advertisement


A look at the Harbeth P3 crossover

Saw that yesterday. I have a pretty low opinion (bordering on zero) of that YouTube channel, but the Harbeth crossover was 'interesting'.
 
Not all audiophile companies sign up for a belief in magic expensive crossover components. Indeed Harbeth has been pretty vocal in the past in not believing in magic expensive audiophiles amplifiers. The LS3/5A crowd seem to believe in complicated crossovers built from normal components and this speaker seems to fit in that camp.

Of course expensive non-linear components in a passive crossover has been both a low quality and expensive way to design and manufacture speakers for several decades now. It persists because some consumers strongly value other things and I would suggest it may be wise to take account of those when assessing the appeal of speakers like the Harbeth P3. Plus of course the chap doing the pontificating is far from a reliable source of information on technical matters.
 
Interesting. The previous speakers I had for 20 years were Marshall designed Monitor Audio R852MDs which I believe had a single component in the crossover, perhaps two? They sounded great but could they have been improved with one of his kits? Do more expensive components last longer, or drift less over time?

My Harbeth C7s sound very natural, grilles off.
 
Also saw that yesterday, and I do wonder why a small 2-way needs that many components. It would be interesting to see the raw driver responses.
 
Also saw that yesterday, and I do wonder why a small 2-way needs that many components. It would be interesting to see the raw driver responses.
That is something I wondered. However I have seen three-way crossovers, from KEF for example, that look rather complex - presumably to do a complex job for the particular drivers.

I am going to speculate that this review illustrates the misuse of technical over-generalization. Generalization that something is always wrong or always right, when the reality is that it depends on specific purpose and detail.

Am I missing something here in my ignorance of the details of loudspeaker crossover design? Are the generalizations actually correct?
 
The channel is obviously pure FUD/marketing to sell boutique “upgrades” etc, but likewise I’m surprised there is just so much in the little Harbeth. I understand it entirely with the LS3/5A as that was designed to fix the somewhat wayward Kef B110, something it did so astonishingly well it remains the benchmark against which mini-monitors are compared against today. It just couldn’t avoid complexity, and it is a real testament to BBC research that it works so well and is so easy to drive.

The Harbeth is decades more recent and was built around Alan Shaw’s (government/BBC-funded) ‘Radial’ driver cone compound so I’ve no idea why it needs so much electrical compensation. I’d have expected a modern driver to be pretty flat with an engineered treble roll-off. I’ve not pulled them to bits but I bet there is way, way less in the crossover of my Spendor S3/5Rs. It would be fascinating to model the P3ESR crossover to see exactly what it is doing! By saying that it consistently gets a Stereophile class B rating and has huge numbers of happy users so who cares?

PS Here’s a Compact 7ES crossover:

2380355519_9e7e0ac736_c.jpg


Again surprisingly complex given fairly modern purpose-designed bass-mid driver. Four iron core inductors. The same number as the 1970s JR149 which also used the B110 (IIRC two inductors in the 149 actually behave as one, just split to get it in the really tight space!). For comparison IIRC the Epos ES14, which predates the Compact 7 by decades, had just one capacitor on the tweeter. The bass driver was designed to run open and naturally roll.
 
Do more expensive components last longer, or drift less over time?

It is hard to know because the evidence that would tell us isn't readily available when it comes to expensive boutique crossover components. What is out there are masses of people talking about how good the various expensive components sound and, in many cases, taking a cut of subsequent sales. Not unlike audiophile cables and similar high margin audiophile accessories.

Having said that, the components in passive crossovers are non-linear particularly at higher voltages. Some will perform better than others and the differences are likely to be audible. The presence of vast amounts of "subjective" information and the lack of technical information demonstrating that expensive boutique components perform better rather suggests one shouldn't expect a strong correlation between price and technical performance. But one should expect the components to perform differently. Air and iron core inductors mentioned in the video differ substantially in performance but where one is appropriate in terms of cost/performance/size/... relative to the other requires context.

If technical performance is significant then opt for active crossovers which will not only avoid significant issues with non-linearity, they will generally be more flexible and able to adjust the response more tightly, can be adjustable when the mechanical properties of the drivers change with age or differ when new, will be cheaper (for manufacturers but not necessarily DIY folk), more energy efficient, etc... Passive vs active crossovers seem to be one of those curiosities where technically there are only pros and no cons on one side. Demand for speakers with passive crossovers seems to exist for people reasons rather than anything technical which is fair enough.
 
Not all audiophile companies sign up for a belief in magic expensive crossover components. Indeed Harbeth has been pretty vocal in the past in not believing in magic expensive audiophiles amplifiers. The LS3/5A crowd seem to believe in complicated crossovers built from normal components and this speaker seems to fit in that camp.

Of course expensive non-linear components in a passive crossover has been both a low quality and expensive way to design and manufacture speakers for several decades now. It persists because some consumers strongly value other things and I would suggest it may be wise to take account of those when assessing the appeal of speakers like the Harbeth P3. Plus of course the chap doing the pontificating is far from a reliable source of information on technical matters.

I don't think he is refering to replacing the existing with "magic expensive crossover components", only "good/better performing" ones.
 
I don't think he is refering to replacing the existing with "magic expensive crossover components", only "good/better performing" ones.
Yes, and replacing components may please the 'speaker owner - that's fine (I have another hobby where I am definitely obsessed by style as well as function).

