Phasing problems with active/passive crossovers

[Andy]

I'm trying to build myself a decent sounding stereo system. What I have at
the moment is a pre-amp feeding into an active crossover which then feeds
into two power amps, one for the woofers and the other for the tweeters and
mid ranges, with a passive crossover onthe tweets/mids. The problem is that
the passive crossover for the tweets/mids shifts the phasing and sounds
awful.

The stop gap solution I have at present is to use a simple 1st order
(passive) crossover on the tweets/mids, and placing a coil in series with
the woofers simply to shift the phase 90 degrees to match the other drivers.
I've used a coil value that SHOULDN'T affect the lineararity (is that a
word?) of the woofs - but my (untrained) ears tell me otherwise. I seem to
be getting a slight roll off on the upper end of the freq. range. Unless I
have other problems with my box design, which is very possible, but I want
to eliminate other possibilities first.

All passive components are supposed to be of suitable quality for
crossovers (from Jaycar).

I built the active crossover myself with a board I purchased from a bloke in
Sydney who makes them. See - http://sound.westhost.com/project09.htm

My question is;
Could I build a 2nd order passive crossover for the tweets/mids and reverse
the polarity on the woofs (removing the in-series coil of course)? Would
that make them in phase or 360 degrees out, and would that matter?

Would I be better off constructing a circuit to go between the active
crossover and the woofer amp to shift the phase?

I'm doing it this way because I was able to get the (reasonable quality) pre
and power amps at a really good price.

Pls note - I have no electronic qualifications but I am fairly technically
minded so simple(ish) answers are appreciated.

Thanks
Andy

 

[Phil Allison]

I'm trying to build myself a decent sounding stereo system. What I have at
the moment is a pre-amp feeding into an active crossover which then feeds
into two power amps, one for the woofers and the other for the tweeters and
mid ranges, with a passive crossover onthe tweets/mids. The problem is that
the passive crossover for the tweets/mids shifts the phasing and sounds
awful.

** While the sound may be poor to your ears saying it is due to x-over
phase shifts is a **very wild** assumption. Your choice of drivers and their
mounting methods or enclosures etc is far more likely to be the problem.

I built the active crossover myself with a board I purchased from a bloke in
Sydney who makes them. See - http://sound.westhost.com/project09.htm

My question is;
Could I build a 2nd order passive crossover for the tweets/mids and reverse
the polarity on the woofs (removing the in-series coil of course)? Would
that make them in phase or 360 degrees out, and would that matter?

** All x-over filters have phase shift accompanying any change in their
frequency response, both active and passive. All drivers have phase shifts
too - accompanying any response peaks /dips. The effect of reversing the
polarity of a mid or tweeter in a 3 way system is normally quite small and
hard to pick.

If you want help then you MUST supply details of your drivers, enclosures
and x-over designs.




........... Phil

 

[Johnny]

I'm trying to build myself a decent sounding stereo system. What I have at
the moment is a pre-amp feeding into an active crossover which then feeds
into two power amps, one for the woofers and the other for the tweeters and
mid ranges, with a passive crossover onthe tweets/mids. The problem is that
the passive crossover for the tweets/mids shifts the phasing and sounds
awful.

The stop gap solution I have at present is to use a simple 1st order
(passive) crossover on the tweets/mids, and placing a coil in series with
the woofers simply to shift the phase 90 degrees to match the other drivers.
I've used a coil value that SHOULDN'T affect the lineararity (is that a
word?) of the woofs - but my (untrained) ears tell me otherwise. I seem to
be getting a slight roll off on the upper end of the freq. range. Unless I
have other problems with my box design, which is very possible, but I want
to eliminate other possibilities first.

All passive components are supposed to be of suitable quality for
crossovers (from Jaycar).

I built the active crossover myself with a board I purchased from a bloke in
Sydney who makes them. See - http://sound.westhost.com/project09.htm

My question is;
Could I build a 2nd order passive crossover for the tweets/mids and reverse
the polarity on the woofs (removing the in-series coil of course)? Would
that make them in phase or 360 degrees out, and would that matter?

Would I be better off constructing a circuit to go between the active
crossover and the woofer amp to shift the phase?

I'm doing it this way because I was able to get the (reasonable quality) pre
and power amps at a really good price.

Pls note - I have no electronic qualifications but I am fairly technically
minded so simple(ish) answers are appreciated.

