A KEF R107 study

I've mentioned in previous threads about R107s that I have parts of a system and it was my intention to build something along the lines. I have the mid/high heads as well as the original LF/MF crossovers. What I don't have are the LF drivers and the KUBE.

This thread documents my venture, which started during a slow week in early Autumn that I was home alone.

I've never been able to obtain the actual drivers, so I started evaluating if other drivers could fit the bill in place of the original 10" paper cone versions.

At the same time I collected visual information on the bass cabinets from threads here (can't find the exact one to link though), articles, brochures, etc.

So, step 1 here, is a plot of some simulations using 4 different KEF B200 drivers and a band pass enclosure.

The picture shows the Transfer Function plots for a 4th order band pass enclosure using WinISD for the following drivers:
Cyan: B200 SP1075
Red: B200 SP1014
Blue: B200 SP1039
Green: B200 SP1022

The simulation parameters are:
Rear Enclosure Size: 20lt
Front chamber size: 7lt
Vent: 12x14cm cross section, 41cm long

You may wonder how did I reach these dimensions. Mainly from the R107 pictures, cross section diagrams and visuals I was able to find. The vent itself is the first to identify as a physical feature and the bottom driver rear chamber was photographed here in an older thread. I've never seen any hints of the top driver rear chamber, so I had to improvise.

Points important to note are the front chamber and vent dimensions. Because the actual unit will have 2 drivers in separate chambers, and because the middle chamber gets to a size around 14-16lt, the simulation was performed using 7lt for the front chamber.

Again, because the actual vent will be 12x29x41 - for two drivers - the simulation was done using half the cross section (12x14).

I confirmed with WinISD that the plot with the above considerations is identical with the TF of two drivers in twice as much enclosures and twice as wide a vent.

Going back to the simulation plots, it looks like the difference between different drivers (and these are pretty different) is minimal. I doubt the difference depicted between, say, the red and blue plots is significant. The T/S parameters are approximated anyway and (IMHO) the WinISD simulator should only be taken as a general hint towards the behavior of a driver in a cabinet.

I've also toyed with different variations of rear chamber size against different front chamber sizes and found that decreasing the rear chamber size is detrimental and decreasing the front chamber size reduces playback level but also reduces the tuning peak at around 150Hz.

So, I worked with these as starting points and worked using a sliding baffle inside the test enclosure to confirm the effect of chamber size against actual measurements.

Wow. I find this whole pursuit super impressive, and will be living vicariously as I follow your progress. Best of luck.

About the drivers

Regarding the use of actual drivers, I also have a few pairs of SP1070 and SP1069 (16ohm versions) of paper cone drivers. My initial thought was that the paper cone drivers should be closer to optimum as the original drivers have lightweight paper cones.

After actual measurements with different drivers, I must admit that the SP1070 (don't know their T/S parameters) behaved worse than the others. And as the 8Ohm (nominal) drivers appeared to behave fine, I decided there is no point in trying 16Ohm drivers there.

Between SP1014, SP1070 and SP1022, the best results measurements gave me (more about actual measurements later on) were with the SP1014. No resonances and a smoother frequency response.

Another departure from the actual R107 design is the placement of the drivers themselves.

As the rear chamber needs to be as large as possible and the front chamber size should be minimized for smoothness of the FR (in favor of efficiency) I mounted the drivers with magnets facing each other. So, the top driver sits face up and the bottom driver is mounted face down.

About the box

I do have a couple of A4 sketches done to scale - will scan them later and post them - but let's go directly to the finished enclosure - pic attached.

The picture is of the completed enclosure (modified test enclosure that was using sliding baffles for optimization).

The piece shown is the complete construction save for the right hand side cover that fits on the rest.

The external dimensions were found to be 79cm tall, 45cm deep and 33cm wide. All outer panels cut from 20mm plywood I had available.

Going from bottom to the top:
You can see the 3 pieces of 4cm wide "flanges" used to affix the bottom cover. The fourth piece is attached to the cover - thought it would be better that way; don't ask why...

Moving up, we have the 4 pieces of 4cm wide flanges again where the bottom driver baffle will be mounted. The baffle will be mounted below the flanges.

Going up again we have the vertical panel that forms the vent (giving a 12cm x 29cm x 41cm vent). That also generates the space where the top driver rear enclosure will be housed.

Since the bottom driver has all the internal space for a rear enclosure, it gets a shallow enclosure. Since the top driver has a smaller cross section for a rear enclosure, the overall size is much deeper. Also the crossover needs to be housed up there.

Going up a bit more, you can see the top flanges for the top enclosure cover. You will notice there are two "steps" there as I need to mount the sealing top cover for the top enclosure (screwed down) and also space for yet another cover that will cover this and be used for mounting the head.

