PERKINS PR–2 PRECISION REFERENCE LOUDSPEAKER

Sec. 0 – Introduction

   
. . . This page will be in development for some time so will be unfinished in places.

. . . It looks now as though I'll write something of a history, as I'm often asked, "How did you get into this business ?" and so on. Having been in high-end audio since its modern day beginnings in the mid-70s my work might be of some interest, replete as it is with techno-adventures and the inevitable war stories.

Notes on the Page Construction
. . . I've provided a lot of backup information, fact-based marketing, and will ultimately list everything in the 'REFERENCES' section at the bottom of the page. In the meantime, links are peppered throughout the text as I decide what to include.
  . . . Items cited in the text are indicated [X], clicking on which will take you to the reference table where the same number is seen to the left as a link. Clicking there will return you to the reference's first mention.

. . . It being a nuisance to download many individual PDFs, the entire collection will eventually be available in one large PDF listed as '00' in the reference table.

 
. . . A quite extensive PDF on the PR-2's development can be downloaded here; noting that the document is a combination of literature produced in 1982 and later updated to include model revision and numerous references.  

   



     

Sec. I – Closeups of the Bass Unit and Tweeter

 
Fig. 5 – The solid Koa tweeter head with the tweeter I spent 10 years developing, the magnet structure for which is seen below. Dimension drawings can be seen in the above-referenced PDF.
Fig. 1 – The felt plug is these days commonly called a phase plug, whatever that might mean. In this case the felt absorbs a weak cavity resonance at the apex of the cone, and is predictably called a cavity resonance absorber.  

   



     

Sec. II – Details of the PEARL 200mm Bass Unit

Fig. 10 – Above is the bass unit built from a raw casting I designed and have made in a foundry just south of Red Deer, AB. Working drawings can be seen in the above-referenced PDF.

. . . As described below, the unit bolts into the enclosure with nine (9) fasteners, one of which goes through a hole (A) in the center pole of the magnet structure. This is done prevent deformation and 'oil can' or 'snap through' resonant modes; as amply illustrated in this deliberately exaggerated FEA animation done in COMSOL by Ulrik Skov of www.iCapture.dk; the basket is resonating at 827Hz


. . . The motor assembly is magnetized using a pulse-discharge magnetizer I scratch built in the early '80s. An intimidating project; I'd never seen one before gathering together some 32,000uF

 

worth of 450V caps, a 2.5" dia. hockey puck SCR, the rest of the sundry bits and just built it. The first firing was a memorable moment.


. . . When back in production the magnet structure will include extensive use of flux demodulation rings that linearize the motor's action and greatly reduce the tendency of non-linearized motors to create what I call the 'pushups-on- a-trampoline' effect where the voice coil/cone's operating center is literally axially displaced in the voice coil gap as though by an applied DC bias. Even-order harmonics are abundantly produced.


Fig. 11 – Directly above is the fairly generic bass unit from an early incarnation. It used a doped Bextrene cone, 38mm voice coil, stamped steel basket and a ceramic magnet structure with the center pole piece coming up only flush with the top plate.


. . . A real problem we identified was that the top 'deck' of the stamped basket was much less than a rigid platform and, as indicated above, could and did get into resonant states known as 'oil can' or snap through' modes.

  Fig. 12 – Seen above is the remedy for the 'oilcan' modes just described: a simple epoxy potting.

. . . This resolved several problems I didn't know existed up to that point as I hadn't yet come up with the constrained layer enclosure material seen immediately below that simply erased a multitude of previously well masked problems.

. . . See the FEA animation previously mentioned for an analog of the oilcaning situation before remedy.
     

   



     

Sec. III – Details of the Bass Enclosure Construction

Fig. 15 – The bass enclosure is a well braced construction made of a 7-laminate, constrained-layer isotropic material (D) developed in-house to exhibit very high internal damping, rigidity and mass density. Its non-resonant performance far surpasses commonly used materials such as MDF and birch plywood. It is the best sheet material yet seen for loudspeaker enclosure construction.

. . . Reinforcements in the four corners (B) behind the bass unit provide for embedded nuts that accept 8 x 1/4" x 3" long mounting bolts. Centered directly behind the driver is a solid 2" thick brace (C) within which is embedded a barrel nut that accepts a 3/8" dia. non-magnetic stainless
  steel bolt inserted into the hole (A) passing through the center of the bass unit's magnet structure seen in Fig. 10 above.

