Audax HD-3P Gold Dome replacement |
Audax HD-3P Piezo tweeter
Gold Changes Peter Comeau explains how to update your World Audio Design KLS3 loudspeakers, if you too have suffered the dreaded 'Gold Dome Wrinkles'... from Hi-Fi World JUNE 08 issue
Since I updated a reader’s KLS10 Gold bookshelf speakers last year I’ve had more than a few e-mails asking whether the SEAS Millennium treble units I installed in place of the gradually declining Audax HD-3P Piezo units could be used in the KLS3 Gold as well. Whilst the simple answer is ‘yes’ I pointed out that a crossover change would also be needed, particularly for the treble unit crossover and possibly, also to the midrange crossover to perform a perfect match.
This dismayed all but one correspondent who was so eager to refurbish his KLS3 Gold MkIIs that he delivered them to the World Designs Lab so that I could perform the necessary measurements and revise the crossover.
Audax Gold Dome wrinkles First, a little bit of background on the problems suffered by the original KLS Gold Audax HD-3P treble units. These work by utilising the piezo effect on a gold sputtered film held in an elliptical dome shape by a pressurised chamber filled with Nitrogen.
Early units suffered from a less than perfect gas seal which causes the Nitrogen to gradually leak away. The first visible sign of this is that, in high pressure (fine) weather, wrinkles or creases appear in the gold dome. The audible effect is of a loss of sparkle and ‘air’ to the sound but, because the loss of pressure is gradual over a long period, the listener may not be aware of the treble loss.
Nevertheless the treble loss is there and, eventually, you’ll need to take action to restore your KLS Golds to their former audible glory. Now, not every Audax HD-3P treble unit suffers in this way. As soon as Audax became aware of the defect the seal was redesigned with the result that later domes will hold their gas pressure. The test is simply to see whether the wrinkles in the dome appear, particularly when atmospheric pressure is high. If not, leave well alone.
When the speakers to be modified turned up at World Designs the problem, in one of the units, was very pronounced and, indeed, treble output was severely curtailed to the point where the speakers were unusable.
Replacing the units with the SEAS Millennium T25CF002 was, physically, not quite so straightforward as it had been with the KLS10 Golds. Because of the close proximity of the midrange and treble units, and the larger outer diameter of the SEAS faceplate, the treble unit has to be moved upwards towards the top of the cabinet. Luckily there is enough room, just, to fit the new treble unit without it overlapping the top of the cabinet!
All that is necessary is to take a file, wood rasp, or preferably router, to the treble unit hole and elongate it to accommodate the magnet and terminals of the SEAS Millennium. In this particular pair of speakers we also took advantage of using a router by recessing the drive units into the speaker baffles. This isn’t necessary for the upgrade but does provide a slightly smoother frequency response as there are no close proximity drive unit edges and steps down to the baffle to cause extra diffraction and interference effects.
Anyway, back to fitting the new treble unit. As I said the SEAS faceplate is larger in diameter than the Audax, at 110mm, so making new fixing screw holes isn’t a problem and it won’t look as though you’ve butchered your cabinet to fit the Millennium in place! Crossover As always I went through a few iterations of the crossover before finalising it. This is often the way as the first attempt never works out as you would expect when you start listening to it. In this case, the first measurements seemed to indicate that a significant change to both the midrange and the treble sections of the crossover would be necessary. That’s because, at first sight, the slope of the upper midrange crossover did not look right for the new tweeter. So it seemed easier to adjust both in order to make sure of good integration between the two.
KLS3 Gold Mk2 Crossover – the original on which the Mk3 is based.
Originally KLS3 Gold had inverted phase between all three drive units. In theoretical terms a second order crossover always will be easier to integrate when the drive units are electrically out of phase. If you’ve followed my articles on speaker development in the past, however, you’ll have seen that what works in theory doesn’t always work in practice when it comes to working with real-life drive units.
