(tapered quarter wave tube)
Copyright 2007-15 Troels Gravesen

TQWT mkII replaced by new TQWT mkIII
- but read this article anyway for background story. 

JA8008 DRIVER    MEASUREMENTS    CROSSOVER    TQWT CABS    KIT-INFO      8008 Applications       

JA8008 + TW034, 95 dB high-efficiency speaker

The TQWT mkII is a revised version of the TQWT construction having a new 2nd order high-pass filter to the tweeter.
For those having built the TQWT you may leave things as they are, or order 2 x three new components to fully upgrade.
Read about the changes
Read about the bigger brother, DTQWT speaker
Also read about the new (2013) TQWT SEAS T35C002 option here.

JA8008, the main driver
(back to top)

Designing your own driver is a major challenge. Having worked with hundreds of different drivers, you eventually come to a point where you would like to design your own driver, hopefully fulfilling a wide range of desired features. Some people may think that cost is a major constraint here, but it is not, unless you want to use a lot of chrome, copper and fancy metal coatings; things that do not necessarily enhance sonic performance. Metal for the motor system, magnets, chassis, voice coils, cones and surrounds basically cost very little. The problem is to get the parts you want - and in the quantities you want. If you order 10,000 units, no problem. Smaller quantities are trouble. Obviously Jantzen Audio did not order thousands of units here, which would have been financial suicide. The driver presented here is not designed to be a commodity product, rather a carefully designed high-end product aimed at people enjoying low-wattage, high-class amplification, but it can obviously be connected to any high-quality system. Be prepared it will reveal possible deficiencies in the system driving the speakers. Be also prepared that it will urge you to use the best crossover components available like those seen below.

Designing a driver is all about compromises! It's no problem increasing sensitivity 3, 6 or even 9 decibels. You cut down on membrane weight, increase magnet size, use low-mass suspensions, use short voice coils, etc. But there are prices to pay for each step you take. Thin cones = risk of cone break-up, increased colouration and lack of detail; short voice coil = low Xmax (reduced cone movement), large magnets = risk of frequency response steadily rising towards higher frequences, etc.

Ideally we want our magnet system to accelerate our membrane instantaneously - but we need the poor thing to stop again - we need damping. I've experienced modern drivers where all measures were taken to increase acceleration and the result was severe peaking towards higher frequences, making crossover construction very difficult, and most unfortunate, the driver had severe ringing due to non-optimised suspensions, very noticeable on e.g. piano notes. The mechanical damping of any driver is extremely important and we have only two things at hand here: The spider and the surround. These devices have to keep the voice coil correctly in the magnet gap and the membrane centered correctly in the chassis. They must have the right elasticity to allow the membrane to move freely within Xmax and they must have correct resilience to quickly restore rest position when no voltage is applied, the latter not least important.

The cone material, shape and mass are obviously important parameters. From the onset of the project there was no doubt in my mind that it had to be paper. Paper cones have been around since Rice and Kellogg and it's another wonder that wood pulp can me pressed into thin curvilinear diaphragms, providing proper strength and internal damping at the same time. Make them more rigid and they may have nasty break-ups - or make them softer and they'll compromise the ability to reveal details.
The paper pulp here is mixed with long cotton fibres to provide the right balance between cone strength and internal damping. Given the right fibre composition, we may not have to coat the membranes at all. Coating is usually modern chemical polymers like PVA glue, which may dampen cone break-up, add mass and reduce sensitivity. I write "may", because sometimes we risk generating severe cone break-ups from coatings. Coatings are sometimes applied to cover basic imperfections in cone composition and geometry. And sometimes coatings can be so thick that we have to ask whether we're listening to a paper cone or a cone made from glue. But don't get me wrong; any pulp cone will hold a certain amount of glue to keep the fibres together and coatings can do a lot of good and produce great results.
Considering the basic simplicity of dynamic drivers, it's a wonder we can make these things sound so great. I mean, good drivers that can sometimes cover more than 6 octaves of musical information. Remarkable.

