Loudspeaker Build

PACPAC
edited November 2013 in Members' blogs
PART 1


Some of you know by now that I've started another loudspeaker project which ultimately will become a commercial venture, so first and foremost, I'm being honest and declaring that.  For the time being though, it's all R&D plus development work, most of which is pure DIY. I thought some might be interested in progress as it will become a sort of blog on how loudspeakers are arrived at from first thought until they hit the dealers floors.  I wonder how many give thought as to just what goes into a decent speaker?  I'll attempt to describe on this thread some of the steps involved if that's ok mods?  Pretty pictures (or not so pretty) will be included as the build goes on.

Right, so where does one start?

Firstly, there has to be what I term as a Statement of Need...a brief if you like which identifies why the design is needed, who it will appeal to, what quality standards it will be constructed to, costs, and performance.

From my own perspective, I've got rather tired of wandering round audio shows this past few years seeing all the same (or similar) narrow little standmount boxes with bold claims on frequency response, insensitive design (electrically) and telephone directory price tags.

After careful consideration including polling some respected dealerships and music enthusiasts (note I didn't say or include hifi enthusiasts...I deliberately targeted musicians of whom I know quite a few), I came up with a remit for a stand mounted speaker harking back to the good 'ol days...think Heybrook HB2, JPWs, Snell (J&K) and similar mid sized wide baffle designs of sufficient size and design to allow drivers to breathe properly without creating peaky one note bass response, and more importantly, on reacreating very pure midrange where it's all at.

This box had to be sensitive enough and an easy enough load to be driven by a wide variety of gear from budget 20W amps to muscle amps, valve amps, including SET designs but be as near to full range as possible and not take up too much room.  The brief was starting to get difficult to visualise!

Cabinet design

It has to be made to very high standard both visually (aesthetically) and build quality.  Compromise always has to be made with speakers and manufacturers who claim otherwise are not telling the whole truth.  The compromise with these is that trying to get full range from something relatively compact meant looking at vented or transmission line designs.  I preferred to go the vented design but well implemented as group delay (acoustic bass delay) can be easier managed although there are other drawbacks (not gone into here).

Enter stage left, the basic outline brief for the new Reference Rhapsody loudspeakers.

From the brief, budgets have to be considered, manufacturing ease, availability of all parts, quality of parts and overall system performance.  The box cannot be designed until the drive units are chosen because the box is sized to the bass unit in terms of required volume for the design.

I won't give away which drive units I chose for now as that is commercially sensitive.  What I can say is that these are a two-way traditional design but using units that mate far better than older units ever could, especially in respect of how low the tweeter can be run, and how high the midbass can be run (the old catch 22 when designing 2 way speaker systems).  The drive units have been selected for exemplary off-axis performance so the "sweet spot" will be very wide indeed.  They will be front ported for placement close to room boundaries as this suits a wider range of rooms and extra care will be undertaken to eliminate port noise and mid range frequency radiation through the port (some sneaky design comes into play here).

My cabs are now designed and sized.  they'll be a shade under 700mm tall and look something like this:

image

I have decided on very hefty construction using solid Baltic Birch ply.

To ensure that the cabinets deliver the same clarity as the Fidelio speakers, an engineering analysis called "finite element analysis" has been undertaken on the panel design to generate the thickness of panels needed, the bracing structure and spacing, plus stiffness needed in the bracing.

Some people will build speakers from mdf and use equi-distant spaced mdf bracing but this is not a good idea.  The idea is to stiffen panels so that resonance rises above mid range and that in doing this, panel amplitude drops below audible threshold, or at least is not significantly additive to driver output.  A 1/4mm panel amplitude across the whole front baffle is like having a whole additional driver adding out of phase muddied information to the mix!


«1

Comments

  • PACPAC
    edited November 2013
    PART 2

     By spacing equi-distant, all sub panels vibrate at the same frequency
    and their commulative output isn't far behind an unbraced cabinet, the net
    effect is similar to just reducing amplitude slightly which could be achieved
    by simply making the main panels thicker and doing away with bracing.

     For this reason, the Rhapsody has panels whose spacing ensures sub panel
    resonance at differing frequencies which significantly reduces panel radiation,
    furthermore, those resonances fall above mid-range and are tiny in calculated
    amplitude...just fractions of a millimetre. Top is also tied to the base via a
    sneaky jig-saw like vertical brace design which connects all horizontal braces,
    and brace stiffness is calculated to be equal to or greater than main panel
    material stiffness to be effective.

