Tannoy 15 inch 3839 passive to active conversion .
Like to invite you on your thoughts about converting my DIY Tannoys from passive to active .
Have been thinking about this for a long time . Although I’m quite happy with the sound of my system at the moment I think there must be more in it .I remember the sound of my Ardens’s back in the 80’s with a Lenco L75 and my Dynaco Pat5 and stereo 400 as more open and precise as my current set up with an Emotiva UMC-1 and a DIY 5 channel L15D Class D Amp .
I still own the Dynaco stereo 400 and it’s no match to the L15D’s . I discussed in other posts about the sync problem in stereo mode of the class D amps , but even then they sound much better .
Offcourse my ears worked better 30 years ago , but I like to challenge them .
My options are :
Hypex PSC2.400
MiniDSP ( which one ? ) with some new Amps like Connexelectronic
Anyone have heard these ?
Or if you have a better solution please let me know .
Cheers ,
Rens
Like to invite you on your thoughts about converting my DIY Tannoys from passive to active .
Have been thinking about this for a long time . Although I’m quite happy with the sound of my system at the moment I think there must be more in it .I remember the sound of my Ardens’s back in the 80’s with a Lenco L75 and my Dynaco Pat5 and stereo 400 as more open and precise as my current set up with an Emotiva UMC-1 and a DIY 5 channel L15D Class D Amp .
I still own the Dynaco stereo 400 and it’s no match to the L15D’s . I discussed in other posts about the sync problem in stereo mode of the class D amps , but even then they sound much better .
Offcourse my ears worked better 30 years ago , but I like to challenge them .
My options are :
Hypex PSC2.400
MiniDSP ( which one ? ) with some new Amps like Connexelectronic
Anyone have heard these ?
Or if you have a better solution please let me know .
Cheers ,
Rens
Which crossover do you use (schematic) ? What are the other details of the implementation ? Have you made some measurements ?
I think the right approach is to identify the problem first and then search for a solution.
I think the right approach is to identify the problem first and then search for a solution.
Charles Darwin posted some comments about older Tannoy dual concentrics. It is not possible to use simple electronic xover for such conversion. HF driver needs some eq to compensate for its colorations. Therefore HF section of el. xover needs that eq.
Google this:
Active crossover for Tannoy speakers
design considerations and suggestions
Google this:
Active crossover for Tannoy speakers
design considerations and suggestions
by John Ridley
No need to google:
tannoy
tannoy
Have a look at these to get an idea of what you are aiming for.
This is not specifically for K3839 but the principle remains the same for all 'pepperpot' Tannoys. The reason for the treble boost is that the horn behaves like a constant directivity one and they all need a 6dB/oct boost from some frequency on up (usually 5-6k). The need for the dip at 2-3k is due to loading issues IMO. In Tannoys passive xovers there is a notch filter which does dual duties: It modifies the slope created by a small cap into a shelf below 5-6k and removes horn honk due to horn-loading issues mentioned above.
I'm running my K3149 analogue active and deal with the above by means of an parametric equalizer but with a digital crossover it becomes quite easy.
The actual crossover point is where the woofer starts to beam at 90deg ie when one wavelength equals the cone diameter or around 800-1000Hz for the 15".
I use a standard 24dB L-R which works fine but Tannoys passives are effectively 12dB and have a certain overlap designed into them. The low pass which is a fairly simple affair (no Zobels or anything) but the series inductor and the parallel caps don't have the same -3dB points.
tannoy
tannoy
Have a look at these to get an idea of what you are aiming for.
This is not specifically for K3839 but the principle remains the same for all 'pepperpot' Tannoys. The reason for the treble boost is that the horn behaves like a constant directivity one and they all need a 6dB/oct boost from some frequency on up (usually 5-6k). The need for the dip at 2-3k is due to loading issues IMO. In Tannoys passive xovers there is a notch filter which does dual duties: It modifies the slope created by a small cap into a shelf below 5-6k and removes horn honk due to horn-loading issues mentioned above.
I'm running my K3149 analogue active and deal with the above by means of an parametric equalizer but with a digital crossover it becomes quite easy.
