reverse engineering a crossover?

BlackDog · 13th February 2005, 10:01 PM

Hi All,

I'm considering tri-amping my current speakers, and thought that as a starting point I ought to determine the crossover frequencies of the built-in passive crossovers.

I can't find the info the specs anywhere, so I thought that there must be a fairly simple way to measure this. - I thought I'd ask some experts before I started fudging around on my own..

I have a PC based signal generator, an old 50MHz 'scope and plenty of multimeters.

Any ideas?

Steve

These are the speakers I'm referring to.

Acoustic Energy AE120

johnnyx · 13th February 2005, 10:54 PM

You need to measure the acoustic response, so a microphone and preamp is essential. If you reverse the connection to one of the units, eg the tweeter, then as you increase the frequency there should be a sharp null at the crossover frequency. I built a mic and preamp cheaply, you don't need anything fancy for this.

Hope this helps.

Bill Fitzpatrick · 14th February 2005, 01:40 AM

As far as I know there is only one way to do this.

Get a response curve of each driver/network combination by using a microphone to measure it.

Remove the driver from the network and connect is directly to an amplifier which is driven by the appropriate active crossover section. The active crossover will have to be altered in such a way as to provide a duplicate of the previously acquired curve.

This will be a major pain in the butt unless you already have the mic and are well versed in the design of active filters.

PB2 · 14th February 2005, 01:58 AM

Quote:

Originally posted by BlackDog
Hi All,
I'm considering tri-amping my current speakers, and thought that as a starting point I ought to determine the crossover frequencies of the built-in passive crossovers.

I can't find the info the specs anywhere, so I thought that there must be a fairly simple way to measure this. - I thought I'd ask some experts before I started fudging around on my own..

I have a PC based signal generator, an old 50MHz 'scope and plenty of multimeters.

Any ideas?

Steve

These are the speakers I'm referring to.

Acoustic Energy AE120

It sounds like you want to retain the original voicing of your speakers, and simply have the advantage of lower distortion from removing the passive components in the crossover, is this right?
There is no need for acoustic measurements, since the only requirement is to duplicate the crossover network transfer function. Further, if the crossover is minimum phase for each section, and most are unless all pass phase compensation is used which is rare, then all you have to do is duplicate the frequency response and the phase response will be right as a result of the properties of minimum phase networks.

Remove the drivers, attach small wires that you can somehow bring out to your measurement gear, and measure the frequency response from the system input to each driver loaded in the system. You would then duplicate these in the crossover before the amplifiers.

There is free PC based software that you can use to automate the process.

BTW, Do you race any particular class of slot cars? We did Deathstars, S16D outlaw 35 and 50 deg rotors, and played with a few C can GP20, S16C, not into it much anymore.

catapult · 14th February 2005, 04:25 AM

Quote:

Remove the drivers, attach small wires that you can somehow bring out to your measurement gear, and measure the frequency response from the system input to each driver loaded in the system. You would then duplicate these in the crossover before the amplifiers.

You need to leave the drivers attached to the passive crossover when you measure the voltage response. The voltage output of the passive crossover depends on the drivers' impedance. Then copy that transfer function to the active crossover which acts as a voltage source.

BlackDog · 14th February 2005, 06:57 AM

Thanks for everyone's input.

I'll just clarify - what I'm looking for is just a fairly accurate approximation of the the crossover frequencies. It does not need to be accurate to 1Hz! These frequencies will be used as the starting points in my subsequent filters, and I'll be tinkering with them to see if I can improve the voicing of the system to my taste.

It seems as if there are differing opinions on what method is best.

Electrical measurement - easy to set up, but probably doesn't take differing driver efficiencies into account

Accoustic measurement - more equipment needed, and I'm sure the only mic I own at the moment has a terrible non-linear frequency response! - So plotting the response curve for each driver would be quite hit and miss. I'll probably try JohnnyX's suggestion, as it sounds as if it'll get me there fastest.

I'm still open to suggestions though..

BlackDog · 14th February 2005, 07:02 AM

Quote:

Originally posted by PB2

There is free PC based software that you can use to automate the process.

BTW, Do you race any particular class of slot cars? We did Deathstars, S16D outlaw 35 and 50 deg rotors, and played with a few C can GP20, S16C, not into it much anymore.

Hi PB2

Have you any idea what the software is called???

As far as the slotcars are concerned, we mainly race S16Ds, but also run a class similar to your GT12, and also an anything goes F1/Indy class.

Our club's website

PB2 · 14th February 2005, 05:10 PM

Quote:

Originally posted by BlackDog
Thanks for everyone's input.

I'll just clarify - what I'm looking for is just a fairly accurate approximation of the the crossover frequencies. It does not need to be accurate to 1Hz! These frequencies will be used as the starting points in my subsequent filters, and I'll be tinkering with them to see if I can improve the voicing of the system to my taste.

It seems as if there are differing opinions on what method is best.

Electrical measurement - easy to set up, but probably doesn't take differing driver efficiencies into account

I'm still open to suggestions though..

You can determine the relative driver sensitivities by comparing the electrical level in the passband assuming you know the frequency response of the system. Or let's say assuming they got it "right" in the original design. I'd do the electrical measurement just to get a ball park idea of what's going on, then go from there. You might find some pass band compensation if the system includes baffle step compensation so keep it in mind.

Crossover points will be about -3 dB if 1st order networks are used and -6 dB if higher order in phase netorks are used but you must also take into account any response shaping. If you measure and post the results we can help interpret them.

Always liked your clubs armature wind chart, nice job.

