Isiris evolution to Osiris Jr? Summary of Isiris build and new update

[JonMarsh]

The Isiris never got it's own build thread, being the culmination of the "Three Way Evil Design Study" - it didn't seem very urgent, as only two pair were expected to be built world wide, due to availability issues with the NS12-513-4A woofer. Of course, that part of the design can be redone with another driver, but that's not what this thread is about. This is about modeling and hopefully eventual modification of my build from the original three way to a four way system with an active low end and a new mid bass/low midrange driver. This got kicked off by some of the discussion in the Unicorn Dreams and Tesseract thread, as this type of four way configuration is at the heart the old Avalon Osiris, the Avalon Sentinel, and the Avalon Tesseract, with various other concepts and acoustical approaches to optimizing power response thrown in. A common thread is an active low end system, and pistonic operation of drivers. Ironically that was also a concept common to the four way system cabinets I gave to Charles Hansen after I got married and they didn't meet WAF criteria at that time.


Text will be added in the near future as time permits this week- go to Post Off Topic to see how things will proceed.



First, to recapitulate the basic details of the Isiris (aka known/named as CA Isis by a good friend in Germany)

[JonMarsh]

Driver measurements




These are gated measurements in room, with long enough gates to show some of the room modes (35H and 170 Hz dips), but still OK for development, in my opinion.

The Aurasound NS12-513-4A




An on-off axis series for the Accuton C173-6-90; 0 deg, 15 deg, 30 deg, 45 deg




An on-off axis series for the Jantzen JDT-1024. Also at 0 deg, 15 deg, 30 deg, 45 deg. Looks like they made some fairly decent design decisions with their diffusor dot; this is about the best performance off axis I've measured for a hard dome. :T




And of course, checking distortion on the JDT-1024 out of curiosity - I'd say it's fairly decent, in the high end Seas/Scanspeak range above 1500 Hz. Third order is very low, 2nd order fairly reasonable. I was NOT trying to do super quiet distortion measurements, just took SPL test data in Fuzzmeaure and did a distortion analysis- ambient noise was higher than I'd like for this test.

[JonMarsh]

This is a follow up to my LspCAD data on a proposed modification to the LF to mid crossover, based on what I was hearing, and going back and redoing and filtering some measurement data.

This will seem a bit OCD to some, so for those of you, "these aren't the data graphs you're looking for, move along now..." :B

I decided to measure both nearfield and at 1 meter on axis, and look at the original response and my proposed modifications, and compare individually for the LF and Mid, and then jointly- thought this might be more useful than just looking at the axial response net at 1 meter. This proved to be quite helpful, as there are clues in the individual response curves that give understanding (to me) for what I hear in the power response which is not well represented in the 1 meter on axis response- another reason I rely on this less and less.


This is the near field LF response, original (blue) and modified (maroon). The modified LF circuit adds 82 uF to the 125uF original C1.



This tightens up the roll off, and the near field shows the corner frequency pulled in and 250 Hz down a bit more than the original- this actually tightens up the phase transition and should result in a better phase match through the crossover region with the adjusted mid.


This is the one meter LF response, original green, modified (red):



This shows some of the room effects along with the increased roll off rate, and the slight tip up coming from baffle step transition near the crossover point; this reaches a better net acoustical point than the original curve.


This is the near field (six inches) midrange response, original (maroon) and modified (red):



The effect of 2.2mH and 4 ohms connected in series, paralleled with the 8 ohm input attenuation resistor is quite apparent; this is needed for baffle step compensation.

The one meter response is shown next for the original filter design (blue) and the modified version (green); as expected. This compensates better for baffle step effects.




This is an overlay of the summed LF and Midrange at 1 meter, original (green) and modified (maroon).



Now, this is probably the LEAST informative measurement- yes, the modified maroon curve for the total 1 meter on axis is a bit flatter, but the sonic difference is more apparent in the power response in room, after listening. What seems to be happening here is that at 1 meter on axis the old network was not summing in correct phase, and I believe this would show up more with more off axis (including vertical) measurements. I think there is less interference summing between the drivers off axis than with the original network, and the improved lower midrange output makes a big difference in many midrange instrumental fundamentals and voice, and in area's that don't show up so much in the on axis measurement- say, in the 300-500 Hz area, by my ear.



this is the total response at 1 meter, same measurement point, for the modified LF and Mid networks, with the tweeter added in:



There's a buch more material I'd like to listen to, but quick checks on two albums have convinced me that this is either "it" or very, very close.

