SpeakerTalk Forum Index SpeakerTalk
This forum has been set up to facilitate discussion of 1970s KEF speakers and drive units. The owner of the Forum has no connection with KEF Audio.
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Simulating the Kef Kube filters
Goto page 1, 2  Next
 
Post new topic   Reply to topic    SpeakerTalk Forum Index -> KEF speakers
View previous topic :: View next topic  
Author Message
theozaurus
Junior Contributor 20+


Joined: 05 Sep 2013
Posts: 21

PostPosted: Thu Jan 30, 2014 10:55 pm    Post subject: Simulating the Kef Kube filters Reply with quote

I've just received one of these: www.minidsp.com/products/minidsp-in-a-box/minidsp-2x4

It can be used for room correction which is quite interesting: http://www.minidsp.com/applications/acoustic-measurements/umik-1-setup-with-rew

However I'm really keen to simulate the Kef Kube so that I can get an idea if it's worth purchasing or building the original design.

The filter blocks I can play with or shown here: http://www.minidsp.com/images/documents/Product%20Brief-%202way%20Xover%20ADV%20plug-in.pdf

Does anybody have any information or the transforms for the original Kube?

Cheers

Theo
Back to top
View user's profile Send private message
speakerguru
VIP Contributor 750+


Joined: 18 Nov 2005
Posts: 958
Location: Green Hut, Tovil

PostPosted: Fri Jan 31, 2014 10:20 am    Post subject: Reply with quote

Andrew Jones and I did nearly all of the work on all of the Kubes. I know AJ has very little information so that just leaves me. Although the KEF Kubes were analogue op-amp designs, I did work with DSP audio eq chips when working on the Sony in-car systems at Ford, so I have worked in both domains.

Most audio DSP chips do not implement the full 2nd order biquad (Friend circuit) which was used in the KEF Kubes. They only allow you to generate variable-Q peaks and dips; very much a rock 'n roll soundman approach to eq, not audio engineering at all. All of which is a long-winded way of saying that you are better off working with analogue and op-amps rather then the superficially more glamorous programmable DSP chip for experimental DIY work.
Back to top
View user's profile Send private message
Garethwatkins
Junior Contributor 20+


Joined: 18 Nov 2011
Posts: 22
Location: Tring, UK

PostPosted: Fri Jan 31, 2014 10:13 pm    Post subject: Reply with quote

I have one of these Mini DSP units. I have used it as a basic eq tool for my Kef 104aB speakers, mainly seeing if I can extend the lower bass "Kube style". Of course the 104aB's were never designed with this abuse in mind but the results were quite amusing. Just had to watch the power a bit. The present job is as a 2 way crossover. The plan was to build a more compact pair than the 104's but with comparable bass extension. I raided the charity shop at the local dump and came away with a pair of old Wharfedale Lintons for 15 which would provide donor cabinets of the correct size. They sounded awful due to a non standard bass unit in one and a rubbing original in the other. Before swapping them out I tried the DSP crossover at 1800 Hz fed into a Quad 306. Still awful. I guessed there was a big peak around 2-3k, so I put in a suitable notch and wow, what an improvement! I have now got some decent drivers and the DSP is working nicely. All I have done is set the crossover freq at 1800 and put in some mild eq based on the manufacturer's published responses. I was thinking of bi-amping the 104's at some point but something tells me I won't improve on the original Xover, plus the fact that the new tweeter really does sound better than the T27 (sorry!)
Back to top
View user's profile Send private message
speakerguru
VIP Contributor 750+


Joined: 18 Nov 2005
Posts: 958
Location: Green Hut, Tovil

PostPosted: Sat Feb 01, 2014 11:54 am    Post subject: Reply with quote

Garethwatkins wrote:
I have one of these Mini DSP units. I have used it as a basic eq tool for my Kef 104aB speakers, mainly seeing if I can extend the lower bass "Kube style". Of course the 104aB's were never designed with this abuse in mind but the results were quite amusing.