The actual crossover point is where the woofer starts to beam at 90deg ie when one wavelength equals the cone diameter or around 800-1000Hz for the 15".
I use a standard 24dB L-R which works fine but Tannoys passives are effectively 12dB and have a certain overlap designed into them. The low pass which is a fairly simple affair (no Zobels or anything) but the series inductor and the parallel caps don't have the same -3dB points.
Thanks Charles ,
I have read all the above , and we had some discussions before about the "sync source" time alignment , and you were right about that, after a couple of weeks after I installed those I took them out again , because it didn't sound right ! I use Phil Short's crossovers , and am quite happy with the sound of the speakers , but like to experiment with some DSP and high quality amps .
I made some in room measurements with REW and the behavior you discribe is very visible . ( Have to do the measurements again , due to Hard disk failure and lack of backup )
The auto EQ function in my Emotiva UMC-1 takes quite good care of the shortcomings of the speakers , but like to get some more control including time alignment .
That's why I like to go the DSP way and not the analog solution .
I am very interested in the Hypex PSC2.400 plate amp , so any feedback or knowledge about them is very welcome here .
Cheers ,
Rens
I have read all the above , and we had some discussions before about the "sync source" time alignment , and you were right about that, after a couple of weeks after I installed those I took them out again , because it didn't sound right ! I use Phil Short's crossovers , and am quite happy with the sound of the speakers , but like to experiment with some DSP and high quality amps .
I made some in room measurements with REW and the behavior you discribe is very visible . ( Have to do the measurements again , due to Hard disk failure and lack of backup )
The auto EQ function in my Emotiva UMC-1 takes quite good care of the shortcomings of the speakers , but like to get some more control including time alignment .
That's why I like to go the DSP way and not the analog solution .
I am very interested in the Hypex PSC2.400 plate amp , so any feedback or knowledge about them is very welcome here .
Cheers ,
Rens
I am in doubt how subjectively intrusive it would be to use 24db LR active xover without eq with pepperpot Tannoy? Would that be a huge coloration, unbearable to listen to?
Cheers Doctor!
Good to know that even I get something right every now and then.;-)
Time alignment shouldn't really be much of an issue with K series drivers. The physical difference between tweeter and woofer coils is essentially the thickness of the magnet.
It boils down to a few microseconds which to the best of my knowledge is below our ears timing resolution but if you can make it perfect why not?
I have no idea about the Hypex or any other class d amps though.
From what I read elsewhere Hypex UcDs sound better than IcePower but not as good as Hypex nCore. Might be worthwhile perusing our Class D amplifier section.
GroundSound looks quite interesting if you don't need a digital input:
Ground Sound
So do the db-Mark processors but I have no idea if there is distribution in your neck of the woods.
Good to know that even I get something right every now and then.;-)
Time alignment shouldn't really be much of an issue with K series drivers. The physical difference between tweeter and woofer coils is essentially the thickness of the magnet.
It boils down to a few microseconds which to the best of my knowledge is below our ears timing resolution but if you can make it perfect why not?
I have no idea about the Hypex or any other class d amps though.
From what I read elsewhere Hypex UcDs sound better than IcePower but not as good as Hypex nCore. Might be worthwhile perusing our Class D amplifier section.
GroundSound looks quite interesting if you don't need a digital input:
Ground Sound
So do the db-Mark processors but I have no idea if there is distribution in your neck of the woods.
I've tried.
It is pretty horrible: Nothing much above 5k but way too much in the 2-3k region, just where our ears are most sensitive. Honk Heaven!
As far as I know the tulip waveguided Tannoys do not require any eq and they should work fine with a simple 24L-R.
No, they do (output on axis falls towards high frequencies as expected).
Attached is a crossover simulation from a System 800 (passive) and System 800A (active)
made by our member dimitri. The voltage to the HF part is about -16 dB at the crossover point for a flat response.
Attachments
Thanks Georg.
So if I got that right they still need the cd horn compensation but not the notch filter?
So if I got that right they still need the cd horn compensation but not the notch filter?
