2021-2023 Isiris Update
Collapse
X
-
-
This is what happens when “style” is emphasized over safety or utility. Sure you never see any other examples of that kind of behavior in modern life…DFAL
Dark Force Acoustic Labs
A wholly owned subsidiary of Palpatine Heavy Industries- Bottom
Comment
-
Interesting, if you are developing a 3rd order for this top end with the TW29B-B-4 maybe it would also be a nice two-way option with a PR in your "No one needs another 2-way".
Was looking at Treols G. Purify BE design, he using a port with a simple 2nd order network. I think your magic stuff might be better.- Bottom
Comment
-
Good points- note that ET has been considering a number of possible configurations for that PuriFi based two way system... this includes...- a tweeter like the SB29BAC, which is very, very similar to the TW29B-B-4, but with less flashy cosmetics and better dome protection. And slightly lower cost, not being a "Satori" model.
- The venerable but hard to source DA25TX008, which we still argue is one of the finest dome tweeters available if implemented with the right crossover- we have a small stock on hand, and more coming from a very remote sector... you'd have to go to Antarctica for it to be any more remote.
- And the TPL75, discussed in his post, which was thought to be a "most interesting" alternative as well as a bit off the beaten track for DIY, so to speak... and besides, he wanted to hear it in at least a POC test build...
- There are even some other stranger possibilities being discussed, like the AST25120, pending some measurements and packaging experiments.
- And last, but not necessarily least, the modified/special back cup design for the GRS PT2522. Needless to say, it's intriguing to come up with the best solution for the minimum dollars, as well as the best overall solution. This applies to the woofer as well... the other logical candidate is the NAA driver being evaluated for the Isiris update, which is more expensive.
the AudioWorx
Natalie P
M8ta
Modula Neo DCC
Modula MT XE
Modula Xtreme
Isiris
Wavecor Ardent
SMJ
Minerva Monitor
Calliope
Ardent D
In Development...
Isiris Mk II updates- in final test stage!
Obi-Wan
Saint-Saëns Symphonique/AKA SMJ-40
Modula PWB
Calliope CC Supreme
Natalie P Ultra
Natalie P Supreme
Janus BP1 Sub
Resistance is not futile, it is Volts divided by Amperes...
Just ask Mr. Ohm....- Bottom
Comment
-
This measurement set is NULL AND VOID!
A configuration problem with the test mixer because of poorly documented monitoring setup has been identified, which resulted in feedback causing ripples in the response curve appearing like time domain issues with the signal measurement.
This is report back on checking out the new crossover boards built by Jonmarsh- LF, Md, and HF. Design changes served a number of purposes... the Chancellor believed that a second set of eyes and hands might bring some "productivity" and objectivity to this long standing effort...- The updated designs reflect an effort to reduce size and expense, and adjust the sensitivity to accommodate an alternate driver for the midrange, the PTT6.5X04-NAA.
- There was an overall topological update to the midrange crossover, eliminating one of the series caps, and adjusting electrical parameters to hit the target acoustic LR3 response. This has been applied both to the NAA woofer and the 6.5" Purify Midrange driver- parts for the latter just arrived today. Testing today was for the configuration with the NAA woofer used as midrange, in consideration of the claimed low distortion for this aluminum cone driver. The location and configuration of the two notch filters has been adjusted.
- The LF crossover uses different coils, Jantzen AWG15, with higher DCR than the large AWG10 Solen coils, which lowers the sensitivity in the 60-500 Hz range by about 1.5 dB, but has little impact in the low bass, due to the rising impedance peaking at about 27Hz. The crossover is much smaller and less expensive as a result.
- In todays, test, the "PURfection" HF crossover was used, and initially tested with the original component values, though system simulation in VituixCAD has already shown the need for a couple of tweaks.
- The "High Value" HF crossover with minimum component count as been simulated and all the parts recently received; system testing will occur as soon as a POC board can be constructed.
- One of the purposes for the crossover board size reduction besides lowering cost (for several interested HTG members, including one with a built set of cabinets!) is facilitating locating the crossover boards in the speaker cabinets with output cables at the top of the woofer cabinet connecting to the midrange/tweeter module. This will be attempted and reported on while updating the 2nd woofer cabinet.
