I've been using the crossover below with KEF B139/B110/T27 drivers for many, many years in a DIY ETI transmission line cabinet and I'm basically very happy with the sound except the midrange has always been too forward for my liking - mostly voices are too strident and a bit 'cuppy'.
So I've also been using a 15ohm resistor in series with the B110 midrange driver and this seems to bring a better balance. Now I know a single resistor is not ideal so I bought some 2.2 and 18ohm resistors today to put in a proper L-pad (B110 is 7.8ohm).
But looking at the crossover again it occured to me that maybe removing the 5ohm and 12ohm tweeter resistors might bring about a similar effect without introducing more components. Or perhaps I could alter some of the midrange network which the instructions say 'reinforces output below 600hz'. Which is the best way to go? (PS I do not use the HF2000 supertweeter section of the circuit as I use the T27 alone.)
So I've also been using a 15ohm resistor in series with the B110 midrange driver and this seems to bring a better balance. Now I know a single resistor is not ideal so I bought some 2.2 and 18ohm resistors today to put in a proper L-pad (B110 is 7.8ohm).
But looking at the crossover again it occured to me that maybe removing the 5ohm and 12ohm tweeter resistors might bring about a similar effect without introducing more components. Or perhaps I could alter some of the midrange network which the instructions say 'reinforces output below 600hz'. Which is the best way to go? (PS I do not use the HF2000 supertweeter section of the circuit as I use the T27 alone.)
Attachments
Last edited:
The crossovers are very messy and difficult to work with so I've decided to completely rebuild them. Jantzen Z-Superiors in the tweeter section and Crosscaps in the mids and bass.
I've also seperated the crossover into a tri-wire (schematic attached).
I'm wondering if the Butterworth tweeter section of the KEF 104aB would be an improvement over the one I use, as it seems to be regarded highly by 104 owners. Would there be any problems substituting it into my crossover (which seems to be very similar to the P. Atkinson 'State of the Art Loudspeaker' crossover) ?
I've also seperated the crossover into a tri-wire (schematic attached).
I'm wondering if the Butterworth tweeter section of the KEF 104aB would be an improvement over the one I use, as it seems to be regarded highly by 104 owners. Would there be any problems substituting it into my crossover (which seems to be very similar to the P. Atkinson 'State of the Art Loudspeaker' crossover) ?
Attachments
KEF Crossover revision
KEF brought out its own computer optimised crossover for the B139(SP1044)/B110(SP1057) combination in their Constructor Series design CS7.
See-http://www.hifiloudspeakers.info/Anatomy/KEFDocuments/KEFConstructorSeries/KEFCS7Leaflet/Gallery/pages/KEFCS7-3.htm
The B110 section uses a larger 1.6MH series inductor with 2.5 ohms intrinsic resistance.The high pass network also includes the AB mod for the T33 tweeter(higher sensitivity version of T27)
The supplied curves from KEF suggest a very flat response.
Having tried this crossover configuration,I found the midrange sounds much better balanced compared to various early magazine designs from ETI,HFNRR etc.
Suitable PCB's may still be available at Falcon acoustics UK.
KEF brought out its own computer optimised crossover for the B139(SP1044)/B110(SP1057) combination in their Constructor Series design CS7.
See-http://www.hifiloudspeakers.info/Anatomy/KEFDocuments/KEFConstructorSeries/KEFCS7Leaflet/Gallery/pages/KEFCS7-3.htm
The B110 section uses a larger 1.6MH series inductor with 2.5 ohms intrinsic resistance.The high pass network also includes the AB mod for the T33 tweeter(higher sensitivity version of T27)
The supplied curves from KEF suggest a very flat response.
Having tried this crossover configuration,I found the midrange sounds much better balanced compared to various early magazine designs from ETI,HFNRR etc.
Suitable PCB's may still be available at Falcon acoustics UK.
Thanks for the suggestions.
Eanee: thanks for that link. I've been to that site many times but never noticed that crossover. My B110 is a SP1003 so its a bit different to the SP1057 used there. Will have to have a closer look at the differences in the specs of those drivers. Did you use the T27 or the T33 with the CS7 crossover?
I've never seen two inductors used in series like that-is that a 3rd order bass section? How was the bass with this crossover compared to the older style crossover?
Its a bit hard to make out some of the values on that page: is L4=3.0mH? And is C5=0.6uF? The combined R of L1 and L2 seems quite high at 1.7ohms.
Eanee: thanks for that link. I've been to that site many times but never noticed that crossover. My B110 is a SP1003 so its a bit different to the SP1057 used there. Will have to have a closer look at the differences in the specs of those drivers. Did you use the T27 or the T33 with the CS7 crossover?
I've never seen two inductors used in series like that-is that a 3rd order bass section? How was the bass with this crossover compared to the older style crossover?
Its a bit hard to make out some of the values on that page: is L4=3.0mH? And is C5=0.6uF? The combined R of L1 and L2 seems quite high at 1.7ohms.
Re Kef Upgrade
Hi Metako,
The KEF Constructor Series CS7 was based on the commercial Cantata which used the B139/B110 plus the more expensive T52,a refined version of the earlier T15.
I used the T27 which is marginally less sensitive but I prefer a less upfront treble balance anyway.
Kef used the 3rd order low pass to deal with the colouration caused by the well known peak the B139 exhibits at the upper end of its operating range.
Similarly the mid section has a 24dB upper rolloff to attenuate the B110 breakup nodes in it's upper range.This does help reduce the infamous nasal quack that the B110 has often been accused of.