From left to right: Phase plug, coated fabric dome, paper dome, felt pad on center pole piece.

Quite some work went into the dust cap. Should it be paper, fabric or a phase plug? Should the dust caps be coated or non-coated, etc.? All these possibilities were tried for measuring performance and evaluated sonically by setting up the drivers in the TQWT cabinet and listening to a wide range of music. Last but not least a thick felt ring was added to the pole piece and this required SEAS to have special tools made for punching out these felt pads.
The phase plug did not measure better or sound better than the domes and eventually a non-coated paper dome was chosen for having a more vivid upper mid. The coated fabric dome sounded a bit dull in comparison. All in all hard choices as the sonic differences were surprisingly small.

I wanted an 8" driver that could handle 40-3000 Hz without severe cone break-up, a sensitivity making 94-95 dB/2.8 volts in the 300-1500 Hz range and an 8 ohms voice coil, providing a proper load for even single ended triode amplifiers. I also wanted SEAS in Norway to make this driver due to general excellent build quality, being Premium or EXCEL. So, with help from Bjoern Idland at SEAS, the driver seen here was constructed. The driver reaches 5.5 kHz, one octave above intended point of crossover, and this without noticeable peaking, allowing simple crossover construction. I couldn't have asked for more. View frequency response in measurement section.
So, why an 8 inch driver? Why not 6" or 10"? Well, I'm sure the 8" driver in an arbitrarily derived ideal size for a dynamic driver. The best balance between bass extension, sensitivity (cone mass) and high-frequency extension. And most important: I've always found an 8" driver to provide the best reproduction of the human voice - and that goes for both male and female voices. Looking back, we often find radio speakers and studio monitors made around 8" drivers, being Supravox in France, greencones in Germany, the BC1 in the UK, etc. Numerous others could be mentioned.

I've often found drivers with foam surrounds superior in performance compared to drivers with rubber surround. I've changed quite a few drivers' surround from rubber to foam and always found an increase in performance, generally a more vivid sound without compromising the drivers' frequency response. Sometimes the frequency response was even improved although this often depended on proper coating of the foam. Foam surrounds are very lossy and have very low mass compared to rubber - even foamed rubber - and often foam will provide 1-2 dB sensitivity increase. As a rule of thumb, the main part of the total mechanical resistance should come from the spider, not the surround and the cone may bend from pushing towards the surround being either too rigid or having high mass. This is in particular important when we are dealing with lightweight cones. Thus, we usually see high-efficiency drivers having either foam or corrugated paper or fabric surrounds, the latter usually coated to preserve the fragile paper and for fabric to make them airtight.
Modern foam surrounds come with built-in UV resistance and with this driver we have further added UV protection from applying a thin black polymer coating. Extensive accelerated UV tests are ongoing, where foam surrounds coated with numerous different compounds are being tested by exposure to strong UV radiation, many times the normal level coming from the sun. Ongoing, because the control sample without coating hasn't even started deteriorating yet after one year. So we feel confident that you will have a driver for years and years to come.

Break-in: As is often the case with high-efficiency drivers, the JA8008 takes some time to break-in before fully rendering its maximum performance, i.e. bass extension. After approx. 100 hours of use over a three-month period, Fs was reduced from 50 to 42 Hz. This from an average of 4 drivers under test. So, be patient after initial set-up.

Drivers and waveguide used in TQWT

I had a mail after finishing the TQWT construction and when I first read it, I thought the guy was kidding: "Would you please describe your feelings about JA8008 driver? It looks like you're writing much more emotional about vintage speakers and this driver is only "one from many" for you. Am I mistaken here?"

My answer:

The JA8008 is my blood, sweat and tears! My desert island speaker system! More than a year of dedicated work. It was my idea to lauch a high-efficiency 8" driver for low wattage valve amps, because there aren't that many around. This was definitely not just "another job" from Jantzen Audio. But I see you are right. I went through my files and I've been so busy telling about all the things going into making a driver and a speaker system, I've forgotten to write something about the system itself and how it performs. Like anything in life, we need emotion, feelings, enthusiasm, etc. I'll see if I can get it on paper. This is the worst part not falling into the ditch of stereotype self-praise - and despite the impossible task of describing sound. It really can't be done. You have to listen before you can know what this driver does.