If I remember well, with domestic 8" tulip dual concentrics, like model 609 (similar to those used in System800) this problem was solved in very simple way: one section of xover was first order, and the other section was second order, so overlap was good. Xover was asymetric.
Oh, I am wrong. I found leaflet for the Sixes series and see that both sections were first order. So, only suitable values of xover components were enough to compensate for peculiarities of drivers.
I also found Tannoy 611 test from German magazine HiFi Vision (March 1993) and it is obvious that first order low pass/first order high pass crossed at 2,5kHz resulted in a 4dB dip at 2,5kHz. No attempt was made to correct this aberration and yet the sound of these drivers was excellent. In fact I had a chance to hear both 609 and 611 with different amps and can say that the sound of Philips CD850/Musical Fidelity A100/Tannoy 609/Kimber 4TC was one of the best sounds I heard. Although it was 20 years ago I still remember how Eagles "Hell freezes over" sounded in my friend's listening room.
Oh, I am wrong. I found leaflet for the Sixes series and see that both sections were first order. So, only suitable values of xover components were enough to compensate for peculiarities of drivers.
I also found Tannoy 611 test from German magazine HiFi Vision (March 1993) and it is obvious that first order low pass/first order high pass crossed at 2,5kHz resulted in a 4dB dip at 2,5kHz. No attempt was made to correct this aberration and yet the sound of these drivers was excellent. In fact I had a chance to hear both 609 and 611 with different amps and can say that the sound of Philips CD850/Musical Fidelity A100/Tannoy 609/Kimber 4TC was one of the best sounds I heard. Although it was 20 years ago I still remember how Eagles "Hell freezes over" sounded in my friend's listening room.
Last edited:
Hi all,
at last I have two K3808 drivers in working condition !!!
Will try 24LR at 1000hz with CD compensation for HF drivers .
Right?
Cheers
at last I have two K3808 drivers in working condition !!!
Will try 24LR at 1000hz with CD compensation for HF drivers .
Right?
Cheers
You still need the notch filter at around 2.5-3k to avoid the horn honk.
If you are analogue try the passive one from the original xover or a parametric equalizer.
If you are analogue try the passive one from the original xover or a parametric equalizer.
The System 600 and 800 do not use a notch filter for HF but the 12inch System 1200
seems to have one (if this schematic is correct):
tannoy
The Dimension TD12 (paper cone, tulip waveguide) has a notch filter too (schematic attached).
It appears that the larger drivers (with lower crossover frequencies) still
employ a notch filter.
Attachments
@Charles Darwin,
thanks for quick replay. I will use a FunktionOneX0-1 loudspeakers management DSP ( made by XTA electronics ) so I will try a notch filter in 2.5-3k like you advise, what notch f. Q factor should try, as a starting point.....
cheers
thanks for quick replay. I will use a FunktionOneX0-1 loudspeakers management DSP ( made by XTA electronics ) so I will try a notch filter in 2.5-3k like you advise, what notch f. Q factor should try, as a starting point.....
cheers
The notch filter varies from driver to driver.
The best way to find a starting point would be to find the crossover schematics for the K3808 and calculate the corner frequencies of the cap'n'coil that make up the filter.
There is also a resistance involved which determines the notch depth.
On mine and going by the less than reliable markings on my analogue eq it's -3dB at 2.6kHz. Q is kinda difficult as it is marked with bell curves rather than numbers/values.
It's a little bit wider than centre position if that helps.
Best way is to measure it but by ear you can just attenuate until the honk disappears but if you notch to deep or wide you get into 'BBC-Dip' territory which does make everything overly 'smooth'.
Georg,
that is quite interesting. I've only delved into the functions of the LittleRedMonitor (K3149) xover. In those as with all pepperpot drivers the notch has two functions:
Modify the 6dB slope so you get a flat(ish) area from xover freq to about 5-6k when you need the 6dB cd horn correction and get rid of the honk.
When going active with those drivers both aspects have to addressed.
It is possible that with the tulip drivers only the former function is needed so when going active only the boost needs to be addressed.