Now, this plot above for the complete system would suggest a soft top end and a slight presence dip. BUT, they don't sound this way at all in this room, even with some Real Traps on the walls. I figure that with the very good off axis response of the system, the net power response is giving them the presence that the curve above could suggest that they lack. Curious, and interesting, to say the least.

Should be pulling drivers out and starting the detailed finishing steps this weekend, and finalizing the BOM.

I'll do a measurement at 30 degrees, too, and add that in this weekend.

[JonMarsh]

Last additional reference plot for now: 0 degrees and 30 degrees, 1 meter

I think this comes as close as I can hope for, for now, to my original planned voicing target- and I like what I hear.






It may not be obvious, but this does fit more or less with my revised target concept, based on Linkwitz's studies with Don Barringer and his Version 2 curve.

[JonMarsh]

Thinking about this at lunchtime today, and having been wanting to put some tuning tricks into the final crossover, my first pass on that in LspCAD looks flexible enough to give me a useful range of results without getting myself in big trouble.

Basically, one switch in the midrange section to had an approximately 2 dB contour/level tune for the upper midrange, and one switch in the tweeter section to adjust the tweeter level by 2 dB. Depending on how the combination is used, different "house curves" can be approximated; which is not to say that my choices for WHAT house curve to approximate won't get tweaked a bit once things are setup in their ultimate home.


This combination I call the "flat curve", for very smooth flat response at 30 degrees off axis from 100 Hz on up.




Transfer function:




SPL curve (predicted)




This I call the "Corner Curve" (or should it be Korner Kurve? :W )





Corner Transfer Function:




Corner SPL Plot:




And this approximate a mild tilt function:





With the Tilt Transfer Function:




And Tilt SPL Plot:



Let's see how this works out in the real world... mmm, Madisound is going to love me, buying more of those Mundorf resistors. Obviously they could be modified to be more pronounced in the effect, but maybe starting off a little more subtle is a good idea for now.

[JonMarsh]

Last, ordered the last batch of resistors needed to build the updates, and some Acoustastuff for the top modules.

[JonMarsh]

Considering a variety of listening tests, on and off axis measurements, and the present plan to rebuild the crossovers in a form factor that can be installed in the cabinets, a certain amount of design massage has been going on, especially since I won't be able to teak levels if the crossovers are internal. (well, maybe I can, at least for the tweeter, but it might be messy...)



[ATTACH=JSON]{"alt":"Click image for larger version Name:\tIsirisJuneUpdate_SPL-June30_zpsbf11c0e9.png Views:\t70 Size:\t49.3 KB ID:\t937121","data-align":"none","data-attachmentid":"937121","data-size":"full","title":"IsirisJuneUpdate_SPL-June30_zpsbf11c0e9.png"}[/ATTACH





The woofer rebuild is complete, the tweeter one close, and about half the parts on hand for the midrange (which has not changed in design, other than fixing some variable resistor configurations or eliminating the options.

This "looks" like it can be a little hot in the 5 K range, but grilles will absorb some there- also, this is using 40 degree off axis data- on axis is softer at 5K, and hotter above 15 K. The rise in the LF with the crossover is likely an artifact of how LspCAD calculates the network response and models drivers- we'll see in the final measurements.

[JonMarsh]

OK, a bit of a teaser- here's a few foils from the presentation...

I'll annotate these in the build thread post...











(this is basically the definition of an LR3 network, a combination of B2 and B1)





(This is the basis for the transfer function of the Modula MT, Natalie P, and the HF crossover of the Isiris).

[Evil Twin]

Clearly it was time for an intervention....

The Chancellor's patience has worn thin- mine dissipated many months ago. Clearly Mr. Marsh has been too pre-occupied with his musical instrument interests of late, and his dedication to this project has waned- certainly tangible results have NOT been forthcoming for some time. It is past time to turn that situation around.


To that end, at the Chancellor's request, I have conducted an updated design review, and also begun addressing procurement deficiencies. Obviously, if certain individuals cannot be sufficiently motivated by the power of the Dark side, or think themselves immune, the only solution left is abrogation.


