Saint-Saëns planar- AKA SMJ-40- time for a project now?

[JonMarsh]

I'm used to having a lot of unexpected distractions, but the last 24 hours have been exceptional! Think in terms of having 4-6 hours stolen from one's daily routine...


However, I did clean out the local Ace Hardware on foam brushes (that wasn't a big effort, just two bags worth!), and the first coat of Arm-R-Seal clear coat did go on this morning- hopefully my technique will be adequate... I usually start with glossy, and finish with semi-gloss or satin. works well in combination.

Can't tell much from this picture, though it does shook shinier in some places!




Now, remember, we're sort of going for a sort of "rustic Idaho" look here, not a "glossy high end slick European" look!


🤣

[JonMarsh]

And another key tool arrived yesterday, actually more than a week earlier than promised- Kudo's to Woodpeck! First production run item...




The minions were besides themselves with glee!

[technodanvan]

Those are really nice to have Jon, I have a set of them from a few years ago.

[JonMarsh]

So, so true! this will be used for radiuses on the MF-HF phenolic input boards

[JonMarsh]

Those are really nice to have Jon, I have a set of them from a few years ago.

BTW, I am starting to get back into the full swing of things since that side track to Seattle and it's complications; so I expect to be doing the tweeter driver measurements soon and getting a sample of the denim insulation out to you. Sorry that my working motto of "slow work takes time" kicked in seriously this last month!

[technodanvan]

No worries (and no hurry!) at all Jon, I have also been busy with other things that have kept me away.

[JonMarsh]

Working on finishing steps for the other MF-LF cabinet now, and the first one has been moved into the assembly "den" for driver installation and crossover testing. There has been a metric ton of distractions and honey-do's this week, including more plants for the backyard and nails in a tire at the edge near the sidewall- going to go ahead and upgrade my whole older set of Michelin's and get this one replaced on the newer set of wheels, pending resolutions of the fact that THAT Michelin in no longer in production, and it's not clear if the replacement one is fully compatible- rules about tire sets and matching can be a necessary PITA!

[augerpro]

Dayton introduced a new AMT: https://www.parts-express.com/Dayton...096?quantity=1

[JonMarsh]

this one looks like it might be usable with a little effort, depending on what the distortion versus frequency and level behavior is like. In general, I find the Dayton, Mundorf, and Sb Acoustics parts rather disappointing. I’ll have to look at it in more detail as soon as I have a bit of “extra” time.

[JonMarsh]

It’s almost as tall as the TPL75, but does not go as low- about 2500Hz corner frequency versus 1500Hz. A lot would depend on the distortion versus level. But it’s much better at least from 3K to 20K than any other Dayton model I’ve seen

[augerpro]

Yeah we really need a reasonably priced AMT. I recently bought an AST2560 for a dipole project and wanted to use the newer v2 (I have v1 and really like them). I don't know if I got a defective one or if there has been another revision but now the low end is completely unusable. I had to send it back to PE. There is nothing under $200 that really impresses me now.

[JonMarsh]

Yeah we really need a reasonably priced AMT... There is nothing under $200 that really impresses me now.

So, so true! If I pushed the upper crossover on the PT5010 higher than I'd care for (3kHz) one might be able to get this new one to work OK, but a 3kHz crossover leaves a rather small CTC.

[Efalegalo]

Yeah we really need a reasonably priced AMT. I recently bought an AST2560 for a dipole project and wanted to use the newer v2 (I have v1 and really like them). I don't know if I got a defective one or if there has been another revision but now the low end is completely unusable. I had to send it back to PE. There is nothing under $200 that really impresses me now.

Yeah - I had a whole thread about this. There was a revision made to the AST2560, but the spec sheet was never updated. This driver was used in The Bordeaux - and the crossover had to be reworked.

[theSven]

I remember reading that thread! It was good you found a working crossover with the driver change.

[augerpro]

Yeah - I had a whole thread about this. There was a revision made to the AST2560, but the spec sheet was never updated. This driver was used in The Bordeaux - and the crossover had to be reworked.

IF it had the same response as yours I'd be fine. But it appears there has been another revision. Or I got a defective unit. Now there is a big spike at 1.5khz then drops straights down.

[JonMarsh]

IF it had the same response as yours I'd be fine. But it appears there has been another revision. Or I got a defective unit. Now there is a big spike at 1.5khz then drops straights down.