The "problem" with high-efficiency speakers (HES) is that people need to spend time reading - actually reading a lot of stuff - about what HES is, what they can do and what they sometimes cannot. People have to realise the relation between size, efficiency and low-end extension to make a qualified judgement of their needs before deciding what to buy. And we must realise that the companies offering 96 dB in a 40 litre cab do not tell these speakers do not play bass below 60 Hz. Thus my articles here: http://www.troelsgravesen.dk/HES.htm.

The HES & HES-II articles are the pre-work going into the JA8008. In particular HES-II. The old SEAS CA21REX driver was the one that got it all started, and if you have a project ongoing for almost 1 year, you simply cannot manage the constant level of the emotional exitement the writer misses. It doesn't work that way. There are times of trouble when it doesn't sound right and times of joy when things work out right.

What I've tried here is to make an overall balanced sound from the JA8008/TW034. Compared to all other speakers I've made, the area where this speaker differs the most is in its smooth midrange and ability to reveal the dynamics of music. The speed and attack it can produce is something none of all the other speakers on this website can manage. It's fast and a drum solo will kick butt, even from a 10 wpc amplifier.
The midrange is special because it comes from a large membrane and it's easy to hear - and understand - what singers sing. This is important if you listen to lyrics in a foreign language, which I mostly do. This speaker can reproduce vocals; it can even play vocals loud without noticeable distortion. I've made distortion measurement and even up to 110 dB @ 1 metre distance it's mostly dominated by 2nd order distortion at low levels. No nasty distortion peaks anywhere.
The bass is different from most vented systems because it is "short". Comparatively it's short and dry like bass from a well designed closed or aperiodic box.

The aim of this project was to create a speaker system that could compete with high-efficiency 8" full-range drivers, but without the shortcomings of whizzer cones necessary in this category. I believe a properly engineered crossover offers significant advantages over the compromised dream of having a single diaphragm doing it all. And I think I've got it. There's no "midrange shout"; there's no 10 dB treble peaks you can't stand in the long run and there's an overall balanced presentation from the ever so important midrange, important for making instruments and vocals sound right. And there's an even power response over a wide area, something we can never accomplish from an 8" whizzer-cone fullrange driver.

Robustness is another important parameter when putting bits and pieces together to make a driver. We often see commercial drivers made from components that really aren't sonically optimal, but durability is an important issue as the least thing a manufacturer wants to experience is his product being discussed for falling apart due to improper gluing, burned voice coils, etc., and having drivers back for re-coning is not something any manufacturer looks forward to. Ideally a driver should last for a couple of decades to fulfill customers expectation to quality and performance. Very few drivers can make 5 decades without suffuring from sagging cones due to loss of spider elasticity. The JBL 123A bass driver is probably the most durable driver I have ever experienced, but a wide range of vintage drivers really do not perform as they should due to age. High humidity and high temperatures are other problems to consider. Last but not least: Young people! Sorry young folks, but you have bad record for frying your voice coils Saturday evening.

I have to include some comments on the large dome tweeter and not least the waveguide used for making an overall smooth transition from upper mid to lower treble. Having a large radiating area combined with a waveguide, providing both acoustic amplification, proper dispersion and reduction in distortion are all key elements in the smooth treble presentation. Treble os often regarded as "something above 3 kHz", which is really not true. It starts much earlier and proper integration of drivers is vital in any successful construction. Forget any other non-waveguide standard 1" dome tweeter for the JA8008. It's been tried and it doesn't reveal the potential of the JA8008 driver. Not even the most exotic and expensive tweeters could handle what this large dome can do.

Back to top.