I guess somebody would have to simulate those xovers to find out exactly what is going on but I'm lacking the software.
The best way to find a starting point would be to find the crossover schematics for the K3808 and calculate the corner frequencies of the cap'n'coil that make up the filter.
There is also a resistance involved which determines the notch depth.
On mine and going by the less than reliable markings on my analogue eq it's -3dB at 2.6kHz. Q is kinda difficult as it is marked with bell curves rather than numbers/values.
It's a little bit wider than centre position if that helps.
Best way is to measure it but by ear you can just attenuate until the honk disappears but if you notch to deep or wide you get into 'BBC-Dip' territory which does make everything overly 'smooth'.
Georg,
that is quite interesting. I've only delved into the functions of the LittleRedMonitor (K3149) xover. In those as with all pepperpot drivers the notch has two functions:
Modify the 6dB slope so you get a flat(ish) area from xover freq to about 5-6k when you need the 6dB cd horn correction and get rid of the honk.
When going active with those drivers both aspects have to addressed.
It is possible that with the tulip drivers only the former function is needed so when going active only the boost needs to be addressed.
I guess somebody would have to simulate those xovers to find out exactly what is going on but I'm lacking the software.
Tannoy Puma has a simple 6db high pass, 3.3uF + 2R2 serial and 22R // .
No notch at all......only a plain 6db high pass centered at 6200 Hz approx., very easy to replicate with a DSP, but I would like to lower X-over frequency to around 1Kz, using 24db LR high pass.
Will play a little with DSP to find notch filter using your reference settings, may a simple equalization will be sufficient.
Now, i need to replace the acoustic damping inside loudspeakers and make new connectors panels with speak-on 4p female to feed drivers separately and then start listening.....
I hope i will share my results in a couple of days.
cheers.
No notch at all......only a plain 6db high pass centered at 6200 Hz approx., very easy to replicate with a DSP, but I would like to lower X-over frequency to around 1Kz, using 24db LR high pass.
Will play a little with DSP to find notch filter using your reference settings, may a simple equalization will be sufficient.
Now, i need to replace the acoustic damping inside loudspeakers and make new connectors panels with speak-on 4p female to feed drivers separately and then start listening.....
I hope i will share my results in a couple of days.
cheers.
Great to hear you got it working again Marco ,
Was wondering if you are satisfied with the lack of bass with the Puma ,
I know they sound pretty "Punchy" as they are , and really "club sound" orientated ( And they did a pretty good job at Tannoy with this one in marketing ! ) but there's no lower bass ! so , subwoofer or reconing with heavier cone ?
Cheers , my friend
Rens
Hola doctordata !!
Have not listened Pumas in standard version too long, but I am confident that, " forcing " slightly K3808·s woofers with DSP, will have good results, if not last world in bass extension a good impact .....
Anyway I have a B&W CM10" ( 500w. ice powered...) to help Tannoy's bass, may will cross them in 40-55Hz range.
Today have made the Speak-on plates ( 1+.1- woofer 2+,2- HF, like we use in PRO installation )
Only need some stuff to put inside boxes ( acoustic absorber ), to begin to play with DSP and listen some MUSIC....at last!!
Will share my finding with DSP settings of course !
Hasta pronto !!!
Have not listened Pumas in standard version too long, but I am confident that, " forcing " slightly K3808·s woofers with DSP, will have good results, if not last world in bass extension a good impact .....
Anyway I have a B&W CM10" ( 500w. ice powered...) to help Tannoy's bass, may will cross them in 40-55Hz range.
Today have made the Speak-on plates ( 1+.1- woofer 2+,2- HF, like we use in PRO installation )
Only need some stuff to put inside boxes ( acoustic absorber ), to begin to play with DSP and listen some MUSIC....at last!!
Will share my finding with DSP settings of course !
Hasta pronto !!!
regarding 3808/9 , after all these years , they're all prone to same problem - increased Fs
read here : Tannoy K3809 – hardened cloth surround issue – solved | Zen Mod Blog
read here : Tannoy K3809 – hardened cloth surround issue – solved | Zen Mod Blog
- Status
- Not open for further replies.