Thanks
Andy

Phase shift is an important consideration in the design of any
crossover. This is one of the things that is neglected when a very
simplistic crossover design is used such as the one on the web site
you mention. The drivers themselves will have phase shift of their
own, and also due to the fact that cone and dome drive units radiate
sound from different points when typically mounted on a flat baffle.

I suggest you try swapping the connection polarity of mids and
tweeters, which might improve the situation if you are lucky. But to
do the job properly you will need to measure the speakers, each driver
individually and as a system. That is the only way to get an
understanding of what is really happening in a particular speaker
design. You can download freeware for speaker measurement at
www.speakerworkshop.com . You will need to use the gated measurement
option otherwise room reflections make it impossible to get meaningful
measurements.

best regards,
Johnny.

 

[Andy]

** While the sound may be poor to your ears saying it is due to x-over
phase shifts is a **very wild** assumption. Your choice of drivers and their
mounting methods or enclosures etc is far more likely to be the problem.

Thanks Phil, I didn't mention it in my post, but I wasn't, altogether,
assuming. I mucked around with different things for quite a while and tried
reversing polarities all over the place but could only make it worse, or no
different. As soon as I placed the coil in series with the woofer the sound
kind of "opened up" and became much clearer.

bloke


** All x-over filters have phase shift accompanying any change in their
frequency response, both active and passive. All drivers have phase shifts
too - accompanying any response peaks /dips. The effect of reversing the
polarity of a mid or tweeter in a 3 way system is normally quite small and
hard to pick.

If you want help then you MUST supply details of your drivers, enclosures
and x-over designs.




.......... Phil

Drivers;

Tweeter - Vifa D19
Mid - Response 5" poly
Woofer - Response 12" Carbon Fibre

All enclosures are sealed and I followed the manufacturers recomendations as
to box volume. Tweeters have an "L pad" attenuating circuit to knock of
about 3db. I used the formula for linkwitz-riley alignment to work out
component values for the passive crossover. Can't remember the values right
now but I will pull the box apart and let you know if this is crucial.
Formulas came from the Loud Speaker Design Cookbook by Vance Dickason.
Enclosures are built so the voice coils of all 3 drivers are on the same
plane.

 

[Andy]

Phase shift is an important consideration in the design of any
crossover. This is one of the things that is neglected when a very
simplistic crossover design is used such as the one on the web site
you mention. The drivers themselves will have phase shift of their
own, and also due to the fact that cone and dome drive units radiate
sound from different points when typically mounted on a flat baffle.

Yep, I tried to take that into consideration when I built the boxes - the
tweeter and mid are on the same baffle and the woofer has it's own which
"juts out" so the voice coils are aligned vertically (they look a bit ugly).

The designer of the crossover circuit seems to believe that it is reasonably
phase coherent though. Do you know of any reason it might not be? As I
mentioned I'm not trained in electronics so I believed what he claimed.
Crossover Freq is 300Hz BTW.

I suggest you try swapping the connection polarity of mids and
tweeters, which might improve the situation if you are lucky. But to
do the job properly you will need to measure the speakers, each driver
individually and as a system. That is the only way to get an
understanding of what is really happening in a particular speaker
design. You can download freeware for speaker measurement at
www.speakerworkshop.com . You will need to use the gated measurement
option otherwise room reflections make it impossible to get meaningful
measurements.

best regards,
Johnny.

Thanks for the link. That's exactly what I've been looking for. I've been
wanting to do an impedance response test to look for any obvious dips or
spikes - or is that barking up the wrong tree?

I've already tried reversing the mids and tweets, but I'm not that lucky

Thanks Johnny
Andy

 

[Johnny]

Yep, I tried to take that into consideration when I built the boxes - the
tweeter and mid are on the same baffle and the woofer has it's own which
"juts out" so the voice coils are aligned vertically (they look a bit ugly).

The designer of the crossover circuit seems to believe that it is reasonably
phase coherent though. Do you know of any reason it might not be? As I
mentioned I'm not trained in electronics so I believed what he claimed.
Crossover Freq is 300Hz BTW.

Depending on your definition, it might be phase coherrant if the drive
units have didn't have any phase shift of there own. Since that is
an unrealistic assumption, the complete system is extremely unlikely
to have in-phase summation between the drive units as required for a
Linkwitz-Riley response.

Thanks for the link. That's exactly what I've been looking for. I've been
wanting to do an impedance response test to look for any obvious dips or
spikes - or is that barking up the wrong tree?

