Shelving 2nd order high-pass?

[Elbert]

5th,


Based on your experimentation and the mysteriously mising Mark-measurements, It sounds like I shouldn't be overly worried about tweeter distortion and overload here..

Having said that, It would be interresting to know what cone, or dome, displacement those 5-6 W would translate to.

Those VIFA units look interresting and quite simmilar to these: Vifa C20WH

That SEAS CA22 driver looks interresting too.. Hmm.. this is starting to sound like a new project in the making...
But at the end of the day, I'd like to keep and make the best use of my W17CY seas woofers, which are after all some fairly expensive (and hard to get!)SEAS EXCEL units which should have potential for serious performance..

I don't know about SEAS, but my measurements were certainly done as you describe. Rather than trying to "normalize" the results, I decided to leave them as-is.


That DSP of yours.. is it something you are actually building??

 

[5th element]

5th,


Based on your experimentation and the mysteriously mising Mark-measurements, It sounds like I shouldn't be overly worried about tweeter distortion and overload here..

The extra measurements were on his blog, but that went down due to "underemployed mischievous teenagers abroad."

Those VIFA units look interresting and quite simmilar to these: Vifa C20WH

I saw that article myself a while ago now and the drivers look almost identical. You can also see how smooth the breakup region is too, very nice for an 8" driver.

That SEAS CA22 driver looks interesting too.. Hmm.. this is starting to sound like a new project in the making...

I've got around half a dozen 'projects in the making' lurking around in side my head, one day I'll find people to give the finished loudspeakers too

That DSP of yours.. is it something you are actually building??

Indeed I am, it's based around an Analogue Devices ADAU1445 chip and that part of the project is sorted and it works fine. All that's left is the 8 channels of digital to analogue conversion. I've been prototyping a 2 channel version for a few weeks now and I'm having a little bit of trouble squeezing out the last drop of performance from the DAC chip itself. It'll be great once I get the thing completed

 

[Elbert]

5th,

That's really impressive, boilding something like that from scratch!

What do you do with PCB's? I imagine that an ordinary kitchen-table single side etching won't do? And what about software for running the thing?

Those projects lurking arround in the head is dangerous, it is only so easy to get buried under half finished projects and very nice second-hand drivers purchased at the spur of the moment on some good idea!

"Fortunately" I can not find the wave-guides John Krutke used...

It would be really interresting to try those together with a pair of SEAS 8" woofer I have..

 

[Elbert]

Browsing arround John Krutkes blog, I found this statement:

"Make no mistake, a waveguide of only 3 or 4 inches in size is still a waveguide. Waveguides CAN be used below their corner frequency of directivity as long as they are paired with a woofer operating below it's off axis rolloff."

This is taken out of its original context, yet I believe the gnereal nature of the statement to be valid and relevant.

This Effectively says, that if NOT crossed over at the corner frequency, then the x-over must be low enough to "meet" a midrange 7 midwoofer before it starts to roll off..that makes sense.

The other way arround; such a wave-guided tweeter operated at or above corner frequency, must be paired with a midreange/ midwoofer that
has a matching directivity in the same range... doesn't this make sense too?

edit:

As it seems to me, the DXT tweeter can only be operated "in its effective range" by matching it with a very small midwoofer, or a midwoofer that has a beaming/ radiation pattern that converges on that of the DXT arround its corner freuency.

 

[5th element]

That's really impressive, building something like that from scratch!

What do you do with PCB's? I imagine that an ordinary kitchen-table single side etching won't do? And what about software for running the thing?

Well a single side wont do, but a double sided one will, these aren't really any more complicated then a single sided board. I suppose the hard part is refining the process enough so that you can make PCBs capable of working with the 0.5mm pitch SMD parts.

I wrote a wiki article on how to do this, but as the DIYaudio site is screwed at the moment I can't get to it easily

Those projects lurking arround in the head is dangerous, it is only so easy to get buried under half finished projects and very nice second-hand drivers purchased at the spur of the moment on some good idea!

I have a whole shelf and wall closet filled with spur of the moment drivers, I want to insert a rolls eyes smiley but cant >.< They didn't cost much mind you and I'm thinking of selling them now instead of making anything out of them.

Browsing around John Krutkes blog, I found this statement:

"Make no mistake, a waveguide of only 3 or 4 inches in size is still a waveguide. Waveguides CAN be used below their corner frequency of directivity as long as they are paired with a woofer operating below it's off axis rolloff."

