Minerva Monitor: "Patience my ass, I'm going to go build something!"

[ergo]

Very nice one again cochinada.

Do you know of any Sketchup tutorial that show this type of modelling. I do not mean speaker units, but generally this type of detailed/complex models instead of a house or a cupboard. I would be interested in the workflow for something like this.
I've been learning Rhino3D with the help of video tutorials and have some idea how this would be built there, but do not have a clear understanding how one approaches this in Sketchup.

[cochinada]

Thanks Ergo. Unfortunately I'm not aware of any such thing. I had to learn by myself but the good news is that one improves and discovers new ways of doing things faster and better as experience grows but it takes time. There are lots of videos on Youtube, some will give you good ideas where others won't and you have always the sketchup group/forum where you can ask for suggestions.

The first time I tried to model a driver it took me days and the results were not very good. It's very important one understands how the object was build as this makes it easier to draw it. Some parts are still very complicated though but like I said, one gets used to it and there are several ways to skin a cat.

Divide and conquer is the key: try to isolate the more complicated pieces as much as possible and then figure out the simplest way to model them. Follow-me tool is very useful, as so is the roundcorner plugin which are amongst my favourites. Then simple things as move by selecting just edges or surfaces of a 3D object can do magic things.

[cochinada]

When drawing a driver I normally follow this method:

1. Start with the plate and respective profiles and screw holes but leave the cone for later.
2. Draw the magnet and find the position of the plate(s) in between.
3. Now comes what often is the more difficult task - the spider supports or pillars or whatever they are called. There are so many different designs that is impossible to give a one fit all recipe. The methods I mentioned on the post above are normally enough and sometimes you need them all. The good thing is that you need only to draw one of this and then just replicate it (use a component).
4. I normally leave the cone for this stage.
5. Draw details like terminals, the fabric on the middle plate, etc.
6. Some trickier round surfaces I leave for the end as normally they need to be done by intercepting two 3D objects or at least that is how I do it (e.g. the round corners between the spider pillars and the main plate).

Use your imagination, copy paste as much as possible and enjoy. :banana:

Some tips I learned the hard way that can help avoid a lot of problems:

a) always use the same number of sides on the circles (at least the bigger ones). I normally use 96.
b) always draw in bigger scale (e.g. 10:1 or even 100:1).
c) the simplest and fastest way of drawing something is ALWAYS the better choice.
d) be particularly alert of when a circle is transformed by Sketchup into an arc.
e) this is a classic but it happens all the time: some lines don't always form a surface automatically and one has to create it. I use 'Make face' plug-in all the time.
f) zoom, zoom zoom! You will find that too often lines are not connected as you'd expect.
g) use guides but be aware they are correct!
h) make sure all your solids are well constructed. If you don't see a volume figure, than you have problems. Correct them at once. I use plug-in 'Inspect and repair solid groups and component'. Sometimes you can have two or more coincident surfaces: just delete all but the last one. The best way to check this is to hide Hidden geometry and keep selecting a surface until you see it all selected and not just part of it. If you do then you have coincident surfaces.

[JonMarsh]

I'm "waking up" a little more as I heal, and it now occurs to me that I should grab those files from Ergo, have a close look, and play with them at lunch.

Sorry I've been slow on the uptake- many distractions, and I'm not out of the woods yet- everytime I've been getting a burst of energy, I've been using it, and then paying for doing that! Makes for kind of uneven days!

[JonMarsh]

When drawing a driver I normally follow this method:

1. Start with the plate and respective profiles and screw holes but leave the cone for later.
2. Draw the magnet and find the position of the plate(s) in between.
3. Now comes what often is the more difficult task - the spider supports or pillars or whatever they are called. There are so many different designs that is impossible to give a one fit all recipe. The methods I mentioned on the post above are normally enough and sometimes you need them all. The good thing is that you need only to draw one of this and then just replicate it (use a component).
4. I normally leave the cone for this stage.
5. Draw details like terminals, the fabric on the middle plate, etc.
6. Some trickier round surfaces I leave for the end as normally they need to be done by intercepting two 3D objects or at least that is how I do it (e.g. the round corners between the spider pillars and the main plate).