As a result, a new design review has been conducted, and procurement activities updates as a result. This will result in an expedited completion of this technological terror, and establishment of a clear benchmark for speaker development in the Empire or the New Republic.

More rigorous criteria have been applied for phase matching during transitions and inverted response profiles, resulting in this modified network design and the documented performance:



This even results in some additional economies of component cost.



The predicted composite response should be most satisfactory.

Impedance behavior and individual network transfer functions comply fully with the preferred targets.







Now it will only be necessary to find a way to supply sufficient motivation to procurement and assembly personnel to get this project back on track... I trust this it shall not prove difficult to motivate those who would prefer not to feel the censure of the Dark side and the necessary penalties.

[JonMarsh]

However, this is the update on the complete crossover, with the updated midrange broken down by components. Note that the tweeter caps could be MR or Z-Superior, and the recently built ones are Z-Superior as a test.

[JonMarsh]

The new woofer boards are finally connected. Still getting ready this AM to do some measurements, but found I must have left the new tweeter crossover boards at home- road trip time!

[JonMarsh]

This is the measured response at the listening position in room for the left cabinet with corrected connections with a 150ms measurement window, using a Half Hamming window shape. Note that the measurement window extends to 50 kHz. This is on axis, where the JDT1024 has a bump up in the 13-17 kHz area; it's actually flattest at 30 degrees off axis.

This is with the newer V2 woofer and tweeter crossovers; the tweeter doesn't really change the response but has better impedance characteristics and less expensive cap sizes! It's been very frustrating to be so busy I couldn't get around to doing this earlier; in fact, I was originally scheduled by the company I'm "embedded" at to be working this weekend, but got my boss to push back on it. Unfortunately, that meant one of our other, more junior guys had to do it. I should feel guilty, but I don't, as he lives down there and has a much shorter commute to Cupertino than I do.

The newer crossover midrange crossover is less like a B3 alignment and more into the LR3 mode, and will exhibit more issues at the lower crossover point if out of phase.

[JonMarsh]

Here's the response with in phase connection of the midrange driver, all other conditions the same as previous measurements.



This looks about like what I expected after the last round of LspCAD sims and the crossover updates that came from that. Note this is at 1/12th octave smoothing.

There's still some other work to be on internal damping of the midrange cabinet and external diffraction control and grill clothes. There's some stuff I'm seeing in unsmoothed data that looks mostly like diffraction effects. Time to squash those. (well, when I next get some free time- we're making progress on backfill hiring and getting me back into a more normal role at work).



Who knows, with the right work, we may get pretty close to this kind of smoothness- just for fun, 1/3 octave smoothing, like you see a lot of guys posting their designs with. Can't blame them, it looks good this way, and may fairly correlate well with the true balance and smoothness of the drivers and network.



We'll have to see how close to this I can get with the next round of work. :W

A refresher: the current schematic:



The new build has split the midrange into two long boards, so they can fit internally in the cabinet. The changes in the crossover are smaller values for the two series caps, particularly the C2a cluster, and L5 going from 1 mHz to 820 mH. Note, I've actually made the crossover less expensive.... mmm, about time, huh? For the tweeter crossover, I normally have that switch closed (actually, hardwired) but it can be used to provide a softer more polite top end. With my current DAC and electronics, I don't think that is necessary or desirable, but I'm sure if I had my old Benchmark DAC, I'd be flipping that switch open in a heart beat.

I did a quick "sound check" before moving on to other chores and errands, just the first cut of these albums...










I'm pretty happy with a quick check on the results! More palpability.... more focus to imaging.


The design will be the basis for the Wavecor Ardent network, too. It may wind up being quite similar.

[JonMarsh]

This belongs in the yet to be created official build thread, but may be useful for reference.

Keep in mind, doing R&D, I've basically built two sets of crossovers...

this is for the current Mk2 configuration, which is also Mike's woofer configuration and similar to his tweeter configuration- this is the component breakdown and cost for a full stereo set, not per speaker. I shudder to think what the markup would be in a commercial speaker, but then, there's probably a reason that a set of Avalon Isis goes for about $72K list.

