Thiel Owners

I interpreted "all pass" as any filter that is not causing a rolloff. As I used the term, there are multiple all pass elements in all of Jim's crossovers. They include various shaping and resonant circuits which are directly opposed to specific resonances in drivers such that the circuit cancels the amplitude, phase and transient problems of the driver resonance. It actually fixes the driver misbehavior. The fixed driver is therefore prepped to roll in or out at its prescribed slope - high or low pass filter. 

These shaping circuits always result in better measurements than the uncorrected driver in all domains. The down sides include cost and whatever electronic veil comes with the extra circuitry. A good listener can hear the effects of the components. Thiel's budgets necessitated mid-level components with mid level artifacts, with the exception of coils. We splurged for 4-9s wire in well made coils. Of course there is room for more improvement there. I am now experimenting with 6-9s foil coils, which are glorious at twice the cost or 4-9s and 6+x the cost of normal production 3-9s ETP copper.

That photo is probably an early or prototype version which goes to shows and seminars. Typically it will morph through a year or more to match the actual settled particulars of its driver complement.
At first glance, coils are 4-9s, best of form. Significant improvement could be made via cap and resistor upgrades from AXON, electrolytic and sandcasts.
Beetle might have an opinion.
Cheers. TT

sdecker - tilting is the less offensive solution. But the changes introduced by either tilting or lifting are very small and either provides better performance than listening at the 'wrong' window. The coherence puzzle has its requirements, and Jim chose the average 3' ear height as part of the solution. The window is not as small as some imagine it to be unless you sit closer than 8' where it falls apart very quickly. Sit at 10' and you get a lot of leeway.

I think it not improper to point out that JA seems to hear what his measurements tell him. At 50" there is a midrange suck-out and some spikes. Notice that he wonders in print if that is what he is hearing. However, those ARE measurement artifacts; they are caused by too-close microphone distance, and disappear at about 8'. Notice that they don't show up in the room-averaged measurements. Nonetheless, he treats them as real.

What we are learning from Beetle's 2.4s and my budding experiments matches what we knew from the beginning - passive parts quality is audible. So I believe the "reticence" is more likely that parts veil rather than the false frequency response artifacts.

Utrak - your room is small, but its proportions are very good, plus you get bass room gain without supporting the deepest and most problematic frequencies. Yes, they are quite efficient and the impedance doesn't drop as low as newer models, plus the sealed bass is less reactive and more coherent than higher-order ports and passive radiators. Sweet. You have kicked me into roping the CS3 into the 3.5 upgrade project. 

Story time: You all may or may not know or remember the deep recession of 1982. The Stock Market tumbled, expendable income plummeted and the emerging high-end industry took a hard hit. Many manufacturers including Acoustat (and I don't now recall other brands) went under. Many manufacturers dropped out of the June 1983 CES in Chicago - which felt like a ghost town compared to prior years. This was our 7th CES (before Las Vegas began), we had yet to make a profit, and we knew we were gambling on a flagship. The concept was the same as the 03 / 03a / 03b, but the execution of the curved baffle, highly customized drivers and next-level XO components put it in a higher league. We hoped folks would notice.

There was always a pre-show walkabout the evening before opening. We had a mostly European purist recording following who always brought master-tapes and headphones to compare notes from studio to showroom. The Danes had been to KEF's new Reference 105 demonstration, claiming "phase correct" presentation. They also did their homework and reported the drivers as being time-aligned, but using high order filters and scrambling their test tracks.

After the niceties, the new works began rolling. Lights out. Spaced omni capture of a live jazz ensemble, ending in a barely-audible triangle tink. It faded forever. Long silence. Applause. And the night went on somewhat like a dream. 

We had come to the show with a tentative retail price of $1250 / pair, a little advance from the 03a at $1150. We, especially Jim, feared a glass ceiling for a dynamic speaker from a fledgling, uncredentialed company at $1,000 / pair in 1983 dollars. These were frontier days, not only in Audio, but in boutique manufacturing generally, and on the heals of the deepest recession since the great depression.  Uncharacteristically, I surveyed the attending guests, about 30 sophisticated insiders and industry associates, asking what a proper price might be for this thing. The 'auction' suggested $2K - $5K. Everyone was elated.
We stayed up half the night discussing everything: the nature of our niche, the disparity between a few deep fans and 'the market' at large, the times, and the real settled costs of putting this thing into production. Note that each baffle was hand-sculpted by yours-truly in what might be called shop-jujitsu. We were pulling this thing out of our hats. Big sales were the enemy of stability and sleep. One key equation was that if I were engaged in hiring, teaching, coaching and managing new cabinet makers to produce this complex cabinet, I wouldn't be available to make baffles. Production is not prototype and we needed production to break into profitability and survive.

When the room quieted around 3:AM, we had decided to introduce the 03b at $1850/pair. 03b? Well, it was the third version of the 03 and our system assigned that name. Jim predicted that we would loose most of our dealers, since our niche was unbelievably inexpensive / unexpectedly good performance.

Next morning before opening, Peter Moncrief of The International Audio Review came to visit - he had heard talk. He was full of the right questions. He got answers he loved. He listened. He smiled. Peter pronounced that we had found the holy grail, that we had created the illusive and rumored impossible "Coherent Source" transducer. We christened it CS3 - Coherent Source, third generation model 3, and had the show of a lifetime. The room was packed full-time. Comparisons to KEF and B&W were flying. Only one east-coast dealer dropped our line - and was replaced by a more appropriate player. The CS3 turned our tide.