     Both front and base panels are extra thick.  All of this adds up to
    one seriously weighty cabinet!

     Driver implementation

     Knowing the Thiele Small Parameters for the mid-bass unit , I have sized
    the cabinet volume to be absolutely optimal for the midbass unit.  This
    means that it will be able to operate without generating that horrible 60 to 80
    Hz false thump thump one note bass (ie the response is free of peaks in bass
    response by virtue of a 
    large enough cabinet volume).  The driver will
    also be able to operate without becoming too power hungry and over-damped in
    the bass (cabinet is small enough in volume to ensure a flat frequency response
    to a handful of Hz above the natural resonance of the drive unit).

     The plotted results of the calculated response for the driver in this
    cabinet are given below and compared directly with the Fidelio speaker response
    (thinner of the two lines) for reference.  This should give those of you
    who have heard the Fidelios some idea at least of how the Rhapsodys will
    perform in the lower registers:

     image

     

    Note how smooth the calculated bass response is for the Rhapsodys and
    how low they dig for a stand mounter!

  • PACPAC
    edited November 2013
    PART 3

    Next lets have a look at the group delay which gives an indication of
    how "fast" the bass should sound.  Good speakers need to perform
    at under 30ms.  Once again, Rhapsodys in the thicker of the two lines:

    image

    Crossover design

    Once the cabinet size and shape has been calculated, the next step is to
    set the best crossover point for the drivers and implement the best crossover
    design to ensure a flat as possible response.  Also, driver impedance peaks in both
    drivers should be electrically well damped to keep everything electrically and acoustically
    balanced.

    It cannot be done by using on-line calculators, so a word of
    warning for anyone considering just that, don't, as without understanding the
    load characteristics of each drive unit, it simply isn't possible to calculate
    values for a crossover to achieve what was set out above.  
    First, the effective
    impedance of each driver must be determined, and in the case of the bass units,
    it's usually the stated nominal driver impedance.  With mid and tweeter
    units, their resonant peaks need to be electrically flattened by introducing suitable shunt resistance into the circuit before effective impedance can be
    calculated and that figure is them used in crossover calculations.Bass drivers also require a zobel to flatten impedance peaks.

    There is then the decision as to whether you then go first order, second
    order, 3rd order or higher order with crossovers.  generally speaking,
    higher order crossovers are better done using active and not passive means
    because i) system response become impossible to accurately model passively and
    ii) expense starts to build with the significant increase in 
    components needed.

     So what is this crossover and what is the "order"?

    It's basically a circuit which rolls off the lower order driver whilst
    rolling in the higher order driver to keep each operating in their most
    efficient acoustic ranges whilst minimising distortion and protecting the
    tweeter from power overload from lower frequency signals.  The higher the
    order, the steeper the roll-on and roll-off which is generally a good thing. A
    first order crosses over at 6dB per octave, a 2nd order at 12dB and a 3rd order
    at 18dB etc etc.  

    With each type comes phase changes as you introduce
    inductors or capacitors into the signal path, so a first order introduces 90
    degrees phase difference between tweeter and woofer, 2nd order 180 degrees and
    3rd order 270 degrees.  This is NOT the same as mechanical phase (ie you
    cannot simply rotate one driver!).  Eg 270 degrees phase difference
    essentially introduces a plus 15 or minus 15 degree angle in the acoustic
    radiation lobing pattern which requires time alignment of one of the drivers
    to correct.  Usually, the mid/mid bass are bright forwards slightly.

    2nd order is usually optimal for passive crossovers as providing the
    right drivers are chosen, the 
    slopes (12 degrees) are sufficient, the 180
    degree phase is easily corrected by inverting polarity of one of the drivers
    and no time delay 
    compensation is required.  Got that? right...moving on...

    There are various types of 2nd order slope just to confuse matters more,
    each having differing crossover characteristics.  For the Rhapsody speakers,
    I've chosen a nominally flat crossover (ie the combined drivers response is
    flattened so there's no crossover peak).

     

  • PACPAC
    edited November 2013
    PART 4
    Implementation

    Its all very well having the understanding and setting things down on
    paper and computer programs, but at some point, a speaker has to be built and
    tested so practical considerations come into play.  these include:

    location of drivers, crossovers, ports, terminals etc etc.  Ports
    need to be designed and whilst software is available, I always do mine the long
    way using the original Helmholtz equations and rearranging for port length, and
    just use software as an additional check.  when working alone on design
    without a team to act as "checker" this is the best way since the
    software becomes the sanity check and you're less likely to come 
    acropper.