A picture of the crossover boars lined up together, HF at the top, LF at the bottom:
Due to the size of the cabinets and CTC distance, measurements today were done at 1-1/2 meters, and at approximately 90 dB output level (checked independently from Fuzzmeasure. Getting these measurements with reasonable low frequency data highlights the challenges with room reflections and acoustic issues like nearby doors leading into other rooms which act at times as out of phase resonators. This highlights the desirability of careful room setup and placement. The resulting comb filtering from early reflections is quite visible in these measurements, compared with near field measurements of the drivers without or with crossovers. However, I can say that what I hear when listening to music from this single cabinet is now reaching my expectations, in spite of the room limitations.
A reminder- the measured response in cabinet of the PTT6.5X04-NAA- cabinet well out into the room, and under 1M measurement distance:
The main breakup modes are at 5kHz and 10kHz, and can be surpressed with notch filters (even as described in PuriFi application note) at 5kHz (series, last circuit to support distortion reduction) and a parallel notch filter for the 10kHz mode. In this system, it's implemented as a midrange, so additional components are needed.
This is the POC crossover initially developed in VituixCAD, component values were combined or updated based on available values:
For example, C12 and C27 were combined into one part, due to the availability of a 14uF CSA cap, and L4 was raised to 220uH and C21 and C48 combined into a 4.0 uF part. Full documentation updates will be prepared soon.
This shows a measurement of the midrange driver at 1.5 meters; reflections showing comb filter effects in the response (which are not present near field) and it was also noted that the 5kHz resonance wasn't quite suppressed perfectly, no the 10K, but distortion sweeps didn't show any resonance amplification- as expected, HD2 and HD3 were better than -60dB from the actual output level.
As the tweeter crossover board hand't been updated yet to match the VituixCAD simulations, Mr Marsch expected the output to be high relative to the other drivers, but unfamiliar as I am with some details, I decided it was best to measure first and adjust afterwards at this point.
As I expected, the HF level above 4 kHz was 2-3 dB high.
This is the original PUR HF crossover schematic:
As a quick update, C6 was reduced to 15uF, and R34 increased to 6.25 ohms by paralleling two 12.5 ohm resistors.
The adjusted response then looked like this:
Now, this is still hot in the 5-20kHz area by a dB or two... and here we are again with a slight presence range dip. Not unlike those popular mini-monitors...
This is the point at which I think I would recommend implementing switches to create options for slight tuning in the level. That said, I listened to a few albums I'm very familiar with, and considering the known polar behavior of the individual drivers, I have to say this curve is quite pleasing to the ear. But fine tuning and evaluation should continue.
Note, I do have a turntable platform, but this system built in fossilized oak is a bit beyond its capabilities.
I did do an indoor walk around the room and this gave me a good opportunity to survey the sonics all over the great room area. Albums included the remastered "Best of Fourplay", Jackson Brown's "I'm Alive" and Strunz and Farah "Rio de Colores". I have to say I really did like what I heard, but never trust the builder to critique his own work. As these are for the living room, and acoustic upgrades to the room are not "encouraged" by the wife, I still find I can enjoy these quite a bit for every day listening as is. Vocal rendition is very good, whether Jackson Brown or Stevie Wonder on "Higher Ground" (Fourplay) or the vocalists on "Chant". At 2 meters or more back they don't sound hot, and the off axis rendition is very pleasing. Where I'm testing them is not at all ideal for low frequencies, as I mentioned; two doorways leading into other rooms; the other side of the living room should be better. But I could certainly get perspective on the low frequency performance by walking around, and an apt metaphor might be that the system sounds like a very well designed six inch two way mini monitor well integrated with a articulate, clean and deep sub that doesn't draw attention to itself when it shouldn't.
I believe many can live with that.
Next up will be building the new midrange board and the "value" budget HF crossover board, and evaluating those in the complete system.
