KEF C60 - first experience with ferro-fluid

[John_BS]

I bought a cheap & battered pair of KEF C60 speakers out of interest (they use a variant of the B200). I subsequently discovered that the tweeters (T33B /SP1210) are a ferro-fluid design, and the sound was decidedly odd, suggesting the fluid was no more. A check in the lounge with pink noise confirmed that the tweeters weren't very happy tweeting.
KEF very kindly told me which fluid had been used, and how much: 0.08mL, which is a tiny amount. The Ebay purchase of 0.6mL came in a 1mL syringe, marked off in 0.1mL main divisions, together with a very small nozzle with which to guide the fluid into the gap.
The tweeters, when disassembled, turned out to contain "ferro-varnish", which had to be removed from the gap with pared-down cocktail sticks and isoproyl, then blown out with compressed air.
Now the exciting part; adding new fluid. Nozzle in place, I carefully plunged until the nozzle filled, then a further 0.1mL. Much to my suprise, the gap filled instantly with fluid, leaving the nozzle completely empty: the magnetic field had "sucked" the contents out!
I blotted some back out, and tried to avoid adding as much to the second unit.
If I was doing this again, I'd transfer the correct amount of fluid to a second syringe, hence obviating the possibility of too much being hoovered out.
BTW, the cross-over, complex even by BBC standards, has a rather limp single-pole low-pass response feeding the B200, which must presumably fall off rapidly above about 3kHz. Note the measured curves are 5dB/div, so there is less than 10dB of voltage attenuation with the drive unit in place.

[Jez1234]

Job's a gud 'n according to the before and after plots How do they sound then?

What's your measuring set up? Nice 1/3rd octave plots there.

[John_BS]

They sound quite good; well balanced, controlled bass, a little colouration on voice (as is the case with every moving-coil loudspeaker I've auditioned to date).
Measurements in lounge: wav-file of pink noise high-pass filtered to remove LF below 100Hz. B&K 2235 sound level meter feeding HP 3561A in 1/3rd octave mode, averaging on. Crossover measured with HP 3577A network analyser (source via Kenwood Power amp) in high impedance mode & connected via scope probes.

[Edward Huggins]

I've never understood why some crossovers have to be so complex. With all that degree of filtering, insertion loss and component degradation over time. What am I not getting?

[John_BS]

Originally Posted by Edward Huggins

I've never understood why some crossovers have to be so complex. With all that degree of filtering, insertion loss and component degradation over time. What am I not getting?

KEF must have gone through a phase of deciding that maintaining a broad resistive impedance across the band was a Good Thing. Hence the nework directly across the input terminals* whose only purpose is to generate R +/- jX to cancel out the -/+jX of the real crossover.
* Now disconnected

[Radio Wrangler]

Originally Posted by Edward Huggins

I've never understood why some crossovers have to be so complex. With all that degree of filtering, insertion loss and component degradation over time. What am I not getting?

It isn't just doing the crossover job, it's also doing some equalisation, fixing any steps, bumps and valleys they found in the frequency response of the drivers as mounted in the cabinets. So you find more complex crossover networks. In some of them you'll even find transformers to modify impedances of some sections.

There's often quite a bit on the development of the crossovers in the old BBC research department reports on their various speaker designs - material that commercial speaker firms wouldn't let loose in public.

David

[Radio Wrangler]

The terminal-impedance modifying network will have an effect when used with amplifiers with relatively large output impedances. This means valve amplifiers with little or no feedback. Low impedance amplifiers get a degree of immunity from influence by speaker impedance.

There's a lot spoken about amplifier-speaker choice and compatibility from people judging purely by the sound, but there can be straightforward engineering reasons behind these effects, and things can be done about them. No rocket scientists need be harmed in these experiments. Nor do shamen need to be hired.

You can have a couple of orders of magnitude of variation in the impedance which the outputs of various amplifiers present to the inputs of crossover networks. It's not surprising if this causes variation in the behaviour of said crossovers and the drivers beyond them.

For domestic hifi speakers, it would make good sense to do two reviews; one with a classic valve amplifier, one with a high damping factor transistor job. It would serve as a better guide for prospective buyers and it might de-mystify things.

David

[Beobloke]

Originally Posted by Jez1234

What's your measuring set up?

Jez, I'm shocked - don't tell me you didn't recognise an HP3561A screen straight away?!

