Celestion Ditton 44 - HF 2000 tweeter replacement?

[Kat Manton]

Hi,

As may be gathered from a S&PW post, the main system in our living room now sounds woolly and the imaging has collapsed into a lop-sided mess. (I've no idea how some people fail to notice a blown tweeter, now it's happened to me it seems obvious!)

The Ditton 44 has the following drivers:

• Celestion HF 2000, 3/4" dome high-frequency.
• Celestion D5, 5" cone midrange.
• Celestion FC12, 12" bass.

Crossover points are given as 500 Hz and 5 kHz. So it strikes me that the midrange driver is doing most of the "work" and the character of these speakers is primarily governed by the bass/mid drivers covering everything from slowly-varying DC to 5 kHz.

The HF 2000 seems to be regarded as a little fragile, I've read reports of them failing for no real reason. In my case it was murder, not suicide; I found a dry joint in an MM preamp/equaliser which turned it into a high-frequency squarewave oscillator with near rail-to-rail output... on +/- 15V supplies... ouch..!

They're fragile, they die or get blown up; that means demand outstrips supply hence they're expensive.

So I'm tempted to replace both of them with something else, with changes to the crossover network to (hopefully) integrate the new drivers so it sounds like a system again.

Some posts on 'diyaudio.com' suggest that the SEAS H0737 19TFF1 is a possible replacement for the HF 2000 in the Ditton 66 (and may be sonically better than the HF 2000 and less easily destroyed.)

The 66 is the 'big brother' of the 44; using the HF 2000 with a different dome mid driver, the same bass driver (along with a PBR) and the same 500 Hz and 5 kHz crossover frequencies.

If the SEAS driver can be used in the Ditton 66 with alterations to the crossover, it should follow that it can be made to work in the 44, doesn't it?

I'm tempted to try it. But I know enough about loudspeaker design to know that I don't know enough about loudspeaker design...

I can't find any data for the HF 2000. But I have one working one and some test equipment and could measure some characteristics, if I'm told what to measure and how to measure it. I have infinite patience for this sort of thing and could produce datasets and graphs if that's useful.

I haven't found a schematic for the crossover, but tracing it out and drawing one should be possible if all else fails (though measuring the inductors might involve a little creativity; I haven't got anything which measures L directly.)

OK... so, do I 'go for it' and order a pair of SEAS H0737 19TFF1 drivers and hope it's possible to integrate them with a few adjustments to the crossover circuitry..?

...or try to outbid everyone else who wants one on the odd occasion an HF 2000 appears (then see if it's possible to live on beans/toast for a month or two)..?

...or another solution involving another driver? The SEAS one looks like it has a smooth and extended response and is a similar soft-dome type to the HF 2000, so it could be a good choice. And as the crossover frequency is quite high, it's really "filling in the top end" rather than operating in the band where hearing is most sensitive. I'd expect it to sound different; but if it sounds subjectively "good" or possibly "better" to me then I'll be happy.

I'm open to suggestions. I'm interested in loudspeaker design and this may present an opportunity to learn more. If the concensus is "get another HF 2000", please send any cheap meal suggestions via PM, though...

Regards, Kat

[threeseven]

Anythings possible even probable !
More recent drivers are always more robust and have higher power handling so it could be a good move.
The issues are going to be crossover roll off from mid to high, does the new driver have higher or lower output in that region so will the crossover need amendment to avoid bumps in the response. Overall output from the new driver, is it higher or lower than the replaced unit, so level adjustment may also be needed.
Then there is sound stage, how will the new driver integrate with the mid ?

Speaker manufacturers expend huge effort on design and whilst not impossible to equal and even improve on their products it certainly isn't easy !

BTW, I had a pair of 66's, very good but needed a solid floor due to the immense sub base output.

[mhennessy]

You're right to be afraid! There's a lot of talk about exotic cone and cabinet materials, magic cables and capacitors, etc, etc, but really, most of what you hear in a well-designed loudspeaker is the crossover design. Not a lot of people are prepared to believe that - to them, I'd say try designing your own speaker!

Obviously, you're already aware of many of the issues. Mechanics and originality would put many people off at the outset, but beyond that, you have issues with the tweeter sensitivity, impedance and response curve...

Some good news: the crossover point is rather high. This is good because the original tweeter is probably working well above its resonant frequency - and that's good because the "slope" probably comes entirely from the crossover filter. Many designs rely on the drivers natural roll-off to form a "pole" in the overall response - that's why some designs might be described as 4th order, yet there's only 2 Cs and 1 L in the filter - "3rd-order electrical, 4th-order acoustical" in the design parlance. Side-stepping this (an advantage of typical 3-way designs) should simplify the choice somewhat...