Was it a bugger to make? Yes. Did we struggle to meet demand, yes. Did we rise to the opportunity? Partly. The CS3 was our watershed. It caused me to take a huge plunge and adopt CNC machining in its infancy - as a first-tier user. We mortgaged everything, and still couldn't manage the $6-figures investment. Banks had no money to lend even if we might qualify.

Then magic happened.

Remember Scott Estes, the Maggie Man reviewing for The Sensible Sound? Scott had become a friend. Kathy and I were commiserating with him about the bucket-load of orders we couldn't fill, the back-order that could bury us . . . and he volunteered news of an unexpected inheritance which he was willing to lend us on the strength of his belief in us. Oh My God. He funded the critical corner, the one that might not have been navigated. The CS3 real-cost equation required a price increase to $1950 at the first Las Vegas CES in January 1984. And that was just fine with everybody.

Of course new learning superceeds old performance and new products always perform better in some ways. But nonetheless, the Thiel CS3 is a true classic in our corner of the history of the development of American High End Audio. And I hope to find a way for it to live on. R is for Renaissance.

Utrak - I can offer a broad-brush intention: more than a dream, less than a plan. Rob of Coherent Source Service and I are working on upgrades for various classic Thiel models. The CS3.5 is now on the list, and the CS3 is likely to come along as a hybrid solution. Those designs are similar to each other and all drivers of both models are obsolete. We are looking for replacement drivers to work in both models with appropriate XO tweaks.

To your situation, cap failure is not likely. Thiel used long-life caps and we know of no age-related cap failures - not yet, probably soon. But, drift is likely. Heat-related (over-driving) cap failures are known - you can see the burned caps and replace them.

I suggest you contact Rob at CSS to work out a strategy, since our Classic Upgrades are not yet real, and there are bridges to cross before they come to market.

JA - I hardly ever get back to Lexington, but that will change with the Renaissance Project. I'll look him up!

Coherent speakers must heed the listening position to test properly. JA knows this. His normal, small room would add too many early reflections to give readable graphs. So he chose 50” which gives false readings for coherent sources. There are other review rags using more legitimate methods. 
Early on, JA explained, even apologized for the misleading readings. But subsequent ownership policy changed that.

your 38” at 9’ is fine.

Pass' Threshold amps do quite a bit right for all the right reasons.

CS3.5 update

As you know, I've been investigating internal hookup wire this spring and summer. It's been quite a long and winding ride. Among the twists and turns has been learning how some wire configurations interface with other wire configurations - for better or worse. Backing up, I wasn't convinced I was sending a neutral signal to the 3.5s, especially with the equalizer in between the pre and power amp. I took a detour with Dick Hardesty's 'series-bypass' method where a wire (in this case) is added then removed from the feed stream. The result was that some 'upstream' cables just don't like driving a particular 'downstream' cable. I propose that just such a condition occurs when a speaker cable feeds an internal wiring harness. Mismatches abound. The bright side is that this test shone light on which cables are more tolerant of their downstream load, and those cables move higher on my interest list. I'll have some final results to report within a short while. Let's just say what many other commentators have said: excellence doesn't necessarily track price. Today I received a second batch of cable from Iconoclast, which interests me greatly.

I will reiterate what I reported recently. Belden's 4694A is a surprisingly good cable for short money ($30s) . It is sold as a 75 ohm digital coax for ultra high definition video with a 12GigaHertz rating. Indeed it shines in that digital role. But, it also shines very brightly as an analog interconnect. I invite anyone to compare it with whatever you're using, and please report your findings here.

Now, back to the 3.5. I settled on driving the 3.5s with that 4694A Belden digital S/P-DIF, plus StraightWire Rhapsody 3 analog interconnects and StraightWire Octave II speaker cables – single wired. The StraightWire cables hold up very well to competition and I have known them intimately for decades.

Now another tangent. A couple years ago, my first crossover experiments were on the CS2.2. Again, I know it very well, using it for listening and musician / mix / master evaluations for decades. I had reported that removing the crossover from the cabinet, optimizing the layout without the EMF of the woofer, and upgrading many critical components – all combined to produce substantial qualitative upgrade in performance. That 2.2 workhorse is back-burnered via for greater ease and lower expense of using the model 1 with its two-way configuration and shorter wire runs.

My report today regards the unexpectedly large improvement in the 3.5 by simply moving the crossover. No new parts, no new layout, still inside the cabinet. The move is from behind the woofer (very bad environment) to shock-mounted on the bottom of the cabinet.

Over the years, the consensus about the 3.5 (among other Jim Thiel designs) is that it gets 'confused' or 'compressed' when the volume gets high or the score complex. Fair enough. There also are reports of an over-analytic, electronic presentation. Fair enough. The equalizer contributes some of those attributes. But I am using a Jim Williams / Audio Upgrade replacement which rivals most amps that would be used with the 3.5.

The test uses signal from the Philips CD80 through the Benchmark stack with a vetted cable chain (with or without the EQ). That stock 3.5 and its mate with the XO on the bottom are side-by-side, both fed the same mono signal. The modified 3.5 takes on a more lively, vivid, sweeter musicality. Surprisingly so.