Originally Posted by Edward Huggins

I've never understood why some crossovers have to be so complex. With all that degree of filtering, insertion loss and component degradation over time. What am I not getting?

Crossover design is a surprisingly complex process, especially when you factor in notch filters and impedance matching circuits that some manufacturers choose to use. It's also worth remembering that, in the days when manufacturers designed and made their own drivers in-house, they could be tuned to their end application well and so a simple crossover was all that was necessary.

However, today, where many loudspeaker companies use off-the shelf drivers (albeit often lightly tweaked to their own requirements), a more complex crossover will give better results by removing the stuff that you really don't want to hear.

I'll be honest, more often than not, I'd rather see a complex crossover than a simple one. Some manufacturers like to boast about how they keep their crossovers as simple as possible and, if they get this right and have the right drive units to achieve it, the results can be very good. However, if you get it wrong (as many do, in my experience!) than you end up listening to artefacts of various drivers' first breakup modes, which is never pleasant.

[mhennessy]

Yes, a complicated crossover is nothing to be afraid of! These designers usually know what they're doing

Looking at a couple of examples - Derek Hughes and Alan Shaw - they are both in the position to (a) design the complicated crossover in the first place, and (b) get it into production. They don't work for massive multinational companies with many layers of bureaucracy and bean-counting, so aren't under pressure to cut corners.

And both start with the drive units. Alan obviously has his proprietary "RADIAL" plastic for the 5" and 8" units, and Derek works directly with OEMs like SEAS and Volt. Of course, back when he was at Spendor, they did their own units.

Things have come a long way in drive unit design, and generally crossovers have less "ironing" to do, but you still have to factor in BSC and diffraction effects, even with "perfect" drivers.

Or, you could just let the Marketing Dept do the work for you. A much better approach, and keeps a whole other department in work too! They get paid so much more than the engineers, so it's really important to take care of them, right Taken to extremes, you could just use a single driver and convince people that crossovers are the work of the devil, what with all that "insertion loss" and other specious things Edward likes to bring up!

(No offence intended, Edward - I know we've had these conversations in the past. Vive la difference )

But yes, this is one of the more complicated designs I've seen. Talk about "taming the beast"!

[Jez1234]

KEF's conjugate matched crossovers were a great idea (now virtually unheard of Yorkshire Hi Fi did much the same) but according to a KEF engineer I spoke to they stopped doing it because it was expensive for one thing and because Joe Blogs buying speakers didn't understand it and ignored it as a plus point.

Their coupled cavity bass loading went for much the same reasons but combined with Joe Blogs wanting to show off impressive woofers to his mates which wasn't possible when they are inside the speaker and not on view! I have a pair of the second from top of the range at the time Reference 105/3 which use both techniques and very fine they are... though unfortunately in abeyance for a few years now due to an unobtainium blown tweeter. They are a 4 way design and have a very complicated crossover spread over two PCB's.

Damping factor is often misunderstood because the resistance of the voice coil is in effect in series with the source impedance which means there is a lot of difference between a damping factor of 1 and 5 but once you get to somewhere around 15-20 there's not much more damping to be had by making that 150 - 200.

[Edward Huggins]

Originally Posted by Radio Wrangler

It isn't just doing the crossover job, it's also doing some equalisation, fixing any steps, bumps and valleys they found in the frequency response of the drivers as mounted in the cabinets. So you find more complex crossover networks. In some of them you'll even find transformers to modify impedances of some sections.

There's often quite a bit on the development of the crossovers in the old BBC research department reports on their various speaker designs - material that commercial speaker firms wouldn't let loose in public.

David

Thanks. Yes, I certainly get the need for EQ when knowingly selecting drivers with uneven characteristics - or as said in a later Post here, in getting them "ironed out".

[m0cemdave]

That's the trouble with Ferrofluid. Sooner or later it dries out and leaves lumps of mess in the gap. Best avoided!

[Jez1234]

I can't recall during all the hoo hah about ferrofluid being the best thing since sliced bread when it first came out any "BTW it will go off, dry up and render your tweeters far worse than the naffest standard tweeters in a few years time"

[steve ss]

Re: KEF C60 - first experience with ferro-fluid

Quote: Jez123
Heptode


I can't recall during all the hoo hah about ferrofluid being the best thing since sliced bread
when it first came out any "BTW it will go off, dry up and render your tweeters far worse than
the naffest standard tweeters in a few years time"

This appears to be a golden opportunity to kick over a few sacred cows!

