Harbeth Monitor 30 Speaker Review

[carlob]

I reconciled myself long ago to the idea that what I hear on my Harbeths (again the SHL5+AE) might not be textbook accuracy. My subjective take, FWIW.

My wife's a viola player, so that makes a good reference point for me, and we go to classical concerts about once a week. So I pay particular attention to string instrument timbre, and it seems unusually right to me. They also sound like much larger speakers, are easy to listen to for long periods of time (not "harsh", "lean", or "bloated", in my own subjective terminology), and I perceive good resolution of multiple instruments in symphonic music. Listening fatigue and instrument resolution are attributes I've had trouble with in other speakers I've used and auditioned. The Harbeths were so different in these ways, using the same playlist used with all the other auditions, that I kind of sat up and took notice in the audition (compromised by audio memory, etc., of course).

I wonder if any of the features/bugs played a role. Different speaker, so hard to know.

I'd share the same subjective opinion, I own SHL5 and Compact7-ESR and I absolutely love them, listening mostly to classical and jazz. Also I am a member of the Harbeth User Group forum since long time and can say that Alan Shaw (Harbeth owner and designer) is an hard core objectivist which in my book is very good. Will be nice to read his comments on this review, even if it's an old design.

 

[Tks]

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

Nice Linux memes

 

[dshreter]

Amir, thank you for measuring this (my) speaker here! Illuminating to see the results for sure.

These speakers are qualitatively known for being clear at low volumes, and maybe the BBC dip helps contribute to that. At closer to reference volumes then these would sound inaccurate.

Off axis has the same error in frequency response but is it fair to say they are well behaved to be correctable via EQ? Or is this the wrong interpretation of the spin?

Last, I want to point out for sake of fairness, the Monitor 30.2 is a different speaker than this the Monitor 30. So the Stereophile plot is interesting to compare but I wouldn’t necessarily go as far as refuting their measurements.

Thank you again for the detailed analysis!

 

[LTig]

Vibration can create chaotic signatures in the impedance trace. For example, vibration is very hard to control in free air measurements for Thiele Small parameters, causing noisy impedance traces that look similar to the Harbeths. I was discussing this with Ken Kantor many years ago, and he shared same experience "You really need to kill any parasitic vibration in the driver to get a clean, symmetrical impedance curve"

I'm not quite sure about vibrations. It's a different field, but usually mechanical vibrations in the vicinity of an NMR spectrometer are seen as sidebands with identical distance to the signal peaks. Here I would expect those sidebands then being at

'woofer resonance peak' +/- 'vibration frequency'.​

This is not the case, being at 63 Hz and 88 Hz they are not symmetrical to the woofer resonance peak at 80 Hz. But I'm not sure whether the analogy between NMR spectra and an impedance plot is valid.

 

[Pio2001]

Please do not kill the messenger...

Still find the range 5kHz to 10kHz a bit conspicuous
View attachment 47555

These variations around 8000 Hz differ from speaker to speaker. Here you have selected speakers that have them at the same frequencies. But the JBL 305P mkII measurement doesn't have them at the same frequencies.

I also have this kind of oscillations in my own measurements :

 

[napilopez]

Curious to see what amir thinks in his listening tests. The Harbeth outdoes the 305P in its preference score by a significant margin with(7.49 vs 6.43) our without a sub(5.03 vs 4.36). Not that that should colour your impressions or anything =]

 

[Ron Texas]

It must be worth it because it can be bi-wired, LOL.

 

[oldsysop]

These are the analyzes that interest me, real speakers and not toys.
Thank you very much Amir !!!

 

[Francis Vaughan]

I'm really pleased Amir got to test some Harbeths so early on, they are one of the design philosophies that is worthy of testing and understanding. Their deliberate use of somewhat resonant thin wood cabinets is important.

IMHO this is where a waterfall plot starts to help elucidate what is going on in a more clear manner. Sure, the same information is present in the frequency response, but it is also present in the impulse response, and we don't ignore the frequency response and just publish the impulse response and say "all the information is there". Representation is important. All engineers and mathematicians understand that a change of bases can take a difficult to interpret problem into one that is vastly clearer to manage and understand.

Coupling of cabinet resonances into the impedance is absolutely a thing. I find it odd how little this is exploited. Something I have started to do a lot is to experiment with box damping and observing the changes in the impedance plot. It can be faster and more reproducible than trying to tease the effects of changes in the frequency response. But that is a different topic.

There is probably a lot more interesting information buried in the test data. Specifically these speakers are designed to radiate significant energy from their cabinet walls. And those walls have a lot of stored energy in them. Hence the waterfall plots. Normally this would be something to root out kill in speaker design. Here it is deliberate. It is part of the signature of the brand, and lots of people like the brand sound. This isn't to be ignored. There is something important to be understood, and it is very likey that this is first time speakers of this design regime have been so fully measured. The results are potentially much more useful than just rating the speaker with some trivialising metric and moving on.