Sensitivity is likely to be higher with the newer model, so you'll need to build a "pad" so that the output of the crossover is presented with the correct impedance in the crossover region. The latter point is obviously important; I'm sure you realise that the impedance varies wildly with frequency, and that's yet another factor that speaker designers have to battle with. The impedance is greatest at Fs, and there's normally a minimum above this - but hopefully this is easier here because of the previous paragraph. You might already find resistors in the crossover, so it might be just a case of adapting their values.

Another factor is that the designers might have accounted for problems with the HF2000 in the design of the crossover. If it had a falling HF response, there might be a CR combination to boost the HF. Or perhaps and LRC trap to catch a peak. Need to get that schematic sketched out! BTW, get yourself a Marconi TF2700 - absolutely invaluable for this stuff

Measuring the result is surprisingly difficult. Sine wave sweeps don't work (standing waves), warble tones are better, but these days multitone signals and FFT analysis are the primary tools. The Behringer ECM8000 is a good, cheap measurement mic, and some of the software can be found as freeware, but you might end up having to use your ears (which would be the final arbiter anyway). If you do have access to a measurement setup, check the depth of the notch when you reverse the polarity of the tweeter - this is a classic measure of integration (prove that the phase of the filters are coinciding).

Finally, I'd recommend a look at this website: http://www.zaphaudio.com/ - John is really refreshing among the audio world, and I think you'll enjoy his straightforward approach. While there isn't an article that answers your question directly, you'll pick up a lot of the subtleties of the art from reading the project write-ups and other pages. I've built one of his designs, and was really impressed with it.

Good luck,

Mark

[roffe]

Found this somewhere..

[Kat Manton]

Hi,

OK, nothing's put me off and I'm 'going for it'

A particular concern with s/h HF2000 drivers is there's no guarantee that voice-coil continuity equates to a working driver. Somewhere on another forum I read of someone plotting impedance curves for four of them. No two were the same and only one appeared to be working correctly. Buying s/h tweeters could get expensive as I might have to obtain several of the things to get one which works. The "working" one I have might not be in the best of health, either.

Note that this isn't a "get them working as cheaply as possible" exercise. I'd like to get them working as well as they should work, if not better (while being slightly less worried about blowing them up again!)

I've already noted that some of the modern 19/20 mm soft-dome tweeters have good off-axis response and wide dispersion. Experience suggests that isn't true of the HF2000. The 44s have a very small 'sweet spot' (by pure coincidence it's at my end of the sofa) and, outside the room especially, "sound like hi-fi not music." (The pair of AR-6 upstairs can still sound uncannily like, say, a guy singing and playing acoustic guitar in the room, even when you're half-way down the stairs...)

All things considered, replacing the tweeters with something non-original seems like the best option. Non-originality may compromise their value, but I've had Ditton 44s for two decades. I replaced my first pair of early black-baffle ones with later veneered-baffle ones as they looked nicer but sounded the same. It's safe to say that whatever other loudspeakers I may end up with, I'm hanging on to these!

Thanks, roffe, for the circuit. Later I'll be disappearing inside mine and will trace out the circuit. Probably worth doing as mine are the later veneered-baffle version, so the odd change may have crept in (though I'd be surprised as all three drivers are the same and there's no sonic difference.)

I guess it would be useful to plot an impedance curve for the working HF2000, too. I have a Sound Technology ST1700B test set (audio generator, millivolt meter, distortion analyser), two good bench DMMs and one of my 'scope collection works. I've also got some nice Russian military-spec high-wattage 1% resistors, which might come in handy.

I also have a Klark-Teknik DN60 real-time analyser (which also contains a noise source) and a reference microphone (this was rescued from a bin at work. Fault? A wire had come off the XLR connector.) What I haven't got is either any idea what the microphone is (it may be a Calrec; it should be a high-quality one at any rate, given where the bin was) or the response chart which should've come with it.

I suppose a probably-good but unidentified reference microphone with no response chart is better than none at all. I'd expect the response to be fairly flat. If I'm looking for large lumps and dips and making relative, rather than absolute, measurements, I guess it'll do?

At the moment, two possible candidate tweeters stand out. One is the SEAS 'Prestige' H0737-08 19TFF 1 I previously mentioned. The other is the Hiquphon OWI.

Mechanically, I don't see anything particularly troublesome. The recess in the Ditton 44 baffle is ~4" (102 mm) diameter. Both the above drivers have slightly smaller ~94 mm flanges which would leave a 4 mm gap (a length of 4 mm square-section rubber drive-belt material would centre the driver and improve the appearance.)