Use your imagination, copy paste as much as possible and enjoy. :banana:

Some tips I learned the hard way that can help avoid a lot of problems:

a) always use the same number of sides on the circles (at least the bigger ones). I normally use 96.
b) always draw in bigger scale (e.g. 10:1 or even 100:1).
c) the simplest and fastest way of drawing something is ALWAYS the better choice.
d) be particularly alert of when a circle is transformed by Sketchup into an arc.
e) this is a classic but it happens all the time: some lines don't always form a surface automatically and one has to create it. I use 'Make face' plug-in all the time.
f) zoom, zoom zoom! You will find that too often lines are not connected as you'd expect.
g) use guides but be aware they are correct!
h) make sure all your solids are well constructed. If you don't see a volume figure, than you have problems. Correct them at once. I use plug-in 'Inspect and repair solid groups and component'. Sometimes you can have two or more coincident surfaces: just delete all but the last one. The best way to check this is to hide Hidden geometry and keep selecting a surface until you see it all selected and not just part of it. If you do then you have coincident surfaces.

Great set of tips, especially for working in SketchUp, which has some quirks compared with the solid modelers I'm used to working with. SketchUp can be tuned by adjusting parameters to work OK with "normal" dimensioning, but you have to go in and tweak several settings. Thanks again for taking the time to constricted these points.

[ergo]

cochinada - thanks a lot for the pointers.

Jon - take is easy and slow, but of course I am happy if you can take a look at my data. This week has been busy and my wife was off to California, but arriving back today. So this weekend I hope to get some progress myself also.
I have been checking the crossover part and their respective effects too and doing some trials to fine tune it a bit.
I also rechecked if the LspCAD is still happy to "emulate crossovers" on Windows 10PC.... with ASIO driver it seems it is and no problems there.
I've done trials in past where I play a measurement signal from file and analyze it later / or use ARTA-s PN Pink noise and RTA. I will try to measure the emulated crossover in the listening position. Biggest question for me is if a flat is the target to shoot for or should the onaxis be bit tilted down on highs. Past experience seems to indicate that waveguided tweeters sound bit too bright if fully flat.
Also it seemed to be better to use a 20deg off axis as the design axis. Which is probablly also a good listening axis for these coax units instead of the spot on axis.... Luckily VituixCAD makes it very easy/fast to choose which one it shows as design axis plot.

[cochinada]

You're welcome John. Take it easy. Don't push to hard and have plenty of rest.

Most problems in Sketchup are caused because of the way a circle (or any curved line like a simple arc) is constructed: n straight line segments. In my earlier models I didn't know we could change the amount of segments and my drivers when examined in large zoom were cumbersome and even my carpenter commented something like "curious that your circles are all made of straight lines". By increasing the default to 96 (instead of 24) things improve a lot but the root cause of the problem remains. Even two circles of the same size after some geometric manipulations can easily become out of synch because of rotation as the vertices no longer coincide and this creates havoc. This happens particularly when Sketchup decides to change a circle into an arc (e.g. after using follow-me tool) and that's why I mentioned tip d)

[cochinada]

cochinada - thanks a lot for the pointers.
...
Biggest question for me is if a flat is the target to shoot for or should the onaxis be bit tilted down on highs. Past experience seems to indicate that waveguided tweeters sound bit too bright if fully flat.
Also it seemed to be better to use a 20deg off axis as the design axis. Which is probablly also a good listening axis for these coax units instead of the spot on axis.... Luckily VituixCAD makes it very easy/fast to choose which one it shows as design axis plot.

You're welcome.

Regarding your doubt, been there, though about that and finally decided to optimize the DI in detriment of the on-axis alone.

Now, don't get me wrong as I'm not an audio expert by no means or measures but I've learned a lot, also from Kimmo who was patient enough to give me a lot of insights about this topic. I have the information somewhere on my thread, namely the criteria to take in consideration about the best possible DI and also why this is so important. I can search for it and post the link here if you're interested.

EDIT: found it here.

and some insights I got from Kimmo of some basic rules for power response:

- Total tilt from 200Hz to 12kHz about 6 dB, or more if calculation is based on time-windowed measurements and speaker is unidirectional (not dipole).
- No dips below upper midrange.
- Straight power response on fundamental region <800Hz is acceptable, but there should not be any rising.
- No dip at upped midrange
- No step or rising towards mid-treble.
- No peaks at lower treble

...So, as you can see not an easy compromise... :-y

[cochinada]

Speaker off axis response: psychoacoustic and subjective importance
Speaker Off Axis: Understanding the effect of Speaker Toe-In

[Steve Manning]

Very nice cochinada .... I'm going to have to swap out the less than stellar ones I used and see how yours turn out in a rendering.