 

[napilopez]

I'm really pleased Amir got to test some Harbeths so early on, they are one of the design philosophies that is worthy of testing and understanding. Their deliberate use of somewhat resonant thin wood cabinets is important.

IMHO this is where a waterfall plot starts to help elucidate what is going on in a more clear manner. Sure, the same information is present in the frequency response, but it is also present in the impulse response, and we don't ignore the frequency response and just publish the impulse response and say "all the information is there". Representation is important. All engineers and mathematicians understand that a change of bases can take a difficult to interpret problem into one that is vastly clearer to manage and understand.

Coupling of cabinet resonances into the impedance is absolutely a thing. I find it odd how little this is exploited. Something I have started to do a lot is to experiment with box damping and observing the changes in the impedance plot. It can be faster and more reproducible than trying to tease the effects of changes in the frequency response. But that is a different topic.

There is probably a lot more interesting information buried in the test data. Specifically these speakers are designed to radiate significant energy from their cabinet walls. And those walls have a lot of stored energy in them. Hence the waterfall plots. Normally this would be something to root out kill in speaker design. Here it is deliberate. It is part of the signature of the brand, and lots of people like the brand sound. This isn't to be ignored. There is something important to be understood, and it is very likey that this is first time speakers of this design regime have been so fully mea Asured. The results are potentially much more useful than just rating the speaker with some trivialising metric and moving on.

This is a really interesting perspective I hadn't known about, using a resonant cabinet on purpose - I appreciate the insight. I've never heard a harbeth before. This is an old speaker and seems to do quite well. Curious to see how newer ones.

 

[carlob]

On the resonant cabinet: https://www.harbeth.co.uk/usergroup/threads/bbc-style-thin-wall-cabinets-why-so-special.215/

 

[ahofer]

It must be worth it because it can be bi-wired, LOL.

This came up on the Harbeth Forum. They added biwire capability in a couple of speakers because their marketing survey said it was a must-have, even though Alan Shaw thinks it adds nothing but a 'tiny theoretical improvement with a difficult speaker load'. So, to paraphrase the old Dentyne ad, 4 out of 5 speaker buyers surveyed demand biwiring for their high-end speakers.

 

[BYRTT]

...Any wind in the lab was self-generated due to previous meal...

... and it was recorded black curve is a starving day over at Stereophile, great thanks the review and hard work.

 

[jhaider]

Advanced Speaker Measurements

Horizontal and vertical polar maps?

 

[SplitTime]

It is a very high resolution sweep (0.73 Hz). The typical published ones are likely at much lower resolution. Do you want me to go in that direction?

Hi Amir,
Thanks again for all your work on these! Please, pretty please, do not go with low resolution data collection - having the higher resolution over time will make it much clearer where speakers are actually, measurably, distinct. That will help us all better understand how the measurements sync up with our individual listening expriences.

 

[Sancus]

Horizontal and vertical polar maps?

I know it's not exactly the same as the ones Amir's plugin generates, but pypolarmap does a pretty good job with this now that we have the data files.

Horizontal Contour

Spoiler: graphs

Vertical Contour

KH80 maps for reference

Horizontal Contour

Vertical Contour

 

[milosz]

Speaker measurements like these do not completely describe what a speaker will sound like. That said, these kind of measurements tell us important things about the speaker, and in particular they can be useful when designing a speaker.

I have not heard any Harbeth speakers except at shows under miserable conditions. There is a brick-and-mortar Harbeth dealer here in Chicago (Decibel Audio) and I have always meant to drop by and have a listen, as there are always favorable comments about Harbeth in various trade reviews.

The BBC dip- is that good with crisps?

 

[spacevector]

It is a very high resolution sweep (0.73 Hz). The typical published ones are likely at much lower resolution. Do you want me to go in that direction?

Perhaps you create an averaged/smoothed version in post-processing and present both overlayed?

 

[jhaider]

Horizontal Contour
View attachment 47584

Vertical Contour
View attachment 47585​

Thanks! Unfortunately I can't get pypolarmap to work on my MacBook Pro, and I don't even know enough about computers to know what to provide to determine why it does not work. See my post in the thread concerning this software.

I'm not sure why polar maps aren't part of the main measurement display. (Any chance you could also add these maps to the Revel center channel review thread?)

Here there is a huge dispersion disruption at the midwoofer-tweeter transition. It is best identified in polar maps.

 

[Sancus]

Thanks! Unfortunately I can't get pypolarmap to work on my MacBook Pro, and I don't even know enough about computers to know what to provide to determine why it does not work. See my post in the thread concerning this software.

I'm not sure why polar maps aren't part of the main measurement display. (Any chance you could also add these maps to the Revel center channel review thread?)

I don't see the C52 Horizontal/Vertical SPL.txt files anywhere... did Amir post them? If not, can't generate these graphs

I completely agree that polar maps should always be generated though, don't really get why they're not part of the main display either.

E: I ran it on Windows so can't help much with Mac, sadly.