4/21/14

I read about the Wharfedale SFB long time ago but thought it was an outdated design. Then I read Troels Gravesens new interpretation of the design and got intrigued. I have heard rely large open baffle speakers sounding good (Audiotronic Megatrend with each speaker having 18 KEFB200 8" drivers and a baffle area close to 2 m2) I have a scruffy packing crate plywood sheet 1200x1600x12mm. I have eight Seas alnico 10" woofers with a Qt of 0.7 and Fr of 34 Hz. And several other Seas vintage drivers. To that I bought 12 meters of wood with the cross section 120x21mm to the cost of about 15 dollar.

I cut the board in half and cut holes for the drivers, then I added a 120 mm deep frame around the perimeter of the boards as well as some extra braces with leftovers.

I screwed the driver in place with 104 screws!
Added a crossover consisting of electrolytic capacitors from West-Germany.
So 4 Seas 25TV-EW per side then 1 Seas 21TV-EW capacitor coupled and finally  3 of the Seas 9TV-LG that is mounted on a metal bracket angling the outer tweeter about 15 degrees to the side.




Mounting the drivers on the baffle lowers the resonance from 34 to 28 Hz. Two impedance peaks at 55 and 70 Hz turns up as well



A Omnimic measurement just in front to one of the dust caps show the dips in response at 55 and 70 Hz otherwise good response from 30 to 300 Hz.




Listening test
Room 6x4 m with the speakers on the standing about 1 m in front of the shorter wall.
Equipment: Philips CD473 NADC320 BEE, the cheapest cables and loudspeaker cables with visible oxidation.
Reference equipment: Maranz CD5001, Cambridge C500 SAnsui BA-F1 Stridbeck TLS (think IMF80) Decent cables.

Music  NADs first test CD, BIS CD-232 Kroumata Percussion Ensemble, Laleh and Rammstein.
I have heard the KODO drummers from the NAD cd and the Kroumata group several times live as well as heard it using my large TLS speakers. I had little expectations for these flimsy baffles.

To my surprise it sound really good, the baffles connect to the room in a way that I last had at home with DIY Klipsch corner horns. The piano on the NAD CD is in the room more so than how it sound with the large TLS. The percussion on Kroumata and Kodo is better with the baffle closer to what I remember of live sound. More than just bass there is a bass that gives space and volume to the music.

Lalehs singing and acoustic music is portrayed well. Finally Rammstein at close to concert level to get that physical experience. Loud and clean and only about 2mm cone motion. Probably not much bass below 50 Hz or so. The baffle vibrated a lot as did all the braces.

Conclusions:
I have to build proper baffles, a modern interpretation of the SFB. I will not have a sand filled baffle as the original Wharfedale nor have 2x25mm plywood as Troedsen. But some lamination of MDF and bitumen as suggested by Hobby HiFI.  Drivers from Norway, eight of them per set adding the kick, the name of the speakers has to be Sleipnir.




1/26/14

I had one final nagging question, does a final 90 degree bend reduce the 5th harmonic at 200 Hz? So at long last I cut out a hole of 190 cm2 in the back board and sealed the bottom of the box.
Note the opening in the narrow back below the driver.  Below is the impedance curves with the two different openings.
 Frequency response at the pipe end does not change. The measurements are months apart and the old one (blue) with the speaker at a wall and the new with the speaker in the middle of the room (green). That 200 Hz peak sure is persistent!


Then I have a new pair of Voigt pipes, designed by late Bosse Hansson of Opus 3 Records. They are designed for Lowther, with 19mm chipboard as the building material and 2x 19mm in front an back they are very different from the flimsy Deccas.
Tuning of the pipe is similar.


In both cases I used a Peerless 8" PP driver that works well in the pipes

Damping inside the pipes are similar with felt covering the walls of the first half of the pipes.

Listening to Adele 21 and some other records the pipes sound quite similar. Playing loud bass and the decca vibrates a lot not only the large front baffle but also the narrow top plate. The other pipe does not vibrate at all or very little.

So that 200 Hz peak is not tamed with a new location for the pipe opening. Nor does the flimsy constructions cause the sound to radically different compared to a massive pipe of similar dimensions. The proper reference would be to build a second DCH but in much thicker material as baffle loading and other factors comes into play here.