Ugly... but typical from what I hear.

Tons of distractions lately, but finding some time to make progress, finish sanding on the 2nd MF-HF module is nearly done.





And just to show my cards a little bit, and give credit where credit is due, my earliest influence and thought provoker for this project:








Courtesy of Bruno Putzeys.

You remember him, right? Hypex? nCore? Kii Audio? PuriFi Audio?

You can see a bit of the lineage with regards to the SPK16 write up at PuriFi Audio. And for those who prefer more conventional, perhaps less "risky" drivers than planar magnetics and AMT, I have an alternative top module in mind that should be possible to build at an attractive price, and using the more restricted range of the current Saint-Saëns concept for mid-high. In fact, I'd call it down right "budget"! Well, except for the cabinet and crossover...

This Kii configuration is not the ideal choice, IMO, as the lower cabinet is basically a subwoofer in this application, for the upper cabinet which is also sold as a standalone speaker, but it was the clear inspiration for me based on other experiments to reduce floor bounce, increase sensitivity, and of course increase SPL capability.

As a former Danish colleague of mine and I would frequently say, "Steal from the best and ignore the rest".

And if you look closely, you will notice that the tweeter waveguide is a DXT style diffraction waveguide.

I also think there's a high likelihood that the woofers in the LF cabinet are of the same Peerless family, 6-1/2" version, tapped for the PS Audio project. That would make them the Peerless SBS-160F35AL01-04





You can get these from PE, at a nice price for their commercial accounts, if you're willing to buy 500 at a pop.

[draki]

Kii bass units available ATM in singular quantity at TLHP (they were until recently 500 pcs MOQ)

https://en.toutlehautparleur.com/brands-page/tymphany-peerless-vifa/grave-medium/speaker-peerless-sbs-160f35al01-04-4-ohm-6-61-inch.html

The mids are also available again (they were off the list for the past two years):

https://en.toutlehautparleur.com/bra...-5-9-inch.html

[JonMarsh]

Cool! THLP is my favorite European distributor! Thanks!

from the info posted, I suspect these are buyouts of some kind, and when they're gone, they're gone. Can't complain about the prices.


On more mundane on topic tasks, I'm half way through the finishing steps for the 2nd MF-HF cabinet.





A lot of other stuff going on tomorrow, but with luck I'll finish it up.

[JonMarsh]

Still in the "Slow Work Takes Time" mode, with all the different things going on, not audio related!

Plus some new concepts and requests waiting in the wings, and updating the Ardent D CAD model for different drivers for a member...

Here's the 2nd "Idaho Rustic" finish cabinet, with the initial urethane clear coat layer applied and dried, along with mandatory cabinet reflections...




Next/final coat will be semi-gloss/satin.


Slow work takes time, but I'm really itching to get the basic input panel (without the rear tweeter) setup and check how it the drivers measure in this enclosure setup, and with the initial POC crossovers. If I'm really lucky, and the Merry Minions have done their work, there won't be significant changes required.





BTW, the Minions got together and voted on their favorite random orbital sander setup, and we have a bit of a dark horse surprise, a model bought at discount when this company produced their first ROS... and as it turned out, the adapters and hoses available made it a no brainer to use with an older Dyson canister vacuum for dust collection.

Mirka Abranet and the new Diablo sanding disks were in a dead heat... but sometimes you can find the Diablo at a local hardware store.

Who'd have thought...

[JonMarsh]

The Chancellor "requested" that I do some minion work for ET this last week- it was interesting mainly to the degree that it confirmed my past router preferences for driver rebates and holes. I report this just as a "Tool Time" reference, there are many possible solutions, and I have another Bosch router setup for this at the moment, but as ET would say, "It is not as satisfactory".


For me, the undisputed champion regarding power and smoothness versus weight and size, and integral dust collection:


Porter Cable 8902






Here, the result of routing for a test cabinet for the FSL-0512R01-08...






The center plug was cut out using a DeWalt DWPS611 with a template follower bit:





No surprises for the test fit..

[JonMarsh]

Font size check- 14 points should now be the default.

[theSven]

Font size check is looking good...

[Evil Twin]

Yes- The appearance is more satisfactory- I needed to check on a large screen visually, as the smaller screen on my Imperial shuttle can be deceiving...

