Go Back   Team-BHP > Under the Hood > In-Car Entertainment


Reply
 
Thread Tools Search this Thread
Old 21st December 2009, 10:04   #16
Senior - BHPian
 
DerAlte's Avatar
 
Join Date: Mar 2007
Location: Bangalore
Posts: 8,076
Thanked: 2,876 Times
Default

Quote:
Originally Posted by naughty001 View Post
no sir its nothing to do with maths, ... add one or two more components to the design ... end result here is the same as having it all in one unit
There is maths behind any electrical behavior, and that is what helps in electrical design. Electrical design is not empirical, even though in life we tend to take a 'solved example' from the past and replicate the solution to another problem (like in any maths exam)!!!

I will give you a simpler example with water pipes:
* If one takes first a 1" pipe, and then tapers down to 1/2", the effective flow rate of water will be that of 1/2"
* If one takes a 1/2" pipe and then puts a 1" pipe, the effective flow rate through the 1" pipe will still be that as it was through the 1/2" pipe

With 1 XO, the voltage at the output is Vo = f(Vi), where f stands for the relationship of the R, C and L of the crossover to frequency ('slope'), and Vi is the input voltage to the XO. f is a logarithmic relationship, hence output drop is expressed in dB, not percentage which would signify linear decline.

With 2 XOs cascaded, the output voltage is
Vo2 = f2( Vo1) ) or
Vo2 = f2( f1(Vi1) ) since Vo1 = f1(Vi1)
If in the above, f1 > f2, then f2 has a negligible effect. If f2 > f1, then f2 prevails. If one wants to add components to an existing design, one cannot add them as f2 - unless one wants to override f1, in which case f1 becomes redundant. If one wants to modify either f1 or f2 by changing components, the above is still true.

Last edited by DerAlte : 21st December 2009 at 10:12.
DerAlte is offline   Reply With Quote
Old 21st December 2009, 10:15   #17
Awaiting Email Confirmation
 
Join Date: Jun 2009
Location: South Africa
Posts: 185
Thanked: 11 Times
Default

Quote:
If in the above, f1 > f2, then f2 has a negligible effect. If f2 > f1, then f2 prevails. All the above is a simplified representation of the actual equations.
Yet this is not what happens because when you cascade the crossovers you will get a slope equal to the total of all the slopes used .... not either/or

I was incorrect in saying its not a matter of the math involved but the simple fact of the matter is that i dont know the math involved to explain it or even to begin to refute your contentions. All i do know is that if you use more than one crossover in whatever configuration ie cascade them in any form or way you will be getting a slope that is equal to the total of the slopes used

BTW please note im not being stubborn for the sake of being stubborn here, because i would like to find out why the slopes do get totalled together and im hoping someone will delve deeper into it and give me the lazy way out by explaining it but i will look for the answer to this myself as well in the meantime

Last edited by naughty001 : 21st December 2009 at 10:18.
naughty001 is offline   Reply With Quote
Old 21st December 2009, 11:40   #18
Senior - BHPian
 
greenhorn's Avatar
 
Join Date: Apr 2007
Location: TN-14
Posts: 6,839
Thanked: 1,318 Times
Default

naughty, deralte's point is that if the crossover frequencies are sufficiently seperated, then the effect of cascading will not be significant.

Suppose you take a 2 way system with an active 12 db/octave crossover with a crossover frequency of 3Khz.

If I were to add a capacitor in series ( a first order HPF to simplfy things), with a cutoff Frequency of , say, 1khz.

Which means that when the passive crossover kicks in, the signal to the tweeter is already 18 db down ~ approximately 1% of the overall signal power, and decreasing. So in this particular case, the effect of this extra crossover on frequency response would be <1%.

Last edited by greenhorn : 21st December 2009 at 11:41.
greenhorn is offline   Reply With Quote
Old 21st December 2009, 12:18   #19
Awaiting Email Confirmation
 
Join Date: Jun 2009
Location: South Africa
Posts: 185
Thanked: 11 Times
Default

Quote:
Originally Posted by greenhorn View Post
naughty, deralte's point is that if the crossover frequencies are sufficiently seperated, then the effect of cascading will not be significant.

Suppose you take a 2 way system with an active 12 db/octave crossover with a crossover frequency of 3Khz.