PB2 · 14th February 2005, 05:14 PM

Quote:

Originally posted by BlackDog

Hi PB2

Have you any idea what the software is called???

As far as the slotcars are concerned, we mainly race S16Ds, but also run a class similar to your GT12, and also an anything goes F1/Indy class.

Our club's website

One is Speaker Workshop, I don't use it since I already have a system, others say it's good:
http://www.speakerworkshop.com/

Too bad there isn't a getting started page for speaker measurement on the web, don't know of one at the moment.

This site is very good but you have to dig there's a lot of good theory, and at one point an easy to build measurement mic:
http://www.linkwitzlab.com/sys_test.htm#Mic
http://www.linkwitzlab.com/

D'Appolito's book is important if your serious about measurement.

Davey · 14th February 2005, 10:00 PM

Try using the evaluation version of SpectrPlus to plot the electrical response of the crossover networks.

http://www.spectraplus.com

Any PC with decent sound card will do the job. Use both channels on your soundcard input...Left for the input connection to the speaker and Right across the actual driver under test. Be careful with the voltage.

If you use the "transfer function" capability of Spectraplus, drive the speaker with white noise and let the system average for a few seconds you'll end up with very smooth curves of amplitude and phase that represent the actual voltage applied to the drivers. It will also let you save this data as a text file for possible future usage as a "target" file for one of the speaker design programs.

This result yields a "baseline" for what the passive crossovers are achieving and gives you a target for an active crossover if you're trying to recreate the original responses. Although, if you're switching to an active setup you can probably improve upon the performance of the original system considerably.

You don't need any acoustical measurements.

Cheers,

Davey.

13th February 2005, 10:01 PM	#1
BlackDog diyAudio Member Join Date: Feb 2004 Location: Wellington, New Zealand	reverse engineering a crossover? Hi All, I'm considering tri-amping my current speakers, and thought that as a starting point I ought to determine the crossover frequencies of the built-in passive crossovers. I can't find the info the specs anywhere, so I thought that there must be a fairly simple way to measure this. - I thought I'd ask some experts before I started fudging around on my own.. I have a PC based signal generator, an old 50MHz 'scope and plenty of multimeters. Any ideas? Steve These are the speakers I'm referring to. Acoustic Energy AE120

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Reverse engineering Krell KMA 160	jwb	Solid State	193	29th April 2012 10:09 PM
Reverse engineering a JVC stereo amp	jovial_cynic	Chip Amps	6	18th April 2008 04:51 PM
reverse engineering a box	Zero Cool	Multi-Way	1	16th July 2007 10:00 PM
Reverse engineering resonant frequency	jonkun227	Multi-Way	13	12th April 2007 05:33 PM

New To Site?	Need Help?
Register to Participate Search Privacy Statement Contact Us	Frequently Asked Questions Did you forget your password? Mark Forums Read


	Please consider donating to help us continue to serve you. Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving

13th February 2005, 10:54 PM	#2
johnnyx diyAudio Member Join Date: Jan 2004 Location: manchester	You need to measure the acoustic response, so a microphone and preamp is essential. If you reverse the connection to one of the units, eg the tweeter, then as you increase the frequency there should be a sharp null at the crossover frequency. I built a mic and preamp cheaply, you don't need anything fancy for this. Hope this helps.

14th February 2005, 01:40 AM	#3
Bill Fitzpatrick diyAudio Member Join Date: Jun 2001 Location: Eugene, OR	As far as I know there is only one way to do this. Get a response curve of each driver/network combination by using a microphone to measure it. Remove the driver from the network and connect is directly to an amplifier which is driven by the appropriate active crossover section. The active crossover will have to be altered in such a way as to provide a duplicate of the previously acquired curve. This will be a major pain in the butt unless you already have the mic and are well versed in the design of active filters.

14th February 2005, 06:57 AM	#6
BlackDog diyAudio Member Join Date: Feb 2004 Location: Wellington, New Zealand	Thanks for everyone's input. I'll just clarify - what I'm looking for is just a fairly accurate approximation of the the crossover frequencies. It does not need to be accurate to 1Hz! These frequencies will be used as the starting points in my subsequent filters, and I'll be tinkering with them to see if I can improve the voicing of the system to my taste. It seems as if there are differing opinions on what method is best. Electrical measurement - easy to set up, but probably doesn't take differing driver efficiencies into account Accoustic measurement - more equipment needed, and I'm sure the only mic I own at the moment has a terrible non-linear frequency response! - So plotting the response curve for each driver would be quite hit and miss. I'll probably try JohnnyX's suggestion, as it sounds as if it'll get me there fastest. I'm still open to suggestions though..

14th February 2005, 10:00 PM	#10
Davey diyAudio Member Join Date: Dec 2001 Location: Bremerton, WA.	Try using the evaluation version of SpectrPlus to plot the electrical response of the crossover networks. http://www.spectraplus.com Any PC with decent sound card will do the job. Use both channels on your soundcard input...Left for the input connection to the speaker and Right across the actual driver under test. Be careful with the voltage. If you use the "transfer function" capability of Spectraplus, drive the speaker with white noise and let the system average for a few seconds you'll end up with very smooth curves of amplitude and phase that represent the actual voltage applied to the drivers. It will also let you save this data as a text file for possible future usage as a "target" file for one of the speaker design programs. This result yields a "baseline" for what the passive crossovers are achieving and gives you a target for an active crossover if you're trying to recreate the original responses. Although, if you're switching to an active setup you can probably improve upon the performance of the original system considerably. You don't need any acoustical measurements. Cheers, Davey.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)