The differences got real in the measurements; there have been persistent anomalies in the frequency and phase measurements of the 3.5. Specifically there is a suck-out (15dB @ 80 to 100Hz) and the top octave droops nearly 10dB. The FuzzMeasure sweeps are taken near full scale – just under clipping. Nearfield single driver measurements are taken at far lower levels. They showed no such anomalies. I assumed the 'problems' were in the room. But no. The 'problems' vanish when the XO is moved to the bottom of the cabinet.

Flash back to the 1982 development of the CS3. This very issue was discussed in the hypothetical. We had noted how much purer the sound was before consolidating the XO into the cabinet. But the XO bottom mount was dismissed due to slightly higher labor cost. Hindsight sadly shows that decision to be significant. That 'behind the woofer' mounting persisted until the removable passive radiator in 1995s CS1.5. (The earlier passive in the CS2.2 was built into the baffle.) That particular 'electronic crunch' follows models with the XO close behind the woofer.

Short of all the labor and expense of a full-blown redesigned outboard crossover, there is a lot to gain by simply moving the XO. Very soon, I anticipate having some firm recommendations for hookup wire upgrade. Those wire and XO position modifications provide significant improvement that is accessible to the DIY owner.

 

jthifi - Stasis was the or among the first Pass designs. Its sliding bias and resulting internal stability obviated the need for global feedback. It is very linear and musical and as mentioned, Thiel used the Stasis 400A (last 70s) as one of its development reference amps.
As time went on, Nelson's topologies became more and more simple, direct, minimalist and with very high quality components. In my opinion they rival the best in the world. The S-300 will not have that extreme finesse associated with later Pass designs. But at its bargain price it will out-do much of what's out there at any price.

Hello Paco - difficult room. Some setup thoughts from my experience: Phase-time coherence introduces problems not heard in normal time-challenged speakers. Your ear-brain strives to make sense of the signal. In non-coincident driver Thiels, the required distance of 8'+ enabled the 3 signals to arrive at the same time via finding the proper vertical position for integration. That serious business is addressed by the coincident driver which solves the mid-tweeter issue, which is the more critical range. Lower frequency waves are so much longer and the ear so much less critical in the bass that they are quite forgiving. So you can sit as close as you want and tilt the speakers as needed.

I have a pet bias for first reflections being very important. The time-correct rendering from Thiels allows your ear to differentiate between direct and reflected sound to a much higher degree than normal speakers require. In round numbers, there is a threshold for arrival transients at about 5 milliseconds. Shorter times try to conflate the reflected sound with the direct sound for a confusing slur. Longer times allow the ear to identify the reflected component as distinct - processing it as an understandable reflection, since your eyes know there is a wall behind or next to your speaker. The side-wall reflection can be readily absorbed or diffracted due to the ricochet pattern . However, the front-wall / behind the speakers is far more difficult due to its large flat-plane pattern. The absorption required to kill the reflected wave would make the room very dead. I strongly recommend having at least 2.5' behind the speaker (x2 for the wave to reach the speaker and continue forward,) more is better.

Low ceilings present problems. Apply the first-reflection strategy to a greater degree. The psyche distrusts overhead anomalies and that ceiling bounce introduces anxiety in addition to inflating the mid-bass frequency response. In musical terms frequency response is extremely important for timbre and identification. But in psychological terms FR is merely academic. First rule: get cozy and safe in your listening environment.
The standing waves between floor and ceiling present problems. I suggest good carpet. Organic hair underpad outperforms any of the plastic foams in spades. Wool carpet likewise outperforms, damping more than an octave lower than synthetics. Alternatively I get good results with rubber-backed commercial carpet tiles with as much thickness and texture as you can stand.

As stated above, near-field listening seriously minimizes room issues. And the x.7 coincident driver takes the compromise out of nearfield listening.

I think that your difficult room cries out for analysis and mitigation. Guys here know of apps and products to address room issues. Good luck. Happy listening. 

jthifi - don't you love it when he talks that way?I am referring to a Stasis design, and not the original A series. Stasis was the dynamic bias which has morphed and grown throughout Nelson's career.
unsound - I now may remember that we used a 400e at Thiel, but it's all a little fuzzy around the edges. Nonetheless, Pass is an extraordinary designer and any piece of his historical progress seems valid to me.

slhibj -  Limited knowledge and experience here. I hope that directly experienced users will comment. Although the FR and other performance and technologies of the SS 10F are appropriate, I worry about the 30 Watt power rating compared with 50 or more (memory fails) for the original. Rob and I have chosen the 10F for a CS2 and 2.2 replacement. But the 3.5s XO is an octave lower (400 vs 800Hz) plus the Equalizer boosts the low end of the midrange driver considerably. Consider that the stock MR were somewhat fragile if driven hard - we replaced hundreds with burned voice coils. So, I'm leery of the the 10F for the 3.5. But my opinion is just that, an opinion without definitive experience.

We (here) need to find a 3.5 replacement. I suggest you contact Rob Gillum at Coherent Source Service to learn his recent experience regarding your problem. And I also look forward to hearing from others here.

