Tannoy 10" Monitor Gold
Copyright 2005 © Troels Gravesen


The Tannoy LSU/HF/III.LZ/8 drivers were made - to my knowledge - with two different chassis. This version with an extra wide chassis allowing the driver to fit in with a 12" front panel cut-out. The shortest diameter is 305 mm. The active cone diameter is 210 mm.
The rounded metal fillets holding the corrugated paper surround were slightly corroded, suggesting the driver has been kept in a rather humid environment. The metal fillets were removed, sanded clean and given a fresh black spray lacquer.
A potential problem with drivers kept in a humid climate is that the paper voice coil may have taken up moisture and cannot move freely in the magnet gap. Fortunately this was not the case with any of these drivers, but I have experienced the phenomenon with a JBL bass driver, LE125A, and I had to lift the voice coil out of the magnet gap, add some varnish to stabilise the paper and cut and sand off paper bobbles that would rub in the magnet gap. Quite a delicate operation.

To lower the resonance frequency, the surround was treated with xylene and given some 40 Hz sine waves for some hours.
I'd very much like to remove the dust cap but the glue used here is hard as rock and doesn't respond to xylene or acetone at all. The reason for using a solvent is that I'd like not just to cut off the dust caps, preferably keeping them in shape in case I want them reinstalled. The dust caps are in a rare perfect condition, but I don't like the idea of having the treble to pass a rather thick fabric cone. I think the dust caps provides some tweeter attenuation (later found out they don't!).

It's hard not to be caught by pure aesthetics when looking at this driver. Like some old cars, old houses, old furniture, a few designs turn out to be just right. We can look at these old objects and say: Yes, this is how it should look, this is how it should feel when you touch it, this is the colour it should have, this is exactly the proportions this product should have to render the beauty of its appearance in space. An add on shoes will mostly display size 38 (European), not the size 44 I use! Size 38 simply has the most appealing proportions and I think this 10" driver is the "38" of Tannoys.

Upgrading this crossover is a must. Getting rid of all the thin wires, rotary switches and replacing
the electrolytic caps by film capacitors is likely to improve sound - at least this is what
is reported by other Tannoy owners. The 1.35 mH coil for the LF units will have to go too.

The Original Crossover:

Crossover schematics.

Not surprisingly, the crossover is different from the 15" MGs and the 25 uF capacitor suggests a lower DC-resistance of the LF units and so it is: 4.8 ohms. Usually an 8 ohm driver has a DC-resistance of 5.5-6 ohm. The HF unit has a rather high DC-resistance of 9 ohms.

The tweeter section is basically a 2nd order filter made from 8 uF and the tapped coil.
The tweeter equalisation is made from the resistors and the 3.3 uF capacitor:
Position 1: 15 ohms series resistance is by-passed by 3.3 uF.
Position 2: 5 ohms series resistance and 10 ohms is by-passed by 3.3 uF.
Position 3: 15 ohms series resistance (capacitor not connected).
Position 4: 15 ohms series resistance and 3.3 uF to ground.
The notch filter: 3.3 uF + 0.62 mH + 12 ohms. I still have to measure the impact of the notch filter.

Frequency response "as is" after setting "level" and "roll-off" for flattest response in wide baffle cabinet.

Frequency response of both speakers compared to Rogers LS35A (green) for same attenuation (this is not at 2.8 volts).
Both drivers are flat in the critical midrange and up to 6.4 kHz, where both drivers have a spooky sharp dip. I don't know what causes this, but something is out of phase here. The dust cap?

Rodolphe from Australia has this comment:
Reading your article about Tannoy III LZ, a coaxial driver, they have dips ands bumps nearly 5khz and 10khz. Bernard, my favourite dealer, who is in love with coaxial drivers, explains this phase shift due to the speaker (rubber or foam) surround. This default disappears out of  the speaker axis. Your dip is very curious but I suppose it's depend of the surround geometry: multi fold not the usual half torus. You have two severe dips : 6.3Khz and 19KHz, but I suppose the sound is very spatial and focalized.
Thanks, Rodolphe. I'll try to do some more measurements, but it cannot be the surround as the compression driver still has this behaviour when I disconnect the bass driver. Furthermore, the bass driver doesn't say much at all at 6.3 kHz.
Even when I measure the compression driver's FR nearfield, I have the dip. So, it must be derived from the geometry of the horn in front of the driver's dome or the dome itself. As can be seen below here (blue graph = minimum phase) , there's an almost 180 deg. phase-shift at 6.3 kHz - like the dome simply starts playing counter-phase above this frequency, i.e. the voice coil travels one way, the dome the other way. Who knows?
The 15" MGs did not display this phenomenon and the compression drivers are quite similar in all these models.