. . . By means of these nine fasteners the bass unit's 8lb cast basket and magnet structure are made integral with the 65lb enclosure, thereby establishing a ratio of 73lb reactive enclosure/bass unit mass to 1oz active cone/voice coil mass greater than 1000:1 or 60dB.

. . . Any rocking motion of the cabinet on its 20lb sand-filled stand is thereby reduced to insignificant levels.

   



     

Sec. IV – Notes on the PEARL Polymer-Graphite 1" Dome Tweeter

Fig. 20 – The tweeter was years of work that in 1986 resulted in the first hard dome that actually worked properly, that is to say without an egregious, high-Q, ultrasonic peak some 20 to 35dB high at 20 to 30KHz.

. . . A 1" dome, it's all-up moving mass is 150mg - less than half that of an equal sized soft dome - its fundamental resonance is 280Hz with a very flat response to well over 20KHz, gently rising beyond 30KHz to a +6dB, 1/3 octave wide peak at 39KHz; numbers almost unheard of to this day. A swept-sine frequency response curve is seen in Fig. 25 below, noting that the minor horizontal divisions are 1dB.

. . . 50 microns (0.002") thick, the formed dome weighs merely 50mg and due to PEARL's unique vacuum-forming technique shows a <±5% thickness constancy from periphery to apex.

. . . The very low fundamental resonance implies a suspension compliance some ten times that seen in conventional 1" devices. Low moving mass, high compliance and low mechanical resistance combine to evince truly exceptional low-level resolving power.

. . . The dome is precision vacuum formed from a remarkable graphite-based polymer that is 70% the mass density of aluminum, has the same flexural modulus (stiffness), 25 times the internal damping and thermoforms with relative ease.

. . . Broadly speaking, this material might be seen as a precursor to modern-day graphene materials, relying as it does on the extreme inter-atomic strength of its laminarized graphite's carbon 'a-b' planes; see below.

. . . First realized by repeated Scotch Tape exfoliation (yes, really) in the 1990s, graphene is a single layer of graphite; a hexagonally bonded X-Y lattice of single carbon atoms extending without theoretical limit across the two dimensions.


. . . Graphite then, is many layers of 1 atom thick graphene stacked one atop the next. While graphene's (carbon) hexagonal inter-atom bond structure is one of the strongest in Nature, its inter-layer strength in what is known as the 'c' or 'stacking' axis is quite weak.

. . . Then and now the weak 'c' axis bond is a major problem in the implementation of the otherwise beguilingly attractive single carbon atom thick, hexagonally bonded 'a-b' layer now well known as graphene.

. . . Rather than attempting to bond atom-thick layers of graphene that in any case wouldn't be realized until the mid-90s, Pioneer's mid-70s workaround was to polymer-bond and calendar laminarize high aspect ratio graphite nano-platelets. See Figs. 21 and 22 below.

. . . An ideal material in this and many other high-performance audio applications, it is the result of genius materials science work by Pioneer thru the mid-70s. Lamentably however, other than Pioneer themselves and their well respected pro audio division, Technical Audio Devices, TAD, no one used it. PEARL, INC. was Pioneer/Mogami's worldwide distributor Shima Trading's only polymer-graphite customer. Pioneer discontinued production in the early '80s and scrapped the very costly, high precision calendar necessary for its production. A great loss . . .

. . . Rebuilt copy of Pioneer Electronics, 'Engineering Research Laboratory's, original 1979 AES Convention presentation is here [5]. A succinct and informative read where a scan through the cited references, some dating back to the 1920s, illuminates the very basic nature of Pioneer's research.

. . . Copy of Pioneer's numerous patents on polymer-graphite and related materials from the late 1970s through the early '90s is here [6], another very informative read.

Fig. 21 – A somewhat schematized but nonetheless accurate illustration of the structure of graphene is shown above.   Fig. 22 – Shown is an edge-on photomicrograph of polymer graphite. The 100 micron scale equals 0.004" or the nominal diameter of a human hair.

   



     

Sec. V – Measurements of the 1" Dome Tweeter

Fig. 25 – Shown above is the anechoic, near-field, swept sine, response of the PEARL 1" polymer graphite dome tweeter mounted in its essentially diffraction-free solid hardwood, early-80s vibration-isolated tweeter head.

. . . The minor divisions are separated by 1dB, clearly showing the first major axial breakup at +6dB x 1/3 octave wide. Note that while other less severe resonances begin to occur at approximately octave-lower frequencies, as they do in any hard
  dome driver, they're not much apparent in the frequency response due to polymer graphite's very high internal loss factor. They do however make themselves somewhat evident in the CSD, acoustic phase and group delay plots below in Figs. 28 and 29.

