Graham Audio LS5/9f Loudspeaker

REVIEW
by Robert E. Greene
Dec 23, 2020

The $7999-per-pair Graham Audio LS5/9f is essentially a floorstanding version of Graham Audio’s LS5/9, which is itself a contemporary embodiment of the medium-sized monitor released by the BBC in 1983. As is typical with the BBC, production of the original LS5/9 design was licensed to commercial manufacturers. And while it uses modern drivers, the quite recently issued Graham LS5/9 is close enough to the original’s specifications to qualify for an official BBC license, all these many years later.

The LS5/9 is thus a BBC speaker in the literal sense, while the LS5/9f is not. BBC monitor designs did not come in floorstanding versions. But the LS5/9f shares the drivers of the Graham LS5/9. And it shares the designer, too: Both were designed by Derek Hughes. Hughes has a deep connection with the BBC tradition. His father Spencer Hughes was the co-founder, with his wife Dorothy, of Spendor Audio, which produced the Spendor BC1 in 1968, a speaker that deserves to be called legendary if ever a speaker has so deserved. And the Spendor BC1 became the BBC LS3/6, another BBC speaker reissued in contemporary form in a design by Derek Hughes, this one for Stirling Broadcast. We are very much in the design presence of the BBC tradition with the LS5/9f.

Long-time TAS readers will have inevitably become aware that I have taken a strong interest in speakers of the BBC ilk for decades. And I still do; I reviewed the Graham LS5/9s with enthusiasm in Issue 270. And the interested reader can find reviews of other Hughes family designs at regonaudio.com, going way back. But there is more to my interest in these speakers than just my respect for tradition and the power of long exposure to the general type of sound they produce.

The BBC and its associated designers and engineers were part of a unique development in audio. Never before, and I rather suppose never again, were so many capable and creative engineers offered the opportunity to work on speaker design with almost unlimited resources and, most of all, with an ongoing opportunity to compare on an almost daily basis their work with live sound, and especially with live orchestral sound.

Of course, any serious speaker designer pursuing the goal of reproducing music as music actually is will compare the results of designs with the memory of live sound and perhaps with the literal sound of individual instruments and voices. But the chance to compare at will the sound of the speaker designs with a literally present orchestra is simply not available to ordinary commercial enterprises.

Others have run live-versus-reproduced demonstrations. Acoustic Research had their live-versus-canned public demos in the early days of stereo. And John Dunlavy played Blumlein-miked recordings of concerts on his speakers right after the concerts for the interested public. But these were occasional matters, not an actual part of the ongoing design work. The BBC work comparing their designs with live music all happened decades ago. But there has not been anything like it since. Today, people design by theory and listening in the abstract. No company today that I am aware of hires in an orchestra on a regular basis to check how they are doing. The expense would be prohibitive for a private company.

One might ask how much this mattered, this constant comparison to live orchestral sound. I believe it mattered a lot. Most speaker companies, then and now, design on the basis of models of speaker behavior and of theoretical ideals. They really have very little detailed information on how well their designs will do in reproducing real sound and what is crucial in a recording/playback paradigm. And “research” in the subject of speakers tends nowadays to be based on preference testing and/or agreement with unverified theories of how speakers ought to work. Unverified, and really unverifiable in the context that no comparisons can be arranged.

The results speak for themselves All you have to do is go around an audio show with a single recording to see how much variation there is among the speakers and how little match there all too often is with anything like live sound. A BBC-school designer (not Derek Hughes) said to me not many years ago that all one had to do to hear that there was a lot wrong with most speakers was to play a single recorded piano note and then play the same note on a real piano. This had to my mind an almost terrifying ring of truth. If a speaker cannot reproduce single piano notes, what hope for a full orchestra?

And Now the LS5/9f
The Graham LS5/9 was and is very good at producing, with the right recordings, sound that resembled the real thing. It suffered, however, from a limitation. While the bass was warm and full as low as it went, it vanished a bit too early to make large-scaled music completely convincing. It just did not have quite as much bass extension as one would like.

Powering it with the Devialet SAM [Speaker Active Matching] system was a revelation. The SAM system extends the bass by an ingenious electronic control of bass behavior, which has a built in limiting so that the speakers are not driven into damage or overload. In effect the speakers are EQed but in a level-dependent way. This really brought home to me how superb the LS5/9 could be if it just had a bit more bass extension. (The review does not identify the LS5/9 by name—I did not want to induce any prejudice about the LS5/9, as it was alone).