This is taken out of its original context, yet I believe the gnereal nature of the statement to be valid and relevant.

This Effectively says, that if NOT crossed over at the corner frequency, then the x-over must be low enough to "meet" a midrange 7 midwoofer before it starts to roll off..that makes sense.

Or rather more accurately, before it starts to beam. This is what I've been saying all along, by using the W17 as low as the DXT will allow. This way the loudspeaker stays omni until the wave-guide on the DXT takes over and the loudspeaker transitions into CD.

The other way around; such a wave-guided tweeter operated at or above corner frequency, must be paired with a midrange/ mid-woofer that
has a matching directivity in the same range... doesn't this make sense too?

edit:

As it seems to me, the DXT tweeter can only be operated "in its effective range" by matching it with a very small mid-woofer, or a mid-woofer that has a beaming/ radiation pattern that converges on that of the DXT around its corner frequency.

Yes, the other way around is to match the directivity of the midrange/mid-bass to the directivity of the wave-guide.

A picture will better illustrate this, my paint skills aren't exactly epic, but I hope you get the idea of what I am trying to say!

With the left hand side part of the image you want the entire omni to CD transition region to occur solely within the DXT and you also want the region below this transition to remain as an omni - that is you don't want any part to have a narrowing of dispersion due to the mid/bass starting to beam.

This is easy to achieve as there is an overlap region between where the mid/bass starts to beam and the lowest xover point the DXT can cope with. This is what I've been saying you should be doing.

The right hand side part of the image, shows the DXT again, but this time below it you've got a small diameter mid/bass. This time round you have to pick the mid/bass and the xover point carefully. I'm sure my representation isn't exactly how it would appear in real life mind you, but the goal is still the same - to cross the two together so you end up with a smooth transition from omni to CD. As you can see, here you cross over after the mid/bass has started to beam and before the tweeter has transitioned to an omni radiator. This is fortunate indeed because it lets you cross the tweeter over higher, relaxing the load it has to handle, and it lets you use the mid/bass into the region where it starts to beam, relaxing the need for a mid/bass that is omni up until a bit before the xover point too.

I will also add in here that when using the small diameter mid/bass with the DXT there is nothing stopping you from choosing a lower crossover point. This will push the xover into the range where the DXT and the mid/bass are both acting as an omni and you'll end up with a speaker similar to the DXT/W17 implementation. While this sounds nice, if you've got a 15" driver mated to a tweeter with a 15" wave-guide, the tweeter wont go down low enough to let you do this, you are literally forced into matching the directivity of the two together.

 

[Elbert]

Still, Double sided DIY prints impress me, damned if I'd be able to get any trough holes to line up on both sides! *grin*

Yes, and what happens when you clear out that lot of drivers? Some weak soul like me will give in and buy it of you! : ) And a few weeks later, the same will happen to you...


A very good and illustrative description of the "problem" you made there!

This allso illustrates some assumptions we are now working from, or at least, I'm working from. I'm assuming that I can actually get away with what you illustrated to the left, (the example for x-ing over to a smaller dia. woofer), but with the W17cy. This is of cource assuming that the dispersion pattern is beaming in such a way that it will "intercept and converge" with the controlled directivity of the DXT at some point. I think that the off-axis plots I did with the 3200 Hz x-over points to this being feasible.

Your example of the 15" scenario is interresting, it clearly demonstrates a situation that doesn't afford extending a waveguide down to get an "omni-overlap". But provided it is possible to overlap in the non-omni range, so to speak, one will end up with less omni and more constant direction.. which I've obviously come to believe is a good thing by now!

Besides this, I'm sarting to itch for designing and building some new speakers with those SEAS 8" woofers and some 6 or 8" wave-guides.. Think I've been bitten by the CD/ waveguide bug here!

 

[5th element]

Still, Double sided DIY prints impress me, damned if I'd be able to get any trough holes to line up on both sides! *grin*

When using the UV transfer method this isn't as hard as you think. With other methods though it would be pretty much impossible, unless you were working with very large parts.

This also illustrates some assumptions we are now working from, or at least, I'm working from. I'm assuming that I can actually get away with what you illustrated to the left, (the example for x-ing over to a smaller dia. woofer), but with the W17cy.

The image on the left was specifically for the W17, showing that you can easily cross over in the region where both drivers are acting as an omni radiator.