. . . Generated in PEARL's lab anechoic chamber seen in Sec. 50 below, the frequency response curve above was taken using 'old school' swept sine methodology with no smoothing, averaging or other data massages.

.

The data below was taken in 1993 with a MLSSA system and on account of the measuring system is somewhat dated.

Fig. 26 – Shown above is the unit's response to a very short interval pulse, which data is compressed in both the X and Y directions. The X axis should display perhaps 4mSec while the Y axis should be in ±dB with a range of about ±40dB. In all then, a not very useful presentation. Nice eye candy though . . .   Fig. 27 – This display shows excellent performance out to the MLSSA system's upper limit of 30KHz.
. . . Seen from 0.34 to 1.2mSec at some -30dB is the dome getting into some ringing from 15KHz to 30KHz, a minor matter but I'd like to deal with nonetheless. Some artifacts are seen from 200Hz to 1.2KHz where a possible reflection crept into the analysis. I didn't do the measurements seen in Figs. 26 thru 31 so don't know the cause.

Fig. 28 – In 1993 acoustic phase was a difficult measurement which to this day is often incorrectly done.

. . . Here we used Heyser's suggestion to arrange compensation for the air path delay such that the phase angle read +90° at the drive unit's fundamental resonance. We were a little off from the actual 280Hz resonant point but at 300Hz we were in the ballpark given what we knew at the time.

. . . The perturbations starting about 13KHz, indicate the onset of  breakup modes almost two octaves below the first, obvious peak in response seen at 39KHz seen in Fig. 25 above.
  Fig. 29 – Group delay, acoustic or otherwise is hardly a topic for a paragraph or two in a figure caption.

. . . For the moment then, here is a Wikipedia link providing a good overview and some references.

. . . I have a lot of information on the topic and will collate into a rational compendium at some point.

Fig. 30 – The tweeter's impedance magnitude, Z, where the tweeter's 280Hz fundamental resonance, is clearly seen. Notice as well the only slight, 20% rise in Z at 20KHz due to the solid copper electroplating of the whole of both the top and back poles of the motor assembly. To prevent oxidation in humid environments, which was discovered to congeal the FerroFluid placed in the magnet's gap, these are subsequently nickle flashed.   Fig. 31 – The electrical input phase magnitude appears above, showing by its maximum ±18° variation a very tractable load for the crossover designer. Note as well the curve's intersection with 0° at 280Hz, another indication of its fundamental resonance.

   



     

Sec. VI – PEARL Designed and Built Vacuum Former

Fig. 33 – Shown with its top and front covers removed, our forming machine (A) was built to precisely vacuum form polymer graphite diaphragms. A closeup of the control panel is seen below.

. . . Typically, vacuum forming involves heating a sheet material to its plastic temperature then sucking it into, or over, a cold die, where it immediately freezes. This has the effect of concentrating material thinning into the areas pulled last to the die, with the result that the formed part's thickness is inconsistent. PEARL's automated process involves first heating the die, then forming, then cooling, then part removal.
  . . . Although time consuming the process result in parts of very consistent cross-section thickness, an important specification.

. . . Seen at (B) is the family of forming die shapes we investigated. Formed by the simple expedient of changing the former seen fitted in the center of the heated/cooled forming head (C), we were able to cost effectively work though many shapes to find an optimum. Once the forming procedure was working well, a task in itself, the process took about a month; while concurrently running the company.

. . . The range of shapes and their method of generation can be seen here.



     

Sec. 30 – Bare Tweeter Dome being Fitted with Voice Coil

Fig. 36 – Dome ready for voice coil.   Fig. 36 – Dome and voice coil glued and ready for surround.


. . . The dome-to-voice-coil glue joint is done by hand with a 3cc syringe and a #28 hypo' needle as the mandrel slowly rotates in a little Unimat jewelers' lathe rebuilt to run at very low speeds. All the jigging parts are Teflon. The Ferrofluid-resistant glue used is a well-known Loctite product used throughout PEARL's the driver assembly processes.
     



     

Sec. 30 – Tweeter Dome Fitted with Ultra-high Compliance, Acoustically Transparent Surround

Fig. 37 – Dome in surround fitment jig, ready for glue.   Fig. 38 – Dome with its ultra-high compliance, acoustically transparent surround fitted, ready for clamping while the glue cures.