I tried adding braces of OSB or chipboard, screwed and glued see pictures. I measured the pipe output and it was no different than the unbraced.

The original HFN&RR article claimed that the removal of the brace for the front panel was a improvement. That with a 9mm plywood baffle. Here I ended up with 11mm OSB-wood glue-11mmOSB-silicone glue-12mm asfaboard.

So to conclude
1. In its original state it works far better than I expected
2. Covering the first half of the pipe in felt helps a lot in the time domain by reducing internal reflections around the cone. It probably also give some panel damping as well.
3. The the bracing shows that the original articles claim that the damping is provided by the thin-walled panel vibrations is not true. Building the cabinets in 12mm MDF or even thicker and more wellbraced cabinet should work well. The well behaved original design is more due to geometry and many non paralel  walls than the walls functioning as absorbers.
4. Moving the pipe opening to the back of the cabinet does not change neither bass output nor harmonics. Having the opening facing a stone slab or a thick carpet makes no difference.
5. It sounds good, it looks cool and the backward facing position works well with many fullrange drivers that sound forward and shrill when facing forward.

11/21/13

The time domain plots clearly showed that getting some damping around the driver is a good thing. Also some vibrations in the OSB could benefit from damping. So I took 12mm bitumen impregnated softboard and glued 10mm felt to it. The sizes are 3 each of 400x120mm and 400x100mm. The final lenght of the boards were trimmed to size by 20-30 mm if needed.
One 100 wide board was glued to the cieling of the cabinet (I felt vibrations there before) The front wall was covered by one 100 and one 120mm piece ( seen far back in the picture below. One 100 wide pice was glued along the back of the closed end below the brace. Then two 120 mm wide pices was glued to the dividing baffle on the side facing the closed end. Finaly the top of the dividing baffle with the cutout was covered with a layer of felt. All gluing was done with silicone glue.

The nearfield response does not suggest any major changes perhaps a slight flattening of the dips at the harmonics.
Now for the pipe
At the pipe and large changes occur even with this very modest amount of damping material.
The sharp peak at 120 Hz is reduced by about 6 dB
The huge peak at 200 is marginaly affected
The peak at 350 Hz is reduced by 9 dB
The peak at 500 Hz is reduced by 4 dB.

I then put a damping layer in the front section by gluing a 60x40cm soft board below the front braces.  This time with ordinary wood glue as I run out of silicone glue.
If I stagger the pipe output with the orginal unlined in green, the lowest trace. Next up the red is with the lining around the driver. With the frontplate is at the top note that the brown trace, note that the trace is smoother than the other two.

If I alight the bass part of the traces on the big improvement was the felt and board lining. The added front board is icing on the cake and the knuckle rap test is much improved. The dip around 150 Hz is a bit flattend out by the front brace

So to conclude I recomend the design. Building it from scratch I would line all parts in the closed en by felt before putting it all together. The front damping is also a good thing to have.

I did some experiments with wool in the closed end but it is really tricky to get anything down there with the braces and the added panels and felt.

That 200 Hz peak is still there. The current opening is about 160cm2 I plan to block it and open a hole at the very back below the closed end. Hopefully this added 90 degree bend will reduce that 200 Hz peak... or not.

After that, it is time for an open baffle I think.





10/14/13

The odd men out of drivers

 This is a OEM driver from the Audio Pro 4-14 dipole speaker, A really good match for pipes and horns.
* Manufacturer:  From Audio Pro 4-40 loudspeaker
* Model:  Paper cone black=greenish color, paper surround with sticky green coating and chassie that looks like 1970s Peerless.
* Piston Diameter = 170.0 mm
* f(s)= 47.10 Hz
* R(e)= 6.65 Ohms
* Z(max)= 34.26 Ohms
* Q(ms)= 1.575
* Q(es)= 0.379
* Q(ts)= 0.306
* V(as)= 59.860 liters     (2.114 cubic feet)
* L(e)= 0.93 mH
* n(0)= 1.57 %
* SPL= 94.07 1W/1m
* M(ms)= 13.81 grams
* C(ms)= 0.83 mm/N
* BL= 8.47
Note the cleas peaks at the harmonics
Good matching with the pipe and good sound as well, the high sensitivity is noted
These are Pioneet TSE20, car speakers that I got as a gift. Fr 48 Hz and Q of 0.6. With the coxial arrangement I could not use the roll of masking tape I usually use for Mms and Vas determinations, but I expect the driver to have a high mass cone. Note the absence of peaks in impedance at the harmonics.