    Driver spacing between midbass and tweeter MUST lie within one
    wavelength of the chosen crossover frequency to avoid comb filtering and having
    a very obvious delineation between midbass and tweeter as one fades and the
    other cuts in, especially close up. Ideally, they should be dual concentric, or
    point source like Tannoys, but that brings a whole host of other difficulties
    which I wont go into here, suffice to say they need positioning close together!

    Crossover boards have to be thought about in terms of materials,
    components, layout, construction and location, and prototypes need readily
    accessible crossovers whose components can readily be changed without writing
    off the boards.

    This is the end of the first instalments, but hopefully it's been of
    interest.  if it's too much mods...shout, but I'm happy to add the rest of
    the stages if there's interest, along with photos of the Rhapsodys as they
    progress 

    :D

  • PACPAC
    edited November 2013
    Onto some of the practical bits.  Crossovers have to be built and PCBs dont exist for the design as yet.  You can get them made but if you decide to change the circuit at all, and want to do anything other than replace values, you're stuffed if using PCBs so for this type of build, the sensible thing is to make up some hard wired boards.  Here. I've used Tufnol board cut to size, and drilled in key areas by placing a template of the circuit diagram over it (drawn to scale).  I've then used copper rivets and tinned solder tags to form the connection points.

    Note that all components can be lifted out and changed from one surface.  All the wiring has been done from the underside, so this makes these very easy to work on.  Values erased from photos due to sensitive nature of this design, and some key components missing (as yet), but you get the idea:

    image
    Rear:

    image
  • Lovely thread Paul.
    Great looking graphs. How reliably do these design graphs tend to translate in to the real thing?
  • No opposition from the mods, Paul.

    We actually launched a blog section that never gets used, so we'd like to see much more of this kind of stuff.
  • He's paid his 50 notes then Dave...?
  • Per post, yeah!
  • PACPAC
    edited November 2013
    Lovely thread Paul.
    Great looking graphs. How reliably do these design graphs tend to translate in to the real thing?
    Pretty reliably from an anechoic perspective Ben but room response can't be reliably modelled, it can only be measured for a number of reasons.  However, the equations used to derive the graphs are based on laws of physics and are very reliable providing the cab systems are not over damped or any other "strange" thing is done with the design to modify response.  My Fidelios measure a little higher than the graph suggests due to in-room response.  They're flat to 30Hz in room in fact, but that's my room.

    What the graphs don't show is the full frequency spectrum.  I'll publish that only when I have it measured.  I can publish theoretical response but this will alter with crossover tweaks.  Suffice to say that the Rhapsodys have been designed to offer almost flat on-axis response.  From just above room bass loading, they theoretically remain flat, within +/- 3 dB all the way to 15KHz before very gently rolling off, and are flat +/- 5dB to 13Khz up to 30 degrees off axis.  That's not a bad result for a quality speaker, and stereo imaging should be exemplary.


    Please feel free to move this to the Blog section Dave if you think it's better lodged there.
  • Sorry Paul

    I wasn't asking about your system. It was for the benefit of anyone else who might want to blog. My bad :-(
  • Paul, I think I'll read this thread several times - thank you for sharing. I hope to hear these one day, as my preference is almost always for smaller monitors with grunty amps, preferably in a near-field setup.

    Are you going to slope the baffle, or base, to time align the drivers, or will your crossover take care of that?

    Also, what are you going to do with regards to diffraction from the cabinet edges? I used to use Jim Goulding's Diffractionbegone pads, but they are not required on my RR3s as the corners are rounded with a large circumference.
  • Hi Alan

    there's no time alignment needed since that is taken care of via the crossovers and makes the design neater externally.  The edges will all be rounded, top and base as well as sides on the front baffle.  I will be using solid Walnut frames on the edges, and these will be machine planed round before sanding.  the final finish will be a water based satin lacquer all round.  This is just for the protptypes but since a high grade (AB) ply is being used, it'll look pretty good self finished. Production versions will be either self finished or veneered to whatever the customer would like, including Zebrano, Olivewood, Walnut, Burr Maple, Ripple Sycamore or whatever is desired.  I can use a wide variety of contrasting or complementary solid timbers for the edges as well.