The second and third pictures show close up details of the construction.

Baffle mounting

The next two pictures show the steps to mount the baffle. The wooden (plywood) flanges are 4cm wide by 2cm thick all round the cross section, glued to the enclosure.

Between the baffle and the mounting flanges, I put one layer of 3mm bitumen sheet to achieve two goals. First, to aid in sealing the gaps and second, to add some vibration isolation between the driver/baffle and the enclosure.

During some dry runs I did, I realized that since the drivers are mounted enclosed, all their energy (front and rear) is "trapped" inside the enclosure and while air goes out from the vent, I had severe vibrations and resonances to tame.

Therefore, the baffles are isolated using 3mm of bitumen sheets and the entire inside of the enclosure is padded with 2 layers (6mm) of the same bitumen sheet to both absorb vibration (damping) and to make the plywood less transparent to the sound pressure waves forming inside. Prior experiments have shown that 20mm of plywood do almost nothing in trapping the low frequencies.

The finished enclosure

The enclosure was finished with (as mentioned) 6mm of bitumen sheet all around the inner surfaces as well as acoustic foam around the rear and front chambers.

I should have taken a picture before sealing the side panel forever - still have another enclosure to build if all goes well, so still a change for more and better pictures.

Attached are some pictures of the top rear chamber where you can see the bitumen sheets (just barely, behind the crossover), the foam, top LF driver and the sequence to seal and cover the top cover.

The first picture shows the inner of the top chamber. I don't feel comfortable will all the length of cables going around the inductors. I need to cut them to the correct size sometime soon.

The second picture shows the polyurethane foam ribbon (self adhesive) that forms the sealing for the top cover ( I used bitumen sheet (3mm) to mount the inner baffles but real sealing foam for the top and bottom covers).

The third and fourth pictures show the top cover in place and the length of the cord going through to the MF/HF head.

The finished system ready to test drive

The finished system is shown in these last pictures. Please pardon the mess of the workshop.

I have used heavy duty castors mounted on the bottom chamber cover to be able to move the thing around. It's pretty heavy - the bitumen sheets add quite a bit - and takes a deep breath to lift it and rest it on the bench - with the head removed of course.

If you look closely in the first pics of this post, you will notice a pair of stacked R104s. I have used these for the left channel and the "FrankenSeven" for the right channel to do some early listening tests and evaluations.

Early impressions

I was weary that 4th order bandpass enclosures are very hard to tune and I bet they are if you are building a KEF Reference system.

The early prototype without any bitumen, without padding and stuffing and with the baffle sliding contraption (I need to take a picture of that before it get's thrown out) and with panels screwed together sounded a bit strange. Lots of resonances were audible to the ear (didn't need to make measurements but measurements confirmed them) to the point I was a bit disappointed.

But, adding the proper baffles and securing them in please with proper sealing and damping as well as adding 6mm of bitumen all around internally changed the performance altogether.

This is when I decided I will stick with these dimensions for rear and front chambers and build it properly.

Once again, I will need to point out that an enclosure designed to be sealed needs to be sealed to sound like ... sealed. And when you crank up the volume, any less than secure joint rattles and resonates like crazy. So no shortcut to properly assembling a unit.

After properly finishing assembly (to the level shown by the pics in the previous posts) the unit sounds impressive. No disco style bass bang but a deep and full extension of the low frequencies.

The interesting point is that there is no measureable (or audible) peak at the 150Hz as shown by the TF plots. I am guessing it's being taken care of by the crossover but need to better understand it first.

After initial measurements, I used a parametric equalizer (Technics SH-9090) to add some active equalization to the low end. Doing that really changes the profile. From a decent low end (like the R105.1 and much better than the R104) it goes into a lung vibrating experience. Chello and Jazz Bass sound like they really do.

Tried this with both a pair of SP1070 and a pair of SP1014. I like the 1014 much better and the measurements show it.

Excursion isn't too much so the system can take a lot of bass boost and still perform decently. However it' not good to use it for LFE in a home theater. It doesn't take explosions and loud thunders well - the cones reach their travel ends.

So this is where the quest has brought me so far. One unit assembled and trying out different drivers - if only to prove or disprove the WinISD plots.

I am going to build another box soon so I can test two variations of drivers in A/B scenarios. I will also soon post some measurements (setup/method/results) to get some comments and feedback on measurement validity and possible improvement.

I am seriously blown away, Im pretty shure this has never been attempted for good reason, it's hard to pull off. You seem to be on track, good luck! Will you eventually veneer the outsides?

Thundermud said:

I am seriously blown away, Im pretty shure this has never been attempted for good reason, it's hard to pull off. You seem to be on track, good luck! Will you eventually veneer the outsides?