Click images to view large

Click images to view large

Tweeter for JA8008
(back to top)

Sourcing a tweeter to match a 95 dB driver was quite a problem, because there simply aren't many around. I didn't want to use horns and ribbons. Horns are too directional and ribbons have poor vertical dispersion and they didn't integrate well. Some recent Vifa, Peerless, ScanSpeak and SEAS tweeters have decent sensitivity, but rarely above 92 dB/2.8 volts. I didn't want a "naked" tweeter; I wanted a tweeter with a modest waveguide because they integrate better with the midrange due to reduced distortion at point of crossover and from providing an even power response. For a system tuned to 94-95 dB sensitivity we do not necessarily need an e.g. 95 dB/2.8V tweeter. Usually the crossover will raise the sensitivity a little, but still this isn't enough.

Adding a waveguide to a dome tweeter will - if well designed - raise sensitivity 4-6 dB in the 2-8 kHz range, but also lower on-axis response above 10 kHz. Most 1" soft-dome tweeters are fairly directional above 10 kHz and applying a waveguide will often produce a more even power response, hence reduce on-axis response above 10 kHz.
Initially the vintage JBL LE26 tweeter, fitted with a waveguide, was used and performed very well, but the LE26 is obviously not a viable option, as it is no longer produced and only found on eBay.

The Dynaudio Esotar was tried, the Monacor DT300 + waveguide, Vifa XT25TG + waveguide, one of the new SEAS EXCEL tweeters, etc., but none of these were as good as the JBL LE26 + waveguide. Nor were the SEAS high-efficiency tweeters T29CF001/002. All very good tweeters in other combinations, but not here.

Eventually the Audax TW034XO-P47N tweeter was tried and turned out to be just what I was looking for. A large 34 mm fabric dome driven by a huge 100 mm magnet. The TW034 is actually a classic, having been around for many years and well known from a range of former British studio monitors. Fortunately, Audax is still producing this tweeter, now in a new and improved version. The dome area is as big as the LE26 and integrates even better than the LE26 with the JA8008 driver. By coincidence, the waveguide will fit both options should you want to experiment.

The waveguide was a new problem, as no existing waveguide would fit the TW034 tweeter. Thus I had to engage a factory dealing with CNC fabrication of everything that can possibly be made on a lathe. A huge 180 mm diameter rod of black POM plastic is stuck into the lathe and in a few minutes turned into nice looking, solid waveguides. No need to fill cavities here with damping material.
I developed the waveguide for Jantzen Audio, thus proprietary to Jantzen Audio, so please do not ask for drawings or specifications. A waveguide developed for a 34 mm dome is unlikely to perform with other (smaller) domes. To my experience every dome will require its own waveguide made to suit the dome's shape, diameter, suspension, etc. In short: A waveguide is not just a waveguide.

Click images above to view large.

TW034 fastened to waveguide.

Final JA8008 production samples. Click image to view large.

JA8008/TW034 applications
(back to top)

Click images to to to website (except Onken)

The JA8008/TW034 driver team has numerous applications as seen above. So far, all except the Onken have been realised.

TQWT: This cabinet is the one presented on this page. The system here is tuned to ~94-95 dB and provides sufficient bass for most applications. Obviously a TQWT tower can be made as seen in the second suggestion, providing a rather tall construction, but with no bends to the tube.
To learn more about transmission line speakers and TQWT cabinets (tapered-quarter-wave-tube) I suggest reading here at Martin J. King's website:
If the tube is tapered it's a horn, in this case a conical horn, and horns may come in all shapes and sizes depending on actual application, from the deepest bass to the highest treble.

The DOUBLE TQWT was my second project and suitable bass drivers were found in the Eminence Deltalite II 2510 10" bass drivers. Minimum impedance can be maintained at 6 ohms from connecting bass drivers in series. What has been my prime target here is not having a high-pass filter on the front driver. The JA8008 takes decent cone excursion and supplemented by two 10" bass drivers the overall performance is enhanced compared to the TQWT. The price to pay for this is size. 150 litres brutto volume is what it takes - but what a sound!

The BR is realised in the QUATTRO project, featuring a 40 liter vented cabinet and an 8 ohms minimum impedance due to the special crossover that enables us to use the low-Qt JA8008 driver in small enclosures.