[Steve Manning]

Mr Marsh ....... don't make me call your girlfriend to make you behave

[JonMarsh]

You're welcome John. Take it easy. Don't push to hard and have plenty of rest.

Most problems in Sketchup are caused because of the way a circle (or any curved line like a simple arc) is constructed: n straight line segments. In my earlier models I didn't know we could change the amount of segments and my drivers when examined in large zoom were cumbersome and even my carpenter commented something like "curious that your circles are all made of straight lines". By increasing the default to 96 (instead of 24) things improve a lot but the root cause of the problem remains. Even two circles of the same size after some geometric manipulations can easily become out of synch because of rotation as the vertices no longer coincide and this creates havoc. This happens particularly when Sketchup decides to change a circle into an arc (e.g. after using follow-me tool) and that's why I mentioned tip d)

Yup! I've heard of this before, and experienced it first hand! It's one of those quirks that prevents me from falling completely in love with SketchUp, useful though it is!

[JonMarsh]

Mr Marsh ....... don't make me call your girlfriend to make you behave

You can call her, you can call me, just don't call us late for dinner... :W

[JonMarsh]

Speaker off axis response: psychoacoustic and subjective importance
Speaker Off Axis: Understanding the effect of Speaker Toe-In

Both good articles, if familiar material. Obviously, for many designers, not so familiar.

[JonMarsh]

You're welcome.

Regarding your doubt, been there, though about that and finally decided to optimize the DI in detriment of the on-axis alone.

Now, don't get me wrong as I'm not an audio expert by no means or measures but I've learned a lot, also from Kimmo who was patient enough to give me a lot of insights about this topic. I have the information somewhere on my thread, namely the criteria to take in consideration about the best possible DI and also why this is so important. I can search for it and post the link here if you're interested.

EDIT: found it here.

and some insights I got from Kimmo of some basic rules for power response:

- Total tilt from 200Hz to 12kHz about 6 dB, or more if calculation is based on time-windowed measurements and speaker is unidirectional (not dipole).
- No dips below upper midrange.
- Straight power response on fundamental region <800Hz is acceptable, but there should not be any rising.
- No dip at upped midrange
- No step or rising towards mid-treble.
- No peaks at lower treble

...So, as you can see not an easy compromise... :-y

These are really ALL solid fundamentals that should be kept in mind for any design- there are some variations on this theme, such as a flat axial and power response and then the 6 dB taper above that, and the other approaches to modifying the high frequency pass band such as Linkwitz favors with a -2 dB/decade response...



But the list you enumerate is very important to natural sound, especially the last three...

[JonMarsh]

Comparing SW223BD02 measurements

OK, did some data review- I think I know where the issue is, but not knowing more about how Ergo measured and recorded results, I don't know why...


This is the composite data from Ergo on his SW223BD02 measurements- a lot of measurements and data, I'm sure this took a lot of work. The curve looks "right", except that I have to question the levels recorded.



This is showing the half space (i.e., infinite baffle) sensitivity in the 800-1200Hz region to be about 75-76 dB, and I just can't reconcile that number. By T/S analysis, by the Wavecor Data sheet, and by my own measurements, I think that should be in the 83 dB range. If you had a "noisy" measurement with no axis averaging, and 1/6 octave smoothing, should look kind of like this:



This was measured in the test cabinet, with the PR, and you can see it contributes some substantial LF output in the sub 40 Hz area.

That's about a 7-8 dB difference overall- about 10dB difference at 100Hz. OTOH, the measurements for the C18EN001 came in quite similar considering- Ergo's were very plausible in detail, and mine matched closer than usual to the Seas measurements. Same with the tweeter- no issue there.

Do you use an SPL calibrator with your microphone before each measurement run? (I have one for my Acco Pacific mic)

Another thing I do that helps prevent "oopsies" (yes, that's the technical term) is do all the measurements at one sitting, with the mic and test gear undisturbed. I also usually do a combined all drivers at once scan, and use that as a check against the summed drivers to ascertain the acoustic offsets, after I've extracted minimum phase data from the driver measurements.