Last edited by Evil Twin; 08 February 2023, 19:16 Wednesday.DFAL
Dark Force Acoustic Labs
A wholly owned subsidiary of Palpatine Heavy Industries👍 2- Bottom
Comment
-
This measurement set is NULL AND VOID!
A configuration problem with the test mixer because of poorly documented monitoring setup has been identified, which resulted in feedback causing ripples in the response curve appearing like time domain issues with the signal measurement.
After more listening evaluations, and consultation with the Chancellor, we have come to the conclusion that further "slicing" and adjustment of component values and full verification of assumptions baed on simulation results is in order before proceeding with the alternative MF midrange crossover development.
While this recent assessment indicates positive progress, it falls short of the preferred result.
But I could certainly get perspective on the low frequency performance by walking around, and an apt metaphor might be that the system sounds like a very well designed six inch two way mini monitor well integrated with a articulate, clean and deep sub that doesn't draw attention to itself when it shouldn't.
Several issues and working concepts should be kept in mind...- Develop a more precise understanding and measurement verification of the impact of the higher DCR AWG15 inductors on the measured crossover behavior and audible impact, if any...
- Evaluate in a definitive matter whether the 4th element (50uF capacitor) as used in the Solen based LF crossovers is necessary or desirable in terms of the measured results for the crossovers using the Jantzen AWG15 inductors; for this previous evaluation, the capacitor was mounted on the crossover board but not connected
- Engage in a more precise measurement of the midrange level and adjust if needed
- Correct the measured HF balance which was still slightly tipped up, and evaluate if needed subjectively
- Compare if there is any subjective difference in the LF performance of the new higher value proposition LF crossover
- Adjust and update measurement windows and refine technique where needed to examine specific issues
- Engage in careful comparisons where no level or gain adjustments are used after an initial setting.
The last measurement from the previous activities shows the issue with HF balance- this was problematic to address due to component supply issues, and it may still be a delaying factor in completing development. This benighted corner of the galaxy is still experience significant supply chain issues...
The first measurement examination is for the AWG15 crossover, without the 50uF shunt element following the 2nd series inductor, and with the 50uF capacitor wired in.
This was measured with the microphone placed between the woofers at 1.5M distance, which is substantially lower than the overall measurement axis or listening height.
The blue trace is the configuration without 50uF, the purple trace is with the 50uF now connected.
Below 200Hz there is nearly no visible difference, though VituixCAD model predicts there would be some lift in the mid bass. The curve above 200Hz is deemed potentially more desirable, with more complete attenuation but slightly better phase characteristics matching up with the midrange driver
Next, both Solen inductor boards were measured, as they had used some different brands of capacitors for some the 2nd shunt element and the notch filter.
The deviation is notable but appears to have no impact on the actual performance of the system, being in a range which is about 50 dB below the nominal output level. There was no measured distortion difference.
Next, comparative measurements were made with a Solen inductor LF board and the Jantzen AWG15 inductor boards, not touching levels or microphone position, only exchanging the connecting cables for the crossover boards:
The results are partly as expected- the upper range of the Jantzen board is slightly higher because a 2.3mH inductor was not available, the closest value being 2.0 mH. The lower frequency range of the Solen board was slightly higher, but not nearly in the range predicted by the VituixCAD simulation.
Moving on to the midrange, I had theorized that instead of two 1 ohm resistors in parallel for the series attenuation for the midrange, I would disconnect one, reducing the midrange level a dB or two, and believing it might be a closer fit to the bass.
Measured with 1 ohm series resistor:
This measurement (mic raised to midway between woofers and mid, at the normal listening height) bore that out.
1 ohm in series instead of 0.5 ohms it is.
Note, it looks like the notch filters could be tuned a little sharper, probably individual component Q issues are reducing the impact compared with the VituixCAD simulation, or a basically a non-optimum resistor value. I do not hear anything problematic on a test sweep or with music, but that will get another review at later this week, after more test resistors can be obtained.
Again, this does not quite agree with VituixCAD. Trust, but verify. Resistor availability is a possible issue, I used 20 Ohms for the notch damper resistor, but 25 might have been better, but was backordered.