If I were to add a capacitor in series ( a first order HPF to simplfy things), with a cutoff Frequency of , say, 1khz.

Which means that when the passive crossover kicks in, the signal to the tweeter is already 18 db down ~ approximately 1% of the overall signal power, and decreasing. So in this particular case, the effect of this extra crossover on frequency response would be <1%.
i do get his point but what is being ignored within his point is that if you change the slope then you could create phase problems since a first order crossover ie 6db/oct changes phase by 90 degrees whilst a second order 12db/oct will change phase by 180 degrees and a third order ie 18db/oct will change phase by 270 degrees

due to complications like this i would say that the end result will become unpredictable for most non-installers, and even for some inexperienced pro installers. So my point is that it isnt worth adding all these complications and potential socalled "traps" for the unwary when they are not really needed

IMHO the only time it is necessary for added caps etc, is if you expect the active crossover to fail and i have not encountered that yet. i have used speakers fully active with no other added protection and if you tune carefully enough you wont have many problems also tuning becomes easier the simpler you keep the setup because you avoid things like cascading equipment and all the potential complications that can occur

while you may like to minimize the effect all of those adjustments will cause because the crossover points seem so close, the resultant phase anamolies could create cancellations that could have a real effect on the frequency response and i have come across this ie where the system has a huge hole at the crossover point and you cannot EQ this out because of the cancellation thats occuring so no matter how much you increase the output of that particular frequency with EQ you never solve the problem because that frequency still cancels and is absent from your final output

remember the crossover point does not cause the frequency to cut off like a steep cliff face, it decreases at the rate of the slope so the frequencies you are attempting to reduce are still present at reducing rates
naughty001 is offline   Reply With Quote
Old 21st December 2009, 14:13   #20
Senior - BHPian
 
DerAlte's Avatar
 
Join Date: Mar 2007
Location: Bangalore
Posts: 8,076
Thanked: 2,876 Times
Default

Naughty, naughty! I guess you will not believe any systematic explanation till you see it on an osci for yourself.

Hook up an XO to a sweep gen as input and see the f-response waveform on the osci! Then, try adding cascading another XO to see the effect.

You are correct from the p-o-v of simple elegant installs.
DerAlte is offline   Reply With Quote
Old 21st December 2009, 14:40   #21
Awaiting Email Confirmation
 
Join Date: Jun 2009
Location: South Africa
Posts: 185
Thanked: 11 Times
Default

Quote:
Originally Posted by DerAlte View Post
Naughty, naughty! I guess you will not believe any systematic explanation till you see it on an osci for yourself.

Hook up an XO to a sweep gen as input and see the f-response waveform on the osci! Then, try adding cascading another XO to see the effect.

You are correct from the p-o-v of simple elegant installs.
well sir you are still not getting my point. In isolation on one set of channels i agree that the cascade will only affect the slope and nothing much else, but the slope will affect the phase and when taken in conjunction with other channels (for example this should be a tweeter highpass then add in the midrange lowpass and those would then have different slopes hence differing phase characterisitics - remember we are concerned with the effects on the system and not only on one set of channels) this could have an adverse effect .... notice i say "could" but why risk that with stuff thats not "really" needed.

either way its not a trainsmash since as i have said i have never had a problem with going active and having speakers needing extra caps or other protection nor have i ever seen a situation which warranted it, and i was using 200 watts RMS on my last set of MB Quart tweeters too prior to my current Dynaudio set and the Quarts are still working perfectly

i can understand though that anyone else investing in expensive equipment wanting to protect their investment so i wont stress the point any further, they need to do whatever they feel will protect their investment ... me personally i dont mind if stuff blows up so i can move to the next step up even though its not great for the pocket, and i understand that not everyone will prefer to work in my "kamikaze" style so no more arguments from me on this topic
naughty001 is offline   Reply With Quote
Old 21st December 2009, 16:16   #22
Team-BHP Support
 
navin's Avatar
 
Join Date: Mar 2004
Location: mumbai
Posts: 22,509
Thanked: 3,885 Times
Default

Quote:
Originally Posted by DerAlte View Post
Not quite! If you draw a response graph, it will be easier to understand. In the scenario you have highlighted, the max. of the 2 slopes - 12dB - will be the applicable slope.
Quote:
Originally Posted by naughty001 View Post
no sir its nothing to do with maths, if you look at crossover design
Quote:
Originally Posted by naughty001 View Post
Yet this is not what happens because when you cascade the crossovers you will get a slope equal to the total of all the slopes used
Quote:
Originally Posted by greenhorn View Post
naughty, deralte's point is that if the crossover frequencies are sufficiently seperated, then the effect of cascading will not be significant.
1. if the tweeter is powered by a good amp without a turn on thump it should be safe. I doubt if a tweeter (even a Dynaudio) can absorb 200W rms but I do not think Naughty's tweeters are doing this. The amp is rated at producing 200W but (I assume) it does not produce but a fraction of this under normal use.