A good match with the pipe and a good sound as well. The tweeters does not seem to work so if I can remove them I can measure TS as well.



Last driver is a PA midrange/midbass driver the Fane Studio 8M. The magnet forced me to have the driver backward
* Manufacturer:  Fane
* Model:  Studi 8M
* Piston Diameter = 167.0 mm
* f(s)= 104.30 Hz
* R(e)= 5.77 Ohms
* Z(max)= 79.19 Ohms
* Q(ms)= 7.394
* Q(es)= 0.581
* Q(ts)= 0.539
* V(as)= 12.970 liters     (0.458 cubic feet)
* L(e)= 0.75 mH
* n(0)= 2.42 %
* SPL= 95.93 1W/1m
* M(ms)= 12.09 grams
* C(ms)= 0.19 mm/N
* BL= 8.87

The lower peak of the tuning is hard to find the 110 Hz is way higher.
Soundwise as evaluated by listening to the driver facing the wrong way. Very high sensitivity, a really brutal attach to transients, bass slam. Deeper bass is missing there is no weight to the bass

While I prefer this sound to the bass mumble of the B200 and the Seas TV-EW there are better drivers for quarter wave pipes.

I will have to do some damping to the walls around the driver so the time to fiddle with damping material is coming up.







10/13/13

The two worst drivers for the DCH I have found so far.

Two closed box drivers from well respected manufacturers. Both of them  good drivers in such closed boxes.
* Manufacturer:  KEF
* Model:  B200 SP1014
* Piston Diameter = 172.0 mm
* f(s)= 37.68 Hz
* R(e)= 7.52 Ohms
* Z(max)= 59.42 Ohms
* Q(ms)= 5.965
* Q(es)= 0.864
* Q(ts)= 0.755
* V(as)= 56.650 liters     (2.001 cubic feet)
* L(e)= 1.20 mH
* n(0)= 0.33 %
* SPL= 87.34 1W/1m
* M(ms)= 23.89 grams
* C(ms)= 0.75 mm/N
* BL= 7.02

* Manufacturer:  Seas
* Model:  21TV-EW
* Piston Diameter = 170.0 mm
* f(s)= 34.99 Hz
* R(e)= 5.99 Ohms
* Z(max)= 70.15 Ohms
* Q(ms)= 7.519
* Q(es)= 0.702
* Q(ts)= 0.642
* V(as)= 52.260 liters     (1.845 cubic feet)
* L(e)= 1.88 mH
* n(0)= 0.30 %
* SPL= 86.93 1W/1m
* M(ms)= 28.65 grams
* C(ms)= 0.72 mm/N
* BL= 7.33


The B200 has some inconsistent mode of measurements compared to the other drivers. The black reference trace is the 9710 (lowered by 10dB) and red for the B200. Note the 5 dB peak driver output at 65 Hz and pipe peaking at 60 Hz. This is how it sound boomy and booring.
 Cast chassi Alnico magnets and a rubber suround, and a lower Q than the KEF it looks like it would be an improvements, does it not

Frequeny wise the it is a similar story to the B200 but with a less pronouced peak of the driver at 65 Hz. Soundwise the difference in bigger. The Seas sounds much worse than the B200, the bass lines that stands out at all times brings a smile for a couple of minutes but after a couple of tunes it really start to become irritating. If I wanted to learn to play bass and wanted the bass lines to stand out at all times it would be OK but otherwise...