Now, leaving one of the Solen LF boards in place for now, updated full range with long window, the disparity of tweeter level is more apparent after having reduce the midrange level slightly.
To bring that down into a more optimum range, and keeping in mind the directivity of the tweeter off axis and tuning a bit also for power response over a +/- 30 degree angle, I estimated that a 7.5 ohm series resistor might be the right value to replace the effective 6.25 ohm resistor from two 12.5 in parallel. This was also driven by the point that I have some 7.5 on hand from previous builds for other projects. 8 might be even more ideal, but no 8 ohm resistors on hand, and backordered at Parts Express still.
This may still be above an optimum level, and will be evaluated once other resistor values are secured.
A full range test with a long window and abundant reflections...
At this point I hooked back up the Jantzen AWG15 coil LF crossovers, and performed some listening checks. This is definitely a much more satisfactory rendition, closer to what I was looking for. The primary test cuts were from Best of Fourplay album (good percussion, punchy bass, and complex vocals, it’s a recent remaster for SACD version, in this case 2020 CD version) and Ken Mo “Slow down” album.
A very satisfactory presentation to my ears, and a more balanced one no longer reminiscent of a high performance mini-monitor and sub.
But I also decided it was necessary to give the Solen "big buck" crossover a listen too, not touching level controls, just swapping out the crossover boards.
Based on the measurements, I wasn’t expecting much if any difference, but it was enough to surprise. And possibly to convey why Magico and others spend the big bucks on the pricey crossover components.
My reluctant conclusion was more impact and articulation in the bottom end. Sounded less like a speaker. Overall balance not all that different, and on some other albums, little impact. But now I’m going to have to look at the internal space carefully on the second cabinet, to see if installing the larger LF crossover in the woofer box is feasible.
I went back and forth several times, still the same conclusion. Any other constructor of this system may want to engage this process for themselves, to achieve the most appropriate results for their situation and system.
This is one way in which "sweat equity" is generated...
Last edited by Evil Twin; 08 February 2023, 19:15 Wednesday.DFAL
Dark Force Acoustic Labs
A wholly owned subsidiary of Palpatine Heavy Industries👍 1- Bottom
Comment
-
- Danny- Bottom
Comment
-
The smaller LF crossover and the others could be installed in the LF cabinet, but the larger one is probably pretty much a no go, unless it would fit in the base, based on my checks. It's possible to do one with just AWG12 Solen inductors, probably. But for now, I'll just use the external binding posts that already exist.
.DFAL
Dark Force Acoustic Labs
A wholly owned subsidiary of Palpatine Heavy Industries👍 1- Bottom
Comment
-
While reviewing the recently measured data and curious behavior of Jonmarsh's test system during Force Meditation, I sensed a disturbance in the Force due to what might be a control integrity issue with the CAM system...
The unusual 60Hz response and the ripples in the measured response and some aberration of the calculated impulse from convolution of the since chirp response suggested the possibility of a signal feedback loop at higher output test levels... perhaps due to cross feed between the measurement channels.
A brief investigation today uncovered exactly this problem, partially due to a poorly documented monitoring control setup which could allow cross feed of both channels with repetitive presses of the monitor button...
Some of the previous example measurements were repeated, and the results are shown here... and comply closely to rational expectations.
LF output with Solen equipped Crossover
NAA Midrange output with recent component updates
Full spectrum Test at 1.5M
These results appear to be most satisfactory....
Of course, these events and outcome might suggest a lesson to be learned from this story...
Never send a moderator to do a Sith Lord's job...
DFAL
Dark Force Acoustic Labs
A wholly owned subsidiary of Palpatine Heavy Industries👍 1- Bottom
Comment
-
A couple of component updates on the midrange notch filter, and some additional LF and full range measurements.
Measurement of NAA Midrange with distortion
Comparison of levels for LF crossovers using AWG15 (Jantzen) and AWG10 (Solen) inductors
Full range measurement with Jantzen Crossover
Full range measurement with Solen Crossover
DFAL
Dark Force Acoustic Labs
A wholly owned subsidiary of Palpatine Heavy Industries- Bottom
Comment
-
The crossover testing and adjustment for the PTT6.5 MF midrange driver has been completed, with most satisfactory results, especially considering a very deliberate effort to minimize the financial outlay for components and simplify the topological structure.