2. Cascading 2 filters can add up to a filter that is a sum of both filters. The most famous example of cascaded filters is a Linkwtiz Riley 4th order filter which is actually 2 2nd order Butterworth filters cascaded.
navin is offline   Reply With Quote
Old 21st December 2009, 21:36   #23
Awaiting Email Confirmation
 
Join Date: Jun 2009
Location: South Africa
Posts: 185
Thanked: 11 Times
Default

Quote:
The amp is rated at producing 200W but (I assume) it does not produce but a fraction of this under normal use.
agreed 100% since the amp rating is a nominal (averaged) rating so i would assume that the actual average power the speaker receives at any point in time would be closer to a real continuous 10 watts only even though the amp is rated at 200 watts RMS and god knows how many watts of ILS (If Lightning Strikes ) PMPO

Quote:
if the tweeter is powered by a good amp without a turn on thump it should be safe
yep ... i didnt have any turn on thump because i used a relay on the remote so despite multiple wires it was still a strong 12v to all the amps ... and phoenix gold amplifiers also have a delayed remote terminal to use instead of the regular remote terminal in conjunction with some resistors if i was suffering any turn on thump so i could have avoided that as well

Quote:
The most famous example of cascaded filters is a Linkwtiz Riley 4th order filter which is actually 2 2nd order Butterworth filters cascaded.
100% spot on, but nevertheless i do get the point about how negligible the effect can be if you are lucky when adding the cap but if you do unfortunately get unlucky you could off course cause some unsolvable phase issues ie only way to solve them being to remove the offending part. so my solution is to not create the potential problem in the first place

Last edited by naughty001 : 21st December 2009 at 21:39. Reason: correcting my atrocious spelling
naughty001 is offline   Reply With Quote
Old 21st December 2009, 22:45   #24
Senior - BHPian
 
greenhorn's Avatar
 
Join Date: Apr 2007
Location: TN-14
Posts: 6,839
Thanked: 1,318 Times
Default

naughty, by that argument, would you not use a fuse when connecting your amp to the battery? The resistance of the fuse would reduce voltage to the amp, and the amp would not draw more current than the rating of the fuse you were going to use anyway

Last edited by greenhorn : 21st December 2009 at 22:47.
greenhorn is offline   Reply With Quote
Old 22nd December 2009, 11:14   #25
Team-BHP Support
 
navin's Avatar
 
Join Date: Mar 2004
Location: mumbai
Posts: 22,509
Thanked: 3,885 Times
Default

Quote:
Originally Posted by DerAlte View Post
There is maths behind any electrical behavior, and that is what helps in electrical design.

With 2 XOs cascaded, the output voltage is
Vo2 = f2( Vo1) ) or
Vo2 = f2( f1(Vi1) ) since Vo1 = f1(Vi1)
Aren't Bandpass and Notch filters are in effect cascaded filters made up of low pass filters combined with high pass filters?

Quote:
Originally Posted by DerAlte View Post
Naughty, naughty!
This reminded me of a song by John Parr (famous for St. Elmo's fire) the chorus went something like this..
"Naughty naughty, loud and bawdy, tttttease me
Take it easy, hug and squeeze me
I'm a naughty, naughty, naughty, I'm a naughty naughty guy"


