Harbeth Monitor 30 Speaker Review

[thewas]

I presume they are just better because they are the product of a lot of refinement over the years. The issue is indeed that all these published measurements are using different methodologies, but I would be surprised if there has not been a meaningful improvement over such a long period of time. Testing old gear can be fun, and may occasionally come up with some pleasant surprises, but we should be careful to imply that tests of old gear tells us much about their modern descendants. Technology moves on.

Some newer models have been measured by Stereophile, for example the "successor" 30.2 40th Anniversary Edition which seems nicely linear in the listening window

only the directivity has similar issues

Source: https://www.stereophile.com/content...-anniversary-edition-loudspeaker-measurements

 

[Thomas_A]

Some newer models have been measured by Stereophile, for example the "successor" 30.2 40th Anniversary Edition which seems nicely linear in the listening window

only the directivity has similar issues

Source: https://www.stereophile.com/content...-anniversary-edition-loudspeaker-measurements

Since there is an on-axis dip 3-4 kHz the directivity energy issue is not as bad as seen in the Stereophile plot.

 

[thewas]

Since there is an on-axis dip 3-4 kHz the directivity energy issue is not as bad as seen in the Stereophile plot.

Yes, although the on-axis dip is just around 2 dB while the off-axis deviation quite higher.

 

[Christoph-ASR]

After discovering that the dimensions (WxHxD) of Harbeth speakers follow prime number ratios—a knowledge gained from my tenure at Mercedes in the acoustics department, where gears with prime-numbered teeth exhibited reduced resonance—I delved into various realms of research. Having previously worked in the department of a Nobel Prize Committee member, I contributed to the development of a groundbreaking frequency analysis algorithm named the Wigner-Ville Transformation. This innovative approach addresses the signal processing uncertainty principle problems associated with the classic Fourier transformation, predating quantum mechanics.

While working on a novel speech recognition system with neural networks (pre-dating Geoffrey Hinton's breakthrough with deep neural networks), I explored the transformation of harmonic resonances into broadband intermodulation distortions. This led me to contemplate the reverse calculation of IMD. Drawing inspiration from the heat conduction equation formulated by Joseph Fourier for analysis, I conjectured its applicability in explaining the characteristics of BBC speakers. However, it is crucial to acknowledge that this explanation is unidirectional and not reversible, presenting an analytical unsolvability.

Furthermore, my experience in a phonetics department reinforced my belief in its capacity to elucidate the intelligibility of midrange frequencies. It's imperative to highlight that this methodology does not facilitate the construction of speakers in reverse. I refrain from referencing the numerous listening tests conducted by the BBC. The hypothesis necessitates further exploration of phonetic properties, a subject to be continued.

Details can be found here: https://www.instagram.com/clauer2014/

Here a Software from me: Sonogram Visible Speech

GitHub - Christoph-Lauer/Sonogram-Visible-Speech: A speech and sound anallysis tool.

A speech and sound anallysis tool. Contribute to Christoph-Lauer/Sonogram-Visible-Speech development by creating an account on GitHub.

github.com

I implemented also a (new) time-frequency analysis, called wigner-kernels, as can be seen in the last image.

 

[mSpot]

Are you using extra tweeters on top of your Harbeths? What were you trying to accomplish and what was the result?

 

[thewas]

a knowledge gained from my tenure at Mercedes in the acoustics department, where gears with prime-numbered teeth exhibited reduced resonance

That prime number ratios are preferred has nothing to do directly with acoustics but so that always different teeth are in contact to minimise/equalise wear. https://infosys.beckhoff.com/index..../ax5000_function_doku_hw2/1153316875.html&id=

 

[Christoph-ASR]

What about the Samambo-Slicing? (Strahlensatz) I'm not entirely sure, but I'm contemplating fortifying and refining my sense of suspicion.

 

[thewas]

What about the Samambo-Slicing? (Strahlensatz) I'm not entirely sure, but I'm contemplating fortifying and refining my sense of suspicion.

What should Strahlensatz have to do with gear ratios and their wear? By the way googling "Samambo-Slicing" doesn't give any result except your above post.

 

[DSJR]

Are you using extra tweeters on top of your Harbeths? What were you trying to accomplish and what was the result?

I was going to ask the same and also to recommend toeing the speakers in to face the listener in the 'hot seat.' The dispersion character in these speakers seems to be lower in frequency than a 'super tweeter' works at, so you could be inadvertently making hf dispersion worse by adding the super-tweeters (what frequency do you know them to come in at as many come in far too low (under 10kHz)...



P.S. I worry about record 'weights' on cheaper turntables as the main bearings aren't always up to it, even Rega ones...

 

[DSJR]

Got to add -

All this discussion about directivity/dispersion not being as good as the fave-rave Klippel wonder-boxes tested here (I'm not being especially combative here) but NOT A WORD on how well the ports work by not taking off or resonating in the upper mids as so many smaller porty speakers do! Surely this has got to help in reproduction quality.