The HF2000 flange has no holes, it's held in by three clamps around the periphery of the driver. These could be retained, removing the need for filling holes in the baffle. They reach more than far enough in to hold a 94 mm flange. The only modification which might be needed is slight enlargement of the cut-out for the rear of the new driver.

As I'm entering new territory here, I think I might start with the SEAS driver (as they're relatively inexpensive.) If I can sucessfully integrate those, I figure I'll have learned enough to do the same with the Hiquphon, if I feel like taking things a stage further. (Or would the lower Fs of the Hiquphon, 850 Hz, vs. 1700 Hz for the SEAS, make things easier..?)

Well, I've wanted to learn a bit more about loudspeaker design. Project 'Super Ditton' is go..!

Cheers, Kat

[radiotircoed]

Interested to hear how you get on with this Kat. I have a pair of 66's in store awaiting replacements. I've been on the lookout for replacement HF 2000's but as you say they are like rocking horse s..t. They do seem to be fragile too...

Bob W

[jamesperrett]

I'd be interested to know why you are looking at the 19mm SEAS rather than the 25mm one which, at first glance, might appear to be more similar to the HF2000. I had a system that used an old version of the 19mm unit where the high frequencies seemed a little indistinct - a bit like tape saturation. I replaced it with a 25mm version and changed the crossover at the same time which resulted in what I perceived to be a slightly clearer sound. Unfortunately, the reason I replaced the tweeters was because the old ones were damaged so I wasn't able to do a proper side by side comparison.

James.

[Kat Manton]

Hi James,

Originally Posted by jamesperrett

I'd be interested to know why you are looking at the 19mm SEAS rather than the 25mm one which, at first glance, might appear to be more similar to the HF2000.

Good question

When I started thinking about replacing the HF2000, I identified two fairly distinct approaches to the problem:

1. Identify a tweeter which is as close a match as possible to the HF2000 but which I can buy new.
2. Identify a tweeter which will work well in this specific system (and which I can buy new.)

With another system, the two approaches might produce the same answer. With the Ditton 44, they don't. It's a 3-way system with the mid/treble crossover frequency at a fairly high 5 kHz.

Smaller tweeters generally provide wider dispersion of higher frequencies but can't be crossed over as low as larger ones. So I might choose a 25 mm tweeter in a 2-way design so the crossover frequency can be selected at a point where the bass/mid driver is still working well.

But the Ditton 44 is a 3-way system with the tweeter crossed over at 5 kHz. A 19 mm tweeter appears to be a better choice than a 25 mm one, in this system (but maybe not in other systems which use the HF2000 but cross over at a lower frequency.)

Generalising somewhat from comparing curves and specifications, what I'd "gain" by fitting a 25 mm tweeter over a 19 mm one is poorer off-axis response, a more 'ragged' response over 10 kHz, less extension over 20 kHz and possibly worse harmonic distortion.

As long as I don't buy loads of s/h HF2000 and test them until I find a pair which match and work properly, I'm "sacrificing originality". Why not "sacrifice" the odd original design compromise while I'm at it?

Is "poor off-axis response" an important feature of the Ditton 44 "character" which I should strive to retain..?

(Incidentally, I'm considering the Hiquphon OWI as the final replacement; the SEAS driver is a cheaper way of gaining some experience. If I find I can't integrate the SEAS driver, I haven't spent as much money on a failed experiment.)

Cheers, Kat

[geofy]

I would have gone for a KEF T27 (or T52 as fitted to my superb KEF 103's).

[valvekits]

I replaced the tweeters in my 44's about five years ago with a round 110mm silk dome tweeter from Skytronic. It was necessary to to slightly enlarge the mounting recess, but this was fairly easy with a router. I can't say that I lost any "Hi-Fi ness" from using these cheap and cheerful tweeters but I did then go on to re-cap the cross-overs which did improve the mid-range to my ears anyway.

Eddie

[Kat Manton]

Hi,

Originally Posted by geofy

I would have gone for a KEF T27 (or T52 as fitted to my superb KEF 103's).

Neither are available new and I'm trying to avoid buying s/h. They are a bit more common s/h than the HF2000, so I've looked at some data for them.

Conclusion? Neither are sensitive enough.

I don't know exactly what the sensitivity of the HF2000 is, but a suggested "drop-in" replacement for the HF2000 is the Coles 4001. This has a sensitivity of 87dB. So it's reasonable to assume that the HF2000 is similar.

