In some passive X-overs there are resistors in parallel with some of the L&C components. I think I've seen them described as being there to prevent "ringing". What do they do? Is there a way for the DIYer to determine when they are required and how do you calculate appropriate values? Jonathan bright
Possbly, the designer has at some stage seen a bump or a dip or something in the frequency response, and figured out that it can be fixed by adding that extra L, R or C. Crossover design i built on basic theory (dBs per octave, butterworth/linkwitz, baffle step compensation etc) but is also a commonly used to fine-tune the frequency response. I suppose that this is what you have seen. It is hard to give general recommendations regrding those extra components, the best way of understanding them is to understand passive electric circuit design.
IMO any explanation that uses time-domain wordings, like "step response" and "ringing" should be viewed with the sceptic eye. I'm not saying they are wrong, but most often they don't help much but rather cause confusion and/or wrong conclusions. This does not apply only to crossovers.
IMO any explanation that uses time-domain wordings, like "step response" and "ringing" should be viewed with the sceptic eye. I'm not saying they are wrong, but most often they don't help much but rather cause confusion and/or wrong conclusions. This does not apply only to crossovers.
They are often included for subtle adjusts to phase and
levels in the crossover region to help with subtle driver
integration issues.
They are not needed to prevent ringing or facilitate
a decent step response, this is mainly controlled by
the Q and order of the crossover alignment.
Not easily. You need good measurements and good modelling
software, and enough skill to only use extra resistors when
they are definetely needed.
sreten.
levels in the crossover region to help with subtle driver
integration issues.
They are not needed to prevent ringing or facilitate
a decent step response, this is mainly controlled by
the Q and order of the crossover alignment.
Not easily. You need good measurements and good modelling
software, and enough skill to only use extra resistors when
they are definetely needed.
sreten.Actually, "ringing" is also a not-uncommon term to describe the oscillation that can occur in LC circuits, particularly in high-order crossovers comprising multiple inductors and capacitors. And a well-placed resistor could certainly be used to shift a troublesome resonant frequency out of band, or otherwise tame a problematic oscillation.
I believe this issue is also one reason why people sometimes substitute multiple small capacitors for a single large one.
Of course it's difficult to tell if that's what the designer intended without an actual example, but theory certainly supports this as a reasonable explanation.
I believe this issue is also one reason why people sometimes substitute multiple small capacitors for a single large one.
Of course it's difficult to tell if that's what the designer intended without an actual example, but theory certainly supports this as a reasonable explanation.
HeatMiser said:
Adding resistors to a badly designed filter is not the way to do it.
The Q (i.e. ringing and step response) of higher order LC circuits
can be completely controlled by correct component values.
Multiple capacitors will not reduce Q, due
to lower losses if anything Q will increase.
There is one case where damping resistors are required :
If you want a response between first and second order.
sreten.
HeatMiser said:
...and ruin the frequency response. This is about what I mean by saying that time domain approaches can lead to the wrong conclusions. Also, a resonance frequency cannot be changed by adding resistors. You'd have to change the L or C to do that.
sreten said:
There is one case where damping resistors are required :
If you want a response between first and second order.
Would you have an example of an 1.5th order filter?
Svante said:
Not a good example really, as the astute will realise a low
Q 2nd order is the more sensible way to approach this.
http://www.users.nac.net/markowitzgd/david/wattcrossover.htm
http://www.users.nac.net/markowitzgd/david/watt-x-o.gif
The CB17RC c/o is basically 2nd order but asymptotes to
first order at high frequencies, why ? who knows ?
But Thorsten is simply copying the original c/o.
sreten.I cannot argue with your statements, however I was merely pointing out what the term "ringing" strongly suggests (a time domain problem, an oscillation) and how a resistor might be used to combat this. I'll leave it to you to seek out the designers of these unseen circuits and tell them they are doing it all wrong.sreten said:
Adding resistors to a badly designed filter is not the way to do it.
The Q (i.e. ringing and step response) of higher order LC circuits
can be completely controlled by correct component values.
Multiple capacitors will not reduce Q, due
to lower losses if anything Q will increase.
There is one case where damping resistors are required :
If you want a response between first and second order.
Really should also note that for bass / midrange units any
resistor of in the area of 3R to 8R in parallel with a series
inductor is for Baffle Step correction.
And in crossovers resistors can be used for correcting the
overall response of the driver, and of course be used for
setting the level of the midrange and / or tweeter.
sreten.
resistor of in the area of 3R to 8R in parallel with a series
inductor is for Baffle Step correction.
And in crossovers resistors can be used for correcting the
overall response of the driver, and of course be used for
setting the level of the midrange and / or tweeter.
sreten.
sreten said:Really should also note that for bass / midrange units any
resistor of in the area of 3R to 8R in parallel with a series
inductor is for Baffle Step correction.
And in crossovers resistors can be used for correcting the
overall response of the driver, and of course be used for
setting the level of the midrange and / or tweeter.
Yes, and add zobel networks too.
HeatMiser said:I cannot argue with your statements, however I was merely pointing out what the term "ringing" strongly suggests (a time domain problem, an oscillation) and how a resistor might be used to combat this. I'll leave it to you to seek out the designers of these unseen circuits and tell them they are doing it all wrong.
I never said they were doing it wrong. Your understanding whilst
electrically correct, adding a resistor reduces Q, setting the L and
C values correctly will reduce Q, adding a resistor is not necessary.
If a resistor is included for a good reason, its not to adjust Q, though
conversely the Q may need adjusting to allow the resistor to be used
correctly.
I agree completely with Svante.
sreten.Here's an example (Infinity RSII crossover http://oellerer.net/infinity_classics/RS_II/RS_II_6.jpg ), using what I suppose is some form of baffle step compensation and freq adjustment in the midrange due to 5" driver being used up to 3kHz:
Attachments
This would be an example where the engineers have probably experimentet quite a bit with the components before arriving at this result. I can see that the woofers are crossed over at different frequencies, and also that the midranges. It seems that only the middle mid is active in its upper range. It also appears as if the speaker is an open baffle design, at least for the mid and high, and that there is a tweeter radiating on the back of the baffle? This could explain parts of the filter as well.
Svante said:
Yes, woofers are crossed 65Hz and 120Hz, top/bottom mid 1500Hz, 'middle' midrange 3KHz and two paralleled tweeters, one at the back, from 3kHz upwards; (anechoic room) freq response is 36Hz-32kHz +/- 2dB but gets down to higher 20's in the room.
This might be a good example for 'there's no cook-book recipe' for crossovers
It actually sounds damn good compared to new commercial offerings I've heard below the $3k range.
I have posted a response to this a while back which shows what happens to the freqresponse when you add a resistor. This is NOT in parallel with the inductor or capacitor but rather in series with the shunt capacitor of a lowpass network. Ill post the pic again.
The resistor in question is R2021, when not included you get the 1st reponse, when the value is optomised you get the second response. The third response is just to high light what happens when u make it bigger in this case about 20ohms is bigger about 6 was needed for the opto response.
The resistor in question is R2021, when not included you get the 1st reponse, when the value is optomised you get the second response. The third response is just to high light what happens when u make it bigger in this case about 20ohms is bigger about 6 was needed for the opto response.
Attachments
sreten said:
I was wondering why outside the crossover region
it reverts to 1st order, from a design perspective.
Ah, Ok, well that beats me too.
Could it be that the RC-branch originally was a zobel, but they noticed that the response got better when the R was decreased, in effect creating a 2nd order filter?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.