Lotsa expensive parts, likely can be done for less than 1/2, but how much remains to be seen- will be targeting cost and size reduction for similar crossover for Wavecor Ardents.

[JonMarsh]

Going totally OCD

Refine, refine, refine. Or go OCD. Where's the difference?

Anyway, more thinking and tweaking in LspCAD, and ordered parts to implement testing modifications to all three crossover sections, though woofer is only a resistor change...


This may just amount to playing around in the error between simulation and measurement, as the measurements look pretty good, but... just in case.



Modifications as basically noted in upper text block,

• R5: from 11 to 16 ohms
• C6: 125 uF to 85 uF
• C4: 6.8uF to 5.6uF
• C7: 15uF to 10 uF
• Some small adjustments to tweeter and midrange level via pad resistors

[JonMarsh]

Updated "production" schematic, parts on hand to modify, but the Wavecor Ardent will take priority this weekend, and many to come.



Sill, I'm looking forward to getting to this- maybe over the Easter break while lacquer is drying? :W




Well, my colleague in crime from Munich will be here on business in the middle of June, and he's been really wanting to hear what he calls the "California Isis", so it behooves me to see if I can pull the latest simulated updates together quickly and get some measurements done, at least enough to finalize the crossover with external construction for now.




So more parts on order this lunchtime, to cover what I don't already have (fortunately do have the Clarity MR in 10uf). I really should have thought of this order yesterday, because then it might have been here before the weekend! or at least earlier today... just a bit too much going on around here right now. Well, at least there's no time to be bored!

[JonMarsh]

Last Update on Original Osiris Crossover - OCD stages on crossover

Have tested the above schematic now- built up a pair of new tweeter crossovers, and updated the value of C2 also, to 72 uF total. It's nice when things measure in room pretty much like the simulation - so there are improved reverse null dips, including depth and more symmetrical, and the overall response is quthe rite smooth if you measure close enough to ignore a lot of the room reflections. Of course, that's not how we listen, so measuring at the listening position on the couch is also an interesting exercise, if not one that often looks all that pretty!






I also worked on room positioning, getting permission to for now pull them out further into the room and change the spacing and relocate the couch. These all have salutary effects, even when you're listening or measuring 6-8 feet behind the couch!

This is the "couch potato" measurement, at 15" degrees off axis from the tweeter, as they are not completely cross-fired.

The goofy thing is, sitting at the work desk behind the couch, listening to on channel of the new Oregon album, or some Jacque Loussier (88.2 kHz material downconverter from ripped SACD) they sound like they have center fill to the middle of the room, even though there's only the left channel playing through the test setup. Hmmm, probably had a glass too many of chardonnay with dinner....:W

[JonMarsh]

Enter the mid bass-low midrange driver candidate:







This is one candidate under consideration- an expensive one, but with some pedigree and the desired efficiency. It also can be tuned to the desired LF alignment in a relatively small volume, but having high suspension compliance, should exhibit good output versus 3rd harmonic characteristics.

I've modeled it with VituixCAD, first preparing a box model for a test enclosure of 25L, and then a high pass alignment for the mid bass cut off, and an approximate low pass alignment similar to the original woofer crossover mating to the Accuton CZ173-6-090 midrange.





LF to MB crossover point is about 65-70 Hz; the Aurasounds in the 110L enclosure are shown with an LR4 low pass crossover (active at 65 Hz) and LF equalization, modeled in VituixCAD. Above shows the detail of SPL response; below the overall design summary. Note, the impedance curve seems to be derived from paralleling the active powered driver with the passive drivers; I suspect there is a way to fix this but I need to review the documentation or get in touch with the developer.

[BobEllis]

I realize ET has set a near term deadline, but audiohobby.eu prices on the Mundorf resistors are also about half of what Madisound gets. I'm guessing shipping will eat a bunch of that, but a single large order should still save a bunch. Janzten inductors look to be about half the price that PE gets, as well but shipping probably eats all the saving. I'm going to have to take a close look at my Ardent XO order. Thanks Renron!

