Good morning everyone. I have some B&W DM602 (S1) boxes, and I have been reviewing their crossover.
I see that it is very simple, with first order for woofer and third order for tweeter.
While I like its sound, I find the treble region a bit tiring.
I have seen a modification for the B&W DM 601 that attempts to correct that problem.
It consists of modification of the capacitor in the high pass, inversion of the polarity of the tweeter and the addition of a resistor and capacitor for second order in low pass.
I show the diagram of DM601. The difference with the DM602 is that this ultimp the high pass filter has no resistors
I don't have measuring instruments, so I ask if it would be beneficial to do the same on the B&W DM602 version 1.
In this configuration, where the high pass is third order and the low pass is first order, how is the positive of the tweeter connected?
Thanks.
I see that it is very simple, with first order for woofer and third order for tweeter.
While I like its sound, I find the treble region a bit tiring.
I have seen a modification for the B&W DM 601 that attempts to correct that problem.
It consists of modification of the capacitor in the high pass, inversion of the polarity of the tweeter and the addition of a resistor and capacitor for second order in low pass.
I show the diagram of DM601. The difference with the DM602 is that this ultimp the high pass filter has no resistors
I don't have measuring instruments, so I ask if it would be beneficial to do the same on the B&W DM602 version 1.
In this configuration, where the high pass is third order and the low pass is first order, how is the positive of the tweeter connected?
Thanks.
Attachments
Last edited:
The original (left hand) diagram shows the positive terminal of the tweeter connected on the positive side - same polarity as the woofer. That seems obvious, so perhaps I've misunderstood your question.
Increasing the value of the 0.47 ohm resistor in that circuit would tame the treble and make it less tiring.
Increasing the value of the 0.47 ohm resistor in that circuit would tame the treble and make it less tiring.
The original of the DM602 model is the same as on the left but without the 0.47 and 1.8 ohms resistors.
My question is reversing the tweeter would be beneficial.
I don't have measuring instruments but by ear I hear some difference but I don't know if it's the right thing to do.
The addition of the Zobel network in the woofer (4.7uf and 1.8 ohms) that would influence.
My question is reversing the tweeter would be beneficial.
I don't have measuring instruments but by ear I hear some difference but I don't know if it's the right thing to do.
The addition of the Zobel network in the woofer (4.7uf and 1.8 ohms) that would influence.
The quick answer is that you connect the tweeter the way round that sounds best to you.
What would the influence of the Zobel be in relation to your problem?
What would the influence of the Zobel be in relation to your problem?
The S1 has only a 'first order' low pass for the bass driver. I put 'first order' in apostrophies because it is not truely first order; the rising impedance caused by the inductance of the woofer will detune the crossover significantly. The rolloff of the bass driver is likely to be only half what it is supposed to, thus the mid output from the woofer will be higher and there will be less suppression of cone and dust-cap breakup modes than a true first order.
The S2 crossover has a Zobel network across the bass driver to correct the voice coil reactance (self inductance) so that the crossover is terminated properly to achieve (hopefully) a true first order acoustic response from the woofer (that being the combination of crossover and driver responses).
Now that the mids and upper mids from the bass driver are attenuated (compared to S1) the tweeter will also need to be attenuated relative to S1 to blend properly across the crossover region, hence the larger attenuator value of 3.3Ω in series with the tweeter.
In an 'ideal' crossover there is 90º of phase shift at the crossover frequency for each order of the crossover. The bass driver first order low pass will be -90º and the third order tweeter crossover will be +270º, making 360º total, bring the two drivers back into phase at the crossover frequency (ignoring the physical displacement of acoustic centers, that is).
I would guess that the reason for the inverted tweeter connection in the S2 crossover is that the physical displacement between the woofer's and tweeter's acoustic centers is approximately an odd multiple of the half wavelength at the crossover frequency (inverting the tweeter gives a similar effect as delaying by a half cycle at the crossover frequency).
If the crossover is 3kHz, for example, the half wavelength is (340m/3000Hz)/2=~6cm. That might be close to the physical displacement of the woofer's and tweeter's voice coils. Of course, I have not made any allowance for the phase response of either driver; these are just crude assumptions.