[tktran]

Kii has released the 1/4 cu foot model- appropriately named 7 (litres). Looks to use the same high excursion 6.5” midwoofers.

https://www.kiiseven.com

[JonMarsh]

Well, that's pretty cool, took out the drivers and electronics for the cardioid LF mod, and just did a more straight forward three way, a bit like what I'm doing in my Modula Extreme Kii inspired design topic. Other than I'm going passive, and some weird woofer and passive EQ tricks.

Thanks for posting the link!

[JonMarsh]

OK, it's time for a little something for Technodanvan....

Yep, you guessed it, it's finally B&G Deep Cup time!

I thought this rear cup design is interesting, though it is paired with a more conventional "front end" lacking the PDR features. Let's see how it does, starting off with impedance plot.





that's a pretty good sized impedance bump at Fs, I've usually only seen this with open back 2522 parts, and lower down, around 600Hz. Qts is up there a bit, as you'd expect.

Here's the polar behavior at 1M, 0 to 50 degrees, 10 degree increments. Note, -10 dB on this graph is about 93 dB absolute. My reaction to the low end and Qts, based on personal philosophy and 4 experiments with modifying the GRS PT2522 parts, is that the enclosure should be smaller, pushing the Fs up, which generally will lift the area above the Fs drop off. Damping is critical, and I found Denim damping material to work best in my experiments.




The best looking plot seems to be for around 20 degree, so I'm showing that here with distortion, which is admirably low.





For comparison, this is my MOD4 PT2522 with custom rear enclosure made from phenolic sheet:





And this is a polar set 0-40 degrees. Note, I suspect that the glitchy stuff in the 7 kHz area may be a by product of the PDR stuff to improve top octaves dispersion. Just a guess- it can be tamed a bit with a notch filter.

[technodanvan]

Interesting stuff Jon! I wonder what they jammed in there. Or could it be hollow?

[JonMarsh]

Interesting stuff Jon! I wonder what they jammed in there. Or could it be hollow?

Only one way to find out, but it’s your part, not mine! I am donating the GRS mounting plates to you, one tweeter mounted already, the other kit in the box.

[JonMarsh]

One more data point to throw into the discussion- this is data for a sample of the MOD3 build, which used a larger rear volume probably not dissimilar to that used in the Deep Cup BG part.

Here is the impedance curve:




Note that the Fb is about the same- around 900 Hz.


Here's the SPL polar plot to 40 degrees:





Note there is a dip in the overall response above resonance, but in this case it is near 2kHz, instead of ~1600Hz, suggesting differences in the rear volume or diaphragm mass. Also, the magnitude of the minima is somewhat higher, only dropping about 4dB, instead of 7. Given the difference in the Fb peak, I'm inclined to speculate that the damping material used in the Deep Cup BG is not as effective as the ATS Denim material I used in my MOD builds. Interesting also that the spread in the 0-40 degree levels for the MOD3 over 3-8 kHz is less than 2dB, whereas the Deep Cup BG part is 3-4 dB. The spread on 10kHz to 15kHz data for 0-40 degrees is even higher.

Now, this version, MOD3, is also quite usable for the intended 2400Hz crossover, if one is willing to compensate for the rising HF response, and it is less "glitchy" in the 5-6kHz area, probably due to the deeper back, about 1-1/2". It has a relatively tight spread in the 3-10kHz area from 0-40 degrees off axis.

Note, these were measured on the same test enclosure, same mounting, same turntable and setup.

Danny is due to come by tomorrow, I think I should give him a pair of the MOD3 and MOD4 PT2522 parts to try out!

[JonMarsh]

One more thing I should probably address- just what is a PT2522 MOD4?

Basically, it's a PT2522c with the back cup removed using a utility knife, to cut through the glue. Then, a rear sub enclosure is built using 1/2" tall Leecraft 1/4" thick phenolic, glued up with JBWeld. Several configurations and volume of rear enclosure were evaluated, 4 in all, and this one proved the best... sort of zeroing in what seems like the best achievable result.




ATS Denim damping material is cut so that the layers are edge on to the front and back.

Then the original PT2522c back is glued back on, using JBWeld again.





Last step, mount in a GRS 2522 mounting plate, sealing the edges with premium GE clear silicone or with Weicom 310 adhesive.