Impedance measurement won't tell you about the acoustic response of
the system. You need to use a microphone. Decent measurement
microphones will cost upwards of a couple of hundred dollars, but
think you will probably get some useful clues by using a cheap PC
microphone. Gated measurements will help mostly for the crossover
between mid and tweeter. For the crossover at 300Hz, you will find it
quite difficult to make measurements using the gated method. Pink
noise and a real-time analyser (RTA) is probably more useful in this
frequency range. There are some RTA software available for demo on
the net, but I am unsure if they are able to playback a pink noise
test signal while measuring.

Since swapping the polarity did not help, you might try bypassing some
of the circuitry on the 4th order crossover board. If you by-pass one
of the 2nd order sections (on one of either hi-pass or low pass
secion) you will shift the phase by 90 degress at the crossover
frequency. Then by swapping the polarity you get another +/- 180
degrees shift, and you might have better luck!

best regards,
Johnny.

 

[Rob Judd]

Andy,

Either use active crossovers or passive, not both.

Rob

 

[Phil Allison]

Thanks Phil, I didn't mention it in my post, but I wasn't, altogether,
assuming. I mucked around with different things for quite a while and tried
reversing polarities all over the place but could only make it worse, or no
different.

** Which clearly demonstrates that x-over phase shift is NOT your problem.

Phase shift at or near x-over frequencies causes either a modest peak or
narrow dip in the combined response - the narrow dip is *inaudible* and
hence the preferred error.

BTW Where did you get the idea putting an inductor in series with a woofer
shifts the phase by 90 degrees ????????

As soon as I placed the coil in series with the woofer the sound
kind of "opened up" and became much clearer.

** That electronic x-over you made likely has a serious fault.

Drivers;

Tweeter - Vifa D19
Mid - Response 5" poly
Woofer - Response 12" Carbon Fibre

All enclosures are sealed and I followed the manufacturers recomendations as
to box volume. Tweeters have an "L pad" attenuating circuit to knock of
about 3db. I used the formula for linkwitz-riley alignment to work out
component values for the passive crossover. Can't remember the values right
now but I will pull the box apart and let you know if this is crucial.

** You have so far failed to explain what the ***PROBLEM ** is -
except that YOU do not like the sound.

It is impossible to make any analysis with something that subjective as the
problem.





.................. Phil

 

[Richard Freeman]

Unless bloke quality)

Yep, I tried to take that into consideration when I built the boxes - the
tweeter and mid are on the same baffle and the woofer has it's own which
"juts out" so the voice coils are aligned vertically (they look a bit ugly).

The designer of the crossover circuit seems to believe that it is reasonably
phase coherent though. Do you know of any reason it might not be? As I
mentioned I'm not trained in electronics so I believed what he claimed.
Crossover Freq is 300Hz BTW.

What Active crossover circuit are you using ? I see in another message you
say you are using a Linkwitz riley Passive but what is the active crossover
? I have had problems with 3rd order butterworth active crossovers (as you
say you can get some weird phase cancellation between the high and low
speakers) whereas when I changed over to a Linkwitz riley Active crossover
everything just sounded a heck of a lot better.
As for Passive crossovers being any good .... It actually takes quite a lot
of work and usually necessitates taking the impedance curve of the
drivers/cabinets into account to get a half decent passive crossover - If
you can, save yourself a lot of drama and try running your speakers three
way active this will at least let you listen to the speaker system on its
own and make sure that it sounds O.K to start with (no use mucking about
with your crossover if your speakers arent up to the job). Other than that I
have found the second order type passive crossovers are probably a better
starting point than most .......

Regards
Richard Freeman

 

[Richard Freeman]

Unless bloke

Yeah sorry I missed this - my other advoce still applies though try running
the speaker three way active and see what it sounds like ....

 

[Andy]

** Which clearly demonstrates that x-over phase shift is NOT your problem.

Phase shift at or near x-over frequencies causes either a modest peak or
narrow dip in the combined response - the narrow dip is *inaudible* and
hence the preferred error.

BTW Where did you get the idea putting an inductor in series with a woofer
shifts the phase by 90 degrees ????????




** That electronic x-over you made likely has a serious fault.


recomendations



** You have so far failed to explain what the ***PROBLEM ** is -
except that YOU do not like the sound.

It is impossible to make any analysis with something that subjective as the
problem.