In summary, the S2 has a more sophisticated crossover that is likely to sound better because of better suppression of undesirable behaviours of the woofer at and above the high end of its passband. If the electroacoustic parameters of the S2 drivers are close to those of the S1 drivers, then the S1's can probably be upgraded by adopting the S2 crossovers. The "correct" phasing of the tweeter will be "select on test". The quick way to do this is with a spectrum analyser.
Measure the on axis response at at least one meter from the boxes. In one connection there may be a narrow and/or deep notch at the crossover. If so the the other way is the correct connection!
If a notch is not discernible, the best connection is that which is smoothest in the octave either side through the crossover region. If the phasing of the tweeter is still indeterminate, then choose the way that sounds best as you walk around the room; this phasing will give the best integration of low and high frequency sections.
The S2 crossover has a Zobel network across the bass driver to correct the voice coil reactance (self inductance) so that the crossover is terminated properly to achieve (hopefully) a true first order acoustic response from the woofer (that being the combination of crossover and driver responses).
Now that the mids and upper mids from the bass driver are attenuated (compared to S1) the tweeter will also need to be attenuated relative to S1 to blend properly across the crossover region, hence the larger attenuator value of 3.3Ω in series with the tweeter.
In an 'ideal' crossover there is 90º of phase shift at the crossover frequency for each order of the crossover. The bass driver first order low pass will be -90º and the third order tweeter crossover will be +270º, making 360º total, bring the two drivers back into phase at the crossover frequency (ignoring the physical displacement of acoustic centers, that is).
I would guess that the reason for the inverted tweeter connection in the S2 crossover is that the physical displacement between the woofer's and tweeter's acoustic centers is approximately an odd multiple of the half wavelength at the crossover frequency (inverting the tweeter gives a similar effect as delaying by a half cycle at the crossover frequency).
If the crossover is 3kHz, for example, the half wavelength is (340m/3000Hz)/2=~6cm. That might be close to the physical displacement of the woofer's and tweeter's voice coils. Of course, I have not made any allowance for the phase response of either driver; these are just crude assumptions.
In summary, the S2 has a more sophisticated crossover that is likely to sound better because of better suppression of undesirable behaviours of the woofer at and above the high end of its passband. If the electroacoustic parameters of the S2 drivers are close to those of the S1 drivers, then the S1's can probably be upgraded by adopting the S2 crossovers. The "correct" phasing of the tweeter will be "select on test". The quick way to do this is with a spectrum analyser.
Measure the on axis response at at least one meter from the boxes. In one connection there may be a narrow and/or deep notch at the crossover. If so the the other way is the correct connection!
If a notch is not discernible, the best connection is that which is smoothest in the octave either side through the crossover region. If the phasing of the tweeter is still indeterminate, then choose the way that sounds best as you walk around the room; this phasing will give the best integration of low and high frequency sections.
Last edited:
Johnmath, I appreciate your explanation. Really very clear and with very interesting theoretical foundations.
I have always seen this very elementary crossover despite the fact that at the time these boxes were not cheap.
I will try to do some of the recommended tests and listen, since I do not have measuring instruments.
Regarding woofer cone breakage, can't you try with a notch filter? Add a small capacitor in parallel to the low pass inductor ?. I say this along with the Zobel network aggregate.
Thank you.
I have always seen this very elementary crossover despite the fact that at the time these boxes were not cheap.
I will try to do some of the recommended tests and listen, since I do not have measuring instruments.
Regarding woofer cone breakage, can't you try with a notch filter? Add a small capacitor in parallel to the low pass inductor ?. I say this along with the Zobel network aggregate.
Thank you.
You can download a free spectrum analyser for iPhone from the App store. For a test signal source you should be able to stream or download "pink noise" to use live, save to a file or burn to a CD. I sometimes use off-band FM tuner noise which is 'pink-ish' (a good approximation to pink noise) above 1kHz to 15kHz or so, and can be used for the tweeter phasing test described in my previous post.
It is possible to design RLC filters to reduce breakup frequencies. However the breakup modes are often very narrowband which means large values of tight tolerance inductors and capacitors that become expensive very quickly.