Enter the LS5/9f. The LS5/9f has the same driver complement as the LS5/9. But it has a larger enclosure. The floorstanding enclosure does not just function to lift the drivers into position. The woofer is actually radiating into a considerably larger interior space than that of the LS5/9. The LS5/9f’s enclosure is not dead-air space. It is functioning acoustically to give the mid/bass driver an actively involved, larger, ported enclosure, providing increased bass extension and dynamic capability.

The added bass extension thus provided does not sound like all that much in numerical description—40Hz versus 50Hz, –3dB. But that is just enough to give one the sense of full-range sound on most music. The bottom note of standard orchestral music is around 40Hz, and the same for the rock-music bass guitar. The LS5/9f is not a bass powerhouse for large pipe organs and earthquakes. But it has the orchestra covered—and rock bands.

This is not to suggest that anyone is likely to buy the LS5/9f because it is a bass powerhouse. But the larger enclosure has moved the response down far enough to eliminate the feeling of missing bass. For most music, you will not feel bass-deprived. The 10Hz additional extension in numbers may not seem like much, but in this frequency range it is, in audible effect, quite a lot.

Moreover, the LS5/9f is somewhat warmer in the lower regions than nominally flat. In room, the bass response is fairly abundant. This is a complex issue which I shall discuss in more detail in a moment. But in any case, the LS5/9f definitely does not suffer from puny little-speaker sound. And the LS5/9fs will also play quite loudly, with plenty of volume for a room of ordinary domestic size. Orchestras are convincing in dynamic scale, for example.

Musical Impressions
As it happened, I spent an unusually large amount of time with the LS5/9f. They arrived for review somewhat before the virus outbreak and, things being as they are, they stayed on partly because I was not finished with reviewing them but also because it was not really practical to return them safely. So, I ended up listening to them a lot. If one of the functions of audio reviewing is to evaluate the long-term satisfaction that a review item offers, I surely gave the LS5/9s a good run. And a most satisfying experience it was.

Listening to a speaker has short-term aspects and long-term aspects as well. The former is based primarily on perceived tonal balance, on frequency response in some sense of combination of anechoic and in-room responses. Changing quickly from one speaker to another brings response differences to the fore, so far to the fore as to all but obliterate other considerations.

Now the LS5/9f does well in the short term. (We shall get to long term in a moment!) They have a slightly idiosyncratic bass character, but overall they are quite “neutral” over the whole range—as speakers go—with certain characteristics that are distinctive. There is a vestige of the excess energy between 1kHz and 2kHz that the LS5/9 stand-mounts have, but that is easily fixed. The really high treble rolls a bit down. The tweeter level is switchable from 0 to nominal + or –. But none of the settings bring up the really high treble to the level that audiophiles have persuaded themselves is “accurate.” If “air” in the very high frequencies is your big thing, you could add a super-tweeter to lift the output from, say, 14kHz on up. And as noted the bass will tend to be warm. Finally, the speaker does best if it is far from side walls and/or there is high-frequency absorption at the first side-wall reflection points. Still, overall one is looking at a very nearly neutral sound, not surprising in a BBC heritage design. (In spite of the roll-off in the extreme top, I found a reduced tweeter level preferable. Otherwise there is a bit extra at 4-6kHz that is better not there.)

But this frequency range-by-range description is far from doing justice to the LS5/9f in long-term listening. For a variety of reasons, the LS5/9fs make recordings sound like live concert music to an unusual and very satisfying extent, far beyond their basically correct balance alone. There are more subtle aspects of their sound that add a lot to one’s musical satisfaction beyond just getting the tonal character of sound correct.

For one thing, these speakers are very coherent. It is all very well to claim that speakers with a lot of drivers and crossovers between them are coherent. But the ear/brain is not easily fooled. And there is in my experience some special attraction to a full-range speaker (or close to it) where one driver produces most of the sound. For one thing, the absence of a crossover means that the lower-order harmonics of midrange musical notes are in-phase with the fundamental and with each other. (Phase differences among fundamentals and lower-order harmonics are known to alter timbre, and such phase alterations always happen except with first order crossovers, or unless DSP correction is used.)