The T27 can be ruled out as 80dB sensitivity is nowhere near. (It won't fit without enlarging the recess either. 108 mm > 102 mm.)

The T52 can also be ruled out as 84dB sensitivity isn't enough. It's also a 39mm dome tweeter so I'm going to assume that off-axis response and extension above 20kHz are worse than the HF2000 unless someone can prove otherwise. (It also won't fit either, without some creative filling.)

The SEAS (88dB) and Hiquphon OWI (87dB) would seem to be a closer match for the HF2000 in terms of sensitivity (if we assume it's similar to the Coles 4001) than either of the KEF drivers.

(I could be tempted to acquire a pair of KEF T27 to use in something else. It was used in the LS3/5a and I'd love a pair, so the obvious thing to do with them is build my own LS3/5a clone. But not yet!)

Originally Posted by valvekits

I replaced the tweeters in my 44's about five years ago with a round 110mm silk dome tweeter from Skytronic..

Crossed off the list, not because it's "too cheap" but because I don't want to enlarge the recess in the baffle; and I haven't been able to find impedance/response curves for it within 5 minutes. (If you can provide a link to a datasheet, I'll take a look. Sensitivity at 95dB is way more than enough. If you left the crossovers unchanged, don't you find them a little bright?)

Cheers, Kat

[geofy]

The KEF speakers are less sensitive than the Celestions, requiring more amplifier power but the trade off is higher sound quality, without trying to be controversial. The KEF T52 gives a very natural top end and at normal listening distance are very detailed without being harsh or bright.

Some of the SEAS are very sensitive which would suit the Celestion higher efficiency drivers better.

[Kat Manton]

Hi,

I've just remembered that I have a configurable active crossover...

This is fairly ancient but 'pro' hardware; has 4 inputs, 8 outputs , 'phase rev' and +/- 6 dB gain trim on all outputs, '0-180º adj' trimmers on all bar the 'hi' outputs, and can be internally jumpered as 4 x 2-way, 2 x 3-way and 2 x 4-way.

'Configurable' equates to looking up R and C values in a table in the manual, pulling all the filter boards and soldering the bits in. Filters are 24 dB/octave Linkwitz-Riley (though I have schematics and could probably figure out other responses/slopes.)

I'm wondering if I should dig out more amplifiers and play around with this...

One (daft?) idea was to configure the crossover for 2-way, crossed over at 5 kHz and feed the bass/mid drivers from one output via their original crossover circuit and the new tweeters from the other. But I guess the original mid crossover slope at 5 kHz plus the 24 dB/octave slope from the crossover would equate to... something else.

Or just run the lot from three stereo amplifiers via the active crossover.

It'd still be nice to get the original passive crossover tweaked for the new drivers (and it'd mean fewer boxes piled up in the living room...)

Something to ponder over, methinks. Active Ditton 44 system? Might be interesting...

Cheers, Kat

[Kat Manton]

Hi,

I've extracted the crossover. The circuit roffe posted above (post #4) is essentially what I've got in my later veneered-baffle 44 (just made up in a slightly different way.)

Tweeter section:

• First capacitor (input side), 4.0 µF (2 x 1.5 µF and 1 x 1.0 µF)
• Second capacitor (tweeter side), 6.0 µF (4 x 1.5 µF)
• Inductor: 0.14 mH

The capacitors are 250V DC, Erie type M312.

The tweeter section (only) is the same as that used for at least one version of the Ditton 66. I found an 'official' diagram, from Celestion, for the 66. This shows a 0.14 mH inductor with part number SP1163 in the tweeter section. The inductor on my crossover has '1163' marked on it.

So I've got one known inductor. None of the other inductor part numbers on my crossover match the 66 diagram (I'll attempt to determine them presently.)

At least we know the component values in the tweeter section with some certainty now.

I've also ordered a pair of the SEAS H0737-08 19TFF 1 tweeters (I could easily have spent weeks looking at various tweeters; better to make a decision and get on with it!)

These came to just under £50 including P&P/VAT, which I think is reasonable considering a pair of 44 typically fetch £100 to £200 and I've seen a pair of HF2000 tweeters fetch £120 (and IMO the HF2000 isn't a £60 tweeter...)

Cheers, Kat

[Kat Manton]

Hi,

Attached, after much repeated measurement of both crossovers; one schematic for the Ditton 44

No curves yet, but I've made some quick/rough measurements of the drivers with them perched on my bench aimed upwards. Not exactly 'free air'...