OPEN BAFFLE OB9: The JA8008/TW034 was tried on an open baffle, supplemented by a high-efficiency 15" driver and performs excellent - the best live-feeling in our living room yet. Usually we would use a high-Qt driver for a single-driver/open baffle set-up. I chose to use a fairly low-Qt driver, Eminence DeltaLite II 2515, by adding a baffle-step compensation circuit, increasing Qt and reducing sensitivity (99.2 dB) to match the 8008/TW034 combo (~94 dB on an open baffle). Initially an electronic crossover was used but next version will be passive and be launched 2011. The OBL-11.

OPEN BAFFLE 11 is finished and can be viewed here. This features a passive solution to the crossover between bass and mid. This is my homage to Mr. Gilbert Briggs, founder of Wharfedale loudspeakers and his famous SFB, Sand Filled Baffle, loudspeaker.
As experienced during the set-up of OB7 and OB9, having a 12-15 inch bass driver on an open baffle is special and we immediately realise what it means to have a bass driver in a box. Anything else but an open baffle sounds "boxy". This 15" Eminence bass driver makes bass physical and a drum kit all of a sudden starts sounding like a - drum kit. Large and highly efficient drivers on an open baffle is special and makes most conventional speakers sound dull and overdamped.

DTQWT-12 was on the drawing board and the ink was hardly dry before an Italian guy took the bait and realised the project. Can be seen here. Be prepared for 152 cm height, 33 cm width and 60 cm depth. 2 x 12" bass drivers will load the large horn and deliver effortless deep bass performance. This project has been postponed over and over again due to lack of time - and not least, where am I going to put these huge cabs once made? Our house is too small for having big speaker systems in every corner. Construction page here.

The ONKEN sketch is showing a well-know application, utilising a 12-15"/95-96 dB bass driver in a 80-160 litres cabinet with vertical front vents. The JA8008/TW034 is on top in either a vented box, a closed box or an open baffle as is often seen. Due to the high efficiency of the JA8008/TW034 duo, the use of electronic crossovers may be worth pursuing here with a point of crossover from 100-250 Hz depending on actual driver set-up. If Eminence DeltaLite 2515 is used as in OB9 and OB11, the passive crossover from OBL-11 is likely to fit in well here.

What the JA8008 history already has told is the importance of having an extended range driver with a excellent midrange reproduction. Around the JA8008 driver, numerous constructions can be made to suit your room, your taste in cabinet design and overall bass requirements.

Crossover for TQWT-mkII
(back to top)

TQWT mkII schematics.

The low-pass section is first-order with a parallel RC-circuit to produce linear impedance. The 0.82 mH series coil is bypassed by a small RC-circuit, providing target roll-off at point of crossover. This simple arrangement together with the driver's natural roll-off provides an acoustic 3rd order topology.

For the tweeter a simple LR2 roll-off is provided. My preferred attenuation setting requires the 2R7 resistor, but some may prefer more.

The chosen crossover design makes an easy 7 ohms impedance, ideal for even 5-7 watt SET amps. Due to the TQWT cabinet used, the usual bass-reflex twin-peak impedance profile is seen in the bass region, but impedance is high where phase angle (dotted line) is capacitive as can be seen from the graph below. Thus, an overall easy load on the amplifier.

Predicted impedance.

Crossover layout for TQWT mkII
Please notice that the orientation of an RC circuit (1.0 uF + 15R) can be either way. RC or CR, doesn't matter, does the same thing. 
I have this question regularly as the photo below is opposite to the drawing above.

Changing your mkI crossover to mkII, download image here.

Back to top.

(back to top)

All kit parts supplied by Jantzen Audio.
Mail at:

All technical questions at: troels.gravesen@hotmail.com

Back to top.

The Finished Crossover

I have learned that some people get confused by the three resistors on crossover board for tweeter attenuation. Some builders have connected all three.
Only one of the resistors should be soldered to the 5.6 uF capacitor. The resistors provide three levels of treble, low value = high level, high value = low level.
Click left image above to view more closely.