Guys - thanks for the grace for my rant. The marble baffle was a work of art as it stood. The concrete of the 6 and 7 has its own story. My last project in the mid 90s was developing a lower cost baffle for the 6 and 7, which was a Hydrostone product combining gypsum with portland cements , is highly machinable, accepts damping agents and paints quite well, and has zero shrinkage. It also requires forced hot air kiln drying, could not be outsourced and all-in, cost more than the CS5 baffle in its final painted form. That never made it to production.
Back story: introduction of mystery character. Walter Kling was one of the 5 partners who started Thiel Audio as our communal venture. Walter was genius at tooling, jigs, fixtures, and so forth. He complemented my skill set very well, and we wouldn’t have created a company without him. He left after a few years for personal reasons and had a career in architecture. For my exit plan after 20 years, I hired 3 people to carry on the rigors of growing production. Walter was central to that transition, and he shone till he left at Jim’s death. Unsung hero, and Jim’s right-hand man.

Walter brought his architectural mind-set, and Jim had always asserted that "Concrete would be the ultimate cabinet material". I disagreed due to its high Q, low impact strength, continual shrinking over time and its quirkiness as a paint substrate.
You probably saw this coming: they chose concrete for the next baffles. And it shrank and cracked and required replacement with the "polymer mineral compound", which is shorthand for aggregated polyester, similar to the CS5 baffle but strictly paint-grade with no marble / pretensions of high-gloss, see-into glamor. I understand that it worked fine. But then what to do with that Hydrostone research and development? After Thiel, I did some consulting including Hales. Remember the mid 90s upper end Hales’ with cast baffles? That’s the stuff.

Go guy! Just for grins, my 3.5 dream that never was is to mate the hotrodded unequalized 3.5 to a subwoofer crossed at polarity correct 2nd order low pass to the natural roll-out of the sealed 3.5 bass, which is perfectly aligned in its sealed cabinet. If you have the skills and inclination to mess around with that project, I'm available to work with you. 

Andy - Unaskedfor advice, given in good faith: I would find it nearly impossible and at least thoroughly frustrating to design a first-order, phase coherent loudspeaker. It took us decades to develop and improve drivers to eventually cover 7 octaves with good enough behavior so that the long tails on the crosspoints would blend properly without over-excursing, resonating, burning out and otherwise misbehaving. The early custom drivers were developed with Vifa, who is no longer in business. Then we developed and made in-house, which are likewise unavailable. Since first-order xo is generally considered unfeasible, I don't know that suitable drivers are out there to be bought. 

I suggest you consider other options. A sweet alternative is the second-order crossover which manages phase integrity within a broader window, but to my ear does the job quite musically. Lots of high-end companies, as the lovely Thiel 02, take that rout.
Please keep us posted on your progress, whatever you do.

Rosami - thank you for your cogent synopsis.
On the bright side, I do have a pair of 3.6s in the HotRod Garage. So that model is more real than hypothetical. Also, the subtle knowledge I am garnering from my experiments applies to all models. It is incumbent on me to ascertain that my upgrades and recommendations are optimized in all ways, including cost efficacy. It is all too easy to pour lots of money into the works to get less than optimum price-performance.

This project has many aspects and elements. I am making headway, but nowhere near wake speed.

Jon - you guys are expressing the fundamental quandary, sometimes stated as ’Mother Nature is a -----’. The laws of physics can only be bent so far before they dig in their heels. The model 3 (as in 03) was developed with very good bass as a principal design goal. Serious engineering gives that 10" driver quite a bit of oomph and, coupled with a well-tuned passive radiator, it produces pretty satisfying bass response in normal rooms at normal levels. Although Thiel is not associated with Great Bass, in fact Jim valued it highly. The Subwoofer Project grew out of how to produce Great Bass properly integrated with Thiel speakers and the room.

The CS2 came more than 8 years later, at the request of dealers and audiophiles who wanted more nimble delicacy, and were willing to give up some bass to get it. An 8" woofer and smaller midrange have advantages in lower mass and break-up and pushing less air. Purity is considerably easier for the 2.
The 3.6 Renaissance can’t change that, but we can address and improve some deficiencies. We can stiffen the few cabinet vibrations; and we can provide cleaner signal paths via higher quality passive components. They will be cleaner and more transparent.

But when the dust settles and the same level of upgrade has been applied to both the 2 and the 3 series, the 2s will still win the purity contest and the 3s will deliver more bass.
The 6s and 7s use more drivers to cover the same audible range, and both deliver better bass than the 3s and potentially better delicacy than the 2s . . . at higher cost.

Spur - Rob Gillum of Coherent Source Service has repair parts and experience for all Thiel models.

To your direct question: Yes, same tweeter. That tweeter was developed fully in-house for the CS5 and also used in the 2.2 as well as the 3.6. Of course it has been eclipsed by further ongoing development of rare earth focusing magnets and such, but it is a very good tweeter incorporating considerable sophistication.  There are probably somewhere around 25K of them out there, so snag them when you can.