Frequency response from crossov

Left: Frequency response from crossover version 1, shown later.
Right: Above measurements of IIILZ speakers and it looks as there are problems in the 6-8 kHz range here too.
The above illustration was taken from this site:
http://www.hilberink.nl/codehans/tannoy6.htmer version 1, shown later.

A closer look at the compression driver.
Tweeter measurements, the phase shift at 6300 Hz.

Left: What is seen above are measurements of tweeter response without crossover attached at various distances. Red = 100 cm, blue = 50 cm, green = 25 cm, yellow = 12.5 cm, purple = 6 cm.

Right: Here's the full picture of tweeter responses vs. distance to microphone. Black = nearfield, i.e. microphone as close as possible to the outlet of the phase plug. No phase shift when we measure nearfield! Now, that's something. Apparently the phase shift is generated in the horn in front of the phase plug. Yes, the dust cap has been removed and now the driver looks like this:

The MG IIILZ has a large voice coil of 2½" ~ 64 mm. The phase shift happens around 6300 Hz = 5.5 cm wave length. Not exactly the same as the diameter of the voice coil, rather the diameter of outlet of the horn inserted in front of the phase plug. There's a straight section of the voice coil in front of the horn and straight sided tubes usually mean trouble. Placing some Gaffa tape - loosely - at the plastic horn outlet to "smooth" the mouth opening to the paper cone this happens:

Red = tweeter response with crossover. Blue = minimum phase. Now, this is good news.. A few strips of tape in the horn mouth opening completely eliminates the total phase shift at 6300 Hz. The only problem is how this is implemented as a permanent solution. A short piece of horn must be made to fit into the
existing horn and make a smooth transition of the plastic horn to the paper cone. This is a longer-term project.

The simple tape experiment suggests an extension of the horn throat. Not an easy thing to make! Ideally it should look something like seen to the right. The insert could be glued to the original horn throat by a small amount of silicone glue. I have to talk to this guy at work who will make anything in glass, metal or plastic.....

Above is seen the impact of removing the dust cap. Bass and tweeter response with no crossover attached. Purple and green is without dust cap. This is a bit of a surprise. I had expected the heavy fabric dust cap to make quite a difference in tweeter response. Not so. Subjectively the removal of the dust cap makes a more "open" sound, but "open" may be a word I use because the speaker is more - well, open! I'm not sure a blind test would reveal any difference.

What would we Danes do without Carlsberg? The "shoulders" of a Carlsberg alu beer can fits perfectly into the horn-mouth and does the trick.

Impedance of both drivers. Very nice; Fs = 40 Hz.

Now, it may very well be that these old corrugated paper surrounds are one of the things that makes a real Monitor Gold, but in this case most of the compliance was derived from the very first - and quite worn - "wave" on the surround where the rest of the surround was fairly rigid. So, the first "wave" was given a thin coat of acrylic lacquer to stabilise the paper, which increased the Fs to some 60 Hz! This was definitely not my intention, but only proved that the first "wave" was taking too much of the heat and that compliance should more or less evenly distributed over the surround. The surround is simply a corrugated extension of the cone material, apparently having the same thickness as the cone itself and it is coated with some slightly sticky damping material. To lower the Fs again the surround was treated several times with xylene solvent to solubilise the lacquer and some of the lacquer was gently removed by a sponge. The xylene treatment took the Fs down to 42-44 Hz again and settled here after some 40 Hz "massage" for some hours while drying after the xylene treatment.