[5th element]

One thing that occurred to me about this was that the S280 is, in and of itself, a very capable bass driver. One approach to this would be to make and model the S280 in its own enclosure to replace the module that the NS12s sit in. Port the S280s, as its specs seem ideally suited, and hopefully tune them to a suitable frequency that would keep excursion low and distortion lower within an optimum frequency band, say giving good extension down to the mid 30s or so (I haven't simmed it so I have no idea if that's feasible). Perhaps place a high pass, flux capacitor style, in there to keep excursion below this under control, or tune for a slight bass hump (in a smaller cabinet?) to increase efficiency where it counts, say 40-50Hz then use the flux cap to EQ the hump back flat.

Then use the NS12 cabs as actual subwoofers, optimally placed for bass performance, rather than in room as stands for the top section. The NS12 subs can then be used as stands for plant pots or other such ornaments with the S280 cabinet perhaps being slightly more compact and SO friendly.

I suppose that turns the idea of a tesseract themed speaker on its head but the basic design of the tesseact is flawed imo anyway. It is perfect for traditionalists who like to keep their speakers as two huge multiway beasts a la the original Focal Utopia. But if you've already got a very capable bass driver within the system, that can play loud and clean where it counts, then why box the subs into the same acoustic space when putting them elsewhere would grant significant gains.

I do like the idea of a more multiway approach, for optimising each speakers pass band a little more, but with drivers that are so capable, at least these days, it isn't really necessary.

I do like the idea of JA Diamond tweeter + C50 2" dome. Cross at 3.5kHz or so to keep the diamond really clean. Then bring in a C90 6.5" or S220 below it. Use the C50 as a classic 'filler' driver crossing in the C90 or S220 at whatever frequency is necessary to provide the bandpass gain required to make the C50s sensitivity deficit a non issue. The C90 might be the best fit for that if it needs to cross over higher than lower. Then bring in the S280 below that. Just more food for thought I guess. And more loudspeakers to build -.-

On a side note those S280s aren't cheap!

[meb46]

Traditionalist - That's exactly where I fall... but at the same time, I do acknowledge that its sometimes to the detriment on the perfect solution

[meb46]

Forgot to mention - Cripes Jon, you HAVE been busy! Spare time on your flight home

[JonMarsh]

It was a pretty intense week last week, on many fronts. Witness also my Munich Report Part One, which is almost finished; there is a Part Two going to go up, too.

Matt, you comments are all quite valid- what I really want to try out here is a way to manage the mid bass to woofer/subwoofer transition in a way that doesn't require a conventional high pass section active crossover (for now; who knows what I might think three or six months from now!) and the sort of kluge idea I'm pursuing now, retaining the existing bass modules, is so I can get something running with a minimum amount of extra work, because there's just too much on the plate this year already in too many areas for me to to do another new major build- but modifying an existing one, creating a mid bass module, could be within the realm of possibility. In fact, I may be able to pull it together to order the S280's around the end of the week. I'm working on ideas for the actual module design right now, to accommodate installing the existing and new/updated crossover boards.

Baby steps, so to speak... expensive baby steps, yes, but it could be quite interesting...

BTW, all the re-cycled Isiris stuff is to lay the foundation for this within the thread, to highlight the degree of OCD behavior which occurred in the development of the crossover, and to also make something closer to a focussed build thread than the original "Three Way Evil Design Study" was.

[5th element]

You'll get no argument from me Jon on that front, where time is short, ideas are many and expendable income can help make some of those ideas a fun and fulfilling reality. Better to spend, get something interesting rolling and complete than to sit on that idea/money for a rainy day that never comes!

The same for the project idea where the simple small cabinet addition is something that can happen in a short space of time than a huge rebuild idea that would take forever!

[JonMarsh]

One thing that occurred to me about this was that the S280 is, in and of itself, a very capable bass driver. One approach to this would be to make and model the S280 in its own enclosure to replace the module that the NS12s sit in. Port the S280s, as its specs seem ideally suited, and hopefully tune them to a suitable frequency that would keep excursion low and distortion lower within an optimum frequency band, say giving good extension down to the mid 30s or so (I haven't simmed it so I have no idea if that's feasible). Perhaps place a high pass, flux capacitor style, in there to keep excursion below this under control, or tune for a slight bass hump (in a smaller cabinet?) to increase efficiency where it counts, say 40-50Hz then use the flux cap to EQ the hump back flat.