Otherwise, maybe there's something wrong with your SW223BD02? was this a sealed box measurement? (don't see the effect of PR output- did you stuff the box? Should only line the LF part of the cabinet).

[cochinada]

I used a lab microphone thats good to 40k, so I also wanted to get the SPL right as I do have the calibrator.... Now getting ARTA to get right levels was a bit of hassle. If one does the microphone calibration, then for example the SPLmeter sub module starts to show correct level right away. The 2 channel IR measurements is different though. In end I had to calibrate both left and right input and then also the microphone SPL -> mV before I got all working.
...
Anyhow. Below are the setup pics + the initial gated measurements entered to VituixCAD. I'll need to follow the process of merging those with the near field measurements and baffle simulation to get it correct in lows also.

OK, did some data review- I think I know where the issue is, but not knowing more about how Ergo measured and recorded results, I don't know why...

This is the composite data from Ergo on his SW223BD02 measurements- a lot of measurements and data, I'm sure this took a lot of work. The curve looks "right", except that I have to question the levels recorded.

This is showing the half space (i.e., infinite baffle) sensitivity in the 800-1200Hz region to be about 75-76 dB, and I just can't reconcile that number. By T/S analysis, by the Wavecor Data sheet, and by my own measurements, I think that should be in the 83 dB range.
...
Otherwise, maybe there's something wrong with your SW223BD02? was this a sealed box measurement? (don't see the effect of PR output- did you stuff the box? Should only line the LF part of the cabinet).

Indeed the most strange thing, now that you mention it, is the absence of the PR effect suggesting something is not right. :scratchhead:

About the SPL level, there are so many things that can go wrong that I'm not too surprised. Could it be that the gain on the amplifier was adjusted between the far field and the near field measurements? I actually did it when I took my measurements because after all with all things constant when you put the mic closer for the near field the levels will rise of course, although a better practice would be to adjust the attenuation on ARTA itself and leave all the knobs untouched.

I know that ARTA is a pain in the ass for calibrating the soundcard and the mic, so much that I eventually gave up (also because I don't have a mic calibrator anyway).
But even if this was not very accurate, I guess one should have the same comparative levels for all drivers and this is not happening, so that's why I'm wondering if the gain levels were adjusted, either on the amplifier or the soundcard or even windows settings during the whole process and particularly for the woofer. The mic had to be moved for sure in order to point to the different centre. Was this placed at the same distance from the new centre or (much) further away?

[ergo]

I did calibrate microphone for the correct SPL, but I did not adjust the level to 1W/2.8V. I used a lower playback level, so it is expected that all 3 drivers show less SPL.
I did keep all the settings/gains unchanged for far field test
Microphone was always at 1m distance on the Coax driver center axis. Also for woofer measurement
For near end I decreased the playback signal from ARTA from-3dB to -13dB, so near field plots were later copmensated by 10dB when merging.

Now if I put datasheet plots and my measured one on same graph and adjusted the midrange to match in level in it's upper end I get that I was off from the normal 1W/1m by approx 7.7dB which matches Jons estimation too.

One plot shoe the datasheet vs short gate vs long (200ms) gate. A bit of drop of my woofer levels versus datasheet is expected as I did not have mic at woofer axis but in Coax axis instead.

Other plot shows datasheet vs long gate vs merged response with baffle step effect. From this it seems that midrange baffle simulation holds ok against the long gate plot, but the baffle simulation for woofer seems way off and over estimates the drop a lot. I'll try to work on this and fix the woofer responses.

[ergo]

I think I found the main issue... a stupid user error form my part. In Merger one should enter the Sd for both near end sources - ie woofer and PR. I think I might have set the Sd values wrong the first time around.

Now with the new merged woofer plots +7.7dB fix offset for all drivers it starts to look better with Jon's cross version without alterations yet.

https://onedrive.live.com/redir?resid=73B3DF257C874FF5!16200&authkey=!AJhXMoXodnq2j8o&ithint=file%2czip

[JonMarsh]

I'm glad you've been able to get to the bottom of this so quickly, Ergo! The driver measurements with all the hoops you jump through in this process are very high quality, but the number of steps certainly increases the chance for an error compared to a long gate direct measurement! Using the latter as a sanity check is always a good idea.