Eazy Peazy!

[JonMarsh]

Had a great 3 hour chat with Danny when he swung by, and the previous posts got me thinking more about that MOD3 version of the PT2522, with 1-1/2" side walls...

So, Danny, and anyone else interested in a 2522 based top end, this one's for you!


Here's the updated crossover, using the PT2522 MOD 3 data- some comments up front are in order:

• The driver data for the tweeter is an RMS averaged composite of horizontal dispersion from +30 degrees to - 30 degrees, at 10 degree increments, generated by Fuzzmeasure post processing.
• Due to the leading phase characteristics of the rising response in the PT5010 in 7.5L enclosure above 2kHz, the PT5010 is modeled stepped back about 18 mm for accurate summing in the crossover region with the tweeter relative to the target crossover functions.
• Note carefully the DCR values for inductors- they impact the crossover behavior, and sometimes are selected for relatively high values, due to damping in the corner transition.
• The midrange crossover shown here is my "junk box build" and of course, capacitors can be consolidated.
• High quality caps are recommended for the tweeter and composite combinations for the midrange, presuming you have a good quality front end. Of course, there's no reason you couldn't go budget and verify the acoustic levels, then experiment with selective upgrades. I would recommend at least Audyn Plus caps or Clarity CSA for the midrange, and if you're feeling flush, Clarity PUR for the HF.
• Measured driver data is used here, but no PR contribution on woofer.

POC Schematic










Reverse null check






I have a test cabinet I can try the mid + tweeter combo out in, and will do that "hopefully" some time soon...

[PANDINUS]

Impedance graph please.

[Evil Twin]

Your request is fulfilled...

This project has unfortunately been put on hold in order to prioritize the CC development for Sven. Not stopped, just think of it as suspended in carbonite....

[a4eaudio]

One more thing I should probably address- just what is a PT2522 MOD4?

Basically, it's a PT2522c with the back cup removed using a utility knife, to cut through the glue. Then, a rear sub enclosure is built using 1/2" tall Leecraft 1/4" thick phenolic, glued up with JBWeld. Several configurations and volume of rear enclosure were evaluated, 4 in all, and this one proved the best... sort of zeroing in what seems like the best achievable result.
...
Then the original PT2522c back is glued back on, using JBWeld again.

If anyone is interested and doesn't want to cut phenolic sheets up and epoxy it together with the original back cup, the attached zip file has a *.stl file for a 3D printable back cup basically replicating what you did. You can edit the original by selecting the blue shaded "rim" and "pulling" to make the cup deeper. (Note, the object saves as a millimeter and has to be scaled up 2540x)

[Evil Twin]

The configuration donated to Technodanvan is 1.5" deep. This is the version modeled above with the recently posted crossover and predicted system response based on a revised crossover.

Another pair of PT2522c with a 1.5" deep back cup is in fabrication today.

[Evil Twin]

The final step in assembly is gluing on the original back plate, with its damping material intact.











This pair is nearly identical in construction to the pair provided to Technodanvan, and will be used for verifying the POC crossover concept shown above.

[Evil Twin]

As some are want to say, "Slow work takes time..."

But the results can be "interesting... most interesting..."

The assembly of the slightly updated MOD3 2522c version was completed, and basic checks and testing...

The interior dimension was slightly modified (longer) and assembly processes and adhesives for the mounting plate were updated, additionally.

The result is not markedly different, in some regards, though in others there is a certain appeal in the differences... as we will see.


The assembled 2522c MODules.





The measured impedance curve, essentially identical for both samples.





As previously, the fundamental resonance at approximately 950Hz.



The zero axis curve is better damped than the factory construction for the PT2522c, but still shows the issues in the 6-7 kHz area that are likely arising due to reflections (and are correlated with a very slight bump in the impedance curve), as well as a peak around 13.5 kHz which may also be correlated with the impedance curve.






Things start to become more interesting when investigating the polar response - here, a plot in 10 degree increments out to 50 degrees off axis.





Keep in mind that one is mainly interested in the response pattern from 2kHz and upwards, given that a typical crossover point may be about 2500Hz.

This plot raises questions in one's mind about how much of the observed behavior is due to some type of diffraction, reflected at the surface of the diaphragm, and reducing in impact as one goes further off axis.