Then use the NS12 cabs as actual subwoofers, optimally placed for bass performance, rather than in room as stands for the top section. The NS12 subs can then be used as stands for plant pots or other such ornaments with the S280 cabinet perhaps being slightly more compact and SO friendly.

I suppose that turns the idea of a tesseract themed speaker on its head but the basic design of the tesseact is flawed imo anyway. It is perfect for traditionalists who like to keep their speakers as two huge multiway beasts a la the original Focal Utopia. But if you've already got a very capable bass driver within the system, that can play loud and clean where it counts, then why box the subs into the same acoustic space when putting them elsewhere would grant significant gains.

I do like the idea of a more multiway approach, for optimising each speakers pass band a little more, but with drivers that are so capable, at least these days, it isn't really necessary.

I do like the idea of JA Diamond tweeter + C50 2" dome. Cross at 3.5kHz or so to keep the diamond really clean. Then bring in a C90 6.5" or S220 below it. Use the C50 as a classic 'filler' driver crossing in the C90 or S220 at whatever frequency is necessary to provide the bandpass gain required to make the C50s sensitivity deficit a non issue. The C90 might be the best fit for that if it needs to cross over higher than lower. Then bring in the S280 below that. Just more food for thought I guess. And more loudspeakers to build -.-

On a side note those S280s aren't cheap!

There would be a lot of possibilities with that approach- including tricks like the guys at Gauder Akustik use- (more on that later- for now, think in terms of a 2nd order high pass tuned around the ported system resonance, and having a tapped inductor or capacitor setup to adjust the Q of the HP filter and thereby provide some bass level output control (say, in 1.5 dB or 2 dB increments, + and -, around nominally flat)

This could be done for the LF, but it could also be done for the mid bass, especially as I'm pondering a passive 2nd order high pass anyway...

But I'm still thinking in terms of fast and not too dirty for now- a quick evaluation for the mid bass. Whole woofer system could be replaced with something else later, have several ideas, including AE drivers and SS PR's. The woofer system being active as previously described.

[JonMarsh]

Fast Path to Test enclosure for mid bass

Good ole PE to the rescue... these types of boxes worked well for my initial SW223BD02 and PR tests- found one that looks nearly perfect for the mid bass test.









Will give me a close fit for baffle width (14.25" versus 15") and the right internal volume for the mid bass tuning I think I want.

May have to modify the bracing design for enough rear space for the driver frame. Something to check before assembling.

[JonMarsh]

This is going in a slightly different direction now, in the case of my personal update. Instead of the S280-6-282, I've ordered a pair of AS190-4-252 -








Specs above are for the 9 ohm version- I'm getting the new 4 ohm ones, spec curves not released yet.

Note that though this driver has a reasonably high sensitivity, at 89 dB for 2.83VRMS, it also has an Xmax of 9mm, so the available travel and linearity for mid bass/low midrange applications should be quite good, even at fairly high SPL.


And I've completed a "proof of concept" modeling using a combination of measured data for the C173-6-090 midrange and JDT-1024 tweeter, and modeled data for the AS190-4-251 and NS12-513-4A Aurasound woofers as used in the Isiris. Driver time delays are based on values measured for the Isiris project with correlation checks against Fuzzmeasure. EVERYTHING will be re-measured and re-checked before a single crossover part is ordered.

Note, the shown tweeter response is the 30 degree off axis response of the JDT-1024- it's a bit hotter above 3 kHz, and goes out way further directly on axis. But 30 degrees response is fairly representative of the power response. With the capabilities of VituixCAD, I will be doing 10 degree incremental measurements with the AudioPrecision setup for all drivers, out to at least 60 degrees.




This network has substantial similarities but also some significant differences from the original Isiris boards on the midrange and tweeter, and new builds will be needed... there goes my lunch money and margarita budget for the next few months! :B

MDF knock down test enclosures have been ordered, too.

And to really keep me in the poor house, Steve Manning found the a US source for the Delignit phenolic beech ply, let's call it wooden based unobtanium, just to be clear, which I'll try for the baffle panels when I build the nice mid bass enclosure, and likely if I can ever get some C168 and do a complete rebuild of the top end. We'll see... can get it in boards 59" long, either 9-1/4 or 18-1/2" wide. If you have to ask how much it costs, it's too much. It's usually used for piano tuning boards, where the tuning pins go in. Seriously hard and tough.