Full success for the stated goals is within our grasp...
A detailed summary report of this project effort to date will be prepared in the form of an "article" and published in the Articles section, along with schematics and measurement results.
To celebrate this milestone, I made a purchase from HD Tracks of the frequent evaluation album, "Fourplay 2020 remaster Greatest Hits" in 24/192 format for Mr. Marsh.
As a "hint", I present this measurement result in room with a very long gating window but 1/3 octave smoothing, as regards the overall balance and response. This is for the "value" version crossover using the PTT6.5M04-NFA-01 driver. with 250Hz and 1800Hz crossover points in LR3. All drivers connected in positive polarity.
The subjective results on a small selection of albums was most satisfactory...
DFAL
Dark Force Acoustic Labs
A wholly owned subsidiary of Palpatine Heavy Industries👍 2- Bottom
Comment
-
- Bottom
Comment
-
All of the current tests are with the TB34B-4 tweeter.
At this time, we will prepare another version of the NAA crossover, moving the upper crossover point down from 2250 (as both were originally designed) to 1800Hz, and may test it with the "value" version of the HF crossover, and probably prepare a modified version of the PURfectionist crossover with an 1800Hz crossover point.
The MF version was better subjectively than expected with the "value" build crossover; I plan to investigate whether this is due to an advantage in off axis polar behavior. The data sheets do not indicate any such thing clearly. Of course, reducing the crossover frequency for the NAA version may change that.
The NAA version has lower measured distortion- in reality as well as the data sheet, but both are well beyond the capabilities of more "typical" drivers.
The test cabinet will also be moved to a location in the great room that will likely be more "neutral" in the mid bass and on down, without nearby room doors creating a bass trap and variable boundary characteristic.
I am also seriously considering giving the crossovers a "subtle" switches level capability for midrange and treble in order to produce a relatively neutral presentation whether fairly close to a wall boundary or out into the room- this will require on the order of a 2dB step, based on current measurements.
At this time, there is no inline filter employed- the intent is to "tune" close to flat on axis, and let inline preamp modifications or subtle crossover switches modify the response from that point.
DFAL
Dark Force Acoustic Labs
A wholly owned subsidiary of Palpatine Heavy Industries👍 1- Bottom
Comment
-
- Bottom
Comment
-
- Bottom
Comment
-
This is a "brief" update with key information- the full working schematic and BOM of the existing build, which could see small tweaks, but this may be up to individual interested builders- the result seems quite satisfactory in general at this time.
This is for the version using the PTT6.5M04-NFA midrange driver- an alternative configuration has been developed in parallel but for now that effort is set aside, including because there is another project under consideration which may use the two samples I have on hand.
System Schematic:
System BOM
PDF format is attached to this post.
Midrange only Plot focusing on cone resonance compensation networks performance
Mid + HF combined
Mid + HF combined- vertical height impacts degree and shape of Reverse Null
Impulse response still shows room reflections in this measurement
Full Range with Jantzen AWG15 inductors- switches set for ~ -2dB on mid and tweeter
Full range with Solen inductor LF crossover - longer gating window, more reflections and room issues. 10 ft from rear wall.
Last edited by Evil Twin; 21 March 2023, 16:59 Tuesday.DFAL
Dark Force Acoustic Labs
A wholly owned subsidiary of Palpatine Heavy Industries👍 2- Bottom
Comment
-
Purifi has a prototype PTT33 Alu tweeter in the works, fyi...
- Bottom
Comment
-
Yes, I read this article a few weeks ago. Shame that it's a bit outside the range of even an advanced DIY effort- these aren't "freeware" tools" (go price out COMSOL and MatLAB- I have the latter) and this is a serious unified engineering effort. Hat's off to PuriFi and for sharing it.