• Bass: resistance 4.0 ohm; res. freq. 26 Hz; Z @ 500Hz 6.4 ohm
• Mid: resistance 4.8 ohm; res. freq. 55 Hz; Z @ 500 Hz 5.3 ohm; Z @ 5000 Hz 18.4 ohm (marked '6 ohm')
• Treble: resistance 9.0 ohm; res. freq. 1678 Hz; Z @ 5000 Hz 10.3 ohm

Hmm... something looks a bit odd...

HF2000 measurements posted elsewhere on the 'net suggest my "working" tweeter is also duff, just not as duff as the o/c one. Resistance and impedance at 5 kHz should be half what I measured.

I suspect that both tweeters were deteriorating and it's remained unnoticed as there were no sudden changes. (I'm even less bothered about acquiring the "correct" tweeters and glad I didn't try to find a replacement, now.)

[Kat Manton]

This is an ex-tweeter. It has ceased to be. (Etc.)

[Kat Manton]

Hi,

Well, the new tweeters still haven't arrived and I'm bored, so I got the camera out. I suspect not everyone is familiar with the Ditton 44 and those who are might not have seen what's inside (or dissected their tweeters!)

So, what is a Celestion Ditton 44?



Introduced in 1971, the Ditton 44 is a large (30" x 14.5" x 10") 3-way "monitor class" loudspeaker with a "substantially level response from 30 Hz to 30 kHz" and "exceptional transient performance, superb controlled bass, accurate mid-range and smooth extended highs". "Distortion is below 1% at bass and middle frequencies, and below 1.5% above 4 kHz at normal listening levels."

They're a bit good then? I think so. My eternal curse is to have heard master tapes played on studio control-room main monitors. That's led me to attempt to put together a hi-fi system which gets me as close to what the engineer heard in the control room, within a limited budget. The Ditton 44 goes a long way towards that without spending a fortune on control-room main monitors (not the little ones you see perched on top of the console, the big ones built into the walls.)

What's inside?



Starting at the bottom, the Celestion FC12, a 12" bass driver using a doped paper cone and rubber suspension. That may be a little old-fashioned these days, but the large magnet and cast alloy frame 'tick the right boxes' and it seems to work. The (43.47 litre) enclosure is sealed, so though the quoted response is "from 30 Hz", the low frequencies roll off more gradually than with a ported enclosure and I (and possibly the neighbours) can attest that low-frequency response is more than adequate.



Moving up, we find a 5" doped paper cone, rubber suspension mid-range driver which Celestion appear to have referred to as FC6 and D5. Again, arguably old-fashioned these days, but a large magnet ("50,000 Maxwell") and cast alloy frame 'tick the right boxes'. Given the crossover frequencies are 500 Hz and 5 kHz, this driver is producing most of the sound. The cabinet incorporates "transmission line rear loading" according to Celestion advertising. This consists of a 154 mm square box, 159 mm deep, filled with green foam-rubber damping material.

At the top, we find the Celestion HF2000 3/4" mylar-dome tweeter. Described variously as "highly acclaimed" and "the world's finest", this is where I think the age of these loudspeakers shows the most. Modern bass/mid drivers aren't that much different to those used here. Modern tweeters appear to have advanced a lot more in forty years.



I've dissected the dead one. Seeing for myself how fragile it is, I'd be terrified of frying another pair if I found exact replacements. Measuring the "working" one suggests that's been damaged. Some of the adhesives used to construct them aren't lasting too well either. Both are starting to fall apart. They look like they're hand-assembled and, while that's not necessarily a bad thing, there's inconsistently applied glue and a certain crudity which suggests mass-production rather than craftsmanship.



Rather than just get them working I hope I can bring these loudspeakers back to being "monitor class" with a different tweeter, properly integrated; then get back to enjoying music on them without the distraction of wondering how imminent another tweeter failure might be.

The next thing to do before the tweeters arrive is ensure I have an amplifier and microphone pre-amplifier which are flat to at least 30 kHz and find out how well my reference microphone works. Would I notice a lump or dip at the 5 kHz crossover point or mismatched mid/tweeter level? I'm not sure my ears would, but my test gear might.

Cheers, Kat

[alex_w]

Kat,

I think I've spotted an error in your crossover schematic. Shouldn't C4 and C6-8 be connected to IN+?

Regards,

Alex

[Kat Manton]

Ah, oops... in my defence I'd been up all night measuring inductors by the time I drew it. (The hand-drawn Biro one I did first is correct...) I'll amend it later, cheers

[dave cox]

Having gone to the trouble of measuring all those crossover components, I wonder what a simulator would make of it ? Of course, there would be some work needed to turn the speaker measurements into some inductance + resistance approximation ...

dc