A better approach sometimes is to identify and remove the causes of the breakup modes. I use an open stethoscope to listen to what sounds are being contributed by different parts of a driver and/or speaker box. Doping or felts can be stuck on dust-caps, constrained layer damping materials stuck on speaker baskets, enclosure walls, etc. Be wary that adding mass to a bass driver will alter its Fs and sensitivity.
However getting the fundamentals right first is the correct approach. For crossover design step one is to design crossover elements, and if necessary Zobal networks so that the driver correctly terminates the crossover, to achieve the achieve the desired acoustic output of crossover+driver for the system design.
Then worry about other problems left over, which will often have been reduced by step one anyway.
If you don't do it this way, the phase response can be all over the place like a madman's breakfast and it will be difficult to get the system to be coherent even if it has a reasonably flat frequency response.
I often use Zobel networks on all drivers and notch filters to smooth out particular drivers. This is quite an involved process and needs quite a bit of knowledge (maths, physics, psychoacoustics, electronic engineering, mechanical engineering, etc) as well as an investment in the right test equipment.
I described how I optimised the crossover for an old KEF Concerto driver based speaker system for a friend here:
3rd order butterworth, proper implementation
It is possible to design RLC filters to reduce breakup frequencies. However the breakup modes are often very narrowband which means large values of tight tolerance inductors and capacitors that become expensive very quickly.
A better approach sometimes is to identify and remove the causes of the breakup modes. I use an open stethoscope to listen to what sounds are being contributed by different parts of a driver and/or speaker box. Doping or felts can be stuck on dust-caps, constrained layer damping materials stuck on speaker baskets, enclosure walls, etc. Be wary that adding mass to a bass driver will alter its Fs and sensitivity.
However getting the fundamentals right first is the correct approach. For crossover design step one is to design crossover elements, and if necessary Zobal networks so that the driver correctly terminates the crossover, to achieve the achieve the desired acoustic output of crossover+driver for the system design.
Then worry about other problems left over, which will often have been reduced by step one anyway.
If you don't do it this way, the phase response can be all over the place like a madman's breakfast and it will be difficult to get the system to be coherent even if it has a reasonably flat frequency response.
I often use Zobel networks on all drivers and notch filters to smooth out particular drivers. This is quite an involved process and needs quite a bit of knowledge (maths, physics, psychoacoustics, electronic engineering, mechanical engineering, etc) as well as an investment in the right test equipment.
I described how I optimised the crossover for an old KEF Concerto driver based speaker system for a friend here:
3rd order butterworth, proper implementation
That may be conflating the mechanical side with the electrical side. A Zobel network is not a requirement nor a method of producing a correctly terminated crossover, for damping a driver, or for producing the correct acoustic response.
You can't remove breakup modes with an RLC filter, and the existence of a response peak may be suggesting there is one.
It is possible to reduce the audible effect of driver resonance and/or breakup with RLC filters, where for some reason it isn't practical to do it mechanically.
Have you ever tried to use a magnesium cone Seas Excel mid-woofer with its 4~5kHz peaks >30dB above the passband response?
There is little reason not to use a filter where a driver suffers audible breakup above its desired passband, Such response tailoring is not uncommon practice in commercial loudspeaker designs.
Have you ever tried to use a magnesium cone Seas Excel mid-woofer with its 4~5kHz peaks >30dB above the passband response?
There is little reason not to use a filter where a driver suffers audible breakup above its desired passband, Such response tailoring is not uncommon practice in commercial loudspeaker designs.
Last edited:
I think there is some confusion here. The two images in post #1 don't refer to two distinct speakers (S2 vs S1). The first image is the factory B&W crossover for the DM601 S3 speaker, whereas the second image is a custom mod for the same speaker. You can read the story here: B&W Bowers Wilkins DM 601-S3 Crossover Upgrade for Better Midrange Perfoprmance
BTW, good use of the existing parts in designing a better crossover, maybe it can be even bettered with different parts (or maybe not). I think that here the acoustic slopes aren't standard ones.