Quote:
Originally Posted by greenhorn View Post
naughty, by that argument, would you not use a fuse when connecting your amp to the battery?
a fuse is in the power section a capcitor will be in the signal path.
navin is offline   Reply With Quote
Old 23rd December 2009, 23:27   #26
BHPian
 
pranava999's Avatar
 
Join Date: Aug 2006
Location: BANGALORE
Posts: 820
Thanked: 19 Times
Default

Wow this thread's turned pretty technical and 'gurucal' A noob like me wants to know if there's any way of identifying good xo?Does such a question make any sense?
pranava999 is offline   Reply With Quote
Old 23rd December 2009, 23:41   #27
Team-BHP Support
 
navin's Avatar
 
Join Date: Mar 2004
Location: mumbai
Posts: 22,509
Thanked: 3,885 Times
Default

Quote:
Originally Posted by pranava999 View Post
Wow this thread's turned pretty technical and 'gurucal' A noob like me wants to know if there's any way of identifying good xo?Does such a question make any sense?
A passive XO is designed for a specific set of drivers (woofers, tweeters, midranges etc..) so a good XO with one set of drivers can be a bad XO is used with another set of drivers. that said look for XOs that use teflon/polyester/polystyere caps and air core inductors in the signal path (in some cases large inductors might have soft iron cores). Ferrite cores or electrolytic caps are not considered good for audio.

Active XOs however is usually quite similar in design. Almost all are op-amp based and it is usually the quality of the opamps used that determines if the XO is good or average.
navin is offline   Reply With Quote
Old 24th December 2009, 00:07   #28
Senior - BHPian
 
greenhorn's Avatar
 
Join Date: Apr 2007
Location: TN-14
Posts: 6,839
Thanked: 1,318 Times
Default

ok. let me add another topic. There is a misconception that higher order crossovers = steeper slopes = better

That is not universally true right ?
greenhorn is offline   Reply With Quote
Old 24th December 2009, 04:28   #29
Awaiting Email Confirmation
 
Join Date: Jun 2009
Location: South Africa
Posts: 185
Thanked: 11 Times
Default

Quote:
Originally Posted by greenhorn View Post
ok. let me add another topic. There is a misconception that higher order crossovers = steeper slopes = better

That is not universally true right ?
IMHO its not a misconception as long as you keep the slope constant ie dont highpass your high frequency speakers at one slope and lowpass the sub at another then you wont have any phase issues

but the reason why the steeper slope is better because let us say you have a speaker with a free air resonant frequency (Fs) of 500hz, there is a general unwritten kind of rule that the usable frequency response of that speaker at a slope of 12db/oct would be one octave above the Fs so that means your crossover point would be a highpass of 1000hz for this speaker, which means that when the speaker wants to naturally resonate that 500hz would be playing 12db softer than the 1000hz, this means that the resonance (vibration) at 500hz would not be noticeably audible

so when you increase the slope to 24db/oct to have that 500hz frequency 12db lower you could use a frequency of 750hz and that 500hz would still be 12db lower which takes that resonance to below and audible level so you are increasing the usable frequency response from that speaker

where this would help is basically if you were designing an active system where you can change the slopes and crossover points and you want to have either a certain degree of underlap or overlap created between the crossover points of certain drivers ... note though normally this would be an underlap because you want to generally leave a gap ie for example a 12 db/octave crossover would allow for the crossover points to intersect where the points that meet are at a -3db or -6db point (depending on the topology of the filter being used) .... im going to use random numbers as an example here so if you have a low pass for the sub at 50hz and use a high pass for the high frequency speakers at 70hz logically you would think that there is a gap in the response but it isnt so because as both drivers decrease there comes a point where the response of both drivers would intersect and at that point everything below would be a replication of the intersecting frequencies which causes a hump in the response in other words the duplicating frequencies become twice as loud since its the same frequencies played by both the drivers and this "fills" in the deliberate gap you left

if you did not leave the gap this means that you would have had a huge noticeable hump at the crossover point in the overall system response curve but leaving the gap flattens this out as described above - steeper slopes also mean that this underlap needs to be smaller

normally with passive crossovers you dont need to bother with all of this because the crossover designer has already taken all of this into account and designed the crossover taking all of this into account as well as sometimes including impedance compensation (normally in the form of a zobel network) as well as L-pads for tweeter attenuation etc. This is one of the many reasons why a passive from one system might not work well with different speakers together with different overall response characteristics of speaker drivers as Navin has pointed out above

when the slopes are constant then you dont get odd phase issues but for example with Dynaudio speakers where the mids have a slope of 6db/oct (1st order) and the tweeters are 12db/oct (2nd order) things become complicated because they have to find a way to compensate for the fact that the phase of the speakers would be 90 degrees different to each other BUT placing these speakers in normal OEM locations ie tweeter on A-pillar and mid in the doors would normally compensate because the large distance between drivers actually shifts the phase again in relation to where your ears typically would be

