Fascinated By GR Research Channel

I know that the proprietor of Harbeth is very much ‘old school’ in that he doesn’t believe audiophile components provide any improvement over industrial grade, and that he personally uses well maintained 1970’s era transistor amplifiers because it’s very difficult to significantly improve on them. Not real big on speaker cables either - very much a ‘wire is wire’ man..

I think you can see that philosophy clearly reflected in the speaker dissection.

Most people wouldn't bother to upgrade any of their equipment. They'd just sell it and buy something else. 

Danny at GR understands that for a subset of the market that is willing to tweak or upgrade, this can be done at a fairly low $$ cost. The real cost is (a) the customer admitting their stuff isn't optimal and (b) being willing to reach out, ship the speakers, take the risk, etc.

IMHO, *that* is his business model. And it's a good one.

I'm another who had Danny do testing and build a crossover for my 1990s Epos ES12 speakers.  He's been a respected designer and engineer for the industry for years on years, so he has a great deal of "insider" experience and prefers to do his own thing.  With "audiophile" speakers, many comment and worry about losing that speakers signature sound with messing with a crossover, thinking that you buy and love the speaker for the sound it made, so what could possibly be wrong with it?  I was VERY concerned about this, thinking "no no, but this was SUPPOSE to sound this way".  Well, yes, it was....but that doesn't mean it is 'right'. 

As most likely know, the Epos line from the original ES14, 11 and 12s were well known for amazing crisp detail and accuracy, strong central focus, a bit lean, but very "British" sounding. Some of this partly owned to the actual LACK of any crossover other than a filter cap on the tweeter. 

Danny uses the same industry testing methods I believe every single manufacturer uses to test speaker designs, box designs, and crossover networks. The testing proved that there were irregularities in the signal response, enough so that what he was seeing on a response I was hearing from the speakers.  Some, were part of the known/loved "Epos sound" but over time were annoying and just not pleasing to me. 

The crossover components, his time, testing, and time with me on the phone and emails were all worth past the $300+ dollars (high quality components, schematic, no rez, wires and silver solder).  I love these speakers  so much already, despite the age (a lot for sentimental value).  Once the new crossovers were put into play and broken in (took awhile), I love these speakers even more.  I was worried that I would lose/miss the "Epos" sound, but in truth what I was getting were bumps and dips in certain frequencies that created some of that "signature sound".  Once properly controlled with a crossover, the design and box really worked incredibly well together: still has crisp accurate detail, still great focus, but with a better blend across the frequency response, more richness, and (ugh) more 'musical'.  

I am able to test the speakers against themselves without the crossovers (I have a second pair left as/is), and it is dramatic how basically it sounds like someone screwed around with an EQ, and I need to rely more on the Rega Sub to 'fill in' with the original crossover-less version.  I lost nothing, but gained a great deal.  They still sound like Epos, only smoother, same detail, a bit more punch (relative term for a speaker with a 5" driver) but a much better listening at all levels, for long periods of time.

What I think is most interesting is the flack GR/Danny takes from both sides of the audiophile community.  He is an actual designer and engineer, musician, with decades of professional experience, and he is a bit of an audiophile.  He gets flack for building crossovers or criticizing speakers with comments like "but did you even listen to them?! That's how they are suppose to sound!", to criticized by the engineering community for pushing 'snake oil' (seriously can we please retire this hackneyed term already?).  He can go from being attacked for only looking at measurements, to NOT looking or depending on measurements for things like speaker cable or power cords.  Just amazing.       
 

And obviously if he can find faults in some of the great speaker designs then it MUST mean his speakers are flawless…

<<<<<<<<<<<<<<<<<<<<<<<<<<<<>>>>>>>>>>>>>>>>>>>>>>>>>>>>

They are. I like ugly girls that are flawless, sound like an oxymoron? It's always in the eyes of the beholder!

Regards

I don't know what specifically motivated these postings, but here is a three-part series by Alan Shaw on this very topic, posted November 8-9 of 2021.

 

The essence of good engineering is cost management. That is to say it is bad design - hence poor engineering - to add a single cent to the cost of a product when a cheaper solution would provide exactly the same user experience, possibly last longer and perform more reliably and use less natural resources. To add needless cost deprives the manufacturer of income that could be usefully directed to R&D and the long term sustainability of the business.

And so this maxim must apply to loudspeakers. It must be good engineering practice for the designer to periodically take a hard look at the selection of components used in, for example, a loudspeaker crossover and to ask four questions:
 

1. Can I save cost in this design by using alternative component technology?​

​

2. Can I improve the audible or measurable performance of the product if I use alternative component technologies?​

​

3. Should I add cost to the product for purely marketing reasons i.e. to make the parts look more sexy, to create the illusion in the non-technical consumer's mind of enhanced pride of ownership (even though there is absolutely no technical or sonic benefit)? You can call this 'go-faster stripes'.​

​

4. Is the selection of these components future proof, are they durable and are they the best use of natural resources?​

​

So, let's consider the selection of the type of capacitor technology; there are two competing capacitor families available to use in loudspeaker crossovers: wet and dry capacitors.