Not a good idea to boost below the box frequency of a reflex system (resonance of the driver moving mass with the box compliance). You really are flogging a dead horse as the port and driver outputs become increasingly out of phase and cancel each other out. If you want to experiment with eq for a reflex box it's best to tune to the lowest frequency of interest and then eq for non-flatness. This can be verified by nearfield measurements (no chamber required). You can easily add mass to the 104 passive radiator by loading the steel tube with plasticene and lead shot.
Back to top
View user's profile Send private message
Garethwatkins
Junior Contributor 20+


Joined: 18 Nov 2011
Posts: 22
Location: Tring, UK

PostPosted: Sat Feb 01, 2014 1:41 pm    Post subject: Reply with quote

I agree absolutely Speakerguru. I was really just testing out the DSP. There is no way I am going to improve on a KEF design with my expertise.
Back to top
View user's profile Send private message
theozaurus
Junior Contributor 20+


Joined: 05 Sep 2013
Posts: 21

PostPosted: Sun Feb 02, 2014 12:53 pm    Post subject: Reply with quote

Wow, I didn't think I'd get to talk to the chap who designed it!

I think this DSP can simulate a second order biquad: http://www.minidsp.com/applications/advanced-tools/advanced-biquad-programming

Does the transfer function there look like the full friend circuit? Do you have any idea what parameters the biquads should be configured too?

Gareth: I've been thinking about using the minidsp as an active crossover for biamping also. Really encouraging to see other people playing with it here in a similar way.
Back to top
View user's profile Send private message
speakerguru
VIP Contributor 750+


Joined: 18 Nov 2005
Posts: 958
Location: Green Hut, Tovil

PostPosted: Sun Feb 02, 2014 10:51 pm    Post subject: Reply with quote

The answer is "yes but...."
If I have the transfer functions of the Kube you are interested in, I will willingly share that information, but it will be in the analogue domain. i.e. a transfer function in S, the Laplace operator, (S=sigma+j omega) or just the coefficients of S in the Biquads. To get to the z coefficients required by the DSP chip, you will have to do your own S to z transform, taking into account the sampling frequency you wish to use and the resolution of the digital filters. Bear in mind that when you want to do eq at low frequencies you need a very high resolution (large number of bits) or else suffer quite large shifts in filter f and Q. When you do eq at higher frequencies, then you may be surprised at how much the skew due to the zero response at the Nyquist frequency (= fs/2) impinges into the pass band. Digital is far from perfect. On the other hand if you sort your resistors and capacitors to better <1%, your analogue op-amp circuit will be within 0.1dB of the calculated response.......


Last edited by speakerguru on Mon Feb 03, 2014 9:53 am; edited 1 time in total
Back to top
View user's profile Send private message
theozaurus
Junior Contributor 20+


Joined: 05 Sep 2013
Posts: 21

PostPosted: Sun Feb 02, 2014 11:06 pm    Post subject: Reply with quote

It would be incredible if you could share the transfer function in S and I'll attempt to convert that (it may involve dusting off some old engineering books).

Long term it would be great to build something similar to the Kube using discrete components. I've been struggling to get all the software setup here to produce a nice PCB.
Back to top
View user's profile Send private message
speakerguru
VIP Contributor 750+


Joined: 18 Nov 2005
Posts: 958
Location: Green Hut, Tovil

PostPosted: Mon Feb 03, 2014 9:51 am    Post subject: Re: Simulating the Kef Kube filters Reply with quote

theozaurus wrote:
...However I'm really keen to simulate the Kef Kube ...

