It seems to me that the whole Exakt controversy has leaked all over the place so I thought I’d try to bring it back here to the thread that was created for it. I have been watching the comments here and there about Exakt and what it is supposed to do. And also about the idea of applying it to the JBL 3677s that I and others here own. I have noticed a fair bit of misinformation and also what seems like a definite agenda on the parts of some people. This is characterized by people pushing the idea of Exact 3677s who don’t own those speakers, telling people they shouldn’t spend their money on JBLs without hearing them (always a good idea but one I didn’t personally follow because of my trust in the ears of the several members who recommended them to me, and the relatively modest investment if I didn’t like them) yet telling people they should chip in to fund JBL Exakt filters without hearing what the result would be, and claiming that Exakt will make their JBLs “perfect” or producing a “linear phase speaker”. Side question: Can you effectively crowd fund a project with a crowd of one?
Now I will admit that living in the US under the current (and hopefully short-lived) administration I have become extremely sensitive to lies and exaggerations. So you’ll have to forgive me if I’m a little harsh but there is no such thing as a perfect speaker nor will there ever be. Nor is there a true linear phase speaker and I don’t think such a thing is likely within the lifetime of most of us. Probably the closest things to a linear phase speaker was the Hill Plasmatronics but only down to about 700Hz where it crossed over to the midbass and bass drivers. The Vandersteen Model 7 is probably the modern speaker that comes closest but still is certainly not perfect (although it seemed pretty good the couple of times I heard it – unfortunately not with my choice of sources or electronics). Some full range electrostatics might also come close except that “full-range pure electrostatic” is pretty much a contradiction in terms.
There are many problems with trying to get a truly linear phase speaker but it mostly comes down to the drivers themselves - although the cabinet assuredly has an important influence too, along with the crossovers, of course. But I thought I would highlight a couple of important problems that keep current speakers from being truly linear phase transducers. Interestingly enough the first problem was brought to light on the Linn forum by an informative question from linnrd and an equally informative answer from Phil Budd (Philbo), who I applaud for his straightforward reply. Here it is (only edited to correct a couple of spelling errors):
Since there are driver anomalies that are amplitude based I was wondering if Exakt systems are meant to compensate for these at a certain SPL, or within a range of SPL.
For example, where the voice coil heats up and changes properties there will be deviations. Can you provide any information at what transient peak SPL this might arise (e.g. 105dB, 100dB, 115dB, or...) and whether the Exakt model compensates for that. If not, below what point should one have the system in order to avoid this.
Also, below a certain level one should expect a greater recalcitrance for the bass drivers to move (higher mass) and a similar question as to what a minimum SPL ought to be for linearity.
In short, is there a minimum, maximum and/or ideal SPL to gain the full benefit of the Exakt compensation of signal level-based deviations of drive units from ideal behavior.
interesting set of questions.
The simple answer is that we tend to design between volumes of 60 and 70 on the Linn volume scale (-10 to -20 dB through the pre-amp). After this we will check the speakers at more moderate and more aggressive playback levels.
We measure the individual drive units for our Exakt systems at a similar level as they go through production.
We are aware that the current Exakt model does not vary depending on playback level, or more importantly, voice coil temperature. This will without doubt add variance to the playback spectrum of an Exakt speaker. We are investigating a few different ways we may be able to account for this variation in our future Exakt products.
As far as linearity with playback level is concerned much of this is still down to the drive units themselves. The motor and suspension systems in a drive unit are probably never entirely linear and will become more non-linear the harder you push them. This will always be a problem with an open loop system. We can try to predict the gross changes and model them but there will still remain some non-linearity. Only by closing the loop around the drive unit (such as our bass servo feedback system) can you start to eliminate non-linearity. Again, this is something we are fully aware of and are investigating ways we may be able to close the loop around both amplifier and drive unit to ensure increased linearity.
Hope this helps.
So as Phil clearly says “the motor and suspension of a drive unit are probably never entirely linear” and they can try to model “gross changes” “but there will still remain some non-linearity”. So clearly a truly linear phase speaker is not currently possible with Exakt and it gets more non-linear as you play it louder or softer. Apparently the more dynamic the music the less linear it will be. This is true of all speakers to one degree or another and it is not something Exakt fixes, or is likely to fix in the near future.
It is interesting to note that despite the importance of this discussion, at least to me, the third post was some of the usual oddball humor from a regular member, which was followed up by others and seems to have completely derailed the thread as it only lasted 9 posts.