The 'Region where the DXT operates as an omni' is supposed to be from about 1500-3000hz.

The W17 acts as an omni up to around 1800hz, so you want to cross over between 1500 and 1800hz..

This is of course assuming that the dispersion pattern is beaming in such a way that it will "intercept and converge" with the controlled directivity of the DXT at some point.

No, it isn't, there is no converging point between the two drivers, that is there is no controlled matching of directivity. Both drivers should be acting as an omnidirectional sound source at the frequency you choose to cross them to one another, hence the 1500-1800hz xover frequency. This is point Zaph was making when he said this.

Make no mistake, a wave-guide of only 3 or 4 inches in size is still a wave-guide. Wave-guides CAN be used below their corner frequency of directivity as long as they are paired with a woofer operating below it's off axis rolloff.

The corner frequency is when the wave-guide tweeter is starting towards constant directivity. A bit below the corner frequency and the wave-guide is acting as an omnidirectional sound source - this is the DXT how we want to use it, it's corner frequency is around 3.5khz and we want to cross below this.

As long as they are paired with a woofer operating below it's off axis roll off.

Off axis roll off is the mid/bass roll off due to beaming. As we've seen with the W17, this extends to around 1800hz before off axis beaming becomes an issue, hence, as Zaph said, we cross over below this point.

So our overlap with the DXT and the W17 is between the lowest point you'd want to take the DXT to, 1500hz, and the highest point you could take the W17 to, that would be 1800hz.

If we were using a 5.5" driver instead of a 6.5" W17, the 1500hz low point of the DXT remains the same, but now, because a smaller driver starts beaming at a higher frequency we would extend the upper xover point to around 2200hz.

Crossing the DXT to the W17 in the 1500hz-1800hz region, where both are operating as an omnidirectional sound source, means that the entire loudspeaker will remain omnidirectional up until the corner frequency of the DXT wave-guide, where it then starts to control directivity.

Your example of the 15" scenario is interesting, it clearly demonstrates a situation that doesn't afford extending a wave-guide down to get an "omni-overlap". But provided it is possible to overlap in the non-omni range, so to speak, one will end up with less omni and more constant direction.. which I've obviously come to believe is a good thing by now!

This isn't quite what I was trying to get at. The 15" wave-guide loaded tweeter will probably not operate satisfactorily in it's omni region and even if it did this would be at say 800hz. A 15" mid/bass driver starts to beam around 400-500hz, it cannot go high enough to mate with the wave-guide in it's omni region, neither can the wave-guide tweeter make it down low enough to mate with the 15" mid/bass (400hz). In this instance you mix the two together, this is where you are actively 'matching' the directivity of the two drivers. At 1000hz the wave-guide is tending towards constant directivity, but it isn't all the way there yet and although the 15" driver is beaming quite a bit by 1000hz it's still got quite a lot of juice left in the off axis plots. Here you want the reducing off axis response of the 15" mid/bass to match up to the reducing off axis response of the wave-guide as it transitions into constant directivity. This is what matching the directivity of the mid/bass to the wave-guide means. It is important to handle it this way with a large loudspeaker because there are direct gains to be had. 1) it lowers the demands of the tweeter and 2) it allows you to use a mid/bass of the same diameter as the wave-guide, instead of a 15" wave-guide and 10-12" mid/bass.

With the DXT this directivity matching would occur at around 3500hz and thus you'd want to use a 3-4" mid/bass driver for the directivity to match. With the DXT though, because the wave-guide is so small, you can easily use it below its region of constant directivity, allowing it to mate satisfactorily with any mid/bass no bigger then a 6.5" providing you cross over before pattern control starts in the DXT and that the mid/bass is crossed over before it starts to show signs of beaming.

Besides this, I'm starting to itch for designing and building some new speakers with those SEAS 8" woofers and some 6 or 8" wave-guides.. Think I've been bitten by the CD/ wave-guide bug here!

I can only agree with you on this. I'd like to get my hands on a ~5" wave guide too! Maybe I could try making one, although I am not really sure how to go about that. Come on SEAS, give us an Excel wave-guide tweeter in a 5" chassis that's designed specifically to mate with your 5" excel midrange drivers.

 

[5th element]

And what about software for running the thing?

The software for running the DSP is a two part thing. The first part is a program made by Analogue Devices called Sigma Studio, this lets you implement crossovers graphically, rather then knowing how to program bi-quads etc. After this Sigma will export a series of arrays that contain the data required to program the registers within the DSP.