Rosami - Good question about who installs what how and where. At this point the situation is more like a stew simmering on the stove than a clearly delineated plan. It is probable that we will develop a bag of tricks, some of which can be implemented by the user, or perhaps regionally certified technicians. As upgrades become higher in complexity, cost and expectation, it is probable that process testing will become relevant, such as right-left matching or actual driver performance to tweak xo values. So, there will probably be various solutions, and probability that consultation will be part of the mix to determine an upgrade strategy for your particular vision.
Regarding cabinet bracing: Thiel cabinets are very rigid, but there are always weak spots in any outcome. I will address those weak spots and devise fixes which must be retrofittable, because we'll be working through driver openings, and possibly be removing a bottom. My emerging method of promise is to use wooden sticks to join adjacent panels in critical places. Since Thiel panels are braced with MDF cross shelves, the Modulus of Elasticity, deformation characteristics, damping, etc. is simple. Dissimilar materials introduce desired complexity. Long-grain solid spruce possesses relevant engineering properties about 10x greater than MDF. Adjacent panel connection introduces differing geometric forces (from shelves) which serves to mechanically damp resonances without introducing mass, as viscous damping strategies do.

All this is to say that I have been working on this cabinet problem, since the cabinets were my work to begin with. Among the tools I have applied is a digital stethoscope which allows me to listen to the cabinet with my ears simultaneously to exporting data for real-time visual (SpectraFoo) analysis along with storage for further FFT based processing.

This Thiel project is encouraging me to develop such methods which will also apply to my musical instrument development work. I love synergy.

Ron - It's good to be here. This project wouldn't be  happening without you guys.

The 3.7 is perhaps at the end of the timeline, since it is a very well optimized design and not in danger of cap failure. Especially, the cabinet is pretty quiet, but not totally - gains could be made. However, the crossover can benefit from passive component improvements. By the time we get to 3.7s we'll know more than we do now.

In the WayBack Machine - Cabinets and their materials. The 01 and 02 were particle board, which is stiffer than MDF, but less well damped - and it doesn't like to be machined. MDF was a new material in the late 70s and we adopted it for consistency, damping and machinability. BUT the 03 and 03a remained multi-ply Finland Birch plywood which was superior in every way except the lack of predictability of its resonances. In the development of what became the CS3, I experimented with making curved plywood panels, which were magic. But we lacked the resources to develop and incorporate that solution into our products. CurvedPly became one of many "somedays" in the dream sequence. I smile that Thiel eventually came back to that solution and pulled it off very well.

Hello rules - I suggest that the 3.7 chamber is probably larger than the 2.7 because the 3.7 XO frequency is lower than the 2.7 judging by the larger capacitor bank at the infeed of the 2.7.

Redesigning a speaker is a massive undertaking. Budget a man-year more or less depending on your knowledge, skills and hardware.

JA - I’m holding one in my hand. Thiel used Dynaudio’s D28 AF (D 2806-2A) in both the CS3 and CS3.5. Its diaphragm is 3mm greater diameter than the ubiquitous D25. The bottom end of the D28 has higher output for that shallow blend with the midrange.

About a year ago when I was re-exposing myself to the State of Thiel, I called the guy offering this filter. I judged that he is seeking to smooth out the measured upper midrange squiggles without understanding the nature of those squiggles as low-level reflections from the driver geometries, and therefore in the time domain. I agree with Rob that the "fix" would harm the sonic presentation in Thiel terms of fidelity to the input, even though his frequency response graph would look smoother. 

I would be interested to know if anyone here has evaluated the filter. My opinion is extrapolated from what I see and what he told me - I haven't heard it. Anyone?

The bottom line is similar - major surgery required.

Dino - I suggest that you contact Rob. Note that if he neglected your prior request, it was not intentional. Rob has the best chance of supplying relevant information for your decision.

Regarding option 3 above: My present work is all preliminary, and I am making parallel progress on various models, measurement models and lab protocols. But a CS3.5R is hypothetical, with no meaningful timing. When it does issue, it will include some valid driver solution such as replacement drivers and/or repair potential.

I am modeling the old drivers for comparison and compensation in the replacement drivers for their Renaissance-life.

bluetone -  the 3.5 drivers and crossovers would be helpful for my modeling efforts. I would offer them for further use once I am finished with them. I also snagged some CS3 drivers and EQ on ebay, and intend to create an update strategy for all versions of CS3 and 3.5 to better than new functioning. 
If applicable, please PM me to arrange shipping. Thanks. 

Rules - my concern is to thoroughly investigate the parameters and then compensate as necessary. I will develop the Thiele-Small parameters of the old and new drivers to judge how the changes interact with the crossover. Furthermore, since first-order crossovers utilize the midrange over a very broad range, the exact placement and shape of the response anomalies are important. The new crossover must match the response profile of the new driver.
All that said, Rob has been shopping for a drop-in replacement and this driver has been recommended to him by knowledgeable sources. Power disparity would matter - 30 watts raises questions, but the 10F is our best present solution. My job is to subject it to further scrutiny.

Unsound - I like seeing you think outside the lines. Indeed that coax is a work of art that should find additional uses. To add some perspective, let me summarize Jim’s design process.