However, when existing components perform sufficiently well then replacing them with "better performing" ones makes no functional difference. That's why I asked earlier about whether / how much over-generalization there is in the video.
 
interesting video, thanks. I found the comments ring quite true also knowing the Harbeth philosophy. They’ve been known for overcomplicated crossovers, big talk and pretence. I am not surprised at all, including cramping inside all the cheapest possible components and hypocritically putting “studio grade ultra pure cable” on the back of the speaker! I thought fair comments on their specific voicing too having owned several of their speakers (2 x P3s).
 
Also saw that yesterday, and I do wonder why a small 2-way needs that many components. It would be interesting to see the raw driver responses.

That is something I wondered. However I have seen three-way crossovers, from KEF for example, that look rather complex - presumably to do a complex job for the particular drivers.

I am going to speculate that this review illustrates the misuse of technical over-generalization. Generalization that something is always wrong or always right, when the reality is that it depends on specific purpose and detail.

Am I missing something here in my ignorance of the details of loudspeaker crossover design? Are the generalizations actually correct?

I would argue that the more electronic components are placed in the signal path the more it (the signal) deteriorates.

It looks like the tiny woofer needs a massive amount of EQ'ing to achieve a flat frequency response.
This is probably one of the reasons why it needs as much power.


A photo of the Striling Broadcast BBC LS3/5a V3 crossover side by side with I presume the V2:

alW4yzd.jpg



Harbeth P3ESR's crossover:

anEjNTO.jpg



And the response of the Kef B110:

lVsK7KT.jpg



It looks like the original LS3/5a crossover design was just as complex as Harbeth's:

j0G30xE.png
 
Yes, and replacing components may please the 'speaker owner - that's fine (I have another hobby where I am definitely obsessed by style as well as function).

However, when existing components perform sufficiently well then replacing them with "better performing" ones makes no functional difference. That's why I asked earlier about whether / how much over-generalization there is in the video.

I've only watched a handful of his videos; I think that sometimes he redesigns the crossover, like Troels Gravesen, but in this case he seems happy with what the crossover is doing in therms of EQ but not with the quality of the parts used.

It's interesting to note that Derek Hughes has completely redesigned the crossover for the latest Version 3 Stiling LS3/5a (image posted above).
 
That is something I wondered. However I have seen three-way crossovers, from KEF for example, that look rather complex - presumably to do a complex job for the particular drivers.

I am going to speculate that this review illustrates the misuse of technical over-generalization. Generalization that something is always wrong or always right, when the reality is that it depends on specific purpose and detail.

Am I missing something here in my ignorance of the details of loudspeaker crossover design? Are the generalizations actually correct?

Without seeing the raw driver responses (on baffle), we can only guess. But, even if you start with a driver that has a perfect frequency response, as soon as you mount it to that small baffle, you will end up with a pretty wonky raw frequency response, due to baffle diffraction. Modelled, it should look something like this (now you can see why the Falcon LS3/5A has a bump around 1000hz):

engMFBx.jpg


I have found with my own designs that when modelling a crossover I start with a very simple design, it then gets more and more complicated, until I'm happy that the subjective sound matches what I see in the model. I then often find myself removing parts from the model, and I'm quite often surprised that everything (phase tracking, frequency response etc) ends up looking better with far fewer parts!
 
Historically, were parts tolerances tighter with BBC LS35A's relative to a P3ESR of today? Just curious as the BBC seemed to be able readily swap out a speaker if necessary, if of same iteration...though I know this is a complicated history in itself.

I suppose I assumed capacitors of today are better made, better tolerances etc.
 
Historically, were parts tolerances tighter with BBC LS35A's relative to a P3ESR of today?

No, but the BBC’s tolerances possibly were. LS3/5As had Kef B110s selected from the far-end of the Kef tolerance range and each individual speaker needed to meet a tight measurement standard when complete. One can argue an LS3/5A is actually a construction and measurement criteria rather than a loudspeaker. A P3ESR, like any modern speaker, should be made of very precise, tightly toleranced and repeatable components. There is no excuse for anything else in this century.

I’m obviously a vintage audio enthusiast, I honestly think so many designs of the past sound better than so many modern designs (and yes, I’d likely take LS3 5As over P3ESRs), but I’d never argue that modern manufacturing tolerances hadn’t improved significantly. The vast majority of modern stuff will be the result of CAD, stamped out by robots, flow-soldered etc. I’d be very surprised if many drivers were hand built anymore, and for pair matching etc that will be a very good thing. For comparison the manufacturing tolerances in vintage Tannoys was so huge you can find very different numbers of spacers between the magnet and compression driver etc. These really were hand built drivers. I’d bet even low-cost mass-market stuff like Kef B110s were made by humans too as the precision machinery to automate such processes didn’t exist back then.
 
Some interesting reading here about the B110 and cone materials in general - http://www.nutshellhifi.com/library/speaker-design2.html

Quote "The BBC-derived designs always employed notch-filter equalization to flatten the Bextrene driver in the midband; the most famous (or infamous, depending on whether you were the listener or the designer) driver was the KEF B110 used in the BBC LS 3/5a minimonitor. Not everyone knows that this speaker, which is legendary for its sweet midrange, employs a deep notch filter with 6dB of attenuation at 1.5kHz to correct the B110."

Frequency response of a B110 here. I think this was the Linn Kan version - http://ww.dibirama.altervista.org/h...f-b110-sp1057-mid-woofer-5-8-ohm-50-wmax.html
Doesn't look like there's a problem at 1.5khz in other versions of the B110 in post #14 either.
 


advertisement


Back
Top