In trying to keep the same integration between the drivers as was evident on the originals I left the electrical phase as it was, in other words the treble and bass units were inverted in phase compared to the midrange. This gave me a good frequency response, once I’d used LspCAD6 to help me adjust the crossover values, but that didn’t translate into a smooth listening experience.
Plugging the new treble unit and crossover into the speakers and lugging them into the listening room I sat down to experience what I hoped was as smooth a response as I had measured. No such luck!
What I hate most about three way crossovers is that when you do anything at all to one of the crossover sections you disturb the other crossover too (audibly, that is). So it was absolutely typical that, having altered the upper midrange crossover, the integration between the midrange and bass units fell apart. Clearly this needed a rethink – a new strategy was called for. New Strategy One thing I noticed, whilst listening, was that the midrange performance was totally different to that with the old crossover. This gave me the clue that I was looking for. It was pretty obvious that, if I tinkered around with the midrange unit crossover, the whole speaker was going to change in character.
Now this wasn’t what the owner wanted. I knew, from a previous conversation, that he was very happy with the overall sound of the KLS3 Gold and he just needed an ‘upgrade’ and restoration of the treble performance, not a whole new speaker! So, back to the drawing board – actually LspCAD6 – and re-look at the crossover details.
To start with I decided to put the treble unit back into phase with the midrange. The reason is one that I’ve found works for me, although looking at other people’s crossovers it apparently doesn’t work for everybody.
This shows the final plot of the midrange/treble crossover.
My philosophy is that the ear is sensitive to the time arrivals of transients. I know that there is a considerable body of evidence to show that the ear does a summation of the total transient response from a speaker and then a type of mechanical Fourier analysis of the waveform to realise the tonality of the sound. But I’m still convinced that the ear also reacts to the initial transient of any sound in a particular way.
It is the initial transient, for example, that gives us directional clues and, hence in hi-fi reproduction, helps position the image in the stereo plane. Thus if the phase is inverted between the drive units, especially at high frequencies, there can be a little confusion for the ear/brain in that it will see the transient from, say, the treble unit going in the positive direction and then that from the midrange unit going in the negative direction.
Effect of reversing the treble drive unit phase indicating excellent phase integration when in phase!
This is particularly important where the drive units are phase aligned (which doesn’t happen very often in modern speakers) but, in my view, is still noticeable even when the output from the treble unit reaches the ear slightly in advance of that from the midrange. Back to the drawing board
So part of the ‘back to the drawing board’ process was aligning the crossover to work with the drive units in phase. A bit more playing around with components and values in LspCAD6 and, hey presto, suddenly I had a smooth response without having to adjust the midrange crossover values far from the originals. In fact the only change is to up the value of the capacitor to ground from 6uF to 8.2uF (8uF was used in the Mk1 crossover).
Again listening showed that this was the right way to go. The character of the KLS3 Gold was now restored plus the missing treble was back again and the stereo image now seemed more solid and realistic.
It took a bit more juggling of component values to elicit the final crossover for the treble section but, at last, I was satisfied with the performance and guessed that the owner would be too. An appointment for collection was made and, while I was waiting, I went back to the listening room for a more extended listen.
Now that the treble was restored to its former glory (as far as I could tell, not being in possession of a pair of correctly working originals for test) it was so beautifully sweet, detailed and transparent that it revealed that the lower midrange had a touch of overt chestiness about it. This clouded female vocals a little and added undue emphasis to male vocals in my opinion. Now this seemed to be coming from the mid-bass crossover and there wasn’t much to suspect there other than a hefty 50uF in parallel with the bass unit. I started by adding a small bypass SONIQS polypropylene capacitor across this 50uF. This seemed to help so I increased the value until the chestiness seemed to disappear.
The value I ended up with was another 18uF on top of the 50uF, not enough to seriously disturb the bass/mid crossover point but just enough to clean up the sound and remove a level of coloration I’d heard.