Examining this trace at 40 degrees off axis is interesting... the behavior is generally most satisfactory.





How many HF units can one find that show this exemplary dispersion at 40 degrees off axis? Or even 30 degrees? Without a waveguide? and at what price?

And how many "mature" Jedi or Sith can demonstrate good hearing capability beyond 15kHz with instrumentation?


Distortion performance is reasonable, though not exemplary...




Similar to many conventional tweeters, HD2 is significantly higher than HD3.


As a reminder, the other candidate for this design, the Beyma TPL75 has somewhat different characteristics, and a much higher price tag.



Smooth on axis, even with this basic test fixture lacking substantive efforts with diffraction control, and generally very low distortion at the standard test level:





As a reminder, a POC crossover set and basic measurements of the PT5010 and TPL75 in a relatively crude test cabinet (compared to the ones now built) shows very reasonable potential...

[JonMarsh]

With Thanksgiving coming up tomorrow, I just wanted to wish everyone a great T-Day with family or friends, and from my minions to your minions!





I'll really be giving Thanks when I can get another pressing project settled from my end, and get back to this one!

[theSven]

With Thanksgiving coming up tomorrow, I just wanted to wish everyone a great T-Day with family or friends, and from my minions to your minions!





I'll really be giving Thanks when I can get another pressing project settled from my end, and get back to this one!

Happy TG to you too, and I'm looking forward to this project picking back up sometime next year... Eventually this will get done, and I will have a pair of these. Still no idea where they will live though, but that's a minor detail 🤣.

[letiennam]

Happy Thanksgiving to you too, and I'm looking forward to this project picking back up sometime next year... and I will have a pair of these.

[JonMarsh]

Santa Claus, AKA Fed Ex, delivered a key tooling accessory today for the machining of the side wall panels for the LF module...

No reindeer...




But a necessary accessory for one of my Bosch routers...

[theSven]

Lets see this new toy mounted to the router :-)

[JonMarsh]

Slow work takes time- a lot of tasks ahead of that one in the queue, especially this time of year!

[JonMarsh]

well, now that I've got my part of the CC mostly completed, I'm back to the Saint-Saëns... with some immediate tasks in progress...

• Develop the LF crossover POC layout and begin construction
• get the single driver woofer + PR out of storage and prepare for testing with the LF POC crossover
• Prepare the cabling for a temporary midrange and tweeter crossover install in one of the built MF-HF cabinets, to see how it works all put together (no acoustic tests in this cabinet yet, could just test the drivers, but it shouldn't be TOO much different than from my MID cabinet tests- cross fingers)
• Get out the new Woodpeck corner routing fixture and check off that task for the input panel for the MF-HF cabinet
• Mount PT5010 and TPL75 and do crossover checks and tweaks as needed
• Etc....
• Etc....

This is the new LF board under construction, will fit in an allotted/planned space in the LF cabinet base.





There's been a lot of secondary stuff going on, too, finishing the "rescue rebuild and setup" for my other 2013 Mac Pro, it's now up to date and running the newest version of MacOS supported on it (for Xenon 6 core E5 processor and AMD D700 GPUs). And some new stuff installed- diving into KiCAD on both this system and my MacStudio (comparing performance, to recommend it for Intel setups)

KiCad EDA

A Cross Platform and Open Source Electronics Design Automation Suite

www.kicad.org

I've been wanting to look at a stable release version of this for some time, see how useful it might be- runs on Windows, MacOS, and Linus. Donation ware, and from what I'm seeing, a well crafted mature product, though this is V7, V8 is in ongoing development.

a sample screen shot:




And on my system...



Slow work takes time, but I just started this AM, and am about half way through a schematic for the full Saint-Saëns crossover.

[JonMarsh]

Making progress on POC crossover assembly and schematics, and did another modeling check this morning.

Updated POC schematic- in KiCAD. the more I use this tool, the more I like it- more flexible than Altium, as you can customize symbols right on the schematic, adding fields and tweaking basic configuration. Don't have to go into symbol editor.