I have also tested drivers with similar high Q but with higher Fr and they fare better. They are not good but the combination of low Fr high cone mass and high Q really bring out the worst in the construction.
* Manufacturer:  Seas   
* Model:  21F-GW
* Piston Diameter = 175.0 mm
* f(s)= 57.87 Hz
* R(e)= 4.76 Ohms
* Z(max)= 15.09 Ohms
* Q(ms)= 2.971
* Q(es)= 1.369
* Q(ts)= 0.937
* V(as)= 42.030 liters     (1.484 cubic feet)
* L(e)= 0.70 mH
* n(0)= 0.57 %
* SPL= 89.64 1W/1m
* M(ms)= 14.62 grams
* C(ms)= 0.52 mm/N
* BL= 4.30


As I said not good but not to bad either.

* Manufacturer:  Philips
* Model:  AD8001W4
* Piston Diameter = 168.0 mm
* f(s)= 59.22 Hz
* R(e)= 4.16 Ohms
* Z(max)= 17.17 Ohms
* Q(ms)= 4.058
* Q(es)= 1.295
* Q(ts)= 0.982
* V(as)= 39.440 liters     (1.393 cubic feet)
* L(e)= 0.76 mH
* n(0)= 0.60 %
* SPL= 89.90 1W/1m
* M(ms)= 12.64 grams
* C(ms)= 0.57 mm/N
* BL= 3.88 
 Familiar octagonal chassi front, very common in high volume low price speakers in the 1970s
 Not a pretty sight with that lift 70-100 Hz. And while it's sound is not to good it is substantially better than the first Seas.
I do not know why the Seas TV-EW sounds so particulary bad. It has the highest cone mass and it could be that the cone that is twise the mass of the Philps exacerbate som vibration mode in the the lightweight cabinet, but this is speculation.


10/8/13

Some really good drivers from Peerless for the DCH
They are not the current stock but they are quite similar to other current Peerless drivers. Data is measured using Dayton Woofer Tester 3.
* Manufacturer:  Peerless
* Model:  850126
* Piston Diameter = 175.0 mm
* f(s)= 48.45 Hz
* R(e)= 6.00 Ohms
* Z(max)= 58.76 Ohms
* Q(ms)= 4.488
* Q(es)= 0.511
* Q(ts)= 0.459
* V(as)= 43.160 liters     (1.524 cubic feet)
* L(e)= 1.91 mH
* n(0)= 0.92 %
* SPL= 91.72 1W/1m
* M(ms)= 20.31 grams
* C(ms)= 0.53 mm/N
* BL= 8.52
*PP cone inverted dome looks like the CC line expected +/-5mm, rubber surround.
Impedance is pretty straight forward, tuning to 50 Hz.

The downward slope in the nearfield of the driver below 100 Hz indicated the low Q of the driver resonance and the good radiation resistance offered by the pipe. The dip in driver near field response (black) is centered at 50 Hz the irregularity is due to a room resonances.
This drver sound really good in the DCH, a very natrual easy balance in the music with both weight and slam in the bass when called upon.

The next one is also good but for some reason not quite as good to my ears at least.

* Manufacturer:Peerless
* Model:  831862
* Piston Diameter = 170.0 mm
* f(s)= 35.66 Hz
* R(e)= 3.19 Ohms
* Z(max)= 22.41 Ohms
* Q(ms)= 2.101
* Q(es)= 0.349
* Q(ts)= 0.299
* V(as)= 60.770 liters     (2.146 cubic feet)
* L(e)= 0.70 mH
* n(0)= 0.75 %
* SPL= 90.87 1W/1m
* M(ms)= 23.72 grams
* C(ms)= 0.84 mm/N
* BL= 6.97
Noninverted fabric dome and foam surround on this one.

The frequency response is exellent very flat up to 2 kHz, it also sound good, but not as good as the other one, why that is and if it will change with some decent amount of damping material in the pipe I do not know, presently.

10/4/13

The near field dip of the drivers in the DCH has had a strange shape, I had suspected that standing waves had something to do with it. Today I took measurements in an other room with the speaker standing on a low table in the middle of the room. Much better! A sharp symetric dip, textbook style

The tip of the dip is further 2 dB down off the scale in this graph. The difference in sensitivity between the 9710 and a Stridbeck/Tyrland is supposed to be 10 dB but it really sounds like 20 dB.  I have to get to work in that 5th harmonic at 200 Hz. But that will have to wait for a while.