Product Review:
Whatmough P33 Loudspeakers
Product Type: Loudspeakers
Reviewed By: Greg Borrowman and tested by Steve Holding
Magazine: Australian Hi-Fi: September 2007
Distributor: Whatmough Pty Ltd
No-one could fail to be impressed by Whatmough’s P33s. They’re finished in a superb Bubinga real wood veneer (a metallic duco is also
available) and stand 1.2 metres tall on a 260×430mm footprint. But more than that, they’re shaped more like pieces of modern sculpture
than loudspeakers.
The Equipment
A true three-way loudspeaker, the P33 design calls for two 180mm diameter dual-layer fibreglass/Nomex cones to deliver the bass. These
operate in tandem, to increase piston area, before crossing to a single 125mm midrange driver. Handling the high frequencies is a 35mm
dual concentric tweeter with a waveguide centre plug. This type of tweeter is a relatively new design, and is sometimes called a ‘ring
radiator’ to distinguish it from the ordinary ‘dome’ tweeter design. Ring radiator tweeters have far smoother frequency responses than
ordinary dome tweeters, and extend high-frequency response beyond 30kHz. This would not be so unusual except that ring radiator
tweeters can also handle far more power than conventional domes, so high frequencies don’t get attenuated after prolonged playing.
Unlike the similar-looking tweeters on some other speakers, the P33 uses a genuine Vifa XT25—though one that’s modified to
Whatmough’s own specification.
Although Whatmough rates the bass drivers as having a diameter of 170mm the all-important dimension is the effective piston area (SD),
which is derived from measuring the diameter of the cone, and adding the width of the roll surround. (A distance often referred to as the
Thiele/Small diameter, after Australian engineers Neville Thiele and Richard Small.) The P33 has a T/S diameter of 135mm, which gives
an SD of 143cm². However, because the two drivers operate together, the SD for the system is twice this, or 286cm². This means that if
Whatmough wanted to achieve the same SD using just a single driver, it would have to have a T/S diameter of 191mm, which would
equate to an overall cone size of around 236mm, based on a similar frame size and roll surround.
The driver does not have a central dustcap. Instead, at its centre is an exposed, bullet- shaped, pole-piece, whose size indicates that the
voice-coil is approximately 30mm in diameter. This so-called ‘exposed voice-coil’ design means that you should always operate the
speakers with their fabric grilles in place, to prevent small particles of dust or grit from lodging between the voice-coil and the polepiece.
The basket supporting the cone is very large, and made from cast alloy. This means that it’s very rigid, and also can’t act as a ‘path’ to
weaken the magnetic field from the magnet. In the case of the P33, this wouldn’t be likely in any case, because the magnet’s field is
constrained by a second magnet attached to the rear of the main drive magnet.
The midrange driver also sports a heavy-duty cast alloy chassis that supports a large, vented, unshielded magnet. This driver has a cone
that appeared to me to be made from coated paper. Unlike the bass driver, it has a conventional dustcap, but it, too, has a rubber roll
surround.
As with all Australian-made Whatmough designs, the internal wires (which are solid five nines copper) are silver-soldered to the driver
terminals, rather than connected with spade/lug connections, which can deteriorate over time, as well as to the crossover network. The
network is all hard-wired using audiophile-grade components, including hand-wound, air-cored inductors (all crossmounted to eliminate
magnetic coupling effects), 250Vd.c. Solen and Bennic capacitors and ceramic resistors.
The P33 is a bass-reflex design, using a front-firing port. The port is quite short (120mm) with an 80mm diameter. The tube is radiused at
both ends and has ‘dual-coat’ interior, where half of the port is very smooth, whereas the other half is quite rough. If you were wondering
about the size of the port, it comes about because within the P33 is a separate sub-enclosure that houses the midrange driver and the
tweeter. This allows Whatmough to optimise the rear-loading for the midrange driver, and also means the rear radiation from the
midrange cannot interfere with that of the bass drivers (or vice versa). The rear of the tweeter is, of course, completely sealed.
The P33 is also available in ‘Signature’ version. According to designer Colin Whatmough, the Signature version of the P33 replaces the
Solen and Bennic polypropylene capacitors with Hovland film and foil capacitors in critical positions. Hovland’s capacitors typically cost
10 times as much as the equivalent Solen capacitors and even more for Bennic. The Signature version also replaces the solid internal cables
with Litz cables made by Cardas in the US. Whatmough says that he uses the Litz cable even though it’s very tedious to wire up, due to
every single strand being individually insulated, because ‘the sonic results of Cardas cables are stunning.’ Cardas also gets a foot in when it
comes to speaker terminals, with the P33 Signature version benefiting from Cardas’ ‘rhodium over silver’ solid copper binding posts.