First, wet (electrolytic) capacitors. These are axial (cylinders with wires out of opposite ends) and are widely used in loudspeaker crossovers, typically having blue, green red or black printed film over their shiny, tubular metal cans. As you might expect, if you cut these open (or they leak) your fingers will be covered in a goo jell. Here are examples:
 

As for dry-film capacitors, there are some package (case) options. Cut these open and the interior working part is completely dry:
 

Above we see an orange cased dipped radial (wires come out of the bottom) capacitor, a plastic-box radial capacitor and a cylindrical axial type - all three are dry-film.

So, the crossover design commences and the designer has to ask himself and resolve certain key issues:

• Does he have any fundamental likes or dislikes amongst these two primary types of capacitors?
• Does he have any negative experience such as premature failure or short working life that could impact on product Warranty?
• What about cost and availability?
• What about capacitance and case size tolerance - is off the shelf ±10% acceptable in the final product?
• How much board area is available for capacitors? Would the tighter packaged (thin, tall) cases of the radial types use less board area than the cylindrical types?
• What do the manufacturer's data sheets specify for working voltage, temperature, current, leakage and life expectancy?
• Does he have some emotional attachment to one type over another?
• Does he believe that one type sounds better or measures better than another? What is the evidence? Anecdotal or personal experience under controlled conditions?
• Does he actually care about the nuances of capacitor technology at mere audio frequencies?
• Is it likely that the International Space Station would use such a type given the extreme dependence on reliability?
• Is his engineering driven by market fancies?

and so on.

If you disassemble these capacitor families, you will find this. Everything you could ever want to know about capacitors, here, empowering the circuit designer to select the optimum type for his circuit. Remember: what may be a good choice of capacitor family for use in a laptop or mobile phone, may be the opposite of the best choice for a loudspeaker.

 

Now we can turn to look at inductors (soils, chokes) as would be used in a loudspeaker crossover.

Again, here the designer has some choices available to him. First, here are two air cored inductors. Air cored simply means that the windings have been made around an invisible tube, which has been withdrawn from the centre leaving a stack of free-standing copper wire. Sometimes, to neaten the stack and stop it toppling over, a non-magnetic plastic bobbin is used, but that would have absolutely no effect on the coil's electrical properties:
 

Inductance, measured in milli Henries is proportional not to the thickness of the copper wire, but to the number of turns of wire around the central core regardless of which layer of the winding it is on so if you want a high value inductance, you will need more turns than if you need a smaller inductance.

In practical terms, a point is reached - the above pictures hint at this - where the number of turns has resulted in a tall stack of enamel insulated copper that is barely strong enough to support itself, and that sets a practical limit on how big an air core crossover coil can be in a loudspeaker application. It's no joke to imagine that if the crossover designer needs a high value air-cored inductor, the inductor itself might fill the entire speaker enclosure!

So a method was needed to 'boost' the magnetising efficiency of turns of copper wire - a magnetic material is put into the core previously occupied by ordinary air and wow! Really high values of inductance are now possible with relatively few turns, and much less copper .... and probably less DC reistance too ....

Here is a winding into which a ferrite material has been slid into the core:
 

Another type is a bobbin core, where the bobbin is magnetically conductive (unlike the plastic bobbin shown in the air core above):
 

and here are very typical iron dust or ferrite-cored coils as used in loudspeaker crossovers:
 

Now we should be able to recognise capacitor and inductors on a loudspeaker crossover (network) board:

Here from an internet image search we can see air-cored coils and dry film capacitors (and a fuse):
 

Here we can see an air cored inductor and two wet-film electrolytic capacitors:

And here is a combination circuit with a steel-cored coil (top left), an air-cored coil (middle) and electrolytic capacitors (red) alongside dry-film capacitors (yellow).
 

It's the designer ability to mix and match components that gives him the power to shape the loudspeaker sound just as he wishes. There is no universal best solution, and a consumer mantra that states that 'all capacitors should be polypropylene' or that 'all inductors should be air core' is just ignorant of engineering reality and demonstrates an emotional response to an engineering matter.

 

So what are the preferences and choices that we have made at Harbeth in our crossover designs?

The first point is ’cost is no constraint’. That means that as far as we are concerned at Harbeth - although I know for certain that competitors work to rigid cost ceilings and this limits the type and numbers of components that their crossovers can have - we will add components and complexity to achieve, in combination with the shape of the cabinet and the characteristics of the drive units, the smoothest measurable frequency response. If that can be achieved with two components great, but that’s a fantasy. If it takes twenty, so be it: we suffer the cost, size and assembly time for the improved performance.

It must be obvious that $10 on material cost is $10 off profit so for a manufacturer to add to the cost of his product, the benefit of improved performance must outweigh the additional cost, and may be reflected in a higher selling price which the delighted consumer is willing to pay for.

The second point is appropriateness to task. Given the power handling of the woofer and tweeter being the defining power limitations in the complete speaker, we can work backwards to selecting appropriately rated crossover components.