Mounting the drivers

This is diy, so you have to do a little more yourself: Center the tweeter on the waveguide as seen on photo.
Mark for holes on waveguide and add a 10 mm mark on the drill. Drill 3 x 10 mm holes in waveguide,
check if dome is properly centered and fasten with the screws supplied with the kit, 2 x 4 pcs M4 x 10 mm. That's it!

Secure leads from drivers like seen on photos. Ripped terminals are a real pain! Wires used are silver plated copper in teflon.

TQWT Cabinet for JA8008
(back to top)

As discussed in detail in the HES article, the price for high efficiency plus low-end extension is size. However, the choice of driver size and basic parameters for the JA8008 allows a modest cabinet size; that is if you think 80 net litres is a fair price to pay for 94-95 dB sensitivity and 40-45 Hz bass extension. We are not going to claim 98 dB sensitivity and 25 Hz bass extension from an e.g. 55 liter cabinet like some manufactures will do, because it doesn't exist.

The JA8008 performs very well in a TQWT cabinet with a modest front baffle width of 280 mm. Height is 1050 mm and depth is 380 mm. Cabinet construction is simplicity itself as can be seen from the drawing below. Front panel is 30 mm MDF and slightly chamfered. Side, top and rear panels are 22 mm MDF and the internal panel is 19 mm MDF, supported by braces between front and rear panel.

Cabinet drawing

Construction details

Cutting side panels at 45 angle at top from pre-vennered MDF.

Test assembly of rear, top and side panels.

Internal panel with bracing.

Left: Gluing top fillet to front panel. Right: Solid 30 x 50 mm oak side bars for front panel. Chamfering is 30 at half width.

The magnet of the Audax dome needs space, thus the front reflector needs some routing.
A 105 mm outer diameter hole is made into the reflector. Route as deep as possible, break off the remains and clean with a knife.

Since the development of the speaker 8 mm grey felt has become available from Jantzen Audio.

Damping material: 8 mm grey felt supplied with kit.

Behind JA8008 driver: 23 x 50 cm, cut hole for bracing. Plus 23 x 35 cm 30 mm acoustilux right behind drivers and on side panels.
Side panels next to JA8008 driver: 4.5/14 x 50 cm.
Below JA8008 driver: 23 x 30 cm MDM3. Use glue or staples.
Top of cabinet: 10 mm grey felt, 23 x 50 cm. This continues all the way to the rear bracing.

In total 0.75 m^2 10 mm grey felt or 8 mm felt material is used.

Routing for the waveguide

The POM plastic material has excellent properties for machining and routing into the waveguide for the JA8008 is
done by slowly increasing router cutting depth; some 0.5 mm at a time.

DO NOT forget to chamfer 8008 driver hole to allow free rear ventilation!


Placement of crossover

Place crossover (here mkI) on rear panel of cabinet above terminals.

(back to top)

Left: The un-smoothed response of individual drivers in TQWT cabinet without crossover attached, here from 400-10,000 Hz - the area most important. As can be seen, there's a huge overlap from both drivers around point of crossover ensuring proper performance in areas where both drivers are operating.
Right: Impedance of drivers JA8008/blue and TW034/red in cabinet. Green (JA8008) and yellow (TW034) is electrical phase.

Impedance plot of 8008 driver in free air.

Left: Summed reponse of drivers driven from crossover. Average midrange sensitivity is 94-95 dB/2.8 V.
Right: Response of individual drivers driven from crossover.

Left: Impedance (red) and electrical phase (blue).
Right: Same as left with higher resolution. Minimum impedance from 100 - 1000 Hz is 7 ohms and 5 ohms above 8 kHz.
Overall an easy amplifier load.

Left: Horizontal dispersion at 0, 10, 20 and 30 deg. Up to 15 kHz we have +/- 1.5 dB over a 60 deg. horizontal angle.
Right: Vertical dispersion at +/- 10 deg. Read = on axis (between TW034 and JA8008).

Back to top.