this is why going active can be a sometimes difficult task but if you understand crossovers properly you can use that knowledge together with some trial and error and generally get things right or you can take the easy way out and leave the speaker set designers settings and use the supplied passive crossover

all of the above only matter if you want a fairly linear frequency response (note i dont say flat) because a flat response at 0dbFS would sound pretty flat and dull but if you really want good imaging and soundstaging the closer you are to flat the better your imaging would be

if you did not understand what ive just said above and consider all of that to be the ramblings of a lunatic (not that far off from the truth, mind you ) then dont worry about it - just keep this in mind it is better to use a steeper slope and there is a technical reason for it

the other reason for steeper slopes being better is that you filter out more of the frequencies that the speaker cannot handle for example as in above if the speakers Fs was 500hz and you used a crossover point of 1000hz then at a slope of 12db/oct 500hz would be 12 db lower and at a slope of 24db/octave it would be playing 24db lower hence the harmful frequency is even more inaudible than usual (has to be good since playing the speaker loud at its resonant frequency would sound bad as well as be damaging)

Last edited by naughty001 : 24th December 2009 at 04:39.
naughty001 is offline   Reply With Quote
Old 24th December 2009, 11:40   #30
Team-BHP Support
 
navin's Avatar
 
Join Date: Mar 2004
Location: mumbai
Posts: 22,509
Thanked: 3,885 Times
Default

Quote:
Originally Posted by greenhorn View Post
There is a misconception that higher order crossovers = steeper slopes = better
Quote:
Originally Posted by naughty001 View Post
but the reason why the steeper slope is better ...so when you increase the slope to 24db/oct to have that 500hz frequency 12db lower you could use a frequency of 750hz ...where this would help is basically if you were designing an active system ...so if you have a low pass for the sub at 50hz and use a high pass for the high frequency speakers at 70hz logically you would think that there is a gap in the response but it isnt so because as both drivers decrease ...

normally with passive crossovers ...as well as sometimes including impedance compensation (normally in the form of a zobel network) as well as L-pads for tweeter attenuation etc.

but for example with Dynaudio speakers where the mids have a slope of 6db/oct (1st order) and the tweeters are 12db/oct (2nd order) things become complicated ...BUT placing these speakers in normal OEM locations ie tweeter on A-pillar and mid in the doors would normally compensate ...

all of the above only matter if you want a fairly linear frequency response (note i dont say flat) because a flat response at 0dbFS would sound pretty flat and dull
"Naughty" is being really really naughty!

higher order crossovers also have greater phase shift. hence they are not nesscarily better. when a crossover is corredtly designed (active or passive) it must account for not only the capabilities of the drivers involved (how high or how low they can go) but also compensate for the anamolies of the system (driver anamolies, the issues cabinets create, and room/cabin compensation). This compensation comes in various forms (Naughty has listed 2 - tweeter vs woofer senstivity & impendace compensation) but also inclucde the vagrancies in the woofer/tweeters frequency response, baffle step (if there is a box involved), edge/difraction compensation, phase alignment, etc...

Obviously there are only so many problems a crossover cna practically correct for - after that the crossover starts to detract from the sound rather than make it better. Hence DIYers prefer to start with drivers that do not have too many issues to begin with. It makes crossover design simpler and also money can be spent on a few good components than many poor/average ones.

in short do not consider a speaker better or worse becuase of it's crossover's slope/topology but consider the entire design as a system and look at the entire design in it's totality.
navin is offline   Reply With Quote
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Crossovers (CUV) - The new trend worldwide volkman10 The Indian Car Scene 95 13th June 2017 14:43
Features comparison of Crossovers under Rs 10 lakhs jessie007 The Indian Car Scene 13 2nd November 2014 14:39
Hiding passive crossovers navin In-Car Entertainment 9 21st October 2008 21:36
Spark, Santro, i10 or others - FE important - 1500kms/month, Rear seat not important monishdesign Hatchbacks 7 11th March 2008 12:27
BMW crossovers the X6 and the X4 rahul_intlad The International Automotive Scene 4 19th March 2007 20:54


All times are GMT +5.5. The time now is 19:38.

Copyright ©2000 - 2017, Team-BHP.com
Proudly powered by E2E Networks