•The drivers are developed against their optimum criteria.•Each driver is measured thoroughly in an infinite baffle, and in the cabinet.•The cabinet including baffle shape, chamber parameters are optimized to align the driver to its infinite baffle state (as much as possible).•Driver anomalies such as resonances, enclosure effects, etc. are evaluated regarding which ones can benefit from electronic (shaping) circuitry.•Iterative process of driver changes (surround compliance, mass, etc.) with XO circuitry to bring each driver closest to its 6dB/octave slopes.•Continual comparisons of components, layout, etc. for optimization within the cost constraints.
It may be clear that changing a major element such as the driver, chamber size or baffle geometry for a different will have interactive effects on most of the design parameters. It was common for a speaker in development to get cabinet changes during development. It is likely that the new coax driver would have required different cabinet geometries, enclosure volumes, etc. to ’work’ as a colorless transducer at a level to satisfy Jim.
Indeed with much smaller changes such as replacing the original CS2.2 midrange with the ScanSpeak 10F, there are significant XO changes required because all the resonances and T/S parameters are different from the original. In steep-slope designs a driver can be ’dropped in’ because most of the anomalous behavior is in the extended overlap zones which are attenuated by the steep slopes. We don’t have that luxury.

Just my little peek behind the curtain regarding how everything is hooked to everything, nothing is simple, and no good idea goes unpunished.

Ron - your 3.7s are State of Thiel Art and I have plenty to learn before claiming to improve them. But, I already am generating ideas.

Guys - we have learned enough to repeat the successful upgrade that Beetle performed. But we have not yet approached the comparative analysis to find cost-effective performance points. Beetle-level parts cost would exceed $1K, and parts cost is not the whole job. So I believe more work is in order.

sdl4 - an outsider you might consider is Morrow. I ended up outfitting my studio with Morrow ICs and speaker runs. My comparisons weren't terribly extensive, but I did read widely and compare against some old Kimber, Straightwire, OCOS, Mogami, MIT, Magnum, and Audioquest, all from the 1980s. I landed at Morrow's "4" series as their cost-effective sweet spot. Morrow direct markets with periodic discount sales and a generous return policy.
If you do give them a try, or if anyone else here has tried Morrow, I am interested in your reports. To my ear it meets your "relaxed, detailed and natural" criteria at a modest price.

Rob - a word for the record about insulation. There's quite a bit to it.Common wisdom is that wood  is 'better' than glas, etc. , which I believe oversimplifies the situation considerably. The various materials act differently, so even at equal cost, specific application criteria apply. I ran extensive insulation comparisons at Thiel in the 80s. We landed at pure wool felt on the back wall of midrange enclosures - nothing else comes close to how well that acts in controlling reflections and so forth. In fact, we called the various engineered foams, batts, etc. "wishful engineering" due to their relatively poor performance. When it comes to filling cavities, polyester fiber-fill of various densities and fiberglas are the front runners. Wool has a bad habit of 'unloading' as frequency drops, which is contrary to what is often optimum. Polyester falls somewhere in the middle. Fiberglas has enough crooks and nodes to stand up to bass resonances without squirming. It outperforms the rest in my experience. That opinion is shared by some other designers of note.

Big problems with 'glas include irritating dust and a sharp hand. But wait! all 'glas' is not created equal. Thiel's 'glas is not builder's fiberglas insulation, it is a clean, graded product for industrial use. I got our fiberglas certified to our insurer's satisfaction.

If you want to compare different kinds of stuffing, I suggest measuring the outcome with REW, etc. to make sure you aren't  compromising performance. Your correct dose will depend on outcome, not on comparative density, mass or volume. Most of the non-'glas products are likely to unload the bass to produce a lower Q with less flat response. Some folks might like that. Jim's goals were toward flat, uncolored bass extension with a 0.7 damping factor, which he considered the ideal. Some folks might prefer a fuller, looser bass from other materials. Let us know what you learn.

catalysis - I second your motion re the subs.

sdl4 - you are correct about bridged amps 'seeing' half the load impedance of a stereo amp. That said, the 2.2 is a fairly benign load, especially as Thiels go.

I recommend you compare your PS to the AHB if you can. I was gobsmacked by the difference between my S300 and the AHB. I'll soon have more to report, I'm expecting delivery today of a BM DAC3B and HPA4 front end. Better than Christmas.

I also have some thoughts about the 'extra' load imposed by the EQ - more on that later.

Andy - the baffle is critically important for how the waveforms propagate into the room. Going back in history, common wisdom dictated the drivers be offset by differing amounts from all baffle edges so as to spread the diffractive effects out, over time. Thiel was an early originator of minimizing diffraction, first by absorption in the 03a and then by rounded baffle edges in the CS3 and beyond. We centered the drivers, like the mouth is centered in the head and a microphone diaphragm is centered in its structure.

There are many conflicting demands of driver placement geometry including unknown listener distance, reviewers ignoring the grille when it is a functional ingredient of diffraction control and so forth. Our seminal statement of the CS3, had the tweeter very high and equidistant L-R-T. The top baffle curve was completed by the top cross strut of the grille frame as were the sides to a lesser extent. Lots of time went into optimizing that system. But in use, the grille was often removed, including for reviewer testing, and the resultant diffraction was noted as a flaw in the design, never as a failure of the user - which left us all flabbergasted and Jim really angry.

As aside to that point. I remember the years-long comments by Larry Archibald, Stereophile publisher, regarding the "early" and "late" CS2s, and the latter's taming of upper range glare and roughness. He never admitted in print that he stubbornly listened to the CS2s for the first year without their sculpted grilles, which controlled edge diffraction as well as incorporated a shallow tweeter waveguide for limiting dispersion of the tweeter's low end to blend properly with the midrange. That blending is far more critical with the large-overlap first order slopes. Bottom line: there were no early or late CS2s, only eventual user cooperation with the design intent . . . and lots of confusion and mis-information in the marketplace.