The next stage uses a microcontroller to program the DSP chip via I2C, this is the part I've implemented myself. Writing the code is pretty simple, all it is, is sending different register changes over the I2C protocol and any bulk transfers are handled via the arrays. All the register changes are created by SigmaStudio, so it's basically an elaborate copy and paste thing.

For example though, to configure the 'program data' of the DSP requires you write 2000+ numbers to the DSP control registers. This would be tedious and take forever if you had to code each number manually as each send requires three lines of code. Instead you point the code towards an array of data that Sigma studio created and as the DSP knows to increment the control register by 1 each time you simply send it the start register then a stream of 2000+ numbers. Still the entire code sequence is 500 lines of code long, most of this is copy/paste and is also a one off thing.

Having writen the code if I now want to change the DSP (for example change from one xover to another) all I have to do is export a new set of files from SigmaStudio, copy these 4 files into the specified folder, recompile the code and reprogram the micro controller, it takes about 1 minute to do all of this, if the required programs are already open.

The other great thing (I haven't done this yet as I want to get the DACs finished) is that I can implement two different xovers in the code and swap between them at the command of a remote control, making it very easy to A/B different designs. As you can imagine I am looking forward to getting it all finished

 

[Elbert]

Again, most impressive 5th!


I can in deed se why you are looking forwards to getting this completed ,the advantages this has as a development tool is obvious! Heck, I'f Im unlucky just getting hold of that non-standard resistor value I usually need for a filter change can set me back several days!

 

[5th element]

Tell me about it I placed an order with farnell for a new set of components to change around the old active xover. I realised a few hours later that I'd miscalculated something, so the next day I had to quickly phone them up and alter the order >.< This is why I chose to use trim pots in the majority of the active xover boards that I had built before.

 

[Elbert]

Finally back!

Yes, been there plenty of times my self, just ordered some parts and realizing I did something wrong just after!

Since last, I've made two very interresting discoveries.. first of all, the L-R balance from the pre-amp was not entirely balanced with the result that the right channel was a bit louder, and the sub is on the left channel (haven't completed the sub for the right channel yet). When this became obvious, it was further obvious why the tonal balance seemed overly bright!

Removed the pre-amp from the set up to check it... and it measured fine!

Put it back and the bvalance was OK, even measured level at the speakers! The only explanation i have for this is a poor contact somewhere...

Further to this, I allso tried to swap the polarity on the sub.. suddenly the sound was much fuller and richer.

I decided to measure this to see if the results correlated with my listening experience.. right so..

Attached is the in room rwa response 1/5th oct smoothed, roughly at the listening position. Red is with phase reversed.

Seems I had an out-of phase situation in the x-over area resulting in a pronounced high-bass dip!

I allso included a phase plot of the same measurement.

The 50 Hz dip is a room mode artefact and the dip around 200 Hz should hopefully be possible to sort out by experimenting with speaker room-placement. The droop in highs can in part be explained by the off axis situation at the listening position, absorbtion from carpet etc..

So 5th, you were in deed on to something when you suggested looking at baffle step allthough it turned out the issue was caused by something else!

Based on these findings, I think I will suspend any further tinkering with midwoofer/ tweeter x-over point, at least untill I've completed the second subwoofer and optimized speaker placement!

 

[5th element]

It's nice to see good phase integration between the subs, even with the smoothing applied it still stands out well. And stand out it does all the way up to around 150hz! It's not a surprise that some warmth was lacking from the sound. Building the second sub should also give you the benefit of loading the room from two separate locations, so it should help even more in smoothing out the lower region.

The integration between the mid and the tweeter has already been shown to be excellent, both in simulation and in measured response, so I don't think there's too much going on there that could cause the large kind of tonal shift you were seeming to have. Usually very small changes in say the 1-3khz region wouldn't be directly noticed upon immediate A/B testing, more the loudspeaker with the more depressed mids would seem a bit easier to listen to over an extended period and neither would seem 'wrong'.

Channel balance is always something that's worth checking, it's something we take for granted too, so we're less likely to expect it to be wrong. It is also worth perhaps checking the response of the active xover through ARTA. Measure the left channel then measure the right, just to make sure that everything is okay. Generally things will be fine, but sometimes irregularities can show up. This is especially useful when you've just made some adjustments to the active xover board. Occasionally you'll get something wrong that would cause something frightful to spring forth from the loudspeakers, scaring both yourself and potentially damaging a driver. I leant my lesson and now I always measure the output of any active xover before hooking it up to an amplifier + loudspeaker. It takes less then a minute to measure all 8 channels so it's worth doing.