The second problem is that the vast majority of speakers use bassmid, or midrange drivers that cover quite a range. Covering a wide range is good in that the fewer crossover points help lead to a more seamless driver blend as does keeping the crossover points out of the middle of the vocal range, where possible. However, it also means that the driver will be making large movements while reproducing frequencies at the low end of its range while also making small movements at the upper end of its range. These small movements will be modulated by the larger motions causing Doppler distortion – a form of phase distortion. This is made even worse when small drivers are equalized to go lower than they go naturally in an active crossover, Exakt or otherwise, as it makes the driver work even harder. So once again, no true linear phase. Not to mention that many, if not most, speakers today are ported and the frequencies coming out of the port will not be phase aligned with the woofer. Note also that the 5” to 7 or 8” drivers most popular for this purpose in modern speakers (in order to keep the cabinet slim) are particularly subject to this problem, as they need a quite large excursion to produce the lower bass frequencies.
These are only two examples of non-linearities in speakers that can’t be fixed by current Exakt designs, there are others such as resonances and edge reflections from cabinets, reflections from metal grilles and many grill fabrics, especially thick ones, and their frames, spacing of drivers which affects the phase of frequencies where they overlap, etc. And this doesn’t even deal with Space Optimization, which is a subject for a whole different discussion. With all this I feel pretty safe to say that a truly linear phase speaker will require technologies not yet created.
Now I want to say that I have no problem with Linn developing Exakt technology and it will be interesting to see where it goes. Indeed it has improved much since the first “lossless” implementations came out. On the other hand I am disappointed that they decided to really work only on that technology (with the exception of the LP12 and its associated parts) and to stop development and production of analog components. But what gets me the most is the oversimplification and claims about the technology that the marketing department has dreamed up and which tends to get parroted by the true believers. The linear phase thing was dealt with above but the “lossless” claims (I suppose Katalyst must be more lossless) and “the source is in the speaker” not to mention the early claims about Space Optimiztion allowing you to put the speakers anywhere in the room and get the same musical quality as in the optimum spots. It is notable that Linn has backed off of most of these claims and I notice in the current Exakt info on the website there is no mention of lossless or the source is in the speaker nor is there the cartoon video showing you how horribly distorted old analog sound is and how perfect Exakt is. So it is good to see a reduction in these types of claims.
As to the current implementations of Exakt I still don’t find them musically involving. I have listened to a number of different versions and so far they haven’t made me feel like I want them in the home. Why is this? Well for the most part I trust my ears and they just don’t create the stirring, emotional musical reproduction I have always looked for in my equipment. They have a balanced and detailed sound but not one that is engaging. In short I find them to be more Hi-Fi than musical, and if that is what you want, more power to you. But as a long time hobbyist I have always been interested in learning what I can about how things work, but only if my ears can verify it. If I can find an explanation that agrees with what I hear and gives repeatable results, such as using a precision torque driver to optimize components, I am happy to find that things work as they should. But if something works that has no logical explanation I don’t worry about it I just go with what sounds good. As far as why I am musically unmoved by Exakt my suspicion is that it is because of all the digital processing that is involved in the chain. In my experience the more stages in a music system are digital the less music gets through. So currently a well recorded, all analog recording played on my LP12 through my Sagatun Monos and Tundra Monos and into my JBLs gives me the most satisfying musical experience. A well recorded digital recording pressed on vinyl and played through the same system comes in second. A quality analog recording converted to digital and played through my RenewDS comes in third and a quality digital recording through the Renew DS comes in fourth. It is interesting to note that any signal that is not 24bit 192kHz when entering a DS, regardless of whether it comes in as analog or digital, has to go through four resamplings before it gets to the DAC and 24/192 data streams still go through three. As Jajo has mentioned elsewhere, each digital transformation is a source of lost information. Going through all those transformations is not doing the music any favors, not to mention all the transformations it goes through in recording, mixing and mastering, which can add several more resamplings. So this is one reason that Exakt might not thrill all of us.
Secondly, it has been my experience that everything a digital signal passes through has an effect on the musicality of the end result. Which Ethernet cable you go through, what direction it is connected in and what its length is all have an audible effect as does the switch you use, the power supply to the switch and even the polarity the power supply is plugged in with. The power cables plugged into the streamer and the server, etc, etc. As can be seen in the thread on the LS-NAS developed by Linnofil, every part of the server including HDs, RAM, motherboard, power supply, internal cables and even the fan also change the music you hear. While this importance of the equipment and cables feeding a DS is something Linn originally didn’t agree with, and may still not for the most part, they do seem to have backed away from this somewhat. At least they are saying that the power supply for the DAC and the quality of the signal going to the DAC is where some of the big improvements from Katalyst derive.