Which Kube?
Back to top
View user's profile Send private message
theozaurus
Junior Contributor 20+


Joined: 05 Sep 2013
Posts: 21

PostPosted: Mon Feb 03, 2014 1:22 pm    Post subject: Reply with quote

The Kef Kube 200 ideally.
Back to top
View user's profile Send private message
speakerguru
VIP Contributor 750+


Joined: 18 Nov 2005
Posts: 958
Location: Green Hut, Tovil

PostPosted: Mon Feb 03, 2014 3:14 pm    Post subject: Reply with quote

The full schematic for the KUBE 200 is available on this site.
As posted here http://www.hifiloudspeakers.info/speakertalk/viewtopic.php?t=1380&postdays=0&postorder=asc&highlight=kube+200&start=15 PostPosted: Fri Jan 04, 2013 11:58 am, I have the full pcb design for the Kube 200 and could get boards made. The set up costs for pcb are quite high so unless we get at least 10 boards made, it will not be economical.

I have also offered the software to extract the transfer functions from the component values of the friend circuits and vice versa. I don't recall anyone taking me up on this. Please remind me if you have.
Back to top
View user's profile Send private message
theozaurus
Junior Contributor 20+


Joined: 05 Sep 2013
Posts: 21

PostPosted: Mon Feb 03, 2014 11:42 pm    Post subject: Reply with quote

I'd be interested in a couple of PCB's for the Kube and could help get some quotes together if that would make things easier.

That software sounds very interesting. I've been looking through the schematics this evening. It appears the "BEC" section is very similar to the Friend circuit you describe, as shown here: http://www.globalspec.com/reference/71146/203279/4-3-general-biquad-circuit-using-a-single-op-amp

The contour topology looks a bit more complicated!

I attempted to cheat this evening and generate the bode plots of the circuit without really understanding whats going on, but the SPICE models have so far vexed me. I figured I could make the DSP match these plots easily enough.

As a slight aside, what does 'o/c' represent for some of the component values?
Back to top
View user's profile Send private message
speakerguru
VIP Contributor 750+


Joined: 18 Nov 2005
Posts: 958
Location: Green Hut, Tovil

PostPosted: Tue Feb 04, 2014 10:53 am    Post subject: Reply with quote

The quote I got in Jan 2013 (no longer valid) was 60 for tooling +229 for 10 boards. I'd guess that if we got you 2 boards and I got 2 for future needs then it would still come out to around 75 per board. I don't know about you but I don't have 150 that I can spare for the good of the KEF fan base. If I had to stock numbers then I would have to put a mark up on the boards to recover some of my time.

The KEF bass extending circuit (BEC) is exactly that Delyiannis-Friend Biquad which, unlike the 4 capacitor Linkwitz circuit, requires only 2 capacitors to realize the full second order biquad function.

The contour circuit is actually a simple 1st order shelf, but variable, which may make it look more complicated than it is.

o/c stands for open circuit (no component)
s/c stands for short circuit (zero Ohm link)
Back to top
View user's profile Send private message
theozaurus
Junior Contributor 20+


Joined: 05 Sep 2013
Posts: 21

PostPosted: Sun Feb 09, 2014 11:18 pm    Post subject: Reply with quote

Sorry for the delay.

Would it be possible to get a copy of the software that extracts the transfer function from the friend circuit? I've been trying to do it this evening based on the equations from http://www.globalspec.com/reference/71146/203279/4-3-general-biquad-circuit-using-a-single-op-amp but I'm not quite confident I've got the right equations.

In terms of the PCB it would be great to get some produced. I could try and get some quotes as well to see if we could reduce the price a bit further to make it more affordable. Is the layout similar to the original? In which case I guess it needs to be around 150mm x 230mm double sided board.

o/c, s/c - makes sense! Are the components still shown so that the filter looks like a standard design?
Back to top
View user's profile Send private message
theozaurus
Junior Contributor 20+


Joined: 05 Sep 2013
Posts: 21

PostPosted: Mon Feb 10, 2014 12:00 am    Post subject: Reply with quote

I've finally managed to simulate the BEC stage. Check it out at: http://www.partsim.com/simulator/#15762
Back to top
View user's profile Send private message
Display posts from previous:   
Post new topic   Reply to topic    SpeakerTalk Forum Index -> KEF speakers All times are GMT + 1 Hour
Goto page 1, 2  Next
Page 1 of 2

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group