It seems to have been forgotten that "ferrofluid" was in fact not a strategy to improve
speaker fidelity but a fix to increase the power handling of cheap tweeters! hence the
term "ferrocooled" or perhaps "ferrofooled".

Given that, it did in fact increase the power handling of the tweeter the mechanism
being the heat in the voice coil was dissipated through the ferrofluid in the gap
to the surrounding magnet assembly!

Though I doubt if much thought went into what was happening to flux density and field distribution in the "gap" due to the presence of the ferrofluid!
and of course the long term consequence of the heating of the colloidal ferrofluid.
All in all (IMHO) not a great idea!

There are suitable replacement tweeters available to replace this that are ferrofluid free!

Er! next.

Quote:
Radio Wrangler
Moderator

The terminal-impedance modifying network will have an effect when used with amplifiers
with relatively large output impedances. This means valve amplifiers with little or no feedback.
Low impedance amplifiers get a degree of immunity from influence by speaker impedance

Quote:
Jez1234
Heptode
Damping factor is often misunderstood because the resistance of the voice coil is in effect
in series with the source impedance which means there is a lot of difference between a
damping factor of 1 and 5 but once you get to somewhere around 15-20 there's not much
more damping to be had by making that 150 - 200.


Ok as mentioned above "Damping factor is often misunderstood" I agree!
It's the impedance ratio of the output device (amplifier) to the driven device
(speaker).
In practical terms a Damping factor between 10 and 30 gives the best sonic performance.
The is no evidence to suggest high Damping factors offer more even with "difficult" loads.

So to my sacred cow!
It is assumed that valve amplifiers notably no feedback type suffer from poor
"Damping factor".

Er! indeed not so!

Why? Well as an example (I'm simplifying this) a medium quality output transformer
assuming simple interleaving has a typical output impedance\resistance less than (say) 1.0 ohm and say the speaker has a nominal 15 ohm voice coil, well we have a damping factor of 15 already without the amplifier even connected to the primary of the output transformer! shock.

Connecting the rest of the amplifier to the primary of the transformer and there will
be a significant increase in the damping factor even without feedback applied.

Also valve amplifiers are not subject to back EMF from the speaker due to the *galvanic isolation and impedance/turns ratio of the output transformer!(*assuming secondary not "earthed"ie no feedback)

This is why certain valve amplifiers can drive "complex" loads such as full range electrostatic speakers without distress while high power high damping factor solid state amplifiers perform poorly!

Alright complex crossovers.

Some can and some do work well, good design or good luck I'm not sure!

Why do I say this?
With good design if you are on top of it will consistently turn out a winner every time!
This is not what I see, patchy results are typical with a good product followed by something less good! Not inspiring.

Things like complex impedance correction (cancel out the -/+jX) does not work in the real (pun intended) world as the network will not track the behavior of real speakers so no point!

There were a few guys Peter Baxandall for one that knew how to make a silk purse and had a common sense grip on the problem.(Wireless World August 1968).

Er! That's it.

I hope I have covered some relevant points!


enjoy
Steve ss.

[John_BS]

Taken to extremes, you could just use a single driver and convince people that crossovers are the work of the devil, what with all that "insertion loss" and other specious things

Mark: you've reminded me about a repair to a pair of B&O speakers, fairly up-market, which turned out to have a solitary inductor in series with the bass unit and a single capacitor in series with the tweeter. i.e. 6dB/octave in both cases.

[mhennessy]

Yes, it can be done. Remember that the electrical filter is only part of the equation, and combined with the natural response of the drive units in their enclosure, the overall acoustic response might well be 12dB/octave in practice.

You can do a fair bit in the acoustical and mechanical domains to tailor the response. Waveguides being an example of the former. Examples of the latter include adding mass to the diaphragm ("doping"), or decoupling, which is often done by selecting the right type of glue.

Of course, it might be trickier to control some of these non-electrical features in mass production, and in use over time, where you encounter a range of temperatures and humidity.

Just some of the reasons why speakers are so fascinating

[Radio Wrangler]

Originally Posted by steve ss

So to my sacred cow!
It is assumed that valve amplifiers notably no feedback type suffer from poor
"Damping factor".

Er! indeed not so!

Why? Well as an example (I'm simplifying this) a medium quality output transformer
assuming simple interleaving has a typical output impedance\resistance less than (say) 1.0 ohm and say the speaker has a nominal 15 ohm voice coil, well we have a damping factor of 15 already without the amplifier even connected to the primary of the output transformer! shock.