We'll see what kind of luck I have merging all this and getting it to work with the AP setup as well- that may be biting off a lot in one step! :W

[JonMarsh]

Speaking about measurements, I bought a smaller lighter amplifier to try out for doing speaker measurements- a Behringer A500. I've tested it, and I think it will be convenient and usable, but certainly not the last word for best in class speaker measurements-

Response is down about 0.5 dB at 20kHz, THD versus level isn't anything to write home about; at least, versus frequency it's pretty consistent, not rising a whole bunch at higher frequencies.




Measured at 20VRMS, 50W at 8 ohms



THD+N versus level. Let's call this a 75W per channel amp on my local AC line.

[dar47]

Okay, why not just get those diy amps put together.:W

[JonMarsh]

Okay, why not just get those diy amps put together.:W

I'm reminded of the opening line of an old Byrds song... "Time, time, time! "

This will be OK for most of the things I need to do, but not for measuring distortion on high performance drivers and extended response tweeters. Troels has a problem with his setup, too- it's soft above 18 kHz, not really useful for looking at extended behavior of tweeters.

I was thinking last night I should take a few minutes (har!) and bench the Aragon Palladium I have sitting in the living room, 72 lb anchor weight that it is...

OTOH, I could just use my NAD M22, but then I'd have to take it out of the main system every time I want to do speaker measurements, and that would also be a PITA.

And my NC400 stuff is out at Steve Manning's though he's been too busy to play with it.

[Steve Manning]

And my NC400 stuff is out at Steve Manning's though he's been too busy to play with it.

Speaking of that ... we still need to figure what we're doing with the front end of those.

[JonMarsh]

Speaking of that ... we still need to figure what we're doing with the front end of those.

Guilty as charged- I owe you a small schematic. I say we just go with the -6dB gain and don't mess with anything else- you'll just need 3-4VRMS to drive them to full output.

[JonMarsh]

I think I found the main issue... a stupid user error form my part. In Merger one should enter the Sd for both near end sources - ie woofer and PR. I think I might have set the Sd values wrong the first time around.

Now with the new merged woofer plots +7.7dB fix offset for all drivers it starts to look better with Jon's cross version without alterations yet.

https://onedrive.live.com/redir?resid=73B3DF257C874FF5!16200&authkey=!AJhXMoXodnq2j8o&ithint=file%2czip

[ATTACH=CONFIG]25572[/ATTACH][ATTACH=CONFIG]25573[/ATTACH]

I should take your C18 and my LF data and see how that looks together, and put together a proposed voicing. Can you put up a link with the revised LF data? That would be even better...

[ergo]

https://onedrive.live.com/redir?resid=73B3DF257C874FF5!16200&authkey=!AJhXMoXodnq2j8o&ithint=file%2czip

New woofer data is in folder
Minerva_EE_try2\woofer offaxis merged - NEW fixed\

[ergo]

I got the emulation to work today and listened to single speaker for about 30 minutes with the VituixCAD ver of your crossover and then very slightly modified one trying to get the phases to match with my measured data.
I do like the sound potential - that is there for sure... the details of voicing - too little listening still. I will be attempting to make some measurements of the speaker in mid room as well as in listening position to see if the balance seen on screen maps to what it is in room.

[JonMarsh]

By the way Jon, when taking measurents if and what value of protection capacitor you use for tweeter? I used 200uF in these. Did not dare to run these tweeters without any cap.

I'm wondering if this isn't a clue to the differences I'm seeing between the low end of your tweeter measurements and mine- though I think 200uF is so large, it shouldn't have any impact.

Now, for about 10 years I have been using test systems which use swept sine chirps- Praxis, Fuzzmeasure, and now Audio Precision. I never use a blocking/protection cap for the tweeter, because the test level is never above 2.83VRMS unless I'm running some specific stress tests, like 5.6VRMS- and tweeters other than ribbons can handle that without issues. I do always setup and adjust levels using an oscilloscope and test first with a midrange or woofer driver to be sure everything is working properly before connecting a tweeter.