TS data
Quite some confusion here:
My initial measurements suggest the following: Fs = 40 Hz, Vas = 80 litre, Re = 5.75 ohms, Qt = 0.5-0.6, Qe = 0.65, Qm = 4.4, Mmd = 25 grams. I'll show the complete data-set once the drivers have been installed and been running for some time.
A Vas of 200 litre would require a very compliant suspension from a 10" driver and my drivers are far away from any resemblance to what is suggested below. Anyone made TS measurements on their IIILZ drivers??

Left: Above data taken from: http://www.hilberink.nl/codehans/tannoy38.htm. Right: My measurements. The differences seen here between declared data and measured data are significant, to put it mildly! Efficiency is pretty close, but that's about it. Declared data would suggest a horn as a suitable compartment for this driver, where my own data points to a closed cabinet or stuffed transmissionline. Suspensions may stiffen over time and account for some of the differences (Cms = 0.43 compared to 1.1 mm/N). I don't think the alnico magnet has weekened over time due to efficiency staying the same despite huge difference in BL factor, 11.2 vs 7.6.

Left: Test cabs with version 1 crossover. Hmm... That Tannoy sound! Even before finishing the crossover they deliver a level of transparency and midrange coherency, a speed and transient attack that few drivers can match.

Right: First time living room set-up. My wife had left a CD on the player: Værker af Hildegard von Bingen. Arvo Pärt: Te Deum. Akademisk Kor/Akademisk Orkester. Soprano:Marianne G Nielsen, conductor: Nenia Zenana. Why not? A quick look at the CD notes do not state where the recording was done...but obviously in a cathedral - and most likely here in Denmark. The recordings also reveal how difficult it is to get things quiet in a cathedral. You can hear cars passing by and doors being closed.

Now, church recordings and female vocals can be a tough test and the Tannoys perform remarkably well. The solos and choruses are portrayed with depth and a wide soundstage and all in all not a problem for the Tannoys. Yes, they can be played so loud that some midrange compression sets in, but you have to play really loud before the Tannoys give up.
Next CD is Uffe Steen: "Dust In My Coffee". A very bad thing! But Uffe has Lennard Ginman on bass and no other than Adam Nussbaum on drums to help with the coffee. This is a magnificent recording from Stunt Records and done in a studio on the small Danish island Samsoe in October 2003. The Tannoys are home here and I cannot think it much better. Having a 10" membrane doing the bass and the whole midrange without any points of crossover leaves most other speakers congested with poor speed and overall coherence.

So, what's playing? The version 1 crossover follows the original quite closely. Fortunately the bass section of both drivers appear to be very consistent in amplitude and phase response and 1.35 mH and 25 uF makes a 2nd order low-pass section. Due to the narrow (!) front baffle, only 38 cm, we have a bump around 750-850 Hz and we need to get rid of this to make an even and smooth midrange and if you think a notch filter takes the life out of a driver, think twice. 10 ohms + 10 uF + 4.7 mH does the trick and we have a smooth midrange.
Below the version 1 crossover is seen. The LspCAD prediction did not correlate too well with actual measurements, so inserted is the CLIO file from this crossover. The compression driver has a remarkable phase shift at 6.3 kHz, where there's a 180 deg. phase shift. 6300 Hz has a wave length of 5.5 cm and I'd be happy to have suggestions on what's the cause of this phase shift.
Phase tracking could be better and the dips and bumps in the 1000-2000 Hz region from the bass drivers makes it troublesome to model the crossover with a point of crossover around 1500 Hz.
There's a lot of LspCAD modelling ahead before the final crossover is ready. Some hours of modelling has left a number of new options for improving the overall response and phase tracking between drivers. More to come here.

Tannoy MG IIILZ crossover version 1.

Initially it was my intension to use the tapped coil for the tweeter section, but to be honest, it's much easier to leave it out. One reason for using a tapped coil is to omit the use of resistors in the signal path to the tweeter, but the original crossover has a number of resistors, so I not sure this tapped coil is such a big deal.

Left: Response of driver from V1 crossover.
Right: Response of individual drivers and summed response (yellow). 1/6 octave smoothing.

Tweaking of the driver itself is tempting: The basket of the Tannoy is anything but a modern spider chassis and damping of the inner surfaces might be tried.
Coating of the large membrane is definitely risky and I very reluctant to try anything here. The problem is that we cannot just order a new driver in case something is irreversibly damaged.