You mention that the pre amp wasn't quite balanced, this could be an issue with the main potentiometer, sometimes the tracks aren't as balanced as you'd like them to be, did the balance change much with the level of the volume control? It'd be worth hooking the preamp up to ARTA and see what's going on.

Over the next few weeks I too will be hopefully on the build, we just fetched 2x 2400x1200x25mm and 1x 2400x1200x18 sheets of MDF. That's a lot of MDF I tell you, never had so much the house. The first job is making 4 x 30cm cubes for 4x Peerless 10" XLS drivers. I've had these in a H baffle for a long while now and I want to give multiple subs with them a try instead. Should be interesting to say the least.

On larger wave-guides, I did a little bit of looking around and found the monacor WG-300, this appears to be easily available, cheap and possibly easy to mate with a variety of tweeters. This website also has some interesting measurements of various common tweeters in different wave-guides. The SEAS27TBFCG mated rather well with it, as did the XT25. I am hoping that it might mate fairly well pressed up to the metal cover on the DXT too. I have a pair of XT25s I could try with them also, size wise it'd match up decently well with 5 and 6" mid/basses. Another project on the horizon perhaps?

 

[Elbert]

Good evening 5th!

So you think the phase integration after polarity switch on the sub looks OK? Well, good news for me then! Again, I'm lacking the experience to make the best judgement of the plots here.

The sound is surprisingly nice with one sub only, especially considering that my listening room is not the best, so I really look forwards to getting that second sub up and running!

Wow, that's a lot of MDF, sounds like you'll be doing a fair bit of cutting and joining in the near future!

It will be very interresting to see (hear) how four subs work out! from a modal distribution point of view, more should, as you say, be better, so it will be interresting to see how it works out in practice!

I did measure both filter cards in-amp and through speaker, and the plots came out remarkably simmilar, so everything should be OK in that department.

And with regards to balance, yes, you are in deed right, assumption is the mother of all F... ups as the saying goes!

The channel level didn't vary too much for different settings, actually less than I expected, especially after what I read in Douglas Self's "Small signal audio design". At least not enough to be a big issue dB wise..

Thanks a lot for the wave-guide link and the tip on available monacor wave-guides!

Allthough this will probably not be for the current project, I think it is inevitable that I will pursue this further in a future project. I find the concept of CD vs room acoustics problems very interresting.

Speaking of.. when I was out travelling, I dropped by a store to pick up some CD's (obsolete medium for storing and playing back recorded music), and there were a couple of loudspeakers playing and sounding quite nice as well. Not that I got a proper listen as this was in a bussy shop..

Behind the grille-cloth I did notice something interresting though, so I i googled the brand: Esprit Altea speaker by Triangle

They have a horn tweeter! Not often I see that in Hi-fi speakers..

I wonder what the off axis response looks like on those?

EDIT: Apparently, the same tweeter is x-ed over at 2800Hz according to the data sheets of the larger models.

 

[5th element]

Good evening Elbert

Speaking of.. when I was out travelling, I dropped by a store to pick up some CD's (obsolete medium for storing and playing back recorded music), and there were a couple of loudspeakers playing and sounding quite nice as well. Not that I got a proper listen as this was in a bussy shop..

Behind the grille-cloth I did notice something interresting though, so I i googled the brand: Esprit Altea speaker by Triangle

They have a horn tweeter! Not often I see that in Hi-fi speakers..

I wonder what the off axis response looks like on those?

EDIT: Apparently, the same tweeter is x-ed over at 2800Hz according to the data sheets of the larger models.

Triangle have been using horns for a long time, I am quite surprised that a music shop would have such costly speakers in them. Certainly you'd think that having high quality music playing, in a music store, would be a good thing, never encountered it here in Manchester though

Stereophile has two reviews of triangle speakers, both that feature the horn tweeter. The first is a flagship design that are a few years old now and are rather unique in their construction. The second is the younger sibling of the loudspeaker you heard in the music store. The off axis seems flatter with their horn, out to a few degrees, then a normal 1" soft dome, although it doesn't hold up as well as DXT.