Fig. 39 – Weighted Teflon clamp on the glue joint   Fig. 40 – The dome/surround assembly affixed to its mounting card, which is then glued onto the magnet assembly top plate and hand aligned.
     



     

Sec. 35 – Classic Holbrook C10 - 12" x 20" High Precision Toolroom Lathe

Fig. 45 – All the metal parts shown above were made with the toolroom lathe seen above. I bought it good condition then took it completely apart to thoroughly clean it and replace every ball bearing in it and to regrind and refit various surfaces.The machine is presently part way through another teardown and rebuild that this time involves a strip down to bare cast iron and a repaint and refit to make it look like the other rebuilt pieces here.

. . . Time permitting I might well have the bed, cross slide and compound ways reground back to new specification and fitted
  with Turcite B, a, " . . . high performance thermoplastic material for use in linear bearing applications such as the guideways of machine tools." Basically the idea is that the material machined away from worn, mating surfaces is replaced with a material purpose-designed for better performance in the guideway application than the original base materials, usually cast iron.

. . . An example of the installation of Turcite B is seen below, note that by fitting this material the reworked parts can be brought back to their original working tool heights and centerlines, a crucially important requirement. A Turcite brochure is here.

. . . Throughout its decades-long history Holbrook consistently produced machines of the highest caliber, albeit at great purchase cost. When new in the '60s the above machine likely cost some $US60 to $100K in today's dollars.   . . . For those with an interest, several Holbrook brochures from the 1950s & '60s are available below:

               
Holbrook Minor - A pristine, extant example.
Holbrook Minor - Operator's & Service Manual.
Holbrook C10-13-16 - Sales Brochure & Operator's Man.
  Holbrook Major - Sales Brochure.
Holbrook D13-15-18 - Sales Brochure & Operator's Man.
Holbrook H15-17-20 - Sales Brochure.
               



               

Sec. 40 – PEARL's Acoustics and Electronics Lab

Fig. 50 – Purus arcu aliquet nulla, pulvinar quisque habitasse lacus conubia Massa Nulla diam. Commodo risus augue nascetur tempus ut sodales lectus luctus fames. Curae; condimentum lacinia habitant. Per fermentum placerat gravida   interdum erat urna, ipsum, netus rhoncus vivamus maecenas aliquam consectetuer aliquet pretium dapibus cras nunc purus. Erat malesuada, vehicula viverra euismod habitasse lobortis neque ligula tempus dui consectetuer habitasse litora.
               
               

 

Fig. 55 – Purus arcu aliquet nulla, pulvinar quisque habitasse lacus conubia Massa Nulla diam. Commodo risus augue nascetur tempus ut sodales lectus luctus fames. Curae; condimentum lacinia habitant. Per fermentum placerat gravida   interdum erat urna, ipsum, netus rhoncus vivamus maecenas aliquam consectetuer aliquet pretium dapibus cras nunc purus. Erat malesuada, vehicula viverra euismod habitasse lobortis neque ligula tempus dui consectetuer habitasse litora.



     

Sec. 45 – 25 foot (7 Meter) Free-field Outdoor Measuring Lift

     
Fig. 60 – The piece to the right allows one to lift a 500lb speaker some 25 feet off the ground in order to make accurate low-frequency measurements in what is called free space.

. . . At the time I built this lift it was likely the only such dedicated apparatus in the Canada. Its construction was inspired by a mid-80s trip to the Canadian National Research Council's facility in Ottawa, Canada where anechoic measurements of my speaker in their inadequate to the very low-frequency measurement task chamber plainly showed me what needed doing.

. . . I ultimately sold it to the University of Alberta's Mechanical Engineering Acoustics and Noise Unit, MEANU, some 25 years ago. An interesting facility housing two reverberation chambers; the only such rooms in the country so far as I know. Built in the mid-70s by Bolstad Engineering of Edmonton, AB, I'm presently giving their original lit some scanner and Photoshop love and will soon upload.

. . . A compendium of articles on low frequency loudspeaker measurement is here.
 
     



     

Sec. 50 – PEARL's Lab Anechoic Chamber

         

         

         
  Fig. 65 – This anechoic chamber is my third. Built from 1.5" laminated MDF and painted colors I had specially mixed to exactly match the well-known Brüel & Kjær two-tone color scheme it breaks down to pass through a typical 32" doorway opening.