[JonMarsh]

Weird! I'm at Starbucks and what should be on the playlist but "Corner Pocket", as played by Count Basie...

Now, I'm more a fan of the Harry James version, but still, it's cool to hear a bit of big band in the morning. Better than the smell of napalm in the morning, for certain...

I'm going to see if I can man up tomorrow afternoon and do some measurements- get out the turntable and the new gear, and have a go at at. This afternoon is a memorial service for one of our neighbors, an old gal that passed away recently after a couple of strokes at 90. Just across the street and one house to the right... GF has known her for decades.

If I have time after the chores and other events, I may look at a Minerva style crossover for this project, too.

[wkhanna]

And to really keep me in the poor house, Steve Manning found the a US source for the Delignit phenolic beech ply, let's call it wooden based unobtanium, just to be clear, which I'll try for the baffle panels when I build the nice mid bass enclosure, and likely if I can ever get some C168 and do a complete rebuild of the top end. We'll see... can get it in boards 59" long, either 9-1/4 or 18-1/2" wide. If you have to ask how much it costs, it's too much. It's usually used for piano tuning boards, where the tuning pins go in. Seriously hard and tough.

wonder if this stuff is similar to what Wilson uses?

[JonMarsh]

Wilson uses various grades of phenolic sheet. Good stuff, heavy, hell on tools.

This stuff I'm talking about is phenolic impregnated thin layer beech ply- it's also known as "panzerholz", and a primary use for it is armoring conventional cars- it apparently does a bang up job of absorbing small arms fire while being no where near as heavy as metal armor plate. Much easier to work with. All comes from the same company in Germany, but it's easier to buy stuff for pianos than for armoring up cars!

[dar47]

Ah, piano source well done. Hate to admit this but my mother in-law who passed a few years back parked her old cheap Canadian 70's studio piano in our sun room and I couldn't give it away so I dismantled as much as I could to recycle and was left with the problem of removing the pin block from the iron. ( not easy) The iron could be recycled so the only way to do it was this, :B



The iron is a 2 man lift never mind with the pin block. Baffles with this stuff are dense.

[JonMarsh]

Doing the Minerva treatment for the Osiris

Yeah, I would think they would be dense... kind of like my head!

Well, got some chores and the memorial service out of the way, spent a little time speculating on a slightly modified version of the crossover using elliptic elements to get LR6 transitions for the midrange and tweeter- doesn't look bad, might have some payoffs in reducing out of band stress on the driver, for not much more in components. This is what I HAD to do in the Minerva to get the distortion looking they way I wanted...





Note, the total impedance looks kind of wonky, but that's because VituixCAD seems to put the active crossover driver impedance right in parallel with the rest- haven't been able to figure out how to disable that so as to only see the passive crossover stuff.

[JonMarsh]

Actually, there are some impedance issues; mid bass network is already tweaked some to optimize Z better; this is always an issue with 4 ohm drivers. However, the "active" network is being connected in parallel. Need to contact him about this...

[5th element]

Well using that Cell driver is certainly going in a completely different direction. I rather liked the appeal of the S280 and its 93dB @ 8 ohm sensitivity.

[JonMarsh]

I did too, on some levels, but considering this as a collaborative project with Mike, and the idea that my mod is a "Jr." Version of his, then using the smaller cell mid bass makes some sense. (He's ordered the big ones for us to test). "Only" 89 dB sensitivity, but very flat, seems clean, and oh, saves me about $800. That's a lot of crossover parts...

Just have to see how it all pans out...

Got more chores and cleaning tomorrow, don't think I'll get to measurements until next weekend. Might fire up some stuff just to get a notion of how the AP works on acoustics stuff.

GF's in a car show tomorrow, EARLY, with her 356, so our whole schedule is sort of thrown out the window for a Sunday...

[Alaric]

GF's in a car show tomorrow, EARLY, with her 356, so our whole schedule is sort of thrown out the window for a Sunday...