Back to the point: tweeter placement evenly spaced to all edges including the top acts more like an ideal point source. Note that the CS3.7's industrial design is extremely polarizing and generally considered offensive. Only those adherents to form following function "get it" and either make their peace with it or actually love it. The driver height scheme is optimized for the broadest vertical listening window for seated listeners, dependent on crosspoint frequencies and driver dispersion characteristics.

So, Andy, even though the 2.4, like the earlier designs, does control diffraction very well, its lower tweeter creates a time discrepancy between side and top diffraction. That effect would be extremely subtle, such that I would be surprised if I could ever hear it. But, your ears are younger than mine.

Prof - Cabinet stiffness including the baffle helps it disappear. I am experimenting with "stiffening" the MDF cabinets, especially the driver mounting recesses with Minwax Wood Hardener. Also that 3.7 cap is far quieter than the 2.7 top. I am extrapolating from how surprisingly active the small top of the 1.6 is. Column ends accumulate pressure and the 3.7 is both very strong and shaped to handle the job.

Unsound - everything matters, and there may be some effects from the forward parts. The cross struts are 1/4" diameter round bar, and not in the path of any drivers. The forward side verticals are 1/2x1/2" MDF, but so far off axis that they present a diamond profile more so than a flat surface. In developing those grilles, we couldn't measure or hear any deleterious effects (although they are hypothetically present.) The fabric itself does add resistance as frequency increases. We voiced for flat with that fabric in the  sound path.

That grille cloth was the most transparent we could find. Present cloth is even more transparent.

Beginning with the CS2, those frames went away, except for the CS5.
Our first CNC mill in 1985 allowed us to machine the more sophisticated solutions beginning with the CS2.2. Later, such as in the 2.4, the 1/16" die-cut steel frame virtually eliminates the frame's geometric advantages.

Prof and you other 02 fans - I just now got my old 02s on line - received from my brother from their life-long use in our parents' system. Drivers are good, coils are good - within their genre. These speakers were made in 1978 before we discovered high-purity wire and baked fusing. These are normal ETP "magnet wire", still the industry standard. We wound these on "Ol' George" - hand-fed with a timer. We then tweaked each coil via weight on a triple beam balance and DC resistance. I checked these today with my inductance meter and they're within 2% of target.
They are considerably better wound than Beetle's Chinese 2.4 coils.

Back to the XO. The resistors are 5 watt carbon composition - all burned to a crisp. The caps are all 50 volt electrolytic - about half of them are blown out, the other half are + 30% out of spec. I replaced all with original grade parts and fired them up.
I haven't heard 02s for at least 30 years. I am impressed! The sonic signature / tonal balance is pure Thiel and they don't screw up the music. They lose a little in direct comparison to the PowerPoint1.2s and the CS1.6s, both 6.5" 2-ways, but really, they aren't shamed.
If we had chosen to take this 2nd order route with our developing company, life would have been immensely easier. However, we may have failed because the territory is so much easier - everybody entering the game played there and it is unknown whether we could have differentiated ourselves.

Anyhow, thank you Prof. Your enthusiasm for the model inspired me to consider it. After hearing it, I am addressing it and will learn a lot in this relatively simple sandbox. The 02R will live. I hope you guys can snag some on the cheap. Rob and I think we made fewer than 1000 between 1976 and 1984 and they would be concentrated on the East Coast and in Germany - go figure.

Unsound - right on! In fact, the model 2's reason for being was conceived as just that - create a huge bang for buck product utilizing technologies and parts whose development costs were amortized at the higher-end.

In the beginning . . . Jim was a very linear, no-nonsense engineer. He projected a 2-way, a 3- way, eventually 4-way, etc. The 3-way was the model 3 (a-ha!), so how to imagine another 3-way? We found merit in the smaller-driver, less bass, less balls idea and committed to allowing, even encouraging trickle-down. The CS2.2 utilized the tweeter developed for our statement CS5.  And so forth. Things got murkier as we developed more products for multiple markets. But indeed the model 2 is in its soul Thiel's cost-performance champion.

Andy - regarding diffraction - please pardon me not knowing its controversy, I've been out of the loop too long. It is not controversial nor trivial, it is a fundamental element of design; it is measurable and directly hearable if the system is transparent. For yourself, or with a friend, listen to yourself speaking or singing. Then put your flat hands beside your mouth and repeat. The difference you hear is diffraction.
However, there are cases where diffraction is masked in playback. Since diffraction is a phase-time anomaly, those systems with compromised phase-time performance might mask its presence or conflate diffraction with the other phase-time puzzles requiring mental reconfiguration. 

Regarding baffleless drivers vs rounded baffles. We experimented with both early-on. Both have their merits. However, the baffleless concept treats the individual drivers as separate units with their own diameter to wavelength behaviors. The rounded (Thiel) baffle treats the unified radiating soundfield as a whole. On balance, I believe (as I would . . . duh) that the unified, rounded single baffle does more things right; its major flaw being increased cost.

Looking to natural acoustics, the mouth or musical instrument forms its soundfield across its entire frequency band, not as separate sources for bass, midrange or treble.