Something that relates to this, and which for me is a total non-starter, is the Digital Volume Control or DVC. I have yet to hear a DVC that does not lose musicality, tunefulness and involvement compared to a good analog volume control, much less to a really excellent one as in the Sagatuns and KK. It appears that Exakt technology is inextricably linked to DVCs. Although Linn was asked about providing Exakt crossover technology with the ability to handle an analog preamp input for volume control and input switching they responded that it was not something they intended to do. My guess is that it would be extremely difficult to get the Exakt system to function acceptably at all with the range of levels that an analog volume control at the input would encompass. Digital systems don’t seem to work well at low levels so dealing with the extreme dynamic of musical signals that could be presented with a 50 or 60dB volume window, on top of the dynamics in the music itself, would likely be overly taxing. Indeed they might need to immediately boost the signal to full level and then use a DVC to mimic the original volume level just to get it to work right. On the other hand I could be completely wrong on this as I am not an engineer. Meridian, for one, seems to have implemented an analog level control in their systems with digital crossover speakers. Indeed they mention specifically that they have done so due to the superiority of analog volume controls! (Interestingly they refer to themselves as the “Masters of Digital Signal Processing” and make similar claims about the phase alignment, etc., of their DSP active speaker systems as Linn does.) Regardless of this, until a truly musical and transparent DVC can be developed, in my mind there can be no Exakt implementation that does not break the hierarchy no matter how good the DACs and crossover implementation that follow the volume control.
A question that relates to this was from beck who wondered why it was so hard to get a digital source (system) to be truly musical yet not so difficult for an analog source (system). I had a theory at the time but didn’t have a chance to put it down, but I think it fits here. It must be remembered that a digital data stream is not music, it is merely a series of bits, 1s and 0s, that are timed and sequenced so as to represent music and to mimic the original analog musical signal once converted back to analog by a DAC. The digital signal is the music cut into little pieces, the higher the resolution of the ADC the more pieces there is for the DAC to convert back and the less granular it is. An analogy might be interesting here. You can take a potato, slice it thinly and deep fry it (or, heaven forbid, bake it) and you will have a potato chip. Or you can take a potato and chop it into tiny little bits, glue it all together and reform it and you have a Pringle! A Pringle looks kind of like a potato chip, except that they all have a certain sameness to them, and it sorta, kinda tastes like a potato chip, but not really much. Digital reproduction is sorta, kinda like analog reproduction but it too has a certain sameness to it. Admittedly it does keep getting better, and the best digital sources can actually provide an enjoyable musical experience to most of us, to a level they certainly couldn’t do even ten years ago, but in my experience it still doesn’t give the musical quality of good analog. So back to why digital is harder to make sound good than analog. My theory is that because digital has chopped the music up into pieces and represented it with digits, it loses some of the musical information and therefore has less actual musical information to start with than an analog recording. This makes it a more fragile thing. And then it goes through a lot of transformations and transportations on its way to your ears as discussed above, each one removing a little of that musical information. So it isn’t difficult to significantly damage a signal that has less information and is more fragile to begin with and there are a fair number of places where it can be damaged. Now analog has a number of places where it can be damaged as well but here is where another factor comes in. Analog signals are subject to analog distortion but those are distortions that our ears and brain are used to so they can overlook or listen around them more easily. In addition we have had over a century of experience refining analog reproduction to reduce the level and significance of analog distortions. On the other hand digital has brought different types and levels of distortion than what we have seen before. Some of these, like pre-ringing, are things that do not occur in nature and immediately trigger a WTF message in our brains. Others are just things like the fact that digital signals have the lowest distortion at high levels and their highest levels of distortion at low levels, whereas analog tends to have the lowest level of distortion at low levels and its highest level of distortion at the highest levels. Since the ear is most sensitive to distortion at low levels and less sensitive to it at high levels it dovetails nicely with the way analog works and not so well with the way digital works. So the combination of a digital signal with less information and more fragility, going through typically more transformations and with less benign distortion products is why I feel it is harder to get it to be as enjoyable as an equivalent, or even lesser specification, analog system.
So there you have it. Exakt is an interesting technology that has already done much evolving and will continue to evolve. Some people obviously like what it does and a number of others, like myself, still find it falls short in the areas of tonality, musical naturalness and emotional involvement. So I’m not particularly holding my breath to see what it would do with a pair of JBL 3677s but I will certainly watch with interest any research Music Lover might put into it.