Having a GF whose vintage Porsche is in a car show excuses a multitude of sins , Jon. And ups your cool factor by orders of magnitude. LOL

[JonMarsh]

Did I mention she has a pilot's license, too? :W

Well, after all the issues this last month, my cool factor could use some upping! :B

[Alaric]

And she's a pilot? And apparently is tolerant of your sparks and wires. You are a lucky man , Jon , March of 2016 notwithstanding.

[TEK]

Respect man! ;x(

[JonMarsh]

I certainly respect her, you can believe that! :T

[JonMarsh]

And she's a pilot? And apparently is tolerant of your sparks and wires. You are a lucky man , Jon , March of 2016 notwithstanding.

OK, she took her 356, and our next door neighbor (a retired helicopter pilot) brought his new (to him) Amphi-Car.

Don't know what an Amphicar is? I remember this from way back as a kid... read about it in something like Popular Science...

https://en.wikipedia.org/wiki/Amphicar

Here they are from opposite sides:




Yes, that's an anchor in front of it...

[JonMarsh]

Well using that Cell driver is certainly going in a completely different direction. I rather liked the appeal of the S280 and its 93dB @ 8 ohm sensitivity.

Some more thoughts or info to share-

This is something I have been somewhat agonizing about - which way to go.

A few things persuaded me in the end to try the AS190-4-252 for now-

• One, while I was in Europe in March, Mike (MEB46) was there, too, and he visited Accuton/GTheil and Partners in Hamburg. They were quite gracious and forthcoming- a big part of that forthcoming was that the new AS drivers series was developed in repsonse to customer input, and represented a key change in direction and design - particularly with regards to achieving an acoustic origin in the same plane as the baffle, and optimizing the radiating area, transient response, and distortion for the size of driver.
• The S280-6-282 would be in production for a number of more years, to fulfill cusotmer needs, but that type of driver design approach would not be used in future products; it won't be receiving any more "love" or attention.
• Just for the sake of argument, if Mike's new system will be the Osiris MkII, and my update will be the Osiris Jr., then there is a certain logic in keeping simlar technology approaches for consistent family characteristics, and based on the known peformance of the AS190-9-251 and T/S parameters of the AS190-4-252, that seemed like a reasonable path to try out. Distortion seems to look very good in the working pass band I'm concerned with.

Will I have regrets or wish I'd chosen differently? Could well be... but that choice can always be reversed in the future. For now, I'm trying to make a forward looking choice based on inputs from others, which is NOT something I do well- usually I'm of the school of "in God We Trust, All others we VERIFY!".

Ultimately I expect to extend this redesign to incorporate the C168, too, but indications are from Madison that it may not be available until much later this year to peons like us.

[JonMarsh]

Practice Makes Perfect?

OK, if practice makes perfect, once I have "real" data, I should be all set for this crossover, 'cause I'm getting lots of practice in right now!


I promise not to post any more intermediate crossover design steps this week.... this is just another pass keeping a few things in mind and looking closely at interactions:

• LR6 equivalent acoutic slopes on the midrange and tweeter, and upper end of the midbass
• New conjugate networks on tweeter, midrange, and mid bass.
• Careful optimization of crossover values relative to net impedance- it's easy to make an 8 ohm driver work with what is really a 6 ohm crossover, but then the net impedance seen by the amplifier is more difficulit. Here, I'm just trying my best to get a 3 ohm worst case load from a 4 ohm driver, in the area where phase angle is low- shouldn't be too hard to drive
• Some of this is just getting comfortable with VituixCAD- because it allows you to see everything at once, it helps avoid getting tunnel vision. I like that.

[sdl2112]

I am interested in the details of the AS190. Like how they do the surround.

I'm not that familiar with vintage Porsche's but your girlfriends look very nice. :T

[JonMarsh]

I'll do as much as I can with pics and what not- they came up with several special construction techniques to maximize the diaphragm Sd and deliver fairly long travel- by a mid bass definition- 9 mm. Have to see just how that works out. It's a bit of a long shot, maybe, but then so were the C173-6-090 back when I bought them!

They're scheduled to deliver on Wednesday, so I don't have too long to wait...


Her 356 is very clean, runs well considering, but it is a money pit. You get all kinds of estimates about what it's worth... Some people say these are in a fair amount of demand. But the numbers I hear bandied about are kind of insane... But then it's more a museum artifact than a car to drive, IMO. Someday I suppose my NSX will be like that...