There are of course some other possibilities, like the influence from difference in baffle dimensions. I'm going to put together some comparison plots today to try to get to the bottom of this. Also, if the weather lightens up a bit, I'll hopefully make more cabinet progress- some I can do even before that, and will probably start that after lunch.

First, just a quick update, composite plots of my measurements at 0 degrees, 15, 30, and 45.


C18EN001 Midrange 0-15-30-45 degrees (Black/Blue/Maroon/Red)




C18EN001 Tweeter 0-15-30-45 degrees




Comparing graphic plots with multiple axis can be confusing, so I'm also going to prepare and post some plots with Ergo's 0 degree and 30 degrees, and my 0 degrees and 30 degrees, in VituixCAD, as these should be more directly and easily comparable. Maybe that will help. Also, I've been working on the crossover, creating a voicing which is compliant to what Linkwitz and some others believe is "ideal", with a net slope of 4 dB between 100Hz and 10kHz, at 2 dB/decade. that should get a short separate post. More to follow today.

[JonMarsh]

Here is the updated crossover, STILL using my old measured data- so this is unlikely to be a "production" version, pending measurements with the new cabinets. However, it is just part of the process of moving the design intention along...





Biggest changes are taking out the original L-R tweeter shunt completely and tweaking the elliptic network slightly; and adding a shunt padding resistor to the midrange, so that as in the case of the Isiris crossover, one can use the series and parallel resistors to alternately tune the passband region and the transition band region.

Other values have been trimmed to obtain the downslope characteristic, while keeping strong reverse nulls and alignment of the phase characteristics.








I've started an Eagle PCB schematic which will break out the individual components and part numbers suggested, but it's a bit early to complete that. :W

[ergo]

Thank you a lot Jon for looking into this.
I'll have some time tomorrow and I'll attempt to redo at least the on axis measurements on my box and this time I'll do the same - ie set the level to 2.83V and run the tweeter without any protection. I'll also try to recheck the ARTA somehhat tedious process of getting the SPL reading right in 2ch sweep measurement.

[augerpro]

Also, I've been working on the crossover, creating a voicing which is compliant to what Linkwitz and some others believe is "ideal", with a net slope of 4 dB between 100Hz and 10kHz,

Harman's research found closer to 6dB (or maybe it was 8dB) from 100 to 10000 as the listener preference. They mention the preferred slope is also directivity dependent, so I wonder if Linkwitz's 4db is more due to his speakers being dipole?

Awesome project BTW!

[5th element]

Active systems have long been a part of my life as well as the dilemma of to protection cap or not. Currently I do not use one. Ironically the only time I've fried a tweeter is when a protection cap was used and in that instance I still have no idea how the tweeter died. It happened when I wasn't listening to music and the insulation on the voice coil wire ended up pretty well melted. Bye bye D2905/9700... Good thing you can get replacement coils for them!

[ergo]

Must have been amplifier going to oscillation to do that sort of damage, but yes odd that no input was able to still trigger that.

With these C18 units it does make me bit itchy as they seem to really be 'all gone' and no new productions coming available. So I don't want to loose my only pair. Jon probably has bought a few extra knowing him

[JonMarsh]

I understand the desire for caution. What I'm being told is that they're out of stock now but not discontinued, available on backorder. Hopefully that's the case...


OK, some FRD file comparisons from measurements...


My C18EN001-T on axis measurement (taken at 1 meter





Ergo's data file for 000 degrees from the most recent download folder for the tweeter section:




My data file for 30 degrees off axis:




Ergo's data file from the most recent download folder for 30 degrees off axis:





There's a significant difference in the levels, and in the case of the 0 degree file, in the shape of the waveguide curve in the area around 2-3 kHz and again at the higher frequencies. I don't really know what to make of it.

I'm still not getting how VituixCAD handles the multiple axis for directivity and power and how that correlates with the main SPL curve; I've just got to spend some more time with the manual.

For now, I also have to wonder if the baffle width and diffraction effects at lower frequencies are playing into this.

Now, one reason I have some confidence in my measurements, is that after designing a crossover based around it, I got results pretty much as you'd expect when measuring the whole speaker in a single moderate gate length measurement.

This is the 1.0 version of the crossover, where I'd decided that the on axis was too hot, and that the tweeter network needed adjustment to reduce the 3-7kHz area. There aren't any measured dips in the crossover region on or off axis as one might expect with Ergo's measurements.