Unusually, the P33s are intended to be tri-wired, and Whatmough is so adamant that they should be so wired that the usual bridging links
are not supplied, so as to encourage users to at least bi-wire. According to the Whatmough’s website: ‘To obtain their optimum
performance, [the P33s] should be bi-wired. This involves using a tri-wireable (six-core) cable or three sets of normal two-core speaker
cables. These cables can be connected to one or two sets of speaker terminals at the amplifier. One pair of cables is connected to the
speaker’s top (tweeter) terminals, one to the middle (midrange and the other to the bottom (bass) terminals… If tri-wiring is not being used,
(hopefully only a temporary measure) a link should be made from the same type of speaker cable that is being used to connect the speakers
to the amplifier. Using this cable, the three positive terminals (red) should be connected together and the three negative terminals (black)
should be connected together.’
However the most obvious (visual and audible) difference between the standard P33 and the Signature version is that the Signature uses the
same ultra-expensive tweeter as Whatmough’s flagship speaker, the Paragon, which extends the response ultra-fl at to beyond 40kHz.
Performance
Despite the height of the P33 cabinets, the tweeters are a very ear-friendly 910mm above floor level when the speakers are standing on
spikes. My initial ‘rough-in’ for siting the speakers in my listening room was so extraordinarily successful, with the speakers delivering
such great bass and a superb stereo stage, that I was congratulating myself on my skill until further attempts at improving things even
further proved beyond any shadow of a doubt that it was the speakers themselves that had permitted my instant success. My further
experiments then simply confirmed that the P33s perform well across a very wide range of positioning options. It doesn’t matter whether
you push them close against a wall or bring them out into the room; angle them in to face the listening position, or over-angle them, or
simply place their backs parallel with the rear wall so they fire straight up the room, they’ll always sound great. Yes, there will be minor
differences in all these positions, but the critical word here is ‘minor’: in every circumstance the P33s really delivered the musical goods,
any small adjustments you decide to make will be for your personal preferences and to suit your room more exactly.
The thing that struck me immediately about the P33s was the strength of their presentation, with the sound surging into the room with a
cohesion that linked the disparate drivers into a single musical whole. This feeling of musical strength extended beyond simply the overall
balance from low bass to extreme treble to encompass the space between the speakers as well. I had no sense of the sounds of the left and
right speakers combining to form a soundstage, just an image of the performers at the front of the room, as if there were no speakers at all.
When there was just a single performer, the illusion was so uncanny it was spooky. Listening to a new-found favourite of mine, John
Couch, it was as if I’d invited him around to my place and asked him to bring his guitar. (I hasten to add that I don’t know him at all: it’s a
metaphor.) I hadn’t heard of him at all before his new CD ‘Desdemona’s Song’ was brought to my attention. It’s available via www.indiecds.com. Couch is a Kiwi currently living in Canberra, where this CD was recorded by Michael Grafton-Green at the Wesley Music
Centre. The Whatmough P33s transferred the sense of acoustic space that’s been captured in the recording directly into my lounge-room,
along with the guitar. I knew I was listening to something special immediately from the sound quality of first soft notes of Isaac Albeniz’s
Asturias (Leyenda) Suite Espanola No 5, Op 47, but when I heard the sound of the first major fret board slip (24 seconds in) my head really
snapped up, because it sounded so live. The superb tone of the P33s (and of Couch’s Philip/Smallman guitar) is rendered beautifully and
rather gently in the following track, Jorge Cardoso’s Milonga. This CD closes with a track that’ll bring a tear to the eye of all ex-pat New
Zealanders: it’s Couch’s arrangement of Princess Te Rangi Paire’s Hine e Hine, better known as Goodnight Kiwi, due to it being used to
signal the midnight shut-down of NZ’s Channel 2 TV for nearly two decades. Fabulous CD on all three counts: its sound quality, Couch’s
performance and the repertoire!