Reading back and comments on the SHL5's like mine having vibrating side walls, mine don't at all at the levels I play at here (I should dig out my sound level meter to see how quietly I play at these days). My fifty year old Spendors are seriously worse with percussive music sources and they always were (so much for damped birch ply boxes back then).

 

[Theta]

This is a review and detailed measurements of the Harbeth Monitor 30 25th anniversary speaker. It is on kind loan from a local member. This specific speaker came out around 2002 I think and cost US $3,200 around that time. Being patterned after the famous BBC LS 5/9 speakers, I imagine the design is still the same to this day. The latest model seems to go by 30.1 designation.

The Harbeth logo was in danger of coming off so the owner pulled it off. Here is what it looks like (I know, very bland):

View attachment 47512​

It barely fit in my little photo booth so not very large. The grill is tricky to remove so I did not attempt to do so. I am told the recommendation is to use it with the grill so that is how I measured it.

The back panel shows ability for bi-wiring:

View attachment 47513​

Love the bit about the speaker being in their log book. Well, its measurements are now in our log book as well.

Let's get into measurements. Since this is a passive monitor, I followed the CEA 2034 recommendation of feeding it 2.83 volts. Temperature was a balmy 57 degrees. Altitude was 13 feet above sea level. Acoustic center of the speaker was determined following ITC-2025 Part B standard (I used my Mitutoyo micrometer). Any wind in the lab was self-generated due to previous meal.

CEA-2034 Spinorama Measurements
As usual we start with our master dashboard of the speaker which is its "spin" data:
View attachment 47514

I have gotten requests for "clean" versions of this picture so I have avoided making notations on it. Let me know if that is a good or bad idea.

There is a lot of controversy around the so called "BBC dip" where around the crossover frequency the on-axis response drops fair bit. Tom Holman of the T in THX recommended that the same dip be implemented in Audessey Room EQ! A mistake in my opinion. Anyway, we see the dip starting around 3 kHz. Getting a 3-D plot of that region gives us this complex sound intensity:

View attachment 47515​

I don't think this is what you want in a speaker. Put another way, it is the classic engineering work of calling a defect a feature. We can see the havoc it plays on our early window (and sound power) directivity:
View attachment 47516

Why would I want to have a suck out in that region of audible spectrum? I let you google forever to read the fights back and forth.

Taking into account the direct and early window reflections to predict the in-room response, we get:

View attachment 47517

The dip is there of course around the crossover frequency. And general unevenness. The overall tilt is quite exaggerated so perhaps good for people with high sensitivity to high frequencies, it would tame the harshness in that region in the recording.

Basic Speaker Measurements
Let's start with impedance and phase measurements:
View attachment 47519

Specification is 6 ohms. Standard requires staying about 80% (?) of this value which would be 4.8 ohm. If you cheat a bit it gets there. Worst case phase is around 100 Hz which impedance around 9 ohms. This make the effect load to be much lower (i.e. the difficulty that the amplifier sees).

Sensitivity ("efficiency") is almost always given as a single number by the manufacturer but it varies based on frequency:

View attachment 47520

In most of the frequency range the actual sensitivity is better than specified 85 dB which is good.

You all seem to like colorful 3-D graphs so here is the so called "waterfall:"

View attachment 47521

I don't trust this graph much. For one thing, these graphs can be manipulated to show any and all things. More importantly though, the Klippel NFS system I use is great for measuring frequency response but on tests like this, its structure can resonate with the speaker, which would show up as its own slices above. Maybe after I test more speakers I get confidence on reliability of this (rather useless) measurement.

Finally, step function seems to be a popular item so here it is, zoomed properly now to show the step response of each driver:

View attachment 47522

The tweeter is always "faster" so it generates a response first. It is swinging negative which means it is out of phase with the woofer which takes its time to generate its response to the right of it at opposite phase. There is discontinuity between the two (around 2.8 milliseconds).

Advanced Speaker Measurements
Here is the rest of the bits for those interested:
View attachment 47524


View attachment 47525

View attachment 47526

View attachment 47592

View attachment 47593
Full spin data is enclosed.

Correlation With Other Measurements
Stereophile measured a much more recent version and reported this for frequency response:

Focus on the black line only. It looks a heck of a lot better than our graph. Almost flat and smooth on Axis! Well, that is not what is really going on.

First the easy part, it has the same bass hump as we do (although their measurements usually exaggerates this due to their close-in measurement technique):

View attachment 47514

We show a dip post our peak in low frequency but they don't. The reason is that they are using gated measurements to eliminate echos and that hugely reduces resolution in a few hundred hertz (to one or two data points). Past that, they do have the dip but they show such a low resolution compressed graph that it is not nearly as visible. We both show the couple of peaks close to 10 kHz.

Overall, given the radically different measurement setups, I say the correlation is very good. Our data is much higher resolution than theirs.

Listening Tests
No time or setup yet for listening tests. Will do some and report back later.

Conclusions
It is interesting to measure these "classic" speakers to finally put some hard data behind their objective performance. As expected the halo is more positive than the objective data indicates. The cost of these speakers is way up there. I appreciate being built in England and needing to have good supply of bear for works after hours but it is a lot of money for this kind of measured performance.