    These won't need a grunty amp...a handful of watts should get them singing at 91dB/1W/1M (system sensitivity).  Load has yet to be worked out, but my best guess at this stage is minimum 4.6Ohm load, nominal 6 to 8 Ohm load.  They'll be easy to drive anyway (T amp anyone?).

    I may try diffraction pads around the tweeter.  Simply placing felt rings around the tweeter baffle is usually enough, but the rounded edges may mean that they're not needed.  We'll see!
  • PACPAC
    edited November 2013
    PART 5 - CROSSOVER CONSTRUCTION, EVALUATION, TESTING AND REVISION


    image

    These are the prototype crossovers for the rhapsody speakers using off the shelf components until I have fine tuned all the values before committing to better quality bits and bobs, but you'll get 95% of the sound quality using the correct value off the shelf components rather than paying a fortune for expensive bits and having the wrong values and not getting to within 50% of the performance required! Hence prototyping relies on selection of the best values for the system design using appropriate budgets and materials.

    When you consider the time, effort and cost that have gone into these crossovers, then the time taken for measurement, listening and evaluation;  then the cost, time and effort into replacing components with revised values until the speakers measure well and sound good, you start to get an appreciation of what's involved just for the crossover section of developing new speakers.  This doesn't include the many hours spend analysing electrical and acoustic performance graphs for the drivers and the decisions needed on the in-room response you're after in the first place in order to select the most appropriate drivers and crossover design in the first place.  Then follows circuit design, drawing the schematics, transferring those to actual scale drawings, choosing materials and components, and finally assembly.

    I'll come onto port design for the next installment.

  • Best of luck with your venture here Paul!!,
  • PACPAC
    edited November 2013
    Thanks James.

    Some further notes on how the speaker design came about (and speaker design in general) continues below:

    Part 5 - Cabinet Tuning

    You’ll see many variations on speaker cabinets from vented
    (ports, transmission lines, aperiodic etc) and sealed boxes.  Which type chosen depends on the bass driver selected, the space that the speakers are to be played in, the choice of music and the bass
    characteristics desired.

    For near field monitors, sealed boxes or rear ported boxes
    are usually selected, whereas larger spaced usually require larger driver area
    coupling to a larger room volume where some form of venting improves bass sensitivity
    and extension.

    There are pros and cons to each eg sealed enclosures
    generally have superior transient speed and less coloured sound but are
    ultimately limited in how low they can play, plus tend to be less sensitive due
    to system impedance.    Vented cabinets
    go much lower but can sound more coloured, with bass  group delay issues and each of the venting options
    has it’s own difficulties in design making accuracy of bass texture a real challenge.

    The Reference Rhapsody ‘speakers will be vented enclosures
    because I want to generate meaningful bass from relatively small enclosures and
    have a sensitive system which can be driven by a wide variety of
    amplifiers.  So why front ports?

    Designing a ported speaker enclosure is not recommended if
    it’s something you’ve never done before. 
    It is far more difficult to get right than a sealed enclosure or even a
    transmission line design.  There are so
    many variables from achieving the proper tuning, port noise, group delay, bass
    depth, and driver/system loading. Get the port length badly wrong or the
    diameter wrong and as well as sounding awful, you’ll not achieve the end result
    desired and could even damage your bass units and amplifier.

    Get it right, and the enclosure should exhibit no port noise
    (chuffing) and result in the driver working optimally generating decent bass
    with good timing.  So many designs get it
    wrong so that ported enclosures have come in for bad press, but they don’t have
    to (the Reference Fidelio proves this point). 

    This sounds arrogant until you learn that the
    prime objective of many smaller ported enclosures is to use something called “choked
    loading” where a peak in response is targeted 
    for 60 to 80 Hz, creating on first audition an impressive big sound from
    small boxes, but on extended listening you soon realise that bass is one note
    and tiring with no real extension or texture. 
    It’s a deliberate marketing ploy and is why so many small boxes all
    sound the same (and so many not so small boxes!).  I have chosen front porting as this allows the
    speakers to be placed close to rear walls without creating boomy and overblown
    lumpy bass response, and calculated port dimensions and shape to eliminate port
    chuffing noise and radiation of midrange information at any audible level.


