Tannoy Precision 8D monitors crossover replacement

[pierrer]

Hi all,

Long message, I hope you wont be put off

The problem:

I use a pair of Tannoy Precision 8D active monitors in my studio, and they have both let me down recently. They are not built anymore but I am used to them so I dont really want to change monitors. Unfortunately none for sale at a price that I can afford at the moment. This is an emergency for me: I don't have speakers to work anymore so I need to solve this quickly.

Originally, their line inputvgoes through a "filter" board. The outputs of the "filter" board are then fed to an "amplifier" board. After testing, I identified the fault to be a fried amplifier card for one speaker, as well as both of those cards fried for the other. Neither Tannoy nor their distributors have replacement cards in stock anymore.

They are built with an 8" driver for the lower frequencies, and a smaller one in its center according to Tannoy's "dual concentric" technology plus a "Super Tweeter" (reproducing frequencies from 21 kHz up to about 50 kHz). Tweeter and Super Tweeter are both 1" drivers:

http://static.bhphoto.com/images/images345x345/Tannoy_8001_3392_PRECISION_8D_Nearfield_Monitor_382462.jpg

 

[pierrer]

Tecnical study:

I have very little knowledge about speaker technology so I did a bit of research and it seems that the way to get the best possible results from a pair of speakers it to bi-amp their drivers, feeding them with signal filtered through active crossovers. So I would like to do that.

I have ordered two of those amp cards from ClassD Audio, to amplify the low frequencies signal and the high frequencies signal sepperately:
http://classdaudio.com/cda-254-audio-amplifier-125w-x-2-8-ohm-250w-x-2-4-ohm.html

Now my problem is to design the replacement crossover filters. In Tannoy's original design, the input signal is supposed to be split by a crossover section at 2,2kHz. There is a set of 20 DIP switches on the "filter" board to modify the frequency response of the system as follows if required (see user manual screenshot attached)

No more info available in any of the documentation available nor directly from Tannoy.
So I fed white noise into the audio in (DIP switches set to flat response) and analysed the two outs of the "filter" board to know what order and type of filters were used.

 

[pierrer]

Here I have attached the spectrum analysis of the output signal assumed to be fed to the Tweeter (green spectrum). The blue spectrum is the signal sent to the board input.

There is no LP filtering applied there at all. And there is a weird "plateau" between 1.5 kHz and a further dip around 4 kHz.

 

[pierrer]

And here is attached the frequency response (green spectrum) of the signal fed to the Bass driver. Again, blue spectrum is what I sent to the board input.

Again, a surprising reading: unsteady frequency dip from 400 or 500 Hz up to the crossover frequency (supposedly 2.2kHz). And a LP curve of apparently around -20 dB/ octave

 

[pierrer]

I'm not sure what to make of those readings, I expected something easier to interpret.

The frequency response of both outputs doesnt seem quite as flat as I expected ; there is no High Pass filtering done on the signal fed to the HF driver, and the slope of the LP filter is surprising too (apparently 20dB/octave or so, as opposed to 24 dB recommended in the articles I read for active crossover?)

Looks like both signals are amplified, so i assume active crossovers are used – probably done with the six MC33079P chips present on the "filter" board. I think each them amplifies a band filtered signal to provide the control over frequency response of the system.

There was also a passive crossover on another board hidden inside the speaker, splitting the unfiltered output signal from the "filter" board between both of the 1" drivers. I have traced the schematic, you will find it attached.

I understand that C1 must provide the HP crossover filter required, in the form of  a passive -6dB/octave. So it seems the crossover between the 8" driver and the Tweeter would provide a funny frequency response around crossover as LP and HP slopes are not correlated.
And C2 must cut frequencies below what is reproduced by the Super Tweeter.


The questions:

Of all of this data, I wonder what is normal or not, and what results from intentional devellopers' choice hence required for optimal sound reproduction, and what is down to compromise.

I tend to think that I should use active crossover butterworth LP and HP filters with a -24 dB/octave slopes at 2.2 kHz, and hook the highpassed signal both directly to the tweeter and to a reproduction of the  passive filter to feed the Supertweeter. And I don't understand what is the function served by L1 and R1? Are there any precautions to take in terms of phase of the signals outputed by the crossovers?

What are your thoughts on all this, how do you understand the data and how would you advise me to procede?

 

[usekgb]

L1 and R1 form a zobel network, usually used to tame an impedance peak in the driver.

 

[pierrer]

Thanks usekgb, i'll lok into what zobel networks are.

Have you had a look at the spectrum screenshots ? what do you think of the dip at 4 kHz ? is it intended, or an issue tolerated by the Tannoy engineers ? What about the unsteady LP slope ?

 

[usekgb]

I really didn't get a chance to.

 

[ricardo]

A good active speaker will use the electronic xover to EQ the native frequency response of the units as well as separate the signals for Bass & Treble units.