Left: Step response. Not particularly smooth, but tweeter and midbass peaks are well integrated.

Right: Tweeter response of both drivers. Quite good I should say, but this didn't mean that the final response from attaching the crossovers also made the same SPL. One tweeter had to be attenuated more than the other. The bump at 3 kHz is taken care of by the tweeter notch filter, almost a replica of the original notch filter.

Crossover, version 3.

I'll skip crossover version 2 and go right to version 3. After a week or so you start feeling at home with a speakers and can take a more objective view of the whole thing. Numerous CDs and LPs have been played and some were good, some were bad and a few were very bad. Oh yes, and a very few were so good that aural nirvana was accomplished. The latter may be a problem because some recordings may be balanced in such a way that an un-balanced speaker may go hand in hand with these recordings and make things sound even better than they were ever meant to do.
The version 1 crossover was targeted to give a quite flat frequency response and from experience it is know that this will leave a number of recordings on the shelf. And again, the BBC dip, prescribing a gradual 2 dB decline in amplitude from 100 Hz to 10 kHz, will make a more tolerant speaker and provide a more natural balance of sound.

The woofer's bump at 1500 Hz is a problem for making a perfect crossover, as the bump is right in the middle of where the point of crossover has to be. So, an additional notch filter was introduced here and made a nicer phase tracking between drivers, but this notch filter is optional and is arguably audible.
Modelling notch filters in LspCAD can be a problem. They never match exactly with real life, hence the correction in component values made in the LspCAD presentation.

The V3 crossover has been running for some three weeks after removing the dust cap and smoothing the horn mouth with some thin plastic and tape, eliminating the 6.3 kHz dip and phase shift. So, it was time to move on. Coating of a cone is a serious matter and there's no go back. You may seriously destroy the driver from the wrong type of coating and produce a frequency response that's hard to correct in the crossover. Coating materials comes in all varieties and only a very few are available to the diy people. Often the coatings offered from speaker shops are simply PVA glue and take great care when using this material. Usually they will add considerable weight to the cone and reduce efficiency. Generally I recommend not to coat your drivers unless you are able to measure what is going on and are willing to sacrifice a driver when things go wrong.

Very soft coatings may smooth frequency response but also reduce the ability to reveal detail where hard coatings may enhance resolution but produce new and serious cone break-ups.
The DAMAR coating used for the bass drivers in the 2.5 clone and SP38 did a good job but also proved to harden considerably and produce new cone break-ups, although in the very upper end above point of crossover.

So, the Tannoy cones were coated once with DAMAR and in the beginning produced a very nice and smooth frequency response in the 1-3 kHz region. However, when fully dry a dip at 1400 Hz appeared and getting rid of the break-up would take serious measures like making the cone into a sliced paper variant. The dip is hardly noticeable and I decided to leave it here because after further fine-tuning of the crossover, version 5, a nice overall frequency response was achieved and for the first time I could get a deep nulling effect from connecting the drivers with the same polarity: Green curve below = actual measurement.

The sound absolutely changed for the better from coating of these 40 years old and soft paper cones. Enhanced transparency, subjectively a more precise bass.

Blue = bass response. Yellow = tweeter response.
Red = summed response with tweeter inverted polarity. Green = same polarity.
Due to the coating the second notch filter for the bass driver could be left out.
Notice the treble dip at 6.3 kHz is gone. Quite a nice response from a compression driver + horn.

The treble from the compression driver and the horn is special. Some like it, some don't. For some people it's too different from the treble coming from a conventional dome tweeter. The treble is incredible fast due to the 100 dB sensitivity of the driver and horn and I think that the best way I can describe it is that we here listen to "treble" rather that a "tweeter". The treble is more directional than what comes from a dome tweeter and toeing in of the cabinets was more important than it use to be. Normally I prefer the on-axis to cross behind my head, but with the Tannoys I sometimes toe them in to cross in front of the listening position.

Tweaking the crossover of the Tannoys has certainly been one of the most difficult tasks ever. The bass driver's irregular response in the 1-2 kHz region for one thing. The compression drivers' broad bump at 2-4 kHz due to the horn loading. Had this bump only continued down to 1 kHz, things would have been easier. But this is probably as good as it gets.


Go to top

Back top intro page