  • PACPAC
    edited November 2013
    Part 5 (continued)
    The ports are positioned to the inside lower face of each cabinet ensuring that their centrelines are less likely to be equi-distant from rear or side walls, again reducing any risks of standing waves creating boomy bass. The means that every pair is a mirror pair.  Many manufacturers’ centre their ports or make them in same handed pairs for economy of manufacture but I believe that this is a compromise that doesn’t have to be!

     So what is a port and how is it’s size calculated?

    Ported enclosures rely on something called the Helmholtz Resonator principle to work.  This is something discovered by a German physicist called Hermann Von Helmholtz and related to how two combined systems (a large vessel and a smaller ported cavity) react  at a certain resonance which is related to the sizing of the port for any particular cavity and resonance point.  It is governed by the equation:

    image


    F = v/2π(  √(A/Vo x L)) x C

    Where F = tuning frequency

    V = velocity of sound in air (343m/s)

    A = Cross sectional area of the port

    Vo = static net of air in the cavity

    Le = length of port

    C = constant determined by the port geometry and speed of sound.

    So knowing these points, you can fine tune a port in theory to generate an in-phase reinforcement of a chosen frequency from the 180 degree out of phase rear compression of the bass driver acting on the internal cavity air mass which causes a sympathetic resonance within the air mass in the port.

    The value of F should not be randomly chosen but is a direct function of the Thiele Small parameters for the chosen driver.  In particular the Qts (total electrical and mechanical damping) of the driver needs to be known.

     

    The tuning frequency is then calculated from

    Fb = Tr x Fas where Fb = box tuning frequency, Tr = tuning ratio and Fas = Free Air resonance of=driver

    In the case of the Reference Rhapsody ‘speakers, the tuning frequency is just under 34 Hz.  Similar equations are used to determine enclosure volume, using driver parameters and also the -3dB point of the speakers (anechoic) which for the Rhapsody ‘speakers result in:

    -3dB point (anechoic) = 37 Hz and xx litres (tbc....lets just say a shade under two cubic feet.  If this sounds familiar, then it’s because many BBC broadcast monitors were sixed at around the same size incorporating 8 inch drivers and it was felt that this gives one of the best tonal characteristics particularly for mid range and natural bass without peaky lumps!_

    So you now see that not only must a driver be selected for sensitivity, impedance load, off axis response, but also on where its own resonant frequency point lies.  A drivers free air resonance usually coincides with peak excursion, maximum impedance and also can be at a point of high cone distortion and low SPL output, so tuning is usually above this point to avoid the driver being forced to the end of its travel in both directions (damaging the voice coils and suspension) and to avoid demanding too much power from the amplifier as impedance drops away again below this resonance point.

  • Some more progress on the Rhapsody Speakers...crossovers now at version "6" (!) after trialling a number of drivers (it's amazing just how many drivers don't meet half the manufacturer's claimed specifications and by a huge margin...all that fitted that description went back).

    Cabinets now well on with CNC machining of baffles completed, internal bracing completed and the main boxes underway.

    Some sexy new drivers now acquired and following initial tests, these will be making their way into the prototypes....

    image
  • Oooo!
    They look like substantial pieces.
  • PACPAC
    edited December 2013
    They're substantially expensive Ben!  Magnets are also very weighty on these.  They use a specially designed woven polypropylene curved cone with superb self damping characteristics and have a commendably low Free Air Resonance so should do bass justice.  As far as I know, I may be only one of two loudspeaker designers currently using these units for commercial loudspeakers in the UK, the choice of which was discussed at great length with the manufacturers.

    I've designed-in impedance flattening circuits for bass and treble units so rising impedance and output close to the crossover point is eliminated.
  • PACPAC
    edited December 2013
    Part 7 - Crossover Fine Tuning

    This is where things can get tricky and where most mistakes are made from DIY enthusiasts having a go for the first time.  Normally, a driver's impedance characteristics are stated by manufacturers.  You'll get things like "nominal impedance" stated along with things called the Thieel-Small parameters, so most people go straight to on-line crossover calculators to determine the values for their units, make up the crossovers and then wonder why things still sound awful!