A MATLAB GUI was written for the optimisation. A netlist of the crossover is input and compiled into equations. Next we import a baffle step simulated in COMSOL using a 3D model of the box and woofer. The Matlab model combines the PR box model with the blocked impedance model and the xover network to give the response form the amplifier voltage input to the SPL output of the box. The SPL is corrected for the baffle step.
Let's take that very brief and unassuming first sentence- if you've never done this, the way they state this makes it sound like something you can whip out between your morning coffee break and lunch time. That's just not the case.
Having championed and commissioned the effort to create a custom MatLAB GUI tool for designing LLC resonant converters using true time domain analysis, and developed advanced SIMPLIS models including accurate time domain magnetics simulation, let me just say this is not a walk in the park. Our goal was to develop an LLC analysis tool based on discrete time domain analysis, not the commonly used FHA (First Harmonic Approximation) approach, which is much like linear loudspeaker circuit analysis, and very inaccurate once you move away from the resonant frequency and need to predict gain and RMS operating currents. In the end we enlisted a consultant organization in the Netherlands, and were able to model the LLC converter accurately based on time domain analysis as described back in 2001 by James Lazar and Robert Martinelli- as I like to joke with colleagues, folks didn't go with this approach because the math was too challenging, but it's the only accurate way to do it.
See the PDF attachment for their paper.
Steady State Analysis of LLC Best Quality.pdf
Then I created an XL tool with accompanying rules of thumb for a simplified design definition approach, then LT SPICE verification situations, which while not as in depth as the SIMPLIS versions, could be accessed with the free LT-SPICE simulator, and which included synchronous rectification for the output stage. This illustration incorporates the key elements in the model schematic as an aid to understanding and usage. I'm still getting direct requests for this model from internal people at my former employer and from end customers.
Back to the PuriFi paper-
One could go through this paragraph by paragraph and comment on various issues identified, but I would remark that a perhaps surprising number can and have been identified and addressed simply using VituixCAD, such as driver impedance interactions with crossover network- this is key for this, project, the update to the Isiris, and to the CC Supreme center channel, which actually uses impedance control zobel networks to assure the intended LF response but also to adjust the response level.
The enclosure modeling is another matter- one might say that for the enclosure development, the most sure path available for the DIY folks is simple brutal overkill with construction materials (maple ply, specialized bamboo materials, for example, like this Isiris build) and caution, and rigorous testing.
The optimization process for the HF level is particularly challenging for the DIY designer, as few have the insight based on an understanding of compression driver and horn throat design to scale this to a shallow waveguide- (heck, most of the companies making those products don't have that- folks like EighteenSound seem to be an exception) in many cases, the best the DIY person can do is experimentation in hardware of the waveguide throat profile and use of tweeters with an integral phase shield or adding this in a 3D printed waveguide structure. The few waveguide efforts I've published have been very focused on consistent directivity up to a minimum of 15kHz, which does take a very optimized approach to interfacing the tweeter/driver to the overall waveguide. And assumptions/tools for waveguide design which use a flat diaphragm model won't get there.
There is little information in this paper, unfortunately, about the actual tweeter design motor and specs, but with the thoroughness of their general approach I would expect it to be fairly exceptional- the interesting question might be whether they offer a "conventional" panel mount build to OEM's or DIY, or only the integrated waveguide with coherer, as they term it. I wouldn't be surprised if the latter may be the case.
In either case, I'm certainly looking forward to seeing what they do....
OTOH, of late, I seem to be moving away from dome tweeters... not a particularly conscious decision, just the result of a lot of recent component evaluations and design work.
the AudioWorx
Natalie P
M8ta
Modula Neo DCC
Modula MT XE
Modula Xtreme
Isiris
Wavecor Ardent
SMJ
Minerva Monitor
Calliope
Ardent D
In Development...
Isiris Mk II updates- in final test stage!
Obi-Wan
Saint-Saëns Symphonique/AKA SMJ-40
Modula PWB
Calliope CC Supreme
Natalie P Ultra
Natalie P Supreme
Janus BP1 Sub
Resistance is not futile, it is Volts divided by Amperes...
Just ask Mr. Ohm....👍 2- Bottom
Comment
Comment