I had my MK1 Kans recapped at Wilmslow 4 years ago. They also fitted new grill cloths and Scanspeak tweeters. When I got them back home I found that the 3 screws on the tweeter faceplate on both speakers were loose, just spinning round and round. Also there was no gasket on the B110 driver, they took the drivers out to get to the crossovers and didn’t reseal the cabinet to make it airtight. Wasn’t best pleased and I rectified myself. Now a great sounding pair of Kans. Thought Wilmslow let me down.
1 Like
I’m afraid Danny is not really a professional and there’s quite a lot about speaker design that he doesn’t appreciate. I’ll give you one example. he was complaining in one of his videos about a speaker with a poor on-axis response, but which was transformed at 8 degrees off axis. Well, 7 - 8 degrees is a fairly standard design angle among what I’d call proper designers. On axis responses are fairly meaningless, first of all because they represent an anomaly where all the path lengths are the same, particularly from diffraction components, so the errors add up when they would be somewhat smoothed out at greater angles. The second reason is that there’s virtually no energy on axis. It represents a dot on the wall behind you, whereas the wider angles are very much larger circles, with a vastly greater area. 7 degrees or so is the first angle at which you get meaningful results for what is being put into the room. A fair number of designers will deliberately keep you away from the on axis response by firing the tweeter above your head or by firing down the room, or both. If they’re clever, they’ll arrange this such that the effective angle you listen at covers a wider area than just one spot and so that what comes off the wall has much the same tonal signature as what you are getting directly.
You will see a change in frequency response, and probably level too if you just stick to the same values. In the original Kans (which are a better design than the iron cored, cut down version) they chose different types of electrolytics in different positions. For example the opening capacitor on the tweeter was a Low Loss and the second was standard. There’s a reason for the LL side as well but the std one will have a higher ESR, which could be significant (though I don’t think I have a 5uF in that size here) and will in any case not be present in a film cap. That means the circuit will be driving a completely different load, essentially shifting both the frequency and the shape of the circuit. It’ll also behave differently with voice coil heating. I’m sorry, but if Danny thinks he can redesign a speaker in an afternoon, he’s very much mistaken. If that were possible, we’d all be doing it.
4 Likes
Oh no 
Compare that to an LS3/5a crossover(!)
Path lengths are clearly not the same on tweeter axis. And of course there’s energy on axis which is why you measure the highest output there and output drops away off axis.
Admittedly there’s things he maybe doesn’t cover so much in depth with his measurements, but addressing spectral decay and phase alignment is surely a valid approach borne out by the positive responses of the people who performed the upgrades.
Plus another difference here is that he’s trying to maximise and preserve soundstage and imaging which I think was never a concern for Linn, so Linn’s emphasis on “musicality” may result in some sacrifices in the soundstage and imaging department.
I myself have reworked three pairs of Tukans as wall mounted surrounds for my Home Cinema room and two pairs of AV 5110’s as the Atmos heights, using REW and a UMIK-1 and built new crossovers using air cord inductors, PP Foil caps and wire wound resistors. Maybe overkill for surround speakers but a fun project
“Path lengths are clearly not the same on tweeter axis.” Maybe you have misunderstood. When I say that the path lengths are the same I mean that if you look a triangle from the tweeter to the edge of the cabinet, and from that edge to the listener or mic, the hypoteneuse of that triangle on both the left and the right hand side are of the same length and so constructively interfere, meaning that they add up. You may not know this but the edges of the cabinets (the full length of the edges) act as secondary radiators. Ie. as though there were a tweeter, or some part of one, at the edge. The reason for this is that as the wave reaches the edge it enters a region of lower radiation resistance because it’s radiating into a wider space. Some of that wave radiates backwards and, because what comes in must equal what goes out, it modifies what goes forward. It’s very like Kirchoff’s laws - in fact it is an application of exactly that. As you go off axis, these path lengths differ more and more, so the compression and rarefaction elements of the sound wave don’t align perfectly any more (because of the additional delay) and so don’t produce such large peaks and dips. The same phenomenon happens with the surface of the radiator itself, so only on axis are the path lengths of the circumference of the tweeter to the microphone all the same. You can then extend this to the whole surface of the tweeter, viewing it as a set of concentric circles. This is why the response rises with frequency when you are on axis or near to it, and where that rise begins is related to the diameter of the radiator . This is widely appreciated. It makes the on-axis response a special case or, as I said, an anomaly. It’s also why you are increasingly seeing responses averaged across an angle in reviews - even John Atkinson is doing it. Those averaged responses are more representative of what is coming out of the tweeter than a single on axis response.