Connecting the rest of the amplifier to the primary of the transformer and there will
be a significant increase in the damping factor even without feedback applied.

Also valve amplifiers are not subject to back EMF from the speaker due to the *galvanic isolation and impedance/turns ratio of the output transformer!(*assuming secondary not "earthed"ie no feedback)

This is why certain valve amplifiers can drive "complex" loads such as full range electrostatic speakers without distress while high power high damping factor solid state amplifiers perform poorly!

Let's take that in sections...

The copper resistance of the secondary of the transformer acts in series with the output and ADDS TO the output impedance of the amplifier. For it to create an impedance ceiling, you may be thinking that it acts in shunt. The transformer presents the output voltage effectively in series with the wire resistance. The wire resistance creates an output impedance you cannot get below, whatever goes on in the rest of the transformer and the amplifier.

The transformer exhibits inductance. In a simple model you can either take it as the primary inductance acting in shunt across the primary OR you can take it as the secondary inductance across the secondary. Don't do both. One is the same as the other due to transformation of the circuit impedance by turns ratio squared and the inductance of either being proportional to turns ratio squared.

There is only the one highpass zero in the transfer function, not two.

THe anode impedance of the output valve(s) is essentially the slope of the anode volts.current curve. This is scaled down by 1/turns ratio squared and appears (in shunt with the transformer inductance) and then in series with the copper resistance to make up the output impedance. With triodes, this comes out as a moderate value, but likely higher than crossover designers take into account assuming modern solid state amplification. With tetrodes/pentodes it comes out appreciably higher making the crossover design assumptions further off the mark.

Application of feedback brings down the output impedances, roughly pro-rate to the factor by which the overall gain is reduced.

Transistor amps usually end in emitter follower stages running significant bias current, creating a low output impedance to begin with. Because of the need to reduce their overall distortion, quite high levels of feedback are designed in and this reduces the output impedance still further.

Book treatments of valve power amps have covered the issue of output impedance and of transformer limitations quite well. I seem to remember it in Fritz Langford-Smith's tome and those of Fred Terman.

David

[Jez1234]

...and as was very neatly demonstrated by Dr E Cherry, in SS amps which use common emitter output stages rather than emitter follower the extra loop gain exactly compensates for the increased native output impedance resulting in the same output impedance as if an emitter follower stage had been used. Which is rather neat I reckon

[Jez1234]

And to slay a few more of Steve SS's sacred cows.. There is more to ferrofluid than improving the power handling of "cheap tweeters"! It helps to damp the fundamental resonance of the tweeter, often allowing them to be used at somewhat lower crossover frequencies and enabling a smoother crossover transition with less need for any response correction from the crossover. The vastly improved cooling of the voice coil is not merely to improve power handling but reduces not only the average temperature of the voice coil but also the changes in resistance due to the temp coefficient of copper. This ameliorates the modulation of voice coil resistance with the envelope of the programme material hence helping to keep the overall response of crossover and tweeter rather more constant than without the ferrofluid.

Damping factor (DF) is really another way of stating the output impedance of the amplifier. If you know the impedance of the speaker with which the DF was stated you can easily work out the output impedance. There is no optimum damping factor except with certain speakers which were either designed back in the 50's using valve amps of particularly poor DF or, rather erroneously IMHO, designed more recently using horrors like SET amps. Speakers should be (and usually are) designed to work optimally with close to zero driving impedance as only then can they be considered a universal design which will give good performance with most amplifiers.
Theoretically you can't have too high a DF as this merely implies a very low output impedance from the amplifier, which is a good thing. By the time we get to a fairly average DF for a SS amp of around 70 then there will be only a very tiny increase in actual damping by making that DF 700.

KEF's conjugate load matching was a very good idea and as much/more for the benefit of the amplifier than the speaker. Speaker designers like to set fiendish obstacle courses for amplifier designers and the KEF system evened out the peaks and dips of outrageous impedance curves and by opposing, largely ended them. The far less reactive load the amp saw gave it a much easier life and in fact the largely resistive load meant that these speakers with very complex crossovers would work very well with old valve amps with poor damping factors (to complete the circle there). My Reference 105/3's have conjugate load matching and are also 93dB/W efficient so worked marvellously with my Leak Stereo 20 valve power amp!