Click to expand...

The plan is to reach a stage where I can think about veneer and finishing touches, like the grille over the vent and foam+cloth to cover the top cover. And of course the "hoods".

For now I'm concentrating on the performance, to confirm I can put together a bass unit to go with the heads and make a good sounding system. And yes, it sounds promising.

Wow! Amazing work. It looks very much like an R107. Keep going!


Sent from my iPhone using Tapatalk

Wow, really nice work. I wish I had the patience, tools, and skills to do cabinet work...maybe some day, but not right now! So I am using my R107's for both music and home theater, and I have often wondered if there is any risk with using them for home theater. I probably didn't read your writing thoroughly enough, but do you see concern using them for home theater? I've had my Kube set to 18 hz (lowest setting). I've considered EQ'ing them with the Kube up to maybe 30 hz, but I do like to take advantage of the speaker's full capabilities. I am also using a single 12" subwoofer with my R107's. I usually don't watch movies at super high SPL's but...sometimes. Just don't want to damage my drivers.

Good work here though! Can't wait to hear the final product and impressions. I have thought about trying to build a JBL Paragon some day, but might just have my Dad's friend build the cabinet (he builds custom cabinets, fireplaces, etc) and I'll add the drivers and electronics.

Hawkeye83 said:

I probably didn't read your writing thoroughly enough, but do you see concern using them for home theater? I've had my Kube set to 18 hz (lowest setting). I've considered EQ'ing them with the Kube up to maybe 30 hz, but I do like to take advantage of the speaker's full capabilities. I am also using a single 12" subwoofer with my R107's. I usually don't watch movies at super high SPL's but...sometimes. Just don't want to damage my drivers.

Click to expand...

The R107s are "full range" speakers (their nominal specs are 20Hz ~ 20kHz) so theoretically they don't need a subwoofer.

Now, in reality, considering that for HT use the subwoofer channel is actually a LFE (Low Frequency Effect) content, I agree it's wise to not put a high quality system through it.

Been a while. Did a lot of things, kept notes and I'm back.

Quick version of testing various drivers is that there's little difference between them. Between the SP1070, SP1014 the better performing ones are (much to my surprise) the SP1014s.

Again (to my surprise) the WinISD simulations seem to confirm this.

I will be posting measurements - eventually - when I can confirm and repeat them. It's a bit hard making valid low frequency (20Hz ~ 150Hz) measurements in a small room.

So for now, I will post some pictures of the build of the second enclosure. It's actually a version 2 based on lessons learned from version 1.

But before I go into that, here are two pictures of the contraption I devised for making tests with varying the size of the sealed chambers. Here it is for the lower chamber: A baffle mounted on a drawer like frame sealed with polyurethane gaskets. The driver mounting opening was done quick and dirty using a jig saw. The final baffles were done much better using a router.

From pieces to outer skin

Here is step 1. making the outer skin out of 20mm birch (3 out of 4 sides) to house the rest of the inner baffles.

The second and third pics show the arrangement and clamps to ensure the side panels are perpendicular to the front side. Next comes the completed 3/4 skin followed by a detail of the dowels used to strengthen the joints. There are 5 of them along the height of the enclosure.

I will post a cross section drawing with dimensions later on.

Inner baffles

Next few pics show dry mounted inner baffles (made to ensure the pieces were precisely cut) as well as gradual assembly.

The first two pics show the bottom driver baffle, before the opening was cut. Pic 3 shows the various pieces once they were cut and trimmed to size.

Pic 4 shows the lower compartment and the frame where the bottom cover will be screwed on. Ideally I should have used a single piece baffle and cut an opening to it, but I cheaped on this. Had too many clippings from various projects and they had to be used somewhere.

Pic 5 shows both driver baffles glued in place and pic 6 shows both baffles and both cover frames.

Adding the vertical separator

What is still missing is the vertical separator that forms the upper chamber and the vent. It sounds easy to glue in place but in reality it's much more complex, if it's to be glued exactly where it has to go.

It took quite a bit of thinking on how to prepare spacers to ensure it goes at the right place. Pic 1 shows them. The first piece of wood marks the space used by the actual cover of the chamber and the second one, raised a bit to form a stop for the vertical separator, is flush with the enclosure and takes the place of the top cover, the one that's supposed to be padded and wrapped in fabric.

Pic 2 shows the clamps employed to secure the baffle in place. It had to be pressed against the driver baffle and the top frame as well as firmly pressed against the side panel.

The last picture shows the completed woodwork. All that's needed is to mount the other side panel and close the enclosure. But there's lot to be done before we can close the lid...

