learning though.
Originally posted by Jack1991 Jack1991 wrote:

So my main concern was that because the crossover i suggested has the "HF Level -3.5db" i just need to make a LPad consisting of 2x resistors to lower it the rest of the way. 

Or would i be better off finding a different pre build passive crossover that doesnt have a HF Level reducer and just use a LPad. Or does it not make a difference?
That crossover will do you fine. Just add the extra L-Pad on the tweeter output and see how you get on. It's actually a good design, because it incorporates that lightbulb/fuse thing to protect the tweeter. If you mount it in the box such that you can see that lightbulb through the port, it makes a handy warning light to tell you when the rig is at its limit ;)

You can always use an EQ on the system to correct any excess hi-mids. A lot of systems run that way. In fact a lot of modern active boxes require really surprising amounts of EQ in the processor, and sound terrible without it...

Studio45 - Repairs & Building Commotion Soundsystem -Mobile PA
Back to Top
snowflake View Drop Down
Old Croc
Old Croc


Joined: 29 December 2004
Location: Bristol
Status: Offline
Points: 3118
Post Options Post Options   Thanks (0) Thanks(0)   Quote snowflake Quote  Post ReplyReply Direct Link To This Post Posted: 29 February 2020 at 6:32pm
make sure the vertical polar response remains sensible through the crossover region - there's no way to fix that with EQ.
Back to Top
 Post Reply Post Reply Page  <12

Forum Jump Forum Permissions View Drop Down

Forum Software by Web Wiz Forums® version 12.06
Copyright ©2001-2023 Web Wiz Ltd.

This page was generated in 0.125 seconds.