Needing something just a bit more meaty to test the bass regions and the extreme treble, I span up Tim Bruer’s rather unimaginatively
titled debut CD (The Tim Bruer Quartet). I did so because previously listenings had revealed that Bruer’s music is far more imaginative
than his album titles, though I have to say that it’s Sean Coffin’s work on tenor sax that I find most interesting on this CD. I didn’t have to
go further than Brett Hirst’s long bass solo
on Like John to hear that the P33s could really deliver in the nether regions of low bass, seeming to go down and down effortlessly and
without any loss of volume at all. The sparse sound of the quartet allowed the tweeter to strut its stuff thanks to Simon Barker’s high-hats,
with the treble positioned exactly where it should be and not even a hint of being too ‘hot’ in the mix. I also couldn’t not mention the sound
of Bruer’s piano which is full and warm. This, plus the fact that Bruer is in top form made me wonder why engineer Ross A’hern leaves it
so distant in the mix at both Sony and Megaphon studios. The tracks recorded at the Q Studios bring it more to the fore, but only because
everything else is pushed back. Still, it’s really only because the Whatmough P33s are so revealing that I was able to pick this so easily.
Vocals through the P33s are sublime. If you need to be reminded of one of the three-way speaker design’s major strengths, just put on a
CD with vocals: male, female or child prodigy, and settle back. The crystal clarity and etched detail you’ll hear from the P33s will be their
own object lesson. I played Joanna Loth’s interesting new CD Slow Secret Smile which has her singing all her own original tunes,
including the title song. Her unforced, natural voice dominates against the sparse piano (Steve Newcomb) and double-bass (Georgia
Weber) backing and the P33s let me follow—and thoroughly enjoy—every note, every subtle nuance, until the end of the CD came around
all too soon (which it actually always does anyway, because at a shade over 20 minutes, it’s far too short, for all reasons…)
Conclusion
In the several months these were in my listening room, there wasn’t a single visitor who wasn’t knocked out by the sound issuing from the
P33s, but I have to say that I was rather surprised at the range of comments about the design, particularly the most obvious element, which
is that ‘cows lick’ on the top. I was surprised because although I prefer the graphite finish, rather than the Bubinga finish that was on my
review pair, I think the speakers look fabulous—and that’s notwithstanding the fact that I am probably appreciating the technical reasons
for the cabinet shape as well as the aesthetics—so I was really surprised to find such a divergence of opinion, sometimes quite strongly
expressed! However, when it comes to sound quality, there was no divergence of opinion whatsoever. The sound is fabulous— everything
you could want in a pair of loudspeakers, and when you consider not only the sound, but the parts that go into making up the whole, you’ll
see that at an asking price of less than five grand, the P33s are a steal.
LAB REPORT
Readers interested in a full technical appraisal of the performance of the Whatmough Performance 33 Loudspeakers should continue on
and read the LABORATORY REPORT published on the following pages. All readers should note that the results mentioned in the report,
tabulated in performance charts and/ or displayed using graphs and/or photographs should be construed as applying only to the specific
sample tested.
Test Results
The measured performance of the Whatmough P33 loudspeakers was excellent. Graph 1 shows the frequency response that results from
feeding the Whatmough P33s a band limited pink noise signal, which requires the amplifier to simultaneously produce every frequency
between 20Hz and 20kHz at the required level. It’s a far more diffi cult test than either the gated sine technique (LMS) or the impulse test
(MLSSA, et al). In this graph, the ‘raw’ unsmoothed trace is shown, measured at 3.0 metres, with a 2.83V input (nominally one watt,
assuming an 8Ω nominal impedance). You can see that it’s spectacularly fl at, particularly in the region spanning 250Hz to 8kHz. Below
250Hz there’s a slight ‘bump’ in the response before the roll-off to 20Hz, and above 8kHz, there’s a slow roll-off to 20kHz. However,
overall, this raw response extends from 22Hz to 20kHz ±4dB.