. . . The widely appreciated B & K color scheme originated with the Danish Army during WWII, when Per Brüel and Viggo Kjær started what became their world leading company.
  . . . As with almost everything else, B & K did things differently. Due to its lack of the large, plane surfaces seen in typical 'wedge' rooms the Cremer 'Acoustic Jungle' seen below has far better performance above about 300Hz, being much less reflective at higher frequencies due to the diffusive nature of its varied-density acoustically absorbent, fiberglass blocks. An important and seeming greatly overlooked performance aspect. An eight paper collection on anechoic chamber design and construction is available here.  
         



         

Sec. 51 – A Large Room at the National Metrology Institute in Japan

Fig. 69 – An example of a very large and very expensive anechoic chamber, this one at the National Metrology Institute of Japan.

. . . Undoubtedly a many-million-dollar facility, and almost certainly a vibration isolated structure built within a very heavy exterior shell.

. . . Comparing the probable about 5ft height of the woman to the apparent length of the wedges, a best guess as to the lower cutoff frequency is a respectable 30 to 40Hz.
  . . . If however, one is prepared to work around northern latitudes' ever changing weather conditions, always point north to keep solar thermal gain minimized, work in the near field at low frequencies and do some signal averaging to better one's S/N ratio, perfectly acceptable results are achievable another octave lower in frequency and at an all-up cost of a few thousand dollars.
     



 
Sec. 52 – A Medium-sized Room at a European Loudspeaker Manufacturer

Fig. 70 – This Danish anechoic chamber consists of a shell of concrete and LECA that rests on coil springs to dampen noise from vibration in the ground. The chamber is lined inside with meter-long sound absorbing wedges of damping material. The chamber is used for all kinds of free-field acoustic measurements. It should be mentioned that 'LECA' (Light Expanded Clay Aggregate) ' is a remarkable material having both sound and vibration damping qualities, very low thermal conductivity and great fire resistance. It can absorb some 15% H20 by weight while simultaneously resisting decomposition by wet- or dry-rot. It is also a hypo- allergenic building material. LECA consists of small, lightweight, bloated particles of burnt clay. The thousands of small, air-filled cavities give LECA its strength and thermal insulation properties.   . . . The base material is plastic clay which is extensively pre-treated and then heated and expanded in a rotary kiln. Finally, the product is burned at about 1100 °C to form the finished LECA product. LECA is an entirely natural, environment-friendly product providing the same benefits as conventional tile but in brick form.

. . . LECA branded product is produced in Italy, Denmark, Switzerland, Norway, Germany, Finland, Portugal, U.K. and Iran. Countries which produce very similar aggregates but with different brand names are: Russia, Poland, Sweden and China making 'Keramzite'; South Africa making 'Argex'; and Spain producing 'Liapour.'

The LECA website is here.
     



     

Sec. 55 – Brüel Acoustics Cremer 'Acoustic Jungle' Anechoic Rooms

     
Fig. 75 – Brüel Acoustics use the Cremer principle which has proved both useful and economic.

. . . The absorbing walls are built up with cubes which towards the center of the room are small and made of special glass fibers with a very low density. Moving outwards towards the room's boundary walls the cubes increase in both size and density. In this way one obtains an extremely good impedance matching with the heavy absorbing material in the internal part of the room. This principle is analogous to an exponential horn in front of a loudspeaker driver unit.

. . . Wedges are normally built of the same material from the base to the tip. In the loudspeaker analogy this corresponds to a linear cone in front of the drive unit so consequently the Cremer room is much better at high frequencies than a wedge room of the same size.
At mid-frequencies the two treatment types are equal, and at low frequencies the wedge type is normally slightly better.

. . . From an economic point of view it can be said that while a wedge room contains more absorbing material than a Cremer room, a Cremer room requires more sophisticated construction work which means higher labour costs.

. . . At B & K we used only Cremer rooms because for us it is very important that the room be a good performer at higher frequencies up to 15 kHz.

. . . With frequencies over 3 kHz a wedge room will always give some uncontrollable phase shifts because the wedges have large plane surfaces whereas Cremer rooms are in effect an acoustical jungle. The lower frequencies below 300 Hz are not so important as the wavelengths are long, allowing the implementation of other methods.A technical review of Brüel Acoustics Cremer rooms is available here.
 

. . . Brüel Acoustics was Per Bruel's retirement hobby horse. He passed at age 100 in 2016.

. . . The company's website is still alive, as might well be the company itself.
     



     

REFERENCES

             

00

  Download the entire collection of references here.        

5

Polymer-graphite Composite Loudspeaker Diaphragm        

6

Pioneer Graphite Composite Patents - 1979 through 1990