................. Phil

Geez Phil ....I tried to be polite and supply the data you said you needed
and you seem to take it as a personal insult. I came here to see if I could
learn something not for an argument. Sorry mate, but you are on my ignore
list. Go ahead, flame away - I won't see it

 

[Phil Allison]

Geez Phil ....I tried to be polite and supply the data you said you needed
and you seem to take it as a personal insult.

** You are an utter idiot - piss off.

I came here to see if I could
learn something not for an argument.

** You are such a pathetic jerk you are completely unable to define the
problem.





............. Phil

 

[Andy]

Impedance measurement won't tell you about the acoustic response of
the system. You need to use a microphone. Decent measurement
microphones will cost upwards of a couple of hundred dollars, but
think you will probably get some useful clues by using a cheap PC
microphone. Gated measurements will help mostly for the crossover
between mid and tweeter. For the crossover at 300Hz, you will find it
quite difficult to make measurements using the gated method. Pink
noise and a real-time analyser (RTA) is probably more useful in this
frequency range. There are some RTA software available for demo on
the net, but I am unsure if they are able to playback a pink noise
test signal while measuring.

Thanks Johnny, I'm pretty sure I can borrow a stage microphone.

I have a program called NCH Tone Generator that can generate white/pink
noise or square/sawtooth and sine waves. If the RTA software can't play and
listen at the same time I can run the generator on my lap top and use my
desktop to measure.(yes?)

Just done a Google search for RTA software and there are dozens! What
features should I be looking for? Do you have any specific recomendations
perhaps? Both my lap top and desktop can boot to Linux BTW.

Sorry Johnny, it was late when I posted and my brain was already in bed. The
reason I wanted to measure impedance response was to detect any resonance
that may be compounding the problem/s. From what I have read almost every
part of the system from source to driver, and even the room can resonate.

Since swapping the polarity did not help, you might try bypassing some
of the circuitry on the 4th order crossover board. If you by-pass one
of the 2nd order sections (on one of either hi-pass or low pass
secion) you will shift the phase by 90 degress at the crossover
frequency. Then by swapping the polarity you get another +/- 180
degrees shift, and you might have better luck!

best regards,
Johnny.

Thanks
Andy

 

[Phil Allison]

Sorry Johnny, it was late when I posted and my brain was already in bed. The
reason I wanted to measure impedance response was to detect any resonance
that may be compounding the problem/s. From what I have read almost every
part of the system from source to driver, and even the room can resonate.

** Andy is fucking nut case.




........... Phil

 

[Andy]

Yeah sorry I missed this - my other advoce still applies though try running
the speaker three way active and see what it sounds like ....

I'd like to be able to. But I'd have to build another (3 way) active
crossover and buy another amp. If I had the bucks that is definitely what I
would do but I have teenagers who keep me poor

What I have tried, however, is disconnecting the tweeters and running as a
straight 2 way active, and the sound was good (albeit missing the highs)
which leads me to believe the active crossover is working OK.

Having said that though - if this gets much more complicated I'm thinking of
starting over by keeping the tweeters, buying some 6 1/2 inch carbon fibre
drivers for woofers and recycling my 12 inchers into an isobaric sub-woofer
using a Subwoofer Amp Module from Jaycar. So I would have a straight 2 way
active, with a subby. I'd have to build another active crossover of course,
as the crossover freq would be much higher. Also I'd have to make some new
enclosures - luckily my brother is a cabinet maker, but I still won't get
out of it for less than about $500 or $600. Although I could convert the old
3way enclosures into a realy cool dog kennel


Thanks
Andy

 

[Johnny]

Thanks Johnny, I'm pretty sure I can borrow a stage microphone.

I have a program called NCH Tone Generator that can generate white/pink
noise or square/sawtooth and sine waves. If the RTA software can't play and
listen at the same time I can run the generator on my lap top and use my
desktop to measure.(yes?)
Sure.

Just done a Google search for RTA software and there are dozens! What
features should I be looking for? Do you have any specific recomendations
perhaps? Both my lap top and desktop can boot to Linux BTW.

I don't have much of an idea whats available now, but I would bet that
there is much better freeware around that when I looked a couple of
years ago!

Sorry Johnny, it was late when I posted and my brain was already in bed. The
reason I wanted to measure impedance response was to detect any resonance
that may be compounding the problem/s. From what I have read almost every
part of the system from source to driver, and even the room can resonate.

Impedance data is useful for checking the tuning frequency of vented
boxes and sometimes severe cone resonances can show up also.

Good luck with your quest!

regards,
Johnny.