    The answer is simple...NEVER use "nominal" impedance to design crossovers as nominal means just that.  It is an average figure for the typical low impedance that the drivers spends most time at and may be miles out in crossover terms for specific crossover points.  Consider the graph below for a typical 8 ohm woofer:

    image
    Right;  say we want a crossover point of 1500 Hz;  look at the graph and the impedance at 1500 Hz is approx 9 Ohms (from a nominally 8 Ohm driver) and rising.  This means that if you designed for 8 Ohms, your crossover frequency would be miles out. the other thing to note is that even with a 2nd order (12dB/Octave) crossover, the rising impedance of the woofer will still result in rising output interfering way past the crossover point which is precisely why first order crossovers are a BAD idea without some way of flattening this impedance rise.

    To do this, we need a thing called a Zobel network which is simply a capacitor and a resistor shorted from positive to negative just before the output terminals.  For the demo driver (above) I've calculated the correct values for the cap and the resistor and the following graph shows the result:

    image
    Now we have a flat impedance from the point after the driver's peak resonance slopes back to minimum impedance and can accurately design out crossover without fear of it being messed up by a rising impedance curve.  In this case we would have used an impedance of around 6.5 Ohms for the crossover. design.

  • PACPAC
    edited December 2013
    Part 7 (continued)

    Now that the woofer is sorted, how about the tweeter?  Have a look at the graph below:

    image

    Look at what is happening just before 1KHz...this is where the tweeters self resonance point lies and we dont want to be anywhere near that at crossover as it would result in both over-driving the tweeter and in horrible distortion approaching resonance.  Normally a tweeter is chosen for crossover at least one to one and a half octaves higher than resonance so for this tweeter, normally a 2.5khz point would be ideal as the lowest point.

    However, there is a way around this.  If we introduce a notch filter to take care of that peak, we can run the tweeter much lower.  I made up a notch filter (sometimes called an L-C-R filter because it uses capacitors, inductors and resistors in the network) so that I could demonstrate the effect on the tweeter.  I have measured the results with the series 2nd order filter in place so as frequency drops the inductor allows current to be shunted to ground, hence the drop in impedance below the crossover point, but note how the large impedance peak has been almost completely squashed:

    image

    Finally, if we combine the tweeter and woofer with all the design considerations in place plus a 2nd order filter arrangement, we get the following for the crossover response:

    image

    Note that we have a nice flat impedance curve once we're past the woofer's peak resonance point with a slight rise at crossover caused by the woofer rolling off (so impedance rises on the woofer circuit and the tweeter rolling in where it's impedance is falling from high impedance below the crossover point) before the tweeter continues at nice controlled impedance level.  This would sound pretty good as providing both units were of similar sensitivity, the acoustic response would be nice and flat, just what we were aiming for.


    Hopefully, what I've demonstrated shows that:
    1.  a crossover cannot be calculated from nominal impedance figures
    2. first order crossovers can be bad news without very careful driver and crossover point selection
    3. second order crossovers are preferable due to steeper electrical roll-off
    4. woofer rising impedance makes it impossible to accurately blend woofer and tweeter at crossover point without flattening the rising impedance
    5. you can run tweeters lower than 1.5 Octaves above resonance is an LCR circuit is employed providing that the tweeter is rated for the power and that you stay outside dome resonance distortion, which varies between tweeters, but never go below half an octave above even with an LCR. An octave would be the recommended minimum
    6. There is a little more to crossover design than first meets the eye.  Establishing and using the most accurate and appropriate values for caps and inductors and crossover points using off the shelf budget caps and inductors etc will result in far superior sound than the wrong crossover values using the most expensive components.

  • PACPAC
    edited December 2013
    8.  Final thoughts on crossovers:

    The above begs more questions than it answers really such as "why then do many manufactures, even some of pretty expensive loudspeakers, continue to use basic 2nd order (no impedance flattening) or first order crossovers?"  the answer lies in several parts.  Firstly, some designers use drive units with highly damped surfaces to combat self resoance so they can be run higher and then roll-off naturally.  Whilst this sounds fine in principle, mechanical treatment of resonance never cures the issues well enough imho and still results in unwanted distortion, plus cannot combat a woofer's tenancy for rising impedance unless crossed over below the impedance rise, which is usually too low for a two-way speaker but possible for a 3-way unit.

    Rising impedance is bad as it presents a constantly changing load for the driving amplifier and it is certainly very bad where many high output impedance valve amps are concerned.  You haven't a hope of achieving a flat frequency response using mechanical means to treat resonance unless crossover points are very low for the woofer.

    Secondly, cost is a factor. Some manufacturers don't bother simply because there's more profit in chucking a few drivers and basic crossover into a box, it really is that simple, and you'd be amazed at the cost of some of these "boxes"!