So you think the phase integration after polarity switch on the sub looks OK? Well, good news for me then! Again, I'm lacking the experience to make the best judgement of the plots here.

You're getting a nice 15dB null at what I presume is roughly the target of the xover between the sub and mains - around 100hz. Measuring this low down in a room isn't an easy thing to. Room reflections will often help to mask a reverse null, smoothing also acts to do the same thing, so the fact you've got a nice symmetrical looking null in roughly the right place is a very good thing indeed.

The sound is surprisingly nice with one sub only, especially considering that my listening room is not the best, so I really look forwards to getting that second sub up and running!

Well most of the work is being done by the mains Most people would say they go down far enough on their own to be largely acceptable. I too have often worked with only a single channel of bass going etc and as long as the level of the sub is adjusted to the fact you're only going on half output, the general listening experience has always been more then enjoyable.

Wow, that's a lot of MDF, sounds like you'll be doing a fair bit of cutting and joining in the near future!

It will be very interresting to see (hear) how four subs work out! from a modal distribution point of view, more should, as you say, be better, so it will be interresting to see how it works out in practice!

Yes Geddes multiple sub idea piqued my interest wel enough that I have been wanting to try this out for a while now. Of course the DSP is vital to system integration, so I need to get that fully functional before any listening can be done. Of course I am quite interested in seeing what the measured performance of a single XLS in a small cabinet is like and I can evaluate that without the DSP.

Thanks a lot for the wave-guide link and the tip on available monacor wave-guides!

Allthough this will probably not be for the current project, I think it is inevitable that I will pursue this further in a future project. I find the concept of CD vs room acoustics problems very interresting.

Indeed me too. In a previous loudspeaker I had swapped between an XT19, XT25 and a standard 1" dome. The XT19 has the best off-axis response, followed by the 1" dome then the XT25. The sound was far more enjoyable with the XT25, which I presume is because is puts out far less high frequency energy into the room.

The XT19 actually sounded like it was spitting loads more high frequency energy into the room, this didn't sound 'bright' it just sounded dirty/grainy/blurry like many reflections were bouncing around. This wasn't anything to do with distortion/phase/level matching as these we're measured for and all adjusted per tweeter. It makes me wonder what the monacor wave-guide might sound like as the DXT pushed the sound into an even better place then the XT25. I also wondered if I was imagining the differences so I swapped between the different tweeters a number of times and as soon as the music started with the XT19 I wanted to turn it off! Gimme the XT25 back.

Monacor themselves have designed the WG-300 to mate directly with one of their tweeters. It would be great if SEAS did a similar thing, have a series of wave-guides say from 4" to 8" that would mate well with a selection of their tweeters. The 27 series of tweeters in the prestige range all appear to have very similar arrangements of dome/surround/faceplate construction, you'd figure it'd be quite easy for them to come up with a decent selection of wave-guides for them.

EDIT - I thought I'd just add in, the triangle loudspeaker exhibits the usual off axis dip around the crossover frequency due to the mid/bass beaming. So certainly an omnidirectional response followed by a region of CD wasn't on their agenda. It always surprises me when manufactures use mid/bass drivers up so high in frequency. It is not hard nor expensive to cross to and to get a tweeter capable of going down to below 2000hz.

 

[Elbert]

I'd never heard about triangle before, so my firsrt assumption was that this was some obscure cheapo brand, allthough that didn't quite match my immediate listening impression. Now I do know better of course, semms to be quite serious stuff.

They were lined up in a big media shop in France, mostly allong with very ordinaryu consumer stuff.

Anyway, thanks a lot for the links to the tests, especially the one that had off axis measurements! Evidently, there is a dip/ glitch there at the X-over frequency, the sort of thing that was expected when I ran the higher XO on my speakers,, must have been lucky there! Perhaps the wave guide was just that much wider or the midbass did just not beam to much..

But even that small off axis dip on the triangles can't be that bad, wouldn't doo if such uppmarket speakers didn't sound good..? Too bad the off-axis plot was onli out to +/- 15 degrees.. Perhaps there is a trade-off between CD and that dip which works out favourable at the end of the day?

My subwoofer x-over is att arround 65 Hz, but then again, as you say, difficult to measure in this regioan as the room happily chimes in with lots of artefacts. I haven't moved the mic all over the place, but to my ears, it seems the improvement after swapping the polarity is fairly consistent throughout the room. This as opposed to the 50 Hz dip which is undoubtedly a strong room mode, if i stand in one corner i get totally 50Hz Oomphed!