I have twice tried to explain to you that those waterfall responses are not really spectral decay. They are worth a glance at, but not much more, because they are entirely related to the frequency response! They aren’t, as you imagine them to be, showing you stored energy or perhaps delayed resonance, except inasmuch as it pertains to the frequency response. So if you have a resonance in the driver, either because of the cone or because of an impedance mismatch, it will show up as a peak and a dip (always the two together because the phase has to get back on track and having had an inflection one way requires an inflection the other way, with those inflections being directly associated with peaks and dips) you will get the characteristics of that resonance showing up in the time domain. Obviously a higher Q resonance will take longer to be damped out and extend further in time. But you don’t need to see that in a waterfall because it’s already there in the frequency response!
As for phase alignment, it’s my view that taking a measurement with the drivers in antiphase and patting yourself on the back because there’s a nice deep ‘V’ in the response showing they are cancelling and so must be aligned, is the very worst way to design speakers! To design speakers properly you need to know the inter-unit time delay and discovering that is not a trivial exercise. You need to take out the element of pure delay from your measurements, and that is not easy because all drivers will have a bandwidth limit and that will show up as a delay of some sort. And if you have a good driver with a nice smooth roll off, that will be almost indistinguishable from pure time delay. In the hands of amateurs this is a recipe for disaster. What is needed here is a Hilbert transform to extract what the phase response should be for that particular frequency response, but even then that is compromised by the fact that you won’t have reliable low frequency measurements, so this has to be pieced together with what the theoretical low frequency response should be, which in turn comes from your impedance measurements.
Re. improving the Tukans, I would say that’s starting from a pretty low bar. As for the Kans, they are from an era when everyone simply copied the KEF tweeter circuits (it showed up in nearly all the well-regarded speakers of the time, whether or not they used the T27s) and crossovers were designed on an ordinary pocket calculator. So, in the Kan crossover, the CR in parallel with the B110 (of 10uF and 10R) gives a remarkably flat load resistance of 8 ohms, even though the B110 has a DCR of 6.7 ohms. The 0.6mH inductor and 4.7uF cap are the theoretical values for a 2nd order Butterworth into 8 ohms at 3kHz - go and multiply them together on a calculator and you’ll see they come out at 3k. It really was as simple as that! It was a complete fluke that they happened to sound as good as they did, and the same is true of a lot of those KEF speakers using the T27 and B110 - as KEF effectively acknowledged when they produced their acoustic Butterworth for their tweeters. In fact, as you can go and simulate on LTspice for free, now that I’ve given you all the values, you’ll see that the Kans don’t actually cross over at 3kHz, but actually cross over at 3.2k - something which KEF will have got right.
As for soundstage and imaging, Danny may think he knows what he’s doing there but I assure you he doesn’t.
I was told that the Kan crossover should be 18dB/octave (3rd order) Butterworth at 2.75kHz but that came from an aging memory so might not be quite right.
1 Like
To remove the passive crossover in Linn Kan and instead use Naim active filters 18dB/octave 3rd order is just absolutely fantastic,the sound is deeper,faster,cleaner and the timing is marvelous.
1 Like
Naim said only a ‘small tweak’ would be required on my olive SBL NAXO but I never went ahead with the service and adjustment in the end. Even so, they produced a very good sound with the standard (unserviced) unit and pair of serviced NAP 110s. Shame I had only limited funds at the time as I would have liked to have hung onto them whilst exploring other avenues.
2 Likes
Yes they roll them off a little higher in the bass,getting an even more tuneful bass , but SBL filters works very good indeed.
1 Like
Back in the day (1988 or so), I ripped my Kans open and activated them. Just used a 42, SNAPS, NAXO and 2 x 110s. I had “Shoebox Syndrome” at the time. The NAXO was just set up for the standard 2 way configuration for SARAs. I had no clue as to a specific Kan NAXO.
I was pretty impressed, (being a fully gone Kan fanboy at the time). I also had Meridian M2s at the same time, both systems had their merits and fatal flaws.