Whilst we’re talking about the mid/bass crossover what about bringing these two units into electrical phase too? Now this wouldn’t, and didn’t, work with the existing crossover and nor does it matter, and here’s why.
At frequencies as low as that of the bass/mid crossover in a three way speaker, the wavelengths are long enough, and the arrival time of the transient slope from the bass unit delayed enough, that the phase of the bass unit is hardly significant at all. I wouldn’t rule it out of the equation completely, because the bass unit is still contributing a small amount of output at higher frequencies, but it’s not worth upsetting the crossover performance just to correct it. So leave well alone! Final listening When the owner arrived, he was ushered into the listening room to make sure all was well. Initially I undid the ‘correction’ I had made to the bass crossover to make sure that this didn’t destroy the character of the speaker that he had been used to.
Initial listening was exceptionally positive. More than just restoring the original treble response I had, according to the owner, apparently improved the clarity and detailing of the treble. The other major surprise was his claim ‘at last, a good, precise stereo image’, which unsolicited comment backed up what I thought about the phase relationship of the two upper frequency units.
‘ What about the natural realism of the sound?’ I asked after a few tracks had passed. ‘Not bad’ came the reply. ‘I’d like to try an alternative, if you don’t mind’ I proffered. That was the point at which I added the 18uF back in. A few more tests of ‘with’ and ‘without’ and the owner proclaimed himself more than satisfied with the more realistic results of adding the 18uF.
Actually, because he was going to reconstruct the crossovers with the new components and values I’d devised, I suggested that he took a selection of SONIQS polyprop capacitors with him in order to optimise the performance in his own room. It’s one thing to do a quick lash up for an impromptu change, but another to actually settle down and listen to that change over a longer period of time. At the time of writing, the 50uF plus 18uF is still the firm favourite, but there’s nothing wrong with experimenting if you want to try different values here.
Tweaks If you want to improve further aspects of the performance of the crossover there are several components that you can upgrade. This applies whether you continue to use the original Audax gold dome piezo treble units or the SEAS Millennium.
Naturally it is worth changing all the capacitors to high quality polypropylenes such as the SONIQS PXX and SAX types we use in our own crossovers. This is particularly true of the 50uF capacitors in series with the midrange unit. Those looking for the best quality may wish to use SONIQS SAX capacitors throughout, but a 50uF value in this range is so large as not to be believed! So the cunning plan is to use a 47uF PXX with a 3.3uF SAX in parallel with it to give the required 50uF.
If you like the sound of the 18uF across the bass crossover 50uF then you don’t have to change the latter as the 18uF bypass capacitor is big enough in value to practically swamp the performance of the original electrolytic 50uF. I’ll leave that decision up to you. In this owner’s speakers he had already changed the electrolytics to polyprops so the decision was already made.
Other than that, the treble level can be adjusted slightly up and down by varying the initial series resistor. I found that 0.68 Ohms gave superb integration and meant that this 3 way design gelled into a big, spacious, music player where none of the drive units stood out on their own but all worked together. If you want more sparkle then try a 0.47 Ohm or, conversely, if you want your KLS3 Golds to sound warmer try a 1 Ohm.
What we’ve ended up with here is a reworking of the KLS3 Gold, previously in Mark II form so I guess, with the SEAS Millennium treble unit and crossover revisions we can genuinely call this KLS3 GOLD Mark III.
KIT OF PARTS What you will need to replace your Audax HD-3P Gold Domes with the SEAS Millennium.
2 x SEAS T25CF002 Millennium treble units
For the Treble crossover - 2 x 0.68 Ohm resistors 2 x 4.7 Ohm resistors 2 x 4.7uF Polypropylene capacitors 2 x 0.11mH air core coil
For the Midrange crossover - 2 x 8.2uF polypropylene capacitors
OPTIONAL UPGRADE COMPONENTS
For the Midrange Crossover - 4 x 47uF polypropylene capacitors 4 x 3.3uF polypropylene capacitors
For the Bass crossover - 2 x 18uF polypropylene capacitors |