The second graph in the series (Graph 2) shows a room-averaged response derived from measuring the frequency response from various
points in a listening room, then averaging all the responses to get a combined trace. After averaging, the trace is then smoothed with a
1/3rd octave filter, which accounts for the smooth appearance of the trace compared to that shown in Graph 1. Using this test method, the
Whatmough P33 returned a frequency response of 30Hz to 20kHz ±3dB and 40Hz to 20kHz ±2dB. You can see that the overall
characteristic of the trace is basically unchanged from the standard on-axis graph, with the response elevated by 2.5dB around 100Hz and
rolling off very slightly above 10kHz. Once again, however, you should note the linearity between 200Hz and 10kHz, where the response
deviates by less than ±1dB. Note that in all the measurements so far, the 20kHz upper limit was imposed by the test method. The P33
tweeter’s response extends well above this, as we will discover…
One difficulty with using pink noise as a test signal is that it has very high energy content at high frequencies, which tends to stress
tweeters, causing them to heat up more than they would when playing music signals, which in turn causes high-end roll-off. In Graph 3,
Newport Test Labs has measured the high-frequency response of the Whatmough P33s using a gated sine test signal instead. This test
signal places almost no stress on the tweeter and also has the advantage that in combination with computer analysis, it allows acquisition of
the response that would be obtained if the speaker were being measured in an anechoic chamber. It also allows a higher degree of precision
in the measurement. (Note, however, that this is only true if sufficient data points are processed. Newport Test Labs measures 500 data
points from which the trace is then derived. Many frequency response graphs are derived using far fewer data points—some only a few
dozen—which delivers a trace that ‘appears’ fl at, but is only so because of the lack of data between the points means the graphing
software has simply been a ‘join the dots’ exercise and is not representative of the speaker’s real performance.) As you can see, even with
500 data points measured, the Whatmough P33’s response is very fl at, being ±1.6dB between 500Hz and 40kHz. You can see that
although the tweeter rolls off very slightly from 10kHz to 15kHz, it picks up again and rises to 20kHz before another minor dip and rise to
40kHz. Remember that this is a completely unsmoothed trace, and the very minor peaks and troughs visible on the trace would be
completely inaudible. By adding the extra high-frequency data from Graph 3 to that already shown in Graph 2, we can extrapolate the
Whatmough P33’s response as 32Hz to 40kHz ±3dB, slightly better than Whatmough’s specification.
The impedance trace (Graph 4) shows the main system resonance at 55Hz, where it peaks at 15Ω. The minimum impedance occurs at
100Hz, where the Whatmough P33 is just 3.5Ω. This means that under International Electrotechnical Committee (IEC) rules, the P33
would be classified as having a ‘nominal’ impedance of 4Ω, which is exactly what Whatmough claims for it. However, because the
impedance modulus lies for the most part above 5Ω it would, however, appear to the driving amplifier as a 6Ω load. Newport Test Labs
measured both left and right speakers and you can see that the traces, which are overlaid, are almost identical, which shows that
Whatmough’s driver matching, cabinet fill and quality control processes are absolutely excellent.
The phase trace (also shown on Graph 4) shows the speakers phase most varies by no more than 30°, so there won’t be any real drive
issues, although at 80Hz, where the impedance is 4.5Ω and the phase angle is –45° indicates to me that you’ll extract best performance
from the P33s if you use a fairly high powered amplifier with a linear power supply that is comfortable delivering at least its rated power
into 4Ω loads. The black trace running along the 3Ω line is simply a calibration control. (Some readers have asked the reason for this. It is
because at such low impedances, it’s quite easy for very small measurement errors to creep in and affect the results. By using the same
cable/connector test set-up to measure a precision resistor, we can be absolutely certain of the accuracy of the graphed traces, which cannot
be said of any graph that does not include such a reference.)
Newport Test Labs measured the Whatmough P33’s sensitivity as 87dBSPL at one metre, with an equivalent 2.83V input. This is a little
shy of Whatmough’s 88dBSPL specification but is actually an excellent result since Newport Test Labs uses an equivalent 2.83-volt input
(i.e., the voltage level that would produce one watt across 8Ω) to drive the speaker and then averages the output over the entire audio band
to reach a single-figure result, rather than stating output at a single frequency which is the standard industry practise simply because it
gives higher SPL figures.
Test Results
The measured performance of the Whatmough P33 loudspeakers was excellent. Graph 1 shows the frequency response that results from
feeding the Whatmough P33s a band limited pink noise signal, which requires the amplifier to simultaneously produce every frequency
between 20Hz and 20kHz at the required level. It’s a far more difficult test than either the gated sine technique (LMS) or the impulse test
(MLSSA, et al). In this graph, the ‘raw’ unsmoothed trace is shown, measured at 3.0 metres, with a 2.83V input (nominally one watt,
assuming an 8Ω nominal impedance). You can see that it’s spectacularly fl at, particularly in the region spanning 250Hz to 8kHz. Below
250Hz there’s a slight ‘bump’ in the response before the roll-off to 20Hz, and above 8kHz, there’s a slow roll-off to 20kHz. However,
overall, this raw response extends from 22Hz to 20kHz ±4dB.
The second graph in the series (Graph 2) shows a room-averaged response derived from measuring the frequency response from various
points in a listening room, then averaging all the responses to get a combined trace. After averaging, the trace is then smoothed with a
1/3rd octave filter, which accounts for the smooth appearance of the trace compared to that shown in Graph 1. Using this test method, the
Whatmough P33 returned a frequency response of 30Hz to 20kHz ±3dB and 40Hz to 20kHz ±2dB. You can see that the overall
characteristic of the trace is basically unchanged from the standard on-axis graph, with the response elevated by 2.5dB around 100Hz and
rolling off very slightly above 10kHz. Once again, however, you should note the linearity between 200Hz and 10kHz, where the response
deviates by less than ±1dB. Note that in all the measurements so far, the 20kHz upper limit was imposed by the test method. The P33
tweeter’s response extends well above this, as we will discover…
One difficulty with using pink noise as a test signal is that it has very high energy content at high frequencies, which tends to stress
tweeters, causing them to heat up more than they would when playing music signals, which in turn causes high-end roll-off. In Graph 3,
Newport Test Labs has measured the high-frequency response of the Whatmough P33s using a gated sine test signal instead. This test
signal places almost no stress on the tweeter and also has the advantage that in combination with computer analysis, it allows acquisition of
the response that would be obtained if the speaker were being measured in an anechoic chamber. It also allows a higher degree of precision
in the measurement. (Note, however, that this is only true if sufficient data points are processed. Newport Test Labs measures 500 data
points from which the trace is then derived. Many frequency response graphs are derived using far fewer data points—some only a few
dozen—which delivers a trace that ‘appears’ fl at, but is only so because of the lack of data between the points means the graphing
software has simply been a ‘join the dots’ exercise and is not representative of the speaker’s real performance.) As you can see, even with
500 data points measured, the Whatmough P33’s response is very fl at, being ±1.6dB between 500Hz and 40kHz. You can see that
although the tweeter rolls off very slightly from 10kHz to 15kHz, it picks up again and rises to 20kHz before another minor dip and rise to
40kHz. Remember that this is a completely unsmoothed trace, and the very minor peaks and troughs visible on the trace would be
completely inaudible. By adding the extra high-frequency data from Graph 3 to that already shown in Graph 2, we can extrapolate the
Whatmough P33’s response as 32Hz to 40kHz ±3dB, slightly better than Whatmough’s specification.
The impedance trace (Graph 4) shows the main system resonance at 55Hz, where it peaks at 15Ω. The minimum impedance occurs at
100Hz, where the Whatmough P33 is just 3.5Ω. This means that under International Electrotechnical Committee (IEC) rules, the P33
would be classified as having a ‘nominal’ impedance of 4Ω, which is exactly what Whatmough claims for it. However, because the
impedance modulus lies for the most part above 5Ω it would, however, appear to the driving amplifier as a 6Ω load. Newport Test Labs
measured both left and right speakers and you can see that the traces, which are overlaid, are almost identical, which shows that
Whatmough’s driver matching, cabinet fill and quality control processes are absolutely excellent.
The phase trace (also shown on Graph 4) shows the speakers phase most varies by no more than 30°, so there won’t be any real drive
issues, although at 80Hz, where the impedance is 4.5Ω and the phase angle is –45° indicates to me that you’ll extract best performance
from the P33s if you use a fairly high powered amplifier with a linear power supply that is comfortable delivering at least its rated power
into 4Ω loads. The black trace running along the 3Ω line is simply a calibration control. (Some readers have asked the reason for this. It is
because at such low impedances, it’s quite easy for very small measurement errors to creep in and affect the results. By using the same
cable/connector test set-up to measure a precision resistor, we can be absolutely certain of the accuracy of the graphed traces, which cannot
be said of any graph that does not include such a reference.)
Newport Test Labs measured the Whatmough P33’s sensitivity as 87dBSPL at one metre, with an equivalent 2.83V input. This is a little
shy of Whatmough’s 88dBSPL specification but is actually an excellent result since Newport Test Labs uses an equivalent 2.83-volt input
(i.e., the voltage level that would produce one watt across 8Ω) to drive the speaker and then averages the output over the entire audio band
to reach a single-figure result, rather than stating output at a single frequency which is the standard industry practise simply because it
gives higher SPL figures.
Whatmough P33 Loudspeakers
Brand: Whatmough
Model: P33
Category: Loudspeakers
Warranty: Five Years
Distributor: Whatmough Pty Ltd