On to the greater concept of controversy: most controversies have little to nothing to do with the subject under scrutiny. They have more to do with the biases or knowledge (including deficits) of those arguing. So much would become so much clearer by using the correct pronouns. When someone says "you can't hear it", I believe they are actually saying "I can't hear it". Fair enough, except that they are projecting their lack of hearing on "you", which includes you and me and everyone else who is hearing it. In my experience, that projector has rarely if ever actually tried to hear the difference being denied, or at least not objectively analyzed and explored the territory. Imagine if you or I were so bold as to tell a conductor or band leader or competent recording engineer that "you can't hear it". If he didn't slap you, he would either dismiss you or ignore you. We earn our keep by hearing it and understanding it and making it better. End of rant. 

Prof - the 02s were made from standard-grade stuff, miniature, standard electrolytic caps,5w carbon comp resistors and normal coil wire. The system will often work with bad parts. At nearly 40 years old, the caps are almost certainly shot, unless they were replaced over the years. Other parts could be OK unless you boogied too hard. The non-technician's look-see is to remove the woofer, shove the insulation aside and see if anything (caps) looks like the 4th of July. 

Yes, I am upgrading at 3 levels: 1) better basic parts (no electrolytics), stabilized coils (varnish and bake), new hook-up wire and jacks, reworked grille to reduce diffraction. 2) add upgraded drivers ( possibly Thiel CS0.5), Mills resistors, better caps including CSA tweeter feed, 3D grille like CS1.3) add sweet passive parts, double jacks, foil coils and sexy golden ellipse cabinet edges front and back. Concepts to be developed in 02 for use in all models. 02s are a very manageable sandbox.

I have to find some more pairs - please keep me in mind.

Rules - I believe that your undertaking will be of great value to Thiel fans. The 3.5 was a pivotal product in Thiel's emergence; but it has lots of room to grow. The cabinet is a significant piece of technology, so re-outfitting its DNA for a new life is exciting. Of the 5000+ pairs sold, I suspect a high percentage are salvageable. Those coils are of the SOTA-aerospace 6-9s variety as are the ultra-bypass nano caps.

JA - please provide a link. I don't see it on YouTube.

unsound - points well taken. Thanks for your linked reminders. And a comment here. My thought was more from the perspective of the extreme difficulty of producing so much bass with the dynamic range and extension possible with today's digital techniques without running out of steam somewhere, especially at the lower reaches of the midrange driver. My hypothetical suggestion would include placing the stereo subwoofer pair to create proper time arrival with the main speakers. The main speakers would be relieved of their heavy chore of reaching all the way to 20 (or 40) Hertz.
All that said, I really love the 3.5s I recently got via this thread, and do consider them, like others here, to be Jim's quintessential work in many ways. I also apply that to the (unsuccessful) CS5 short series. Of note is that those speakers are truly minimum phase transducers, all the way to the deepest bass - and the sound underscores it.

More later. A guest just arrived.

A Renaissance 3.5 would require all new drivers. The extant woofers are bulletproof and decent, but lack the sophistication of later designs. They are a 1985 solution. The replacement midrange needs to be a full range driver, which is findable, we just haven't yet succeeded - same with the tweeter. The XO upgrades I have been developing this past year all apply to the 3.5 to up the performance league dramatically.
The EQ could also be executed better. It is all discrete, and therefore upgradable, but it is likely that a talented circuit designer could do the job better today. Now, IF the EQ could be a requirement rather than an option, then other possibilities emerge. The midrange / tweeter could run on one channel with no boost to its low end. The equalized woofer would run on the other channel for good power balance. The crossovers could be incorporated into the EQ before the power amplification stage. It is possible that the woofer, or all the drivers could run crossoverless for direct, lossless feed, or with driver-specific and tweakable elements installed in an outboard XO near the cabinet. Such an active-crossover analog solution would squarely fit Jim's approach and sensibilities, and indeed one of the original design pilots which was deemed unfeasible in the company's fledgling state.

As an aside from history, I really like the 3.5 cabinet which shares technologies with the CS2 and 3 but obsoleted by the 2.2 and onward. Those early cabinets used MDF only for the baffle for its sculptability. The walls are 1-1/8" industrial particle board laminated both sides - for about 1.5x the stiffness of later 1" MDF cabinets. For production management reasons we landed on 1" MDF as our sole cabinet material. But the particle board is better.

Obviously, I am personally more excited by this potential upgrade than by pairing the non-eq 3.5 with a sub. But it would take lots of doing by someone(s) with youthful talent and vigor to pull it off. BTW: the CS3 cabinet is functionally identical to the 3.5; the combined 3 and 3.5 models sold about 7000 pair, and the 03, 03a adds another 2000 pair build of 20mm FinPly (better than BalticBirch or ParticleBoard). Hmmm.

Power recommendations are really sketchy, since the manufacturer doesn't know your loudness requirements or room size which are the largest factors in the equation. Regarding the sensitivity increase, I remember that driver refinements allowed lowering of series resistors in all 3 drivers to gain the 1dB increase. The minimum requirement was increased because all driver failures were caused by overheating voice coils (mostly midrange), which is invariably caused by clipping.

I would expect all sonic characteristics to improve with doubling your amplification. Your challenge will be self-control. The cleanliness of the response might tempt you to crank up the volume. The equalizer boost does make serious demands on the midrange driver, so keep it modest, or keep the EQ at 40Hz or off if you're gonna rock.

Let us know how it shakes out.