    Thirdly, some claim that the less crossover components in a circuit the better.  This is simplified thinking at best.  Whilst it is better for passive crossovers never to go beyond 3rd order because predictability of response becomes increasingly complex and almost impossible to calculate with any accuracy, up to 3rd order does far less to hamper the signal than all the circuitry that has come before in the amps and source equipment!  Think about that for a little while...

    There is no excuse, simple as that, for loudspeakers not to use properly designed crossovers to achieve low distortion, a flat impedance load and a flat acoustic response.  Anything else, despite what makers may claim, is not high fidelity simply because it is not possible.  All voice coils exhibit inductance which changes impedance with frequency and even with changes to temperature!  All drivers have self resonance points.  You may through very careful selection, choose drivers that will work with the minimum of crossover components, but event then, they will never mechanically compensate for what can be achieved by proper electrical design.  
  • Nearly there!

    image
  • I want to hear them!
  • We'll have to arrange another Audio Chews get together over here Dave for a Rhapsody day!
  • May have to do that.

    On the matter of bake-offs, I must host one at the new Ugly Towers once we're in and settled. Then you can all have a laugh at the sounds I like.
  • We'll have to arrange another Audio Chews get together over here Dave for a Rhapsody day!
    Lovin' that idea!
  • ...I have the added security knowing that Rhapsodys will be too hefty to sneak out under one's arm  :D
  • Not sure if you're aware Paul, but i've this rather special pair of refurbed old 70s speakers, that i'm already very happy with...
  • Not sure if you're aware Paul, but iI had this rather special pair of refurbed old 70s speakers, that I was very happy with...and then you go and do this to me!
    Fixt!
  • We'll have to arrange another Audio Chews get together over here Dave for a Rhapsody day!
    Sounds good to me
  • They live!


    image

    image


    image


    image


    They have to have the most natural mid-range of just about anything I've heard to date.  Good HF detail and imaging too.  LF goes plenty low (impedance plot shows 34Hz cabinet resonance tuning point and F6 in-room is under 30Hz).  Very pleased with these.  I few minor crossover tweaks left to get them spot-on but they're remarkably close as they stand.
  • edited December 2013
    I'm sensing pride. Relief and pride.
    :-)
  • Neither to be honest Ben.  They're almost there but need a little tweaking...just a few percent reduction in a couple of cap values and a slight increase in a few resistor values.  Measured cab tuning point is within 0.5% of calculated which gives confidence in the methodology used.  I'll be less self critical when I get them spot on and whilst not far off, they need tweaking time.  It's no different for any commercial speaker development and can take many months. I'm just very fortunate to have got them so close first time round so no real headaches really.

    I think they'd look pretty stunning in a ripple sycamore veneer, that or a burr walnut.  may try one of the two...
  • Fortunate...?
  • I've been promising myself a large glass of Armagnac sufficient time to read this thread properly. Maybe sometime over Christmas.

    Have you ended up with a modern-day pair of Snells, Paul? There's absolutely nothing wrong with that. I quite like Snells :-)
  • I've been promising myself a large glass of Armagnac sufficient time to read this thread properly. Maybe sometime over Christmas.
    Oh the festive merriment of the Uglymusic household during the yuletide period...
  • I've been promising myself a large glass of Armagnac sufficient time to read this thread properly. Maybe sometime over Christmas.
    Oh the festive merriment of the Uglymusic household during the yuletide period...
    Hic!
  • edited December 2013
    "Sorry darling, can't hang around for the Xmas pud', there's this splendid thread on 'chews I've been eyeing up, that requires my attention..."

    image
  • Paul, I notice that a feature of your designs is a high driver area - to - front panel area ratio (not sure if that's the right way round, but you know what I mean) compared to the "average".
    I guess this might mean that you belive in the utility of large and carefully calculated ported cabinets...?
    I also guess that your answer will be long, and beyond my comprehension.

  • PACPAC
    edited December 2013
    Sorry for tardy responses.  Dave first:

    They're very similar to the Snells in the mids Dave but with way more bass attack and a smoother (more linear) treble reponse.

    Ben:

    The modern trend for narrow baffles and small footprint area has a number of advantages: Aesthetically, it can be more pleasing but ultimately they "all look the same" and for the most part "they all sound the same!" Imaging can be great too...