I haven't adjusted the sub to compensate for half output, the more the reason for completing the other one!

For your sub project, will you be adjusting delay/ phase for all four subs individually??

I guess I'll have to make do with two for the time being, after all, being 140 L closed boxes with 15" dayton woffers, it's completely overkill for my smallish apartment as it is, but oh boy what a deep and wonderfull bass, even from just one!

If I remember correctly, John Krutke was favourable to the XT25 as its phase-plug apparently improved top octave matters in larger wave-guides..

The easy route would obviously be to go for matching monacor waveguide and tweeter, but personally I must admitt I have a soft spot for SEAS products, and has anybody seen the monacor factory, or is it just a "brand name"?

BTW, what do you make of those oval 18Sound wave guides? To me the CD looked quite impressive, perhaps more so than for the round wave-guides?

EDIT:

My stupid, I was looking at the vertical plot.. Yes, the dip is visible on the horizontal plot too bout only relatively far out to the side where lots of "other" things are happening too.

 

[5th element]

Anyway, thanks a lot for the links to the tests, especially the one that had off axis measurements! Evidently, there is a dip/ glitch there at the X-over frequency, the sort of thing that was expected when I ran the higher XO on my speakers,, must have been lucky there! Perhaps the wave guide was just that much wider or the midbass did just not beam to much..

I have to say I was quite surprised that your 3.2khz xover showed virtually no signs of dip through the xover region. Maybe something would have shown up if you'd done measurements out to 90 degrees in 5 degree increments.

But even that small off axis dip on the triangles can't be that bad, wouldn't doo if such uppmarket speakers didn't sound good..? Too bad the off-axis plot was onli out to +/- 15 degrees.. Perhaps there is a trade-off between CD and that dip which works out favourable at the end of the day?

Are we looking at the same plots? The off axis measurements of the triangle loudspeakers go out to 90 degrees laterally.

if i stand in one corner i get totally 50Hz Oomphed!

Gotta love room modes eh?

For your sub project, will you be adjusting delay/ phase for all four subs individually??

Currently I only have one amplifier channel built as there are some design kinks with it that need to be sorted out. This will be seen to after the two current projects get finished. So at the moment I only have 1 amp for 4 drivers. The DSP also only has 2 channels for the bass, so at the most I can have 2 drivers per amp and 2 channels of DSP. This isn't ideal and I have thought about building 4 amplifier channels and adding another two channels of DSP if it's necessary, that will then allow for a lot of flexibility towards integrating the subs.

I guess I'll have to make do with two for the time being, after all, being 140 L closed boxes with 15" dayton woffers, it's completely overkill for my smallish apartment as it is, but oh boy what a deep and wonderfull bass, even from just one!

140L It makes the little XLS enclosures that I'm building seems positively tiny! Here's the first one I made yesterday.

Of course these need a large amount of EQ applied to get a decent bass response out of them.

This is why 4 of them make sense. According to simulations, 4 of them with room gain and boundary reinforcement should net around 105dB @ 20hz just meeting the THX specification, not that I watch many movies mind you

If I remember correctly, John Krutke was favourable to the XT25 as its phase-plug apparently improved top octave matters in larger wave-guides..

Indeed, the XT25 gave the smoothest response when interfaced with the monacor wave-guide. It'll be interesting to eventually try them out some time.

The easy route would obviously be to go for matching monacor waveguide and tweeter, but personally I must admitt I have a soft spot for SEAS products, and has anybody seen the monacor factory, or is it just a "brand name"?

Monacor have been around for many many years, I think wikipedia said since the 60s although I as I can't read anything other then english I am only guessing here.

BTW, what do you make of those oval 18Sound wave guides? To me the CD looked quite impressive, perhaps more so than for the round wave-guides?

EDIT:

My stupid, I was looking at the vertical plot.. Yes, the dip is visible on the horizontal plot too bout only relatively far out to the side where lots of "other" things are happening too.

I do not know whether it is strictly important in keeping the vertical off axis response identical to the lateral. You can see in 18Sounds datasheet that the horns control of directivity is different for the vertical and lateral off axis. Lateral having a decent stab at CD beyond about 1500hz, with the vertical only doing so from about 2.3k. Personally I'd rather have a circular WG and CD down to 1500hz from all angles. After all it is only 8" wide in its largest dimension. I guess for outdoor venues the vertical axis upwards = the sky, so it's less important and below you've usually got people and grass/chairs etc, which'd be pretty good at scattering any reflections.

