Linn Forums

Current time: 2018-06-18, 19:53 Hello There, Guest! (LoginRegister)

Linn Forums / Linn Music Systems & Hi-fi Separates / Network Music Players & Music Streamers / Exakt v / Exakt Filter Variants

Post Reply 
Exakt Filter Variants
2017-08-21, 11:44
Post: #1
Exakt Filter Variants
Hi Folks,
thought I would take a few minutes to explain the filter variants, there have been quite a few questions on different threads.

timster posted a link to our Exakt version info:

(2017-08-19 06:33)timster Wrote:  If you want more information on the filters, there is a (slightly techy) description to be found here:

If you click "Exakt Version Info" it toggles between Basic and Advanced. Advanced has info on Compilers, Dimensions, Firmware etc.

It is interesting if only for the Notes bits.

For many this link will raise more questions than it gives answers, so I will start by clarifying the information contained therein.

Lets look at the oldest Exakt speaker, the Klimax 350, which we launched in September 2013.

On the Exakt Version Info page, click on linn, scroll down and click on Klimax350 and delve into 'Klimax Exakt Filter, Klimax Exakt 350, 1 channel'. Inside we see seven filter variants.
The keen eyed will notice that Version 1.0 (original design), is dated 16 Sep 2014, a year after launch. This was the date that the current web service was initiated, prior to this the data was hosted elsewhere.

Versions 2 and 3 are essentially the same design. Some changes were made between version 2 and 3 to the file format of the design and this was reflected by the change in version number.

The keen eyed will also note that there are two copies of Version 3, one dated very recently. This was required to be done in order that the current compiler would see version 3 as viable and would permit its selection and deployment in the latest Konfig.

For many speakers there is also a version 4 which is not available for selection in Konfig. This was essentially the same as the 'independent highest frequency modelling' filter. Changes were made to all design files to ensure they fell in line with the generic measurements stored in the cloud for the drive units employed by the filter. This was put in place to guard against cloud drop out. It shouldn't ever cause a problem, so version 5 can be considered as a 'belt and braces' replica of version 4.

So, that deals with the confusion in the dates and filter replication, now lets describe the actual filters...

In the original release of Exakt the drive unit models were rather inflexible. The roll-on and roll-off of each drive unit were modeled separately but could only be deployed to correct for both the magnitude and phase of the natural response, or do nothing. At the lowest frequencies there was very little we could do due to the very large processing requirements, so for most speakers we left the natural roll-on alone. Furthermore we didn't have a viable model for the roll-on of ported systems or capability to correct for any fixed internal electronic filtering.

It should be noted that the original release for many supported speakers is not available as a selection using the current version of Konfig. The file format is very different to what is used today and is not readily supported by the Exakt compiler.

The first and most sizable change to the Exakt filters happened in October 2015 when we introduced 'Independent drive unit modelling', often described on this forum as 'fourth order filters'. This change introduced the current file format used for all Exakt designs and allowed a huge increase in the processing flexibility of the Exakt engine. Each drive unit could now be modeled with a greater flexibility. Improvements included; ported system models (fourth order), support for fixed electronic filters (e.g. subwoofers), independent modelling of magnitude and phase or phase only for low frequencies, better processing by application of new windowing techniques (allowing extension of linear phase to the lowest frequencies), and probably a few more that I have forgotten...

More recently, around Christmas, 2016 we improved the model for the highest frequency unit, 'Independent highest frequency modelling', these quickly became known as the 'Christmas filters'. This allowed for extension of the linear phase characteristics of all speakers to the uppermost frequencies, without the risk of excessive magnitude changes.

In the last couple of months I have taken the opportunity to re-examine the Klimax 350. The basic filtering for this speaker had been untouched, apart from the changes due to 'Independent drive unit modelling' and 'Independent highest frequency modelling'. We have learnt a lot over the last three and a half years of what can be achieved with Exakt. I have also invested significant effort in improving our acoustic measurement capabilities (bought a very posh laser measurement system, built an anechoic chamber, written a new measurement suite) which gives us a much clearer picture of what the loudspeaker is doing. With Katalyst in the speaker we decided it was time to re-examine the 350. The results of this work are described as 'Extended drive unit modelling'. Previously we treated each unit in general terms, correcting for the roll-on and roll-off behaviour using a model based approach. The extended models treat any consistently measurable acoustic aberrations in the pass band of each drive unit while also adding more precision to the fit of the general models. The result is a response from the speaker which runs more closely to linear phase than before. Previous filters resulted in a phase response to within +/- 15 degrees of linear over the full range of the speaker. The extended models get closer still to a truly linear phase response.

A lot to digest, and very little in terms of how these changes will manifest themselves in the listening experience (I will leave that up to you to decide). Needless to say, we strongly believe that each step has improved the performance of our loudspeakers from both a measurement and listening perspective.

Comments welcome Wink

Find all posts by this user
Quote this message in a reply
2017-08-21, 11:55
Post: #2
RE: Exakt Filter Variants
Many thanks for this, Phil. This is just the kind of information we need to understand what we are doing when we are trying alternative filters (or, as in my case, deciding that we probably don't need to). Is there any chance that this thread could be made 'sticky' at the top of the sub-forum?