The Bass xover response is exactly the sort of thing I expect.  The dip above 400Hz compensates for a rise in the Bass Unit.

I'm less sure about your measured Treble xover response.  There should be 'some' sort of HP roll-off in the electronics and this has to be matched with the Passive xover too.

Ask Tannoy for a Service/Tech manual and also show them what you measured.

BTW, passive xovers are also used to EQ the native response of the units.

 

[pierrer]

I have the "technical" manual - not quite a service manual... I've attached it here if you want to take a look. Tannoy wont disclose any more documentation, I have asked. But I will send them my measurements and post their reply here.

A good active speaker will use the electronic xover to EQ the native frequency response of the units as well as separate the signals for Bass & Treble units. The Bass xover response is exactly the sort of thing I expect.  The dip above 400Hz compensates for a rise in the Bass Unit.

So let me rephrase to make sure I understand: you reckon the dip at 400 Hz is intended and aiming to correct a non-linearity in the bass driver's frequency response ?

I'm less sure about your measured Treble xover response.  There should be 'some' sort of HP roll-off in the electronics and this has to be matched with the Passive xover too.

What do you mean?

 

[ricardo]

I have the "technical" manual - not quite a service manual... I've attached it here if you want to take a look. Tannoy wont disclose any more documentation, I have asked. But I will send them my measurements and post their reply here.

Alas it doesn't have any info that's useful to us.  It's really a User Manual

So let me rephrase to make sure I understand: you reckon the dip at 400 Hz is intended and aiming to correct a non-linearity in the bass driver's frequency response ?

Yes.  All good speaker designers do this ... including Alex Garner, Tech Director at Tannoy whom I worked with at Wharfedale in Jurassic times.

I'm less sure about your measured Treble xover response.  There should be 'some' sort of HP roll-off in the electronics and this has to be matched with the Passive xover too.

What do you mean?

You can do part of the xover passively and part of it active.  eg for a 18dB/8ve treble xover, you use 12dB/8ve passive and 6dB/8ve active.

But these sections need to be carefully matched and are NOT what naive design methods come up with.  The 2 sections also need to compensate (EQ) the response of the drive units.
_______________

The LF response you show in #3 is what I'd expect to see.

However, the HF response in #2 isn't.  You might be measuring it wrong but it could mean the HF filter on the filter board is wonky.  The response you show will overload early, make nasty noises and may damage the treble unit(s).  Send them your measurements and ask for comment.

 

[pierrer]

I did follow your advice Ricardo, and sent them my measurement to Tannoy. They confirmed what you said : both bass and treble signal spectrums show an intended equalisation on the designer's part, to compensate for response variations in the drivers.

I will ask if they agree to disclose the actual frequency response of the drivers... If they do, I can apply appropriate software filtering on my master send prior to entering the crossover section to compensate, and then go through those linkwitz-riley type crossovers :
http://archive.siliconchip.com.au/cms/A_30278/article.html

This way I can also apply a filter at 40 Hz to avoid subsonic overload of the woofer in the crossover section as well.

Would you see any potential issues in proceeding this way ?

 

[ricardo]

I will ask if they agree to disclose the actual frequency response of the drivers... If they do, I can apply appropriate software filtering on my master send prior to entering the crossover section to compensate, and then go through those linkwitz-riley type crossovers :
http://archive.siliconchip.com.au/cms/A_30278/article.html

This way I can also apply a filter at 40 Hz to avoid subsonic overload of the woofer in the crossover section as well.

Would you see any potential issues in proceeding this way ?

Yes.  The first is you MUST retain the passive xover on the treble units .. otherwise you stand a good chance of blowing them.

Secondly, by using the silicon chip xover, you are completely changing the design.  Rather than asking Tannoy for frequency response of the drivers, you are better getting some way to measure acoustic frequency response yourself [*].  You can then check if the complete system actually does what you want it to do.

Can you check that both filter boards give the same response on the treble channels?  If so, they MAY be OK.  They will be suboptimal but will work.

Speaker Design is VERY addictive so have fun.
____________

Alternatively, try & get the schematic of the filter board from them.  This will be a good starting point even if you go the full Speaker Design route.

[*]  Have a look at ARTA and also read the Clio manual on how to measure acoustic frequency response.

 

[pierrer]

one of the filter boards is fried so I have to trust the measurements done on the working one. Tannoy were unable to  provide the drivers frequency responses, and i cant make measurements of the whole system (no test room and gear).

However, they saw no problem with using active crossover for the tweeter signal (i think it makes sens as the tweeter out of their filter board show a gain applied to signal, hence an active signal path before the passive crossover board).

Also, they clarified that L1 and R1 simply make the filter for the Super Tweeter a second order. Nothing to do with Zobel networks.

I will build everything and see what it sounds like in the end... Thanks for your help Ricardo.