But there is a price to pay: If the driver is large enough to produce bass in a reasonable way, it will inevitably become “beamy” at higher frequencies. Two-way speakers with a high crossover point tend to sound their best for only one listener. But the benefits of a single driver, reproducing almost the whole range of piano fundamentals and their lower harmonics, for example, are considerable.

The LS5/9s sound like a point source even when you sit quite close to them. And this gives one, among other things, ideal stereo, aided by the excellent pair matching (in my review-sample pair anyway, but I rather suppose this will be true in general).

Another feature of the LS5/9f is that the mid/bass/driver is well behaved out-of-band. Polypropylene drivers can have the important property of not having “edge” arising from narrow-band high-Q break-up modes. In the contemporary world, a lot of manufacturers are using drivers of harder materials, which tend to have such “hard” breakup-mode behavior. This may not look like much in measurements of overall balance, with little resolution frequency by frequency. But it makes a difference. Such bad-driver behavior tends to give a hard and edgy sound, which at first might sound “detailed” but wears out its welcome over time.

The LS5/9f is very much free of this effect, much to its credit. The sound is non-edgy, independently of how one might set the tweeter controls or the balance by EQ adjustments. Nothing wears one down. Pianos, for example, have the character of real pianos, with no untoward “bang-iness.” This becomes more and more appreciated as time passes.

Another feature that makes the LS5/9fs welcome long-term visitors—or potentially permanent residents—is the bass energy already referred to. This is a point that audiophiles seldom discuss honestly. Harry Pearson used to say that he liked his midbass “lean.” Unfortunate preference! The truth is that when one plays a recording at less than live volume—and one really has to do this for various reasons—the famous Fletcher-Munson curves pull the perceived bass level down relative to the rest. This is a very large effect. Back when people seemed to care more about the exact perceived balance of audio, preamps had a “loudness” control for this reason!

Think of it this way. Up close to an orchestra, where microphones are typically placed, it is very loud. I have measured 100dB for short intervals in, say Scriabin, from the first row. You really do not want to have this level at home. In fact, you really do not want to have it at concerts, either. There is a good reason why most of the audience sits back a bit by preference. But if you drop the level of the recording by x dB, the bass level drops in relation to the mids by surprisingly close to that same x dB. The equi-loudness curves are spaced nearly twice as close together in the bass as in the midrange. Turning the overall level down reduces perceived bass.

The inconvenient truth here is that if you want to hear natural balance at plausible levels, a certain amount of bass excess over flat response is needed. I am not a fanatic about this. Indeed. when I was working with the Sigtech people years ago setting up their DSP room correction, they commented that I seemed to like flatter response than most people.

Still, everyone ought to know this problem is out there. All the room-correction people offer down-sloping “target curves.” Experiments about this in audio practice date at least back to Ortofon’s “target curve” before DSP correction existed.

Another feature of this is that it is much easier to attenuate bass then to boost it. If a speaker seems too warm for some particular recording, you can tone down the bass easily enough whereas turning bass up by EQ is a tricky business (except if one does an adaptive protection system like the Devialet SAM). All this adds up to the fact that it is a good thing having speakers with adequate bass energy—and maybe a little bit more.

Yet another aspect of the lower frequencies is that almost all floorstanding speakers tend to develop a dip between 100 and 300Hz from the Allison effect of the floor. Look at contemporary in-room measurements of most floorstanding speakers. There is rarely adequate energy in this crucial frequency range. This is one thing that makes audio sound also stupidly wrong to people that have an acute memory of actual music. But the effect is so widespread that a lot of people who spend more time listening to reproduced music than to real music have come to feel that it is normal—and even to like it!

One can EQ this away—the Allison effect affects the power response and is thus correctable with EQ. But it helps to have a speaker that has some extra energy in the sub-300Hz range. As it turns out, through correct placement and some judicious EQ, I found it quite easy to get the LS5/9f to sound really correct in this region. Indeed, with careful enough placement, it did this quite well without any EQ.

I do not want to leave any impression that the LS5/9f is “boomy” in any sense. Not so. But it does give music an appropriate warmth and fullness without turning the volume up excessively and in spite of “the usual floor dip,” as Martin Colloms used to call the frequently occurring hole between 100Hz and 300Hz. (Incidentally, the correction of this hole is almost always the largest effect of DSP room correction, the one thing where even really smooth speakers tend to change a lot when “corrected.”)