Main system: [Basik/Basik+/K5/Lejonklou Gaio >][Roksan Kandy Mk III >] KEDSM > Akurate Exaktbox 10/1 > Linn Silvers> A4200 x 2 and A2200 > K600 > Akubariks
Second system: Kiko
Find all posts by this user
Quote this message in a reply
2017-08-21, 12:31 (This post was last modified: 2017-08-21 12:33 by Donald.)
Post: #3
RE: Exakt Filter Variants
Thanks Phil - this is a first class response that helps to clarify a number of outstanding questions posted recently. Many of us on the Forum can't seem to to get enough of this stuff, not as an end in itself but as a means of better understanding/explaining what we are hearing as part of our listening experience! Smile

All you need to do now is run a Linn Home Katalyst Master Class - same as you did when Linn first introduced Exakt - a game changer for some of us. Big Grin This of course could be a webinar with the opportunity to view later - first in a Series?? Big Grin

Oh and the usual sign off - remember it's all about the Music so "Just Listen".

Unidisk SC
LP12 | Kore | Akito/Adikt | Trampolin 2 | Lingo 4.
Akurate DSM/2 | Katalyst Akurate Exaktbox 10 | 2 x Akurate 4200/1 | Exakt dual-mono Keilidhs
Raspberry Pi 2 Model B | IQAudio Pi-DAC+ | Audiolab 8000A | AR28s.
Find all posts by this user
Quote this message in a reply
2017-08-21, 12:38
Post: #4
RE: Exakt Filter Variants
Thanks Phil. Your explanations are very clear and very helpful. Has your re-examination of the 350 given rise to any learning that could benefit other speakers e.g. Akubariks?
Find all posts by this user
Quote this message in a reply
2017-08-21, 12:46 (This post was last modified: 2017-08-21 12:47 by Phil Budd (Philbo).)
Post: #5
RE: Exakt Filter Variants
(2017-08-21 12:38)drsean Wrote:  Has your re-examination of the 350 given rise to any learning that could benefit other speakers e.g. Akubariks?

Find all posts by this user
Quote this message in a reply
2017-08-21, 14:59
Post: #6
RE: Exakt Filter Variants
(2017-08-21 12:46)Phil Budd (Philbo) Wrote:  
(2017-08-21 12:38)drsean Wrote:  Has your re-examination of the 350 given rise to any learning that could benefit other speakers e.g. Akubariks?


Brill! Look forward to hearing if they ever get implemented Smile
Find all posts by this user
Quote this message in a reply
2017-08-21, 15:46
Post: #7
RE: Exakt Filter Variants
Thanks Phil,

This is excellent information, and helps to explain the future direction of the development of these Exakt filter modelling.

So to paraphrase, or this case, paraphase, we have:

I.D.U.M (Independent Drive Unit Modeling)

I.H.F.M (Independent High Frequency Modeling)

E.D.U.M (Extended Drive Unit Modeling)

I look forward to the next level of releases:

U.D.U.M (Universal Phase Unit Modeling)

W.E.A.L.L.D.U.M (Web-based Exakt Allocation Linear Leveling Drive Unit Modelling) Tongue

I can only guess that the next belt and braces version will be a belt, braces, and jockstrap version for extra support? Big Grin

Seriously, it sounds like great things ahead

Congratulations on you and the team, doing an incredible job of improving the sound quality of my speakers and all other Linn owners.

I look forward to every new release of filters, and have been blown away by the improvements in sound quality that I hear.

Exakt Surround 5.0:
Katalyst Akubariks, AEDSM, AkuKat Exaktbox 6/4200 for 225 Centre, Majik Exaktbox I for 212 Surrounds, OPPO 203 on StillPoints, Cat8
Power: Environmental Potentials 2460, Shunyata Denali, Zitron Cobras, Alpha Digitals
Room Acoustics: SPACE, SubDude HT's, Acoustic Panels
Find all posts by this user
Quote this message in a reply
2017-08-21, 16:48 (This post was last modified: 2017-08-21 16:52 by wickie8166.)
Post: #8
RE: Exakt Filter Variants
Thank you Philbo, I am looking forward to Filter 6.0 and the Extended drive unit modelling for Akubarik EXAKT (Katalysed or not ;-).

BR, Frank

LINN: LP12, EKOS 2, Krystal, Radikal, AEDSM, Akubarik Exakt, Karik III
POWER: MFE Power Conditioner NF-3, Goldkabel executive SuperCord Gold
NAS: Synology RS814 (8TB), Minimserver
SAT: VU+ Solo2 Twinreceiver HD
Find all posts by this user
Quote this message in a reply
2017-08-21, 17:58
Post: #9
RE: Exakt Filter Variants
Thanks Phil,

much appreciated.
What is interesting though that for the first time I don't agree with 'newer filters are better' and I know a number of people share a similar view. For me 3.0 is much more lively and fun than 6.0.
Any ideas what happened (and why)? I know this is a difficult question but e.g. have you experienced the need to re-adjust the gains or bandwith or anything of that nature with the new filter versions? Or is it just me having clogged ears Big Grin

Find all posts by this user
Quote this message in a reply
2017-08-21, 18:56
Post: #10
RE: Exakt Filter Variants
Thanks Phil

So is the reason that I only see 2 filter options on my 350s because I have upgraded to Katalyst and the original filters of 1-5 don't apply any more? Obviously I can't truly remember any more but I do remember being gutted when it went from 3.0 to the next one because it was a definite dip for me which couldn't be reclaimed by Space changes etc. I think it was the one before 3.0 somehow pushed something in the mid range right back but 3.0 just made everything sing. As I say I am on Katalyst 350 now and the last change was an uplift. Still think (for different reasons) there is some tweaking to do as I have lost those perfect female vocals but I do think that's a Space issue.

1 - independent highest frequency modelling


2 - Extended drive modelling
Find all posts by this user
Quote this message in a reply
Post Reply 

Forum Jump:

User(s) browsing this thread: 1 Guest(s)