And the updated version of the crossover (last built one, that is) looks OK, too, at least on the test axis.



well, enough! I should go out and work on that test cabinet- :W

[JonMarsh]

Harman's research found closer to 6dB (or maybe it was 8dB) from 100 to 10000 as the listener preference. They mention the preferred slope is also directivity dependent, so I wonder if Linkwitz's 4db is more due to his speakers being dipole?

Awesome project BTW!

You know how it goes, Brandon.... sometimes gentlemen of honor and principle may just have to agree to disagree and season to taste!

Some logic is that the wider the dispersion, the greater the desired slope down... but with a Dipole there shouldn't be as much reflected low frequency energy, and that would argue against using a shallower roll off. Siegfried worked with several people and several technical sources evaluating this... I'm more of a Linkwitz fan than a Harmon fan (for now) so I'm going to try this as a target.

Next month, who knows?

[JonMarsh]

Making more sawdust... or router dust, as the case may be!

Just doing all this computer stuff gets on my nerves after a while... if there isn't some sawdust involved somewhere sometime, something is wrong....

Today's task was to trim the second side panel, follower trim it, bevel, and so some overall sanding, just enough to get it ready for handling and test, not for finishing at this time.

The little Craftsman 5" battery powered circular saw has been a champ- it came with a very decent thin kerf blade with nice sharp carbide teeth that have stayed sharp, in spite of the material I'm using them on- that's the first step in trimming the side panel, cutting things down to about 1/8" overlap.





Next step is the flush trim, just like laminate, only bigger, using a new Hitachi 2-1/4 HP router I picked up during a binge buying session on power tools after my last medical procedure (we each act out in our own ways...)





The Hitachi is a gem; like the M12V's tucked away in storage in Livermore, they are glass smooth- they make a Porter Cable feel like the bearings are getting close to giving out in comparison. Probably some Samurai craftsman in the background lineage of the production designer, or someone in Hitachi Heavy Industries that is just used to over doing everything...





Speaking of Porter Cable, next up for the bevel edge cut on the side...





You see, this was an immediate A/B comparison, though not a blind comparison- only a fool uses power tools with his eyes covered!


Remember, these are just the 2nd generation acoustical prototypes, and will be finished and used in a utility application such recording or possibly taken into the work lab if we ever get back on that Class D project there. But from horseback, they probably don't look too bad. And they are hella stout! (front to back brace of BB Ply, sides are combo of BB ply and LBL, tops and bottom and fronts are double LBL) .





For now, I'm leaving the cut out plate for the passive radiator in there; it will have the holes for test leads going through it, since we don't really care about the sub 50 Hz stuff for designing the crossover, and I can finalize the connectors and wiring later on.

[augerpro]

but with a Dipole there shouldn't be as much reflected low frequency energy, and that would argue against using a shallower roll off.

When I was looking into designing a dipole I assumed the same thing based on the physics, but just thought I'd mention it anyway. I'd like to build that dipole, too bad I'm in an apartment downtown Denver now, so no garage. High-ish ceilings in my place though, which means I can still maintain a clean 6.5 ms window.

[JonMarsh]

How's the downtown Denver lifestyle these days? For me, it's really hard to imagine living without a garage or some kind of shed/storage/work space. A lot of people do that, though...

[Juhazi]

I got the emulation to work today and listened to single speaker for about 30 minutes with the VituixCAD ver of your crossover and then very slightly modified one trying to get the phases to match with my measured data.
I do like the sound potential - that is there for sure... the details of voicing - too little listening still. I will be attempting to make some measurements of the speaker in mid room as well as in listening position to see if the balance seen on screen maps to what it is in room.

[ATTACH=CONFIG]25582[/ATTACH]

My guess is that this will be a bit too bass-heavy. But it might as well be just an issue with near-far merging in simulation. We will see!
This kind of on-stand speaker will suffer from floor reflection cancellation somewhere 200-300Hz typically. Please tell your measuring distance for in-room/listening place measurement!

[ergo]

My guess is that this will be a bit too bass-heavy. But it might as well be just an issue with near-far merging in simulation. We will see!
This kind of on-stand speaker will suffer from floor reflection cancellation somewhere 200-300Hz typically. Please tell your measuring distance for in-room/listening place measurement!