------------
As always, questions, comments, recommendations, etc. are welcome.

Had some great sushi at my favorite spot when I went to pick this and other speakers. It was over US $60 for lunch! While I feel satisfied, I also feel poor. So please help lift my spirits by donating using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

Low Tech, High price, good review, here in Europe Harbeth speakers often win best of show, I have often wondered why, they sound Ok, like any old bextrene cone kef speakers you can buy for a couple hundred quid on the used market.

 

[ahofer]

Low Tech, High price, good review, here in Europe Harbeth speakers often win best of show, I have often wondered why, they sound Ok, like any old bextrene cone kef speakers you can buy for a couple hundred quid on the used market.

That’s a very old model, the newer ones are quite different.

 

[sktn77a]

That’s a very old model, the newer ones are quite different.

Not according to the manufacturer...... they are "subtle improvements on a sound original design".

 

[ahofer]

Not according to the manufacturer...... they are "subtle improvements on a sound original design".

Alan says a lot of (sometimes contradictory) stuff. Linearity is a lot better.

Here is the contemporary version of the same speaker. Compare to the OP

Harbeth Monitor 30.2 40th Anniversary Edition loudspeaker Measurements

Sidebar 3: Measurements

www.stereophile.com

 

[Willem]

That’s a very old model, the newer ones are quite different.

And do not use bextrene drive units.

 

[sktn77a]

And do not use bextrene drive units.

Neither did the M30 - it used the Radial 1. The M30.1 and on used the Radial 2.

ahoffer - you can't compare the M30.2 FR plot from stereophile with the M30 FR plots in the OP - they are take under totally different measuring circumstances.

 

[Deleted member 21219]

Neither did the M30 - it used the Radial 1. The M30.1 and on used the Radial 2.

ahoffer - you can't compare the M30.2 FR plot from stereophile with the M30 FR plots in the OP - they are take under totally different measuring circumstances.

If you wish someone to notice your post, presenting their username or handle like this

ahoffer

won't do it. Instead, use the "at sign". (@) and reproduce their username exactly as shown,

@ahofer

and they will get a notice.

Jim

 

[MattHooper]

Low Tech, High price, good review, here in Europe Harbeth speakers often win best of show, I have often wondered why, they sound Ok, like any old bextrene cone kef speakers you can buy for a couple hundred quid on the used market.

I guess it depends what you’re looking for in a speaker, taste, etc.

I’ve been at shows were Harbeth loudspeakers sounded to me more natural with the human voice than any other speaker I heard there at any price.

 

[milosz]

Seems to me that Amirm's judgement of the measurement of a speaker is "how close does it come to overall flat response, low distortion and even directivity" - this is the criteria for a neutral speaker, such as might be used as a tool in a recording or mastering studio to judge what's been captured in the recording. Seems to me that such "linear" speakers can often sound quite good in domestic listening situations, but not everyone wants this kind of sound. Harbeth, the BBC, and many other classic British speakers offer a different "tuning" - which is to say different colorations. Rather than try to achieve perfect linearity (which is impossible anyway) they opted for characteristics that suited their purpose. These classic British speakers all evolved out of research done by the BBC which was carried out in order to find speakers that suited the BBC's purpose- which is to say RADIO BROADCASTING. Since in radio we don't see the speaker's face / lips, we need a little acoustic help in order to ensure intelligibility. Hence the reason it's always said that voices on these speakers sound better than voices on other speakers - these speakers act as filters to improve intelligibility, to emphasize the voice so the guys at the mixing desk can get the settings right for the radio audience. And so on top of favoring the vocal range the BBC wanted to have a decent overall response, so monitoring classical music would be OK too.

I had a pair of Rogers LS3/5a's and they have a pleasant sound on vocals, folk, small ensemble jazz, and quite a lot of classical except maybe bombastic large-scale orchestral pieces. But not that great for modern IDM, rock-n-roll etc. And forget about hip hop or reggae- not nearly enough bass performance.

 

[Floyd Toole]

research done by the BBC which was carried out in order to find speakers that suited the BBC's purpose- which is to say RADIO BROADCASTING.

I visited the BBC research labs years ago and was toured around by Harwood. The BBC has a history of useful research, but budgets had been much reduced, Their loudspeaker designs were of course aimed at radio broadcasting, but also for use in their small, acoustically dead studios, listening at various distances, some quite short in the several ones that I saw. The engineers heard mainly direct sound, well represented by the on-axis frequency response. The measurements therefore show flattish on axis and typical two-way off-axis dip as the woofer becomes directional just before crossover to the tweeter. This defect is more audible in typically reflective domestic rooms where off-axis sound is a significant part of what is heard. Whether it flatters certain sounds or not, it is not as neutral as it could be in normally reflective rooms. Modern two-way designs often employ shallow waveguides on the tweeters to permit crossover at lower frequencies, better power handling, and a better match to the directivity of the woofer/mid transducer at the crossover frequency. This is one reason why three-way loudspeakers are a step-up design.