    Wider baffles...baffle step losses occur at lower frequencies so more upper bass attack, decent sized cabs can be used for (imho) advantages to bass quality (depending on driver parameters).  In this case, the driver parameters dictated the optimum cab dimensions and I plumped for the "golden ratio"  for all cab ratios as it ensures few issues with standing wave formation and is meant to be aesthetically pleasing. I think the cabs would look much nicer with a high quality nicely figured veneer so bear that in mind...these are prototypes and a lot of cash has gone into developing them so the extra for nice veneers was unimportant to me at this juncture.  What matters most is sound quality and with tweaks and all needed, they still sound way more natural and balanced (a Snell trait?) than many modern slim fashion boxes.

    The Fidelios are wide because the drivers are 12 inches diameter to start with!


  • Your speakers sound great. What role do the large front baffles / cabinet volumes play in that?
    Also they look good to me.
    But, I can't remember seeing many speakers from any era with a front panel approx 8 times as big as the surface of it's drivers. My Goodwoods are driver area : panel area equal at best. What is the large front area / total cabinet volume of your speakers adding to their great sound?
  • Plenty do Ben....look at Horning speakers, older Snells, Audio Note speakers...all have large baffle areas compared with speaker area in one or more models.

    The volume is dictated by the Vas of the driver (volume of air equal to driver compliance) and Qts (combined electrical and mechanical damping factors).  There's a mathematical formulae for dictating net cabinet volume from those parameters.  The original drivers I chose had a net vol of over 50 litres, and when i discovered (through measurement) that manufacturer's data was wrong (I contacted them and provided my data to them....following which they agreed with me) I had to choose an alternative driver.  The cabs were underway by then but as luck would have it I found a better main driver and simply adapted the tuning of the cabinet to the driver for the flattest response possible.  It seems to have worked out well.

    The cabinet could have been made narrow and tall as is the fashion but the crossover design would probably have had to include baffle step correction (for the same bass efficiency) so the design would have resulted in a slight loss of efficiency but perhaps with "better" imaging. I was careful to choose drivers with great off-axis response so my wide baffle design retains slightly greater efficiency but with still  good imaging.  I wanted (deliberately)  to appeal to buyers who valued some great 70's speaker designs and with the ability of a wider baffle to show-off some stunning veneer choices makes these a prospective attractive furniture piece.  I guess "music lovers" will value how they sound whilst "hifi lovers" will seek out skinny fashion towers. As I'm not in mass production and these are all hand built to order, I'd rather appeal to music lovers who value the sound qualities of the design.

  • Sorry for tardy responses.  Dave first:

    They're very similar to the Snells in the mids Dave but with way more bass attack and a smoother (more linear) treble reponse.


    That sounds encouraging.
  • Plenty do Ben....look at Horning speakers, older Snells, Audio Note speakers...all have large baffle areas compared with speaker area in one or more models.
    Then I really must get out more.
  • PACPAC
    edited December 2013
    Well, the past few days have been spent tweaking the crossovers.  The slight niggles I was having turned out to be due to the tweeter being run a little too low even though I used a circuit to squash impedance due to its low resonance.  Upping the crossover point a few hundred Hz did the trick.  Now sounding pretty fine all round.  A few inductor tweaks (it's amazing just how a change of 2% on inductor values can alter sound) sorted some slight mid emphasis so they're now very balanced.  Ports also tuned and measured cab resonance point is 36Hz. If I've learnt a lesson from this, it's just how vitally important it is to get inductor values spot on, or at least within 1%.  NEVER buy 5% tolerance inductors. I ended up winding a few of my own and trust me that it's not much fun spooling 12ft of copper onto a small core!
  • Sounds like you're there...?
  • Better photo:

    image

    Looks like it Ben.  A few tweaks have them nicely voiced and they get better with every hour on that suspension.  Had a few musicians and hifi nuts in to compare Fidelios and Rhapsodys, and it was a split field with the musicians liking the bass of the Fidelios but everyone agreed that the Rhapsodys are more accurate and detailed plus great off-axis.  the bass will improve with running in but its already there without dominating or booming.
  • PACPAC
    edited January 2014
    A few pictures which are being used for the launch of these speaks :-)

    Enjoy!

    image

    image

    image

    image

    image

    image

    image

    image

    image

    image

    image

    image

    image

    image

    image

  • impecible joinery there Paul
Sign In or Register to comment.