Having plotted the response in my room without the Kube in circuit I have a noticeable peak at 50Hz and dip at 100Hz and a peak at 4kHz.
NAT, thank you for posting the Kube frequency response above. The dips and peaks seems to be an inverse of the response that I measure with my Mk 1s without the Kube in circuit, which seems like a strange coincidence. Maybe this is the reason for the HF section in the later Kube ? This peak at 4k, could be flattened out, I guess.
I have been experimenting with my crossovers and found a really easy way to open up the top end. A 10uF radial wound capacitor to the T33 and 1.5 ohm 5W resistor replaced the upper MF/HF section completely. The MF crossover is in the lower unit anyway, so you only have to remove the MF capacitor and inductor from the lower unit LF section and install it in the top unit ( or external would be fine ) This looks like an easy solution to bi-amp and use the Kube just for the LF units, thus keeping those cheap Kube opamps away from the mid and high frequencies.
However, I'm now looking at replacing the Kube with a professional active EQ like Behringer, so that I can get the exact response to compensate for problematic room modes.
I realise the MF/HF unit has some conjugate loading components that I have eliminated, but the response was more even using a first order HF roll-off set at the 4k point ( with 10uF foil capacitor ).
It also reduced the peak at 4Khz. My friend also has MK1 107s. We were able to swap the original pods in and out with the pods with the altered crossovers. In a blind test we were both able to identify and preferred the simpler HF arrangement. I find it harder to tell the difference between foil capacitors and electrolytic capacitors, than I do differences in crossover design, so I don't put much of this down to the replacement with better and newer capacitor type, though I accept this could only help. I used Ampohm capacitors from CPC.
I'm pleased with the results. Next it has to be professional EQ treatment to the bass units. The 50Hz peak I have just isn't right.
One thing this sort of exercise teaches you is there is far more important factors in hi-fi to get right than issues of interconnect choice. I recommend everyone get themselves an SPL meter and some Room EQ software. Best to design your crossover for your room, I think.
NAT, thank you for posting the Kube frequency response above. The dips and peaks seems to be an inverse of the response that I measure with my Mk 1s without the Kube in circuit, which seems like a strange coincidence. Maybe this is the reason for the HF section in the later Kube ? This peak at 4k, could be flattened out, I guess.
I have been experimenting with my crossovers and found a really easy way to open up the top end. A 10uF radial wound capacitor to the T33 and 1.5 ohm 5W resistor replaced the upper MF/HF section completely. The MF crossover is in the lower unit anyway, so you only have to remove the MF capacitor and inductor from the lower unit LF section and install it in the top unit ( or external would be fine ) This looks like an easy solution to bi-amp and use the Kube just for the LF units, thus keeping those cheap Kube opamps away from the mid and high frequencies.
However, I'm now looking at replacing the Kube with a professional active EQ like Behringer, so that I can get the exact response to compensate for problematic room modes.
I realise the MF/HF unit has some conjugate loading components that I have eliminated, but the response was more even using a first order HF roll-off set at the 4k point ( with 10uF foil capacitor ).
It also reduced the peak at 4Khz. My friend also has MK1 107s. We were able to swap the original pods in and out with the pods with the altered crossovers. In a blind test we were both able to identify and preferred the simpler HF arrangement. I find it harder to tell the difference between foil capacitors and electrolytic capacitors, than I do differences in crossover design, so I don't put much of this down to the replacement with better and newer capacitor type, though I accept this could only help. I used Ampohm capacitors from CPC.
I'm pleased with the results. Next it has to be professional EQ treatment to the bass units. The 50Hz peak I have just isn't right.
One thing this sort of exercise teaches you is there is far more important factors in hi-fi to get right than issues of interconnect choice. I recommend everyone get themselves an SPL meter and some Room EQ software. Best to design your crossover for your room, I think.
Aha! As I suspected... it seems the original Kube accounts for the mid-band dip:
That image was from the 1987 Stereophile revisit to the 107 by martin colloms:
http://www.stereophile.com/floorloudspeakers/550/index4.html
The 107/2 Kube measures thusly (from the 1991 review):
as per the Meridian plot.
So... it seems the Kubes are slightly different regarding the mid-range and also equally the two versions of the 107. Do you know if that is correct AndrewJ?
Surely this mid-band peak can be altered at the crossover if we're not going to use the Kube.
That image was from the 1987 Stereophile revisit to the 107 by martin colloms:
http://www.stereophile.com/floorloudspeakers/550/index4.html
The 107/2 Kube measures thusly (from the 1991 review):
as per the Meridian plot.
So... it seems the Kubes are slightly different regarding the mid-range and also equally the two versions of the 107. Do you know if that is correct AndrewJ?
Surely this mid-band peak can be altered at the crossover if we're not going to use the Kube.
It would be a very interesting revelation if the Kube response might solve the mid-range peak in the MK 1S. I'm sure most of us thought that the Kube only altered the lowest frequencies. It might explain why many of us thought the sound more natural with the Kube in play, despite the resulting increase in noise and all the other disdavantages .. op-amps .. more components replicating signal etc ..
The bandpass peak at 110Hz is not an issue in my room, it measures a dip at this frequency that measures flat when I disconnect this 110Hz notch. ( this might be down to the room ). So I think i should be able to remove one of the parallel capacitors from the 110Hz crossover bandpass which when recalculated results in a 50Hz notch instead. So I have 120uF and 1mH. The 50Hz peak that I have can be tuned further with resistor setting, but I will try without first.
It appears to be a complete stroke of luck, but I would be interested to know if anyway else has a similar peak at 50Hz ( with and without Kube setting ).
I get this peak with and without the Kube in circuit, so I think this is purely a room mode problem. This finding may be of no significance to anyone else.
At the same time when I looked into this, I found the reason for the 100Hz dip. It only occurs in one loudspeaker. The other has a dip at 110Hz ( please remember this is the loudspeaker and room response I'm measuring )
I suspected the 110Hz bandpass notch was shifted in this LF unit and right enough, the capacitor value has drifted, possibly through age, possibly it's always been like it. For me, the 110Hz bandpass in the crossover can be eliminated altogether. The 50hz room peak is far greater than the peak at 110Hz peak in my room.
So this tells me that any attempt to modify the crossover has to take into account the room and may require a different EQ to those employed in the crossover.
Who'd be a crossover designer, my experience is that room modes can be more problematic than peaks / dips attributed to cabinet design.
This is why I am considering going with an active EQ for the bass units. For example, who is to say that their won't be a natural 110Hz dip in your rooom, which might cancel out a peak attributed to the loudspeaker design.
Damn room, I could do with a room based on the golden section principle. How many of us actually have them ? Not a lot I imagine.
The bandpass peak at 110Hz is not an issue in my room, it measures a dip at this frequency that measures flat when I disconnect this 110Hz notch. ( this might be down to the room ). So I think i should be able to remove one of the parallel capacitors from the 110Hz crossover bandpass which when recalculated results in a 50Hz notch instead. So I have 120uF and 1mH. The 50Hz peak that I have can be tuned further with resistor setting, but I will try without first.
It appears to be a complete stroke of luck, but I would be interested to know if anyway else has a similar peak at 50Hz ( with and without Kube setting ).
I get this peak with and without the Kube in circuit, so I think this is purely a room mode problem. This finding may be of no significance to anyone else.
At the same time when I looked into this, I found the reason for the 100Hz dip. It only occurs in one loudspeaker. The other has a dip at 110Hz ( please remember this is the loudspeaker and room response I'm measuring )
I suspected the 110Hz bandpass notch was shifted in this LF unit and right enough, the capacitor value has drifted, possibly through age, possibly it's always been like it. For me, the 110Hz bandpass in the crossover can be eliminated altogether. The 50hz room peak is far greater than the peak at 110Hz peak in my room.
So this tells me that any attempt to modify the crossover has to take into account the room and may require a different EQ to those employed in the crossover.
Who'd be a crossover designer, my experience is that room modes can be more problematic than peaks / dips attributed to cabinet design.
This is why I am considering going with an active EQ for the bass units. For example, who is to say that their won't be a natural 110Hz dip in your rooom, which might cancel out a peak attributed to the loudspeaker design.
Damn room, I could do with a room based on the golden section principle. How many of us actually have them ? Not a lot I imagine.
Yes, the KUBE's did not only address the bass tuning, but in certain models were also used to provide midrange and high frequency EQ. This is why when the KUBE is not used, the response will not only lack low bass, but will also have a very skewed sound balance.
The reasons for this is that with some amount of active EQ combined with the passive EQ in the network, the overall efficiency of the system can be maximised. A passive network alone will always result in lowest overall efficiency.
The down side is that if the KUBE is eliminated, the speaker will not have a flat sound balance.
Now we did not do this for all kubes. Some were add-ons, therefore they were designed for systems that had already been neutrally balanced, so the Kube jsy extended the bass response.
Of course, with the example of the R107, if the kube is bypassed, then because of the huge change in sysem response then of course you will hear a big difference in sound, though most of this will be due to the change in frequency response, NOT the elimination of the op-amps.....
The reasons for this is that with some amount of active EQ combined with the passive EQ in the network, the overall efficiency of the system can be maximised. A passive network alone will always result in lowest overall efficiency.
The down side is that if the KUBE is eliminated, the speaker will not have a flat sound balance.
Now we did not do this for all kubes. Some were add-ons, therefore they were designed for systems that had already been neutrally balanced, so the Kube jsy extended the bass response.
Of course, with the example of the R107, if the kube is bypassed, then because of the huge change in sysem response then of course you will hear a big difference in sound, though most of this will be due to the change in frequency response, NOT the elimination of the op-amps.....
I agree, I think this confirms that the Kube should be used with the 107s, or you would have to re-design a crossover to compensate for not using it. I'm not sure that the efficiency of the system is maximised by a combination of crossover and active EQ, just that Kef decided to use a combination of the two. Whether this is the best approach, I'm not so sure.
If you are going to employ active EQ, then why not only use active EQ, where there will be obvious benefits to amplifier loading with no passive components between amplifier and loudspeaker.
Actually I'm being a little unfair, why not employ active EQ for all loudspeakers.
The 107s are some of the best used loudspeakers pound for pound, but there is potential for further improvement. It's not just opamps I am concerned about, it's the principle of EQ between my pre and power amp that affects all audio frequencies. I make exceptions for bass frequencies, as they have a tendancy to be more problematic, due to the lowest frequencies having higher energy content, but using the Kube between my source and 805 Class A SE valve amp does have it's disadvantages..
If you are going to employ active EQ, then why not only use active EQ, where there will be obvious benefits to amplifier loading with no passive components between amplifier and loudspeaker.
Actually I'm being a little unfair, why not employ active EQ for all loudspeakers.
The 107s are some of the best used loudspeakers pound for pound, but there is potential for further improvement. It's not just opamps I am concerned about, it's the principle of EQ between my pre and power amp that affects all audio frequencies. I make exceptions for bass frequencies, as they have a tendancy to be more problematic, due to the lowest frequencies having higher energy content, but using the Kube between my source and 805 Class A SE valve amp does have it's disadvantages..
It's not just opamps I am concerned about, it's the principle of EQ between source and amplifier affecting all audio frequencies. I make exceptions for bass frequencies, as they have a tendency to be more problematic, due to the lowest frequencies having higher energy content.
I think I already improved the skewed response in the 107/1s MF/HF unit when not using the Kube, by using a simple first order HF high-pass cut-off at 4k ( HF capacitor of 10uF ), eliminating the MF/HF crossover, and just utilising the MF 6.4 mH inductor and 200uF ( 2*100 ) capacitors
in the LF crossover unit for the B110 unit.
There may be advantages with the bi-amp capability of the 107/2s, but as mentioned, there may be less control over the response when not using the Kube with the MF unit.
I think I already improved the skewed response in the 107/1s MF/HF unit when not using the Kube, by using a simple first order HF high-pass cut-off at 4k ( HF capacitor of 10uF ), eliminating the MF/HF crossover, and just utilising the MF 6.4 mH inductor and 200uF ( 2*100 ) capacitors
in the LF crossover unit for the B110 unit.
There may be advantages with the bi-amp capability of the 107/2s, but as mentioned, there may be less control over the response when not using the Kube with the MF unit.
I am about to embark on a project using a DEQX PDC2.6 as a digital replacement for the crossovers in my 107's. My Plan is to tri-amp the speakers using Arcam power amps ( I just need to get the third amp 
) At the moment I am using the PDC just as a fancy graphic equaliser between pre and power amp. Even so it has made a large difference to my system. For those that are new to the DEQX. it not only acts as a graphic equaliser but also adjusts phase response for each driver. Each speaker / driver is calibrated, then room equalisation can be added to the system.
According to DEQX I should completely bipass all components from the crossovers, and "just weld the cables from the power amps on to the drivers"
Any thoughts from much more experienced people here?
How steep should I program crossovers The PDC default is 96 dB/octave ! And any ideas as to crossover frequencies, I can see from this thread that bass to mid is normally 150Hz I am intending to document my experiment on the excellent kef speaker forum, Would anyone be interested in seeing the documentation here?
) At the moment I am using the PDC just as a fancy graphic equaliser between pre and power amp. Even so it has made a large difference to my system. For those that are new to the DEQX. it not only acts as a graphic equaliser but also adjusts phase response for each driver. Each speaker / driver is calibrated, then room equalisation can be added to the system. According to DEQX I should completely bipass all components from the crossovers, and "just weld the cables from the power amps on to the drivers"
Any thoughts from much more experienced people here?
How steep should I program crossovers The PDC default is 96 dB/octave ! And any ideas as to crossover frequencies, I can see from this thread that bass to mid is normally 150Hz I am intending to document my experiment on the excellent kef speaker forum, Would anyone be interested in seeing the documentation here?
I don't know if it was mentionned before but someone on the Vintage Kef forum has posted the schematics for an active crossover that comes from "unofficial" KEF sources. It was designed for the 105.2 but perhaps may be adapted to other 3-way speaker systems such as the 107.
As far as the schematics for the Kube, here are the ones for Kube 200. I have that Kube running with my 105.2 and they make the bass go much deeper while clearing out the lower mids. And it has two contour adjustments.
http://www.hifiloudspeakers.info/Anatomy/KEFDocuments/Kube.pdf
An externally hosted image should be here but it was not working when we last tested it.
As far as the schematics for the Kube, here are the ones for Kube 200. I have that Kube running with my 105.2 and they make the bass go much deeper while clearing out the lower mids. And it has two contour adjustments.
http://www.hifiloudspeakers.info/Anatomy/KEFDocuments/Kube.pdf
Anybody still there after four years...? 
I have a pair of 107's sans Cube and they have a pronounced hiss at upper mid/high frequenies.
Is that to be expected withoput the Cube? I understand that the Cube basically is for extention of bass response but that it also can change the complete tonal balance?
BTW Did anybody succeed in tracing out that mid/high xover in the turret?
Jan
I have a pair of 107's sans Cube and they have a pronounced hiss at upper mid/high frequenies.
Is that to be expected withoput the Cube? I understand that the Cube basically is for extention of bass response but that it also can change the complete tonal balance?
BTW Did anybody succeed in tracing out that mid/high xover in the turret?
Jan
Many different Kubes were supplied with different speakers, some were sold as an upgrade, other Reference series included Kube from first delivery, those References did not sound good without Kube, shrill trebles too I remember. Incoherent sound too. I think I remember introduction 107 at Transtec, Kube was part of package, demo included placement of a gulden (coin) on its side on top of 107 bass cabinet and then turning up volume throught the Nakamichi PA-7's, it did not fall off, or moved
I don't believe that your tweeter hiss is going to change with the Cube. That's likely noise from your electronics.
I saw experts that used to work for KEF posting here: http://www.hifiloudspeakers.info/speakertalk
Schematics and such are available....
I saw experts that used to work for KEF posting here: http://www.hifiloudspeakers.info/speakertalk
Schematics and such are available....
Hi Guys...
I need bit of a info....
Without listening ...and while overseas i jumped at opportunity to grab a pair of mint KEF 107 (without listening lol) . Was not much of an issue because I was after mint pair of these for a while because I heard them with my current amp Mark Levinson 27.5 while back and was blown away with results (hence search for them).
Now they are sitting in my room and I am 20.000km away for another 6 months without chance to listen to them and that's killing me
This is a bit of an upgraded 107 as it is recapped/refoamed/rewired(Kimber) and instead of T33 it now has a custom Raal 70-10 ribbon (made to spec to be a "drop-in" replacement)
X/O is said to be kept original , just recapped.
Unfortunately , NO KUBE.
Anyhoo , what I would like to know is has anyone measured frequency response on bottom end of R107 when used without KUBE ? Out of boredom I felt the urge to know what I am to expect when I get back home.
Owner of Raal and upgrader of this pair said to previous owner they were linear to 30hz at low end after upgrades so would like to confirm that and your experiences with low extension without KUBE.
Thanks
I need bit of a info....
Without listening ...and while overseas
i jumped at opportunity to grab a pair of mint KEF 107 (without listening lol) . Was not much of an issue because I was after mint pair of these for a while because I heard them with my current amp Mark Levinson 27.5 while back and was blown away with results (hence search for them).Now they are sitting in my room and I am 20.000km away for another 6 months without chance to listen to them and that's killing me
This is a bit of an upgraded 107 as it is recapped/refoamed/rewired(Kimber) and instead of T33 it now has a custom Raal 70-10 ribbon (made to spec to be a "drop-in" replacement)
X/O is said to be kept original , just recapped.
Unfortunately , NO KUBE.
Anyhoo , what I would like to know is has anyone measured frequency response on bottom end of R107 when used without KUBE ? Out of boredom I felt the urge to know what I am to expect when I get back home.
Owner of Raal and upgrader of this pair said to previous owner they were linear to 30hz at low end after upgrades so would like to confirm that and your experiences with low extension without KUBE.
Thanks
Hi Guys...
I need bit of a info....
Without listening ...and while overseas
Now they are sitting in my room and I am 20.000km away for another 6 months without chance to listen to them and that's killing me
This is a bit of an upgraded 107 as it is recapped/refoamed/rewired(Kimber) and instead of T33 it now has a custom Raal 70-10 ribbon (made to spec to be a "drop-in" replacement)
X/O is said to be kept original , just recapped.
Unfortunately , NO KUBE.
Anyhoo , what I would like to know is has anyone measured frequency response on bottom end of R107 when used without KUBE ? Out of boredom I felt the urge to know what I am to expect when I get back home.
Owner of Raal and upgrader of this pair said to previous owner they were linear to 30hz at low end after upgrades so would like to confirm that and your experiences with low extension without KUBE.
Thanks
Umm.... would you like to post photos of these when you're back (maybe there are some online from an advert? ) . I'd love to see how well the ribbons fit in.
I have a pair sitting in storage but rescued from being thrown away for blown bass drivers and kube. So I can't say I've listened to them other than tops ages ago... Love their look and still on my list for activisation and tweeter upgrades (a scanspeak 95000 is supposed to be a good drop-in, though the front tweeter baffle hole needs widening by 2mm).
But from memory - maybe from this thread, or at least another on here - the Kube was designed as part of the crossover in effect but all about fixing the electrical impedance and overcoming the physical impedance (? - for want of a better term) of that coupled bass cabinet design. With it, an amp sees a flat 4ohm load across the board, without it, I think the impedance is really hard to drive and will therefore effect frequency response.
Now.. how that will effect say an NCore amp with Hypex's claim of impedance not being a problem at all, is possibly a good question.
From an earlier post in this thread, it looks like I should correct myself:
AndrewJ : "So, it's main function is not to make an easier load to the amplifier, but to make an easier load for the [crossover] network."
And maybe because of that, they would respond extremely well to being made active with a digital line-level crossover?
AndrewJ : "So, it's main function is not to make an easier load to the amplifier, but to make an easier load for the [crossover] network."
And maybe because of that, they would respond extremely well to being made active with a digital line-level crossover?
Last edited:
For some time, until I was able to find the 107 KUBE, I was sort of compensating by using a parametric equalizer for them.
The one I used is a Technics SH-9010 and it only needed the first two bands "energized". A large boost in the 20Hz range and a second one in the 40Hz range. Both with a small Q (less than 0.5).
Nothing scientific, just by ear and a bit of measurements, placing the mic above the port, close to the B110.
The one I used is a Technics SH-9010 and it only needed the first two bands "energized". A large boost in the 20Hz range and a second one in the 40Hz range. Both with a small Q (less than 0.5).
Nothing scientific, just by ear and a bit of measurements, placing the mic above the port, close to the B110.