Deadening resonances

A few years ago I did an experimental enclosure (180lt) to measure the vibrations of an enclosure when a large woofer (12") is mounted on it. The results showed that even if the driver is totally enclosed in 20mm plywood, the sound comes out of the panels like there is very little there to stop it. I assume this is the reason why many consider sealed enclosures sound boxy.

Anyway, bracing is needed but I also confirmed that affixing deadening material to the inside of the panels does a lot to reduce emissions - almost to the point they cannot be measured.

In that experiment, 3 layers of 3mm bitumen pads glued to the entire inner surface added a substantial reduction in emissions. The 4th layer was an overkill (and also killed my back trying to lift that huge box onto the bench).

So, since this is a much smaller cabinet (70lt compared to 180lt) I decided two layers should be enough.

So here we go:

I cut the pieces from a large roll (10m x 1m) of 3mm bitumen pads. I tried to cut them as precisely as possible and to cover the entire inside of the baffles.

As you can see, I have also added two layers to the top side of the lower driver baffle. It's a rather large piece and the blank part of the baffle forms the end of the vent. Don't know how much it adds (or removes) but looks more uniform when viewed from the mouth of the vent.

First layer is glued on using contact glue, then secured in place using staples. Second layer follows using glue and staples again. It's important to glue the pads (rather than just loosely staple them). I perhaps overdid it with the staples (I did run out of them in the process).

I also thought I should mount some pads to the edge of the vertical separator that forms the edge of he "front compartment" for both drivers. I applied 3 layers of pads on the inside edge, 2 layers of pads to the inside of the upper chamber and one layer along the vent side. In total, the vertical separator now has 3 layers of pads, aiming to tame higher order resonances.

The third picture shows the completed task along with the remaining side panel with its own complement of pads. It was a bitch to cut the pads to size, glue them and ensure the panel would mate with the rest of the cabinet. I did draw the limits on the panel while glueing the pads but didn't realize the inner baffles had become a bit too thick compared to their original thickness. :sigh:

Still more to come...

There's a bit more work that has been done - but not yet documented.

I've cut the pieces of foam sheets that will go inside. Have not mounted them yet. Need to mount the crossover, route the cables through the top baffle, drill the hole for the head cable as well as the bigger hole for the head to rest on the top covers.

Will probably complete the second unit over the weekend and post pictures of the pair - along with a drawing to scale.

Promise to give you more details, results with various drivers tested, A/B tests with my main rig's R105s as well as chapter two, making them look like loudspeakers...

It's been a month and a bit since I last posted here. I've been pretty busy with the project.

Second enclosure finished, both units auditioned in the listening room (in place of the R105.1s and ... I was blown off.

Both enclosures veneered (cherry), trimmed and lacquered. Made grilles for the vent mouths and the loudspeakers are in the listening room since this afternoon. I hasted a few things as I really wanted to have them in place for the holidays.

Pictures will follow tomorrow.

First impressions: The overall sound is good and loud. It may be that the treble is a bit bright - I'm fearing that perhaps the ferrofluid has dried out, but that needs to be confirmed.

I've listened to orchestral music, Rock and Jazz. The revelation was listening to a couple of Jacques Lousier CDs that are well recorded and offer ample low frequency content, be it bass or drum solos.

Listened to them for a couple of hours. Then plugged in the R105 again. In a few words, the R105s are more "tame". In the high frequencies that may be a good thing (not that bright) but in the low frequencies, I can compare the two as I would compare the Calinda or R104 with the R105. The bass response is there but attenuated and your brain needs to fill it in a bit.

I must repeat that I have no KUBE and the woofers are not B250 but the plain old B200 SP1014. Regarding equalization, I did sense a bit of improvement when I raised the low end of the spectrum by 12dB up to 33Hz and 6dB at 40Hz. More than that and it started to get boomy.

The new pair has already taken place in the listening room and the R105s need to move elsewhere, probably to the living room.

One note: My first tests were done with a Technics SU-V6x integrated that gives out 100W at 8Ohms and also 100W at 4Ohms. It did get a bit warm when used for an hour or so but not hot to the touch, even though I had volume set between 2 and 3 (10-11 o'clock).

My tests today were performed using an SE-9200 power amp that offers just a bit over 90W at 4Ohms. It's heavier, older and surely better built but showed to be underpowered to drive them at loud levels. It's the first time I've heard the amp clipping during drum playback and not the speakers reach their excursion limits. Even the VU meter lights started to dim to the rhythm when I pushed too hard.

Clearly I need more power to properly drive them (to the limits) although they sound terrific (but don't shake the couch) with just 20-30W peaks. Keeping the amp output to 60-70W peaks the sound levels get realistic but perhaps too loud if you live in a condo. I was listening in the basement of a concrete building with plenty of closed doors in between and my other half came down from the first floor to complain.