What It All Adds Up to Musically
The Graham Audio LS5/9f, with its warm balance and slightly rolled-off extreme top, is not typical of contemporary speaker designs. One thinks of Peter Walker’s dictum that a speaker succeeds or fails by 10kHz, though the actual top end –3dB point is 16kHz (manufacturer’s specification). Nor does the LS5/9f make an overt attempt to break new ground in terms of peculiar or unusual radiation pattern or driver type. But it is, in fact, unusually adept at making recordings sound musically correct. With everything set up just so—not too close to side walls and with first sidewall reflection points damped, correct placement to counteract Allison effect, tweeter level adjusted right (attenuated in my room), and perhaps a slight cut between 1kHz and 2kHz, something remarkably close to what is actually recorded emerges.

Piano recordings are startling in their realism. No bangy sound, no exaggerated ringing resonances, just the solid sound of real pianos. This is a rare and for me very valuable aspect of speaker sound. My old favorite, Freddy Kempf playing the Kreisler-Rachmaninoff Liebesleid on BIS [CD 1042] sounded almost precisely like a real piano, far more so than with most speakers. So did Volodos’s recording of his transcription of the slow movement of the Rachmaninoff Cello Sonata [Sony Classical SK 64834]. And Songs My Mother Taught Me [Mobile Fidelity CD877] with violinist Arturo Delmoni and pianist Meg Bachmann Vas, recorded by David Hancock using the legendary Cambridge ribbon microphones, offered excellent piano sound and superb violin sound as well, with a remarkable truth to timbre. Flat response is crucial for this but the phase coherence between fundamentals and harmonics also contributes to this effect of tonal truth. Flat response is crucial for this, but the coherence between fundamentals and harmonics arising from a single driver covering much of the frequency range also contributes to this effect of tonal truth. And the absence of cabinet and driver colorations also plays an important role. Here “neutral” is not a cliché buzzword but essentially the literal truth.

The same truth to timbre extends to more complex music. Orchestras sound really right, instrument by instrument. My standby favorite, Telarc’s Ravel-Borodin-Bizet disc [Telarc 80703], sounded right in timbre and also very detailed. The thin-wall damped-enclosure approach developed initially by BBC’s research pays off here: The speaker sound very “compact” in the sense that things stop when they should, and nothing rings so that the real ringing of music from reverberant halls or the like is revealed extraordinarily well. (The trumpet solos on the Telarc bounce around the hall with a tactile realism.) This “compactness” is I think likely attributable to very clean decay behavior in the midrange. In any case, you will know it when you hear it. The LS5/9f make most speakers sound somewhat ill-defined in the mids.

This is not all just a question of how classical music sounds. Sonic qualities apply across the board. And the impressive truth to timbre, non-resonant and highly resolved character, and spatial character were all pleasing evident in, for instance, “I’m Ready” from Dave Wilson’s Cruising with the DeSotos recording, which is anything but classical. It does, however, have natural voice recording and saxophone, too, and on the LS5/9fs, these were shining through. And the whole thing had the “compact,” well-defined sound I have been trying to describe, the sound associated with very rapid and clean decay, in the midrange especially.

The Overall Picture
Much as one may admire speakers designs which depart from conventional behavior in various ways, seeking new approaches to the whole question of speakers in rooms, the truth is that wonderful things can happen by carefully perfecting the right traditional approaches combined with modern driver technology. And so it is with the LS5/9fs.

I had no excuse at all for adding yet one more BBC School speaker to my permanent collection. But I was tempted.

Specs & Pricing

Type: Two-way, bass-reflex floorstanding loudspeaker
Enclosure: Damped thin-wall construction, birch plywood
Drivers: Diaphnatone polypropylene (developed and manufactured by Volt) mid/woofer, Son Audax HD13D34H tweeter
Frequency response: 40Hz–16kHz, +/-3dB
Nominal impedance: 8 ohms
Sensitivity: 89dB (2.83V, 1 meter)
Max. output: Over 104dB per pair, 2m
Recommended amplifier power: 50–200 watts unclipped program
Dimensions: 35 x 105 x 37cm
Weight: 25 kg
Price: $7995 standard cherry veneer finish (other custom veneers slightly higher)