As can be seen in the measurements (see link above), the DM601 S3 mid-woofer don't show a breakup, so there is no need to implement an RCL network for breakup suppression. This doesn't imply that this is also true for the DM602 S1 mid-driver, however.
The crossover of the OP speaker (DM602 S1) can be find here: https://bwgroupsupport.com/manuals/bw-service
It is similar to the DM601 S3 crossover but it miss the tweeter padding resistor and also the R2 shaping resistor in the tweeter HP filter (this can significantly alter the roll off of the HP filter).
Thinking to fit a custom mod for a DM601 S3 to a DM602 S1, which use a different and bigger woofer, without having measurement equipment is risky, and will significantly lower the resale value of the speakers. Another point to take into account, the custom mod but also the factory crossover both have a high crossover point, something acceptable for the 601, but IMHO not for the 602. However if the S1 tweeter is similar to the S3 one, it could have a Fs too high for an ideal mating to a 6.5" mid.
To recap: if you have nothing to lose (i.e. you don't fear to lower the resale value of the speakers), you can try the mod. Keep in mind that the sensitivity of the S1 drivers can be (slightly) different than the S3 drivers, in particular the 6.5" can be more sensitive than the 5" one, so you may try slightly different values for the tweeter padding resistor.
Ralf
BTW, good use of the existing parts in designing a better crossover, maybe it can be even bettered with different parts (or maybe not). I think that here the acoustic slopes aren't standard ones.
As can be seen in the measurements (see link above), the DM601 S3 mid-woofer don't show a breakup, so there is no need to implement an RCL network for breakup suppression. This doesn't imply that this is also true for the DM602 S1 mid-driver, however.
The crossover of the OP speaker (DM602 S1) can be find here: https://bwgroupsupport.com/manuals/bw-service
It is similar to the DM601 S3 crossover but it miss the tweeter padding resistor and also the R2 shaping resistor in the tweeter HP filter (this can significantly alter the roll off of the HP filter).
Thinking to fit a custom mod for a DM601 S3 to a DM602 S1, which use a different and bigger woofer, without having measurement equipment is risky, and will significantly lower the resale value of the speakers. Another point to take into account, the custom mod but also the factory crossover both have a high crossover point, something acceptable for the 601, but IMHO not for the 602. However if the S1 tweeter is similar to the S3 one, it could have a Fs too high for an ideal mating to a 6.5" mid.
To recap: if you have nothing to lose (i.e. you don't fear to lower the resale value of the speakers), you can try the mod. Keep in mind that the sensitivity of the S1 drivers can be (slightly) different than the S3 drivers, in particular the 6.5" can be more sensitive than the 5" one, so you may try slightly different values for the tweeter padding resistor.
Ralf
Thanks Ralf, I clearly misunderstood. My posts still stand, but are not necessarily directly relevant to the OP's 602s.
However there clearly is breakup or bad behaviour in the DM601 S3's mid-bass driver in the 10-12kHz region, and it impacts both in the frequency response and the spectral decay. The revised crossover with the Zobel across the driver deals with the mess pretty well.
The mods to the high pass section attenuate the level of the tweeter which appears to be a few dB hot in the first measurements, and also tilts the slope downwards a little at the upper end.
The latter will possibly help reduce the characteristic 'tish-i-ness' of the classic supersonic metal dome breakup at ~30kHz that bends the clean 'ting' of a triangle or cymbal into an unnatural 'tish' sound. I used to keep a few percussive instruments in my hifi shop auditioning rooms just to demonstrate how that kind of percussion should sound.
It's no wonder the mod'ed 601 sounds better, and there may be some clues there about how to improve the 602.
"it impacts both in the frequency response and the spectral decay."
What does it mean when both are affected at the same time?
What does it mean when both are affected at the same time?
The comment is that the effect of the misbehaviour of the mid-bass at 10-12kHz is visible in both the frequency response and the spectral decay here: B&W Bowers Wilkins DM 601-S3 Crossover Upgrade for Better Midrange Perfoprmance
These are clearly highlighting an issue that will be audible that is reduced by by the upgraded crossover design.
These are clearly highlighting an issue that will be audible that is reduced by by the upgraded crossover design.
Last edited: