WD25A EX – the final crossover, by Peter Comeau


In the March issue I introduced a ‘top-of-the-range’ version of WD25 called the EX, the design of which was prompted by a re-visit by an owner of the original Dynaco A25 on which the WD25 was based.


If you haven’t read the previous write-ups on WD25 and the Dynaco A25, suffice it to say that the Aperiodic concept introduced in the ‘70s by the Scandinavian Dynaco range has never, to my knowledge, been successfully re-incorporated in a commercial loudspeaker. The WD25 is an attempt to recover the performance of this classic design using up-to-date materials and sound quality.


Initially the sonic character of the WD25A, the stand-mount that ‘copies’ the concept of the Dynaco A25, was brought as close as I could remember to that of the original. Our memory of sound, however, is not without failings and, you have to remember, back in the ‘70s I was using vinyl as a primary source and without the advances in turntable performance that we have today – not to mention Compact Disc!


As such I was delighted to hear from one Hi-Fi World reader who still had a pair of Dynaco A25 in everyday use and would I like to hear them? You bet I would! So a couple of years ago our reader arrived, with Dynacos under each arm, for a sonic shoot-out.


In the comparison that followed, while I was delighted by how close I had got to the performance of the originals, Mr Dynaco owner was not so pleased. In particular he preferred the warmth and full blooded bass of the originals, whilst the more revealing midband of the WD25A did not impress him. So we agreed to disagree and, perhaps, meet again once I’d brought the desired elements of warmth and bass power into the equation.


In fact it was to be eighteen months before I got round to having another ‘go’. By this time I’d used the SEAS Millennium treble unit in a couple of speakers, one a rejig of Noel Keywood’s KLS3 Gold, successfully replacing the Audax pressurised Gold Dome unit.


The Millennium had impressed me, as well as others, so much so that I wondered whether it had a place in WD25. Of particular interest was the way it managed to reveal all the treble and midrange detail required while not sounding ‘bright’ or ‘forward’ in the way many modern treble units have to do. So it sounded right up Mr Dynaco’s street in that I could bring back the warmth of the bass without compromising on treble detailing.


The basic crossover design was outlined in the March issue but, at that point, I had little idea how much work was left to be done. Accordingly a date was set for a revisit of the Dynaco A25s and, this time, Mr Dynaco owner was happy to leave them with me for a day and a half to aid analysis of the comparison.


On first listen it was clear, once again, that I was close, but not close enough. The first step was to bring up the bass power by reducing midrange level. A small increase in coil inductance in series with the bass unit, from 0.68mH to 0.76mH, was enough. However this, naturally enough, left the treble over-exposed and the crossover audibly obvious, so it was back to the crossover drawing board to see what could be done.


Increasing the treble attenuation a touch, with 1 Ohm from the input to the crossover circuit itself, and 0.68 Ohm in series with the treble unit, put the treble level in the right area. But this also altered both the crossover point and its phase relationship with the bass unit crossover where the increase in inductance actually lowered the crossover frequency.


With the comparison with Dynaco A25 now firmly lodged in my memory the process of tweaking and adjusting went into full swing. I thought it best to re-establish the crossover point and investigate the acoustic slopes of the crossover as a result.


Figure 1. Acoustic slopes of LF and HF drive units and crossover aligned to

quasi-Linkwitz 3rd order targets. Crossover frequency 2kHz.


Using LspCAD v6 I am able to feed the measurements of the drive units in the cabinet into a ‘virtual’ crossover to see its effect before listening to the result. The acoustic crossover slope that I’ve found successful, time and time again, is what I like to call a 3rd order Linkwitz characteristic. In theory there is no such thing, the Linkwitz-Riley topology pertaining to even order filters only, but the characteristics of a Linkwitz filter, namely a -6dB crossover point and ‘perfect’ phase integration between the low and high pass sections, can still be simulated in LspCAD.


So, working to this ‘quasi-Linkwitz’ 3rd order slope we see an acoustic response from the drive units that looks like Figure 1 (having found the ideal crossover point for these drivers to be 2kHz).  The circuit for this crossover is shown in Figure 2.


Figure 2. Crossover to achieve quasi-Linkwitz 3rd order slopes as shown in Figure 1.

Now you may be a little puzzled as to why the electrical crossover looks remarkably like a second order if these are third order slopes. The answer lies in the way the electrical crossover works with the natural roll-off of the drivers.  The bass unit, for example has a gradually decreasing output beyond 1.5kHz which approximates to a first order slope. Add this to our second order electrical crossover and we get a third order acoustic filter. As the meerkat says “Simples”!


LspCAD is a great tool, but only when you use its results to feed into your own listening process. The technique is to use LspCAD to predict the measured results, then listen, make adjustments, feed these back into LspCAD, adjust to match the target slopes, listen again and so on. This is a process of iteration. Sometimes the results come quickly, sometimes it takes days or weeks to achieve the ‘perfect’ performance.