I have tried the SP1003 version of the B110 and felt it required a larger value of series inductor I/L/O 1.6mh,to equal the response of the smoother SP1057 midrange.
2mH-2.4mH would be a good starting point.
KEF were very specific about abiding by their recommendations if results were to be guaranteed.This include the use of cored inductors and reversible electrolytics eg ALCAP-
"The combined R of L1 and L2 seems quite high at 1.7ohms"
My understanding of their recommendation is "less than 1 ohm","less than .7 ohm" which is easy to achieve with eg.high power iron dustcore inductors.My experience is that a little series R doesn't do any harm anyway as in this case the B139 has quite a prodigious output regardless.
Hi Metako,
The KEF Constructor Series CS7 was based on the commercial Cantata which used the B139/B110 plus the more expensive T52,a refined version of the earlier T15.
I used the T27 which is marginally less sensitive but I prefer a less upfront treble balance anyway.
Kef used the 3rd order low pass to deal with the colouration caused by the well known peak the B139 exhibits at the upper end of its operating range.
Similarly the mid section has a 24dB upper rolloff to attenuate the B110 breakup nodes in it's upper range.This does help reduce the infamous nasal quack that the B110 has often been accused of.
I have tried the SP1003 version of the B110 and felt it required a larger value of series inductor I/L/O 1.6mh,to equal the response of the smoother SP1057 midrange.
2mH-2.4mH would be a good starting point.
KEF were very specific about abiding by their recommendations if results were to be guaranteed.This include the use of cored inductors and reversible electrolytics eg ALCAP-
"The combined R of L1 and L2 seems quite high at 1.7ohms"
My understanding of their recommendation is "less than 1 ohm","less than .7 ohm" which is easy to achieve with eg.high power iron dustcore inductors.My experience is that a little series R doesn't do any harm anyway as in this case the B139 has quite a prodigious output regardless.
Attachments
Last edited:
Noting your comments on the midrange, I built some bookshelf speakers using the B110 as a bassmid a few years back.
First measurement of the frequency response of the B110 certainly raised my eyebrows. Far from flat, and with leaks that really needed treatment. My rusty memory says a 5B peak at about 1KHz and a dip at 500Hz. After a lot of mucking around I concluded the measurements were real.
The LS3A used these drivers, and looking at the schematic of the crossover includes a trap to tame the peak this drivers rather unruly response.
Have attached one version of the LS3A crossover that I downloaded just now from a random site - note it the trap will punch in something over a 6dB notch at resonance.
The thing that really surprised me is that these are quite famous speakers, and drivers!
First measurement of the frequency response of the B110 certainly raised my eyebrows. Far from flat, and with leaks that really needed treatment. My rusty memory says a 5B peak at about 1KHz and a dip at 500Hz. After a lot of mucking around I concluded the measurements were real.
The LS3A used these drivers, and looking at the schematic of the crossover includes a trap to tame the peak this drivers rather unruly response.
Have attached one version of the LS3A crossover that I downloaded just now from a random site - note it the trap will punch in something over a 6dB notch at resonance.
The thing that really surprised me is that these are quite famous speakers, and drivers!
Hi googlyone. That LS3A crossover is a complex one. Interesting the different ways of tackling the same problem.
Hi Eanee. Thanks for those clearer pics. So the T27 will work OK in the CS7 circuit without modification. I wonder if the T27 tweeter section of the 104aB crossover (see attachment) would be a better to use with the CS7 mid and bass section. I'm slowly learning how to use crossover simulation software so hopefully soon I'll be able to see how they would work together. Thanks for the suggestion re the need for a larger series mid inductor too.
Hi Eanee. Thanks for those clearer pics. So the T27 will work OK in the CS7 circuit without modification. I wonder if the T27 tweeter section of the 104aB crossover (see attachment) would be a better to use with the CS7 mid and bass section. I'm slowly learning how to use crossover simulation software so hopefully soon I'll be able to see how they would work together. Thanks for the suggestion re the need for a larger series mid inductor too.
Attachments
kef mods
Hi metako,
Yes I think the KEF 104aB high pass section, being set at a reference level of approx 85dB, would be just fine for the B139/B110/T27 setup.
The 18dB high pass slope is pretty important to avoid exciting the rather high fundamental resonance of the T27-1.5kHz or thereabouts IIRC.
Hi googlyone,
It's interesting to read a little of the history of the BBC LS3/5a.
Apparently the various licensed manufacturers had real problems trying to ensure their production remained within BBC specs due to the variability of the raw drivers,especially the B110 SP1003 version.I recall Martin Colloms testing the LS3/5a in Hi Fi Choice and finding a huge 6dB peak at 1.2kHz.Late in the day KEF introduced the improved B110 SP1228 version along with calibrated matching crossovers to solve this variability problem.
Hi metako,
Yes I think the KEF 104aB high pass section, being set at a reference level of approx 85dB, would be just fine for the B139/B110/T27 setup.
The 18dB high pass slope is pretty important to avoid exciting the rather high fundamental resonance of the T27-1.5kHz or thereabouts IIRC.
Hi googlyone,
It's interesting to read a little of the history of the BBC LS3/5a.
Apparently the various licensed manufacturers had real problems trying to ensure their production remained within BBC specs due to the variability of the raw drivers,especially the B110 SP1003 version.I recall Martin Colloms testing the LS3/5a in Hi Fi Choice and finding a huge 6dB peak at 1.2kHz.Late in the day KEF introduced the improved B110 SP1228 version along with calibrated matching crossovers to solve this variability problem.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.