Much as I loved the speed, superior to what I remembered from the active Isobariks I ran a few years earlier, (all bolt down electronics), the lack of “Oomph” was a disappointment.
Yeah guests would marvel at the huge sound emanating from the Kans, (helped along by a Koetsu Onyx on the LP12), but I had pretty high expectations of what a speaker should do.
I did try the Kans with a Krell KSA50 and some Stax class A monos. Then they gained scale. And for one glorious night I had a pair of Krell KMA100s driving the little beasts. That WAS sensational, although totally mad.
4 Likes
Sadly true!
The passive Kan crossover is 3rd order on the tweeter section and, from a voltage perspective, is 2nd order on the bass. Current-wise, which is what tends to count when looking at the impedance, it’s a 3rd order because of the inductance of the voice coil (0.44mH, from memory, though it might be a bit more with a more detailed model). It’s a number of years since I did a simulation of the Kans - and I’d have to dig out an old PC to look again - but I’m pretty certain the -3dB points were at 3.2kHz - and I seem to remember that was where the voltage responses crossed over as well. That’s in line with what’s supposed to happen with odd order Butterworths (ie. -3dB at crossover with the phase responses 90 degrees apart, as opposed to in phase at -6dB with the Linkwitz Rileys, which are always even order). There’s a fairly large time delay between the drivers - maybe 50uS, IIRC - which means they don’t add up quite as they should but I seem to remember that it was one of the better or more benign delays you could have. They are connected with the tweeter in inverse polarity giving a 1st order overall function.
2 Likes
Does that mean anyone running active Kans should connect power amp to tweeter in reverse phase, or have I misunderstood?
That’s a very interesting question. I wouldn’t have thought so, but at some point the manufacturers will have had to agree which of them is going to invert the phase. It would (might) require an extra stage in the active x-o so I’d have thought the obvious starting place would be at the speaker, meaning that the tweeter will already be connected in inverse phase. I’m afraid I don’t even know which way round Naim connect their tweeters and somehow have never asked. Time delays aside, two 3rd order Butterworths add up to the same flat response whichever way you connect them. 3rd Order Bessels don’t sum to Unity so I’d guess Naim pull them slightly apart to compensate for this (the error isn’t gross, anyway) and there will be a notional difference between each summing.
Having said that, I may be able to help here anyway because the differences in sound are quite easy to hear. The all +ve connection will first of all sound pretty OK, being quite immediate and maybe more realistic and less coloured. The other way round, with tweeter in inverse polarity, will likely sound a bit coloured or even phasey initially, but that’s the one with all the magic. Like many things Naim, that supposed colouration will evaporate almost immediately (you literally won’t hear it on the next record you play) and you’ll find there are so many more cues to it being real that you won’t want to lose that. Scale, too, should be very much better.
For a definitive answer it’s probably worth putting this question to Neil. One good reason for having a Naim answer is if anyone wants to convert passive Kans to Naim Active. While the internal cabling after the crossover to the drivers doesn’t generally matter much in a passive setup - the amplifier doesn’t really see the cable through all the crossover components - it will make a difference in an active system and it may well be worthwhile changing it to NACA5. So if anyone wants to do that it would be nice to know exactly how it should be done. FWIW (and it’ll be fun if I’m wrong), I’d be very surprised if they invert the phase in the Naxo 2-way. They might just do it on the 3-way - inverting the midrange in that case - but even then I’d be surprised.
Thanks Christian.
I was just curious - I had a rare pair of Linn-converted active Kan 1s but have sold them on now. Don’t know if @Igel or anyone else is interested or not. Most folks have done the passive-to-active conversion themselves.
1 Like
On Linn Kan 2 you need to
revert the phase on the tweeters when active.
2 Likes
I understand the Kans used stock Kef B110 drivers. I don’t know if you or anyone restoring Kans is aware, but Falcon Acoustics are now making the B110, my understanding being that it is to exactly the original spec (the original designer from Kef joined Falcon). - so if wanted you can have new.
Linn modified the KEF drive units (doping, drilling vent holes, etc.) so they won’t be exactly the same. I’m not sure if all of the mods are recorded or not. Also, they would only be applicable to Kan 1s prior to when Linn changed to the SP1057 unit in '85.
2 Likes