This is a PROOF OF CONCEPT schematic, that is, developed from individual driver tests in POC cabinets, PRIOR to mounting and testing in the full blown cabinets (coming soon). It's subject to change, but the test results so far have been very promising, including a combined test of midrange and tweeter, so I'm not expecting big changes. This will convey an idea of the complexity and potential cost; of course the latter varies depending on the components you chose to use. These values are all commercially available, though some are less common. Midrange and tweeter test crossovers have been built and undergone initial testing. The woofer crossover is in assembly and hopefully will be finished today. The big caps for the Fb LCR zobel network arrived yesterday. 200uF Solen.

NOTE THAT THE TWEETER CROSSOVER includes an input LCR shunt network whose purpose is to flatten the overall input impedance seen by the driving amplifier, and reduce input current phase perturbations. This can be located anywhere in the system. It's small, and compensates for input impedance in the tweeter network.





This is the modeled on axis response in VituixCAD for the measured driver data and crossover network, not including PR output.




This is the modeled Group Delay and phase tracking for each crossover section. Based on what I measured for the midrange + tweeter test, this seems accurate, and of course, inverting the phase on the midrange produces deep symmetrical notches for both crossover points.




I think this is the part, where after all this work, I can say, "Umm, it's about time to have some fun now!!"

[theSven]

Great to see progress progressing on this build again. We definitely need to play some Saint-Saëns conducted by Seiji Ozawa on these speakers once they are ready. Sad he passed on yesterday and his style will be missed.

[technodanvan]

600 uF per low frequency section, that's gonna take up some space! I have a handful of the Solen 200 caps for another project so I know their size well.

[JonMarsh]

The LF crossover is laid out to fit on the bottom of the LF enclosure between the side walls...

I was joking with Steve today, that I must not have the magic Chris Brunhaver touch to come up with crossovers as small an inexpensive looking as the FR30 (I have pictures). But on the other hand, as large reflex speakers go, these should have a pretty benign LF impedance, including relatively small current phase issues...








If you don't believe me, model this configuration and crossover WITHOUT the LCR zobel... why, the input impedance looks as bad as some Magico's and other well known "high end" brands!

Updated to clarify impact on input impedance.

[JonMarsh]

The LF crossover assembly is proceeding smoothly- I do have to note that the "right" way to do this, if one was doing any quantity beyond 2 or 3, is to layout a single layer PCB with wide traces and 3 oz copper (the heaviest normally available as a standard process) and cut the assembly labor by about 90%.

Of course, that assembly labor is where a big hunk of the "sweat equity" in DIY comes from, right?

And using KiCAD as a documentation tool might substantially facilitate that effort- since I can run it on almost any of my Macs, whereas my Altium Designer install is "frozen" on the WIN10 Bootcamp partition on one of my older Mac Pro's.

[JonMarsh]

OK, as my wife might say, it's been a heavy Monday, a lot to do on a variety of fronts, but having gotten started early (before 5) I can now relax after 1 once this post is done!

The construction of the POC build LF crossover is finished, turning out as planned. Note, this one was also assembled mostly using Cardas Trieutectic solder, which is lead free. It requires a higher temp setting on a temperature controlled iron, and a somewhat different technique; it doesn't flow the way lead-tin rosin core solder does, but nice shiny strong joints are possible, stronger than with solder using lead. What it doesn't do well in my experience so far is flow well into stranded wire, so I cheated and used Quad Eutectic for those conductors.




Here's the test location, the easiest placement available quickly to get a plane boundary of sort (fairly heavy doors). testing at standard levels (2.83VRMS) I don't figure to be rattling anything- a single driver has the same impedance as the 4 driver array, but 6 dB less sensitivity. This is just a sanity check to be sure things are looking, shall we say, vaguely in the ballpark!





SPL + Distortion plot




As expected, baffle step compensation looks a bit heavy, due to the somewhat wider baffle width. No changes until we have a big boy cabinet built and ready to test. HD2 is a bit higher than HD3, but level on both is pretty decent, though not up to PuriFi standards. Neither is the pricing or necessary enclosure size!


Just for fun, recorded the impulse step behavior-




I also measured the input impedance at the crossover, though this is just the LF part. Given how well this correlates with the design simulations with 4 drivers, gotta say VituixCAD works pretty nicely...





DATS has some issues doing current phase through a crossover, but the impedance amplitude looks as expected.


Next steps are getting drivers loaded into one of the MF-HF cabinets, and checking out the POC build crossover with that configuration.


Slow work takes time... does it ever!