In this case the results were good, very good, with high quality source material. But that’s not everything. Good quality recordings can often fool you into believing that the speaker is ‘perfect’, but I find that a few poorer quality discs, what I call my ‘problem’ CDs, highlight errors that may otherwise not show up.

These ‘problem’ discs can be relatively modern recordings, such as Tears For Fears ‘The Seeds Of Love’ or older master tapes transcribed to CD like The Rolling Stones ‘Forty Licks’.  The latter should be a fascinating trip through rock and roll music and recording techniques from the ‘60s onwards, even though the early singles are definitely ‘lo-fi’ they should still be listenable.


Largely it comes back to this question of balance between bass and treble. If the treble output is a tad too high the speakers will sound bright and detailed but overbearing and ear-rippingly harsh on these early recordings. Similarly a touch too much bass and the music slows to a crawl, percussion sounds heavy and bloated and the early recordings come across as coloured and wooden.


Then there is the midrange. One way of overcoming the compressed, mono mixed, squeezed-through-a-tube vocals of the ‘difficult’ recordings is to recess the midrange a little producing that apparently spacious ‘depth’ so beloved of audiophiles.


This may clean up the sound of difficult records and make them sound warmer but the overall effect is to push vocals too far into the background leaving them swamped by the rest of the band. Clearly that isn’t right, either, it’s just an acoustic trick.


What is needed is good projection of the midband without making it forward. I won’t deny it, it’s a difficult juggling act, but it can be achieved with judicious balancing of all three elements so that the speaker sounds seamless across the audio band. When that point is reached you lose the bass-midrange-treble conflict and start listening to the music instead of the speaker.


In this case the initial listening results indicated some hardness in the treble crossover region which was ameliorated by ‘softening’ the turnover of the filter by increasing the series resistance of the treble coil.  Traces of treble excess on sibilance and stringed instruments led me to turn to adjustment of the bass section to bring the upper midrange output of the bass unit into line with the treble.


After extensive listening I ended up significantly reducing the capacitance across the bass unit which, as well as increasing midband output, had the unexpected bonus of subjectively boosting the bass power too, probably as a result of the component choice open to me with the new, preferred, value.


Figure 3. Measured results from WD25A EX showing acoustic crossover and response on treble unit axis.


Figure 4. Family of responses of WD25A EX at 1m. Blue - on axis with treble unit, Red – 15

degrees horizontal and 30 degrees horizontal. Note close correlation of sonic character

and even power distribution in room right up to 10kHz.


Figure 5. Response with drive units In Phase (red) and Out Of Phase (blue).  Deep notch

for OOP response confirms accurate phase integration throughout the crossover.


Figure 6. ‘Easy drive’ impedance of WD25EX. Overall an 8 Ohm load with no abrupt phase

changes. Minimum impedance of 4 Ohms in the treble region will not cause amplifiers

any problems.


Once the actual values of the crossover have been worked through it’s time to consider component selection.  Coils are a no-brainer. We use the best quality ferrite core for the bass because of its low resistance and an air core for the treble for low distortion.


Incidentally if we were also to use an air core for the bass section the coil resistance, in series with the bass unit, would jump from 0.2 Ohms to 0.8 Ohms with a corresponding loss of sensitivity of 1dB and a reduction in amplifier damping factor. Having listened to both types of coil the problems outweigh the benefits when using an air coil for bass with this design.

Of more interest is the capacitor performance. Clearly superior quality polypropylene capacitors are de rigueur for the treble section, but what about the bass? In the past we’ve used a reversible electrolytic bypassed with a polypropylene at 10% of the value of the electrolytic, and this does work well. An electrolytic on its own produces a level of cardboardy and wooden colouration through the midrange simply because its resistance does not change linearly with frequency so the capacitor losses at high frequencies disturb the crossover dynamically.


Figure 7. Final schematic crossover of WD25A EX derived from listening tests.

With the level of musical transparency produced by the WD25A EX, however, even a bypassed electrolytic in the bass section did not perform as well as using all-polypropylene capacitors here. Luckily the lower value of capacitance required by this crossover does allow us to use polyprops without too much of a space issue – a pair of SONIQS 8.2uF proved ideal.


The final crossover circuit is shown below. It proved an exhausting and, at times, frustrating achievement but, I have to say, well worth it in the end. A similar prototype crossover for the WD25T floorstanders was demonstrated at the Bristol show to great acclaim, so we have the proof this combination of drive units is the best WD25 yet.


But what about Mr Dynaco owner – what does he think of the WD25A EX? Reader, he bought a pair.


Add your comment

Your name:
  The word for verification. Lowercase letters only with no spaces.
Word verification:


Hi-Fi World, Powered by Joomla!; Hosted by Joomla Wired.