Room measurements done at 2.5m. Speakers sit about 20cm off wall and also listening position is only some 40cm from back wall. The setup is along the long wall while I know that it would be much more desirable to put it along shot wall and much further from back wall.

[ergo]

Ok - I've gone to use "bigger guns" - ie replaced ARTA based setup with CLIO 11 FW. The only negative with CLIO is that I can't get it to behave together with my laptop running win 10 and PC-card FW interface.
It does work on my HP desktop PC which also runs win 10..... anyhow, due to this I had to do measurements in my listening room instead of a bigger family room. The gating needed to be 4.1ms due to that

I think levels wise this setup is getting it very close. There is still a discrepancy especially in 500...3k range even though this time I did not use any protection cap.

My measurements are with
ARTA + GRAS 40AM mic
CLIO + GRAS 40AM mic
CLIO + LinearX M53 mic
With M53 I'm pretty confident that CLIO has a bug and never applies the mic cal file. I tried all tricks and copied the MICA.cal file to the Clio.exe folder etc - I never see any change. The plot also indicates that mic cal has not been enabled.

Still my two mics give resonable good match - so the conclusion I guess is that the baffle shape and maybe a bit of unit to unit variance must be the cause here.

[ergo]

I also used CLIO-s sinusoidal measurement for low frequency (no gating there)
I measured woofer
1) mid room - mic at 1m
2) left speaker position - mic at 1m
3) right speaker position - mic at 1m
Then averaged these 3 LF measurements and tuned the far field / near field merging level based on this. So hoping it is getting closer to truth now

After all the hassle the match between Jons V1.3 cross + my Clio data looks already very close to Jons simulation

https://onedrive.live.com/redir?resid=73B3DF257C874FF5!16208&authkey=!AMU3mh4MSxcz4Bw&ithint=file%2czip

[sdl2112]

Making more sawdust... or router dust, as the case may be!

Just doing all this computer stuff gets on my nerves after a while... if there isn't some sawdust involved somewhere sometime, something is wrong....

Today's task was to trim the second side panel, follower trim it, bevel, and so some overall sanding, just enough to get it ready for handling and test, not for finishing at this time.

The little Craftsman 5" battery powered circular saw has been a champ- it came with a very decent thin kerf blade with nice sharp carbide teeth that have stayed sharp, in spite of the material I'm using them on- that's the first step in trimming the side panel, cutting things down to about 1/8" overlap.





Next step is the flush trim, just like laminate, only bigger, using a new Hitachi 2-1/4 HP router I picked up during a binge buying session on power tools after my last medical procedure (we each act out in our own ways...)





The Hitachi is a gem; like the M12V's tucked away in storage in Livermore, they are glass smooth- they make a Porter Cable feel like the bearings are getting close to giving out in comparison. Probably some Samurai craftsman in the background lineage of the production designer, or someone in Hitachi Heavy Industries that is just used to over doing everything...





Speaking of Porter Cable, next up for the bevel edge cut on the side...





You see, this was an immediate A/B comparison, though not a blind comparison- only a fool uses power tools with his eyes covered!


Remember, these are just the 2nd generation acoustical prototypes, and will be finished and used in a utility application such recording or possibly taken into the work lab if we ever get back on that Class D project there. But from horseback, they probably don't look too bad. And they are hella stout! (front to back brace of BB Ply, sides are combo of BB ply and LBL, tops and bottom and fronts are double LBL) .





For now, I'm leaving the cut out plate for the passive radiator in there; it will have the holes for test leads going through it, since we don't really care about the sub 50 Hz stuff for designing the crossover, and I can finalize the connectors and wiring later on.

Very nice Jon. No one would guess those were just prototypes. Nice collection of routers! I've been thinking of getting another but my DW621 does so well at everything but its nearing 20 yrs old. :E

[sdl2112]

Here's a htguide site question.

I noticed when I reply to an embedded message with an image attached within htguide the image is reduced like a thumbnail. This also happened when I replied to a message that had an image hosted by photo-bucket. The last message the images were full size when I replied and they are hosted by flickr. I hate to clutter the thread with redundancy, is there a setting or procedure other then removing the pics....just a curiosity.

Thx,

Scott