 

[Elbert]

Ah, back from the annual 1st of May motorbike trip to the continent!

Good point about looking in to what goes on off axis even out to 90 degrees! The 30 or so degrees is probably the most interresting area with respect to the balance of the direct sound hitting the listening position. beyond that, the sound will progressively contriburte more to the diffuse/ reflected sound-field with increasing angle, which is at the end of the day quite important in a domestic listening environment.

I'll definitively extend my off axis measurements on the next occasion!

I guess if you first manage building one DSP, then simply building as many copies as you need is no problem? As you say, that should make for some impressive sub-integration possibilities!

Nice little subwoofer there! handy size if you have four of them to distribute arround the room! Actually, your calculated/ expected in-room response is not to far off from what I get with two of my "monster" subs..

The reason I'm "questioning" Monacor a bit, is that their driver line-up seem a bit to varied to come from the same factory, and I remember seing some drivers which I have allso seen elsewhere under a different brand name or no particular brand name at all.. you don't see that with Seas or ScanSpeak..

The "eliptical" disperison of the 18Sound horn is perhaps an advantage for exactly the same reason you stated.. just replace sky with reflective ceiling and crowd with floor.. From this, I think this one actually looks more interresting than a spherical wave-guide. Perhaps the narrower vertical pattern will allso reduce rhe magnitude of comb-filtering/ phase issues between midwoofer and tweeter?

 

[5th element]

Ah, back from the annual 1st of May motorbike trip to the continent!

Good point about looking in to what goes on off axis even out to 90 degrees! The 30 or so degrees is probably the most interresting area with respect to the balance of the direct sound hitting the listening position. beyond that, the sound will progressively contriburte more to the diffuse/ reflected sound-field with increasing angle, which is at the end of the day quite important in a domestic listening environment.

Indeed it might also be useful to do more measurements within the 0-30 degree range, say in 5 degree increments, just to see if anything pops up. Nothing should mind you, but if you've got the equipment setup for it, might as well spend a little bit more time making measurements.

I guess if you first manage building one DSP, then simply building as many copies as you need is no problem? As you say, that should make for some impressive sub-integration possibilities!

Well the DSP side of things is all sorted, I've built it with the capability to handle 16 channels. The only limit will be on the number of D/A channels I have. First off it'll be 8 channels, 6 for the tweeters, upper mids and lower mids, although the lower mids are bass drivers in their own right and will extend down to around 60hz without much issue. Originally the two remaining channels were to be used for stereo bass, although I am thinking of summing the stereo signal into a mono one and having two different xovers for the different subs. Adding a 'cheap' 8 channel DAC isn't hard either if I want to have a different filter for each sub. Currently I only have one amplifier channel for the subs! But that can easily be extended too.

Nice little subwoofer there! handy size if you have four of them to distribute arround the room! Actually, your calculated/ expected in-room response is not to far off from what I get with two of my "monster" subs..

Yeah you just need the amplifier power to be able to do it.

The reason I'm "questioning" Monacor a bit, is that their driver line-up seem a bit to varied to come from the same factory, and I remember seing some drivers which I have allso seen elsewhere under a different brand name or no particular brand name at all.. you don't see that with Seas or ScanSpeak..

We don't see that from SEAS or SS maybe because they don't want us to see? Monacor produce a lot of low cost drivers for cheap boom boxes and TVs etc so that increases the catalogue quite a bit. I am not saying that SEAS and Scan do that too, but I am sure they make far more OEM variants then we ever get to see in their standard line up and more's the shame!

The "eliptical" disperison of the 18Sound horn is perhaps an advantage for exactly the same reason you stated.. just replace sky with reflective ceiling and crowd with floor.. From this, I think this one actually looks more interresting than a spherical wave-guide. Perhaps the narrower vertical pattern will allso reduce rhe magnitude of comb-filtering/ phase issues between midwoofer and tweeter?

You'd have a choice with the 18Sound driver if you want. Mount it horizontally for better lateral pattern control, or mount it vertically if you prefer to reduce the ceiling reflections.

Comb filtering shouldn't really be changed because the woofer should be greatly attenuated by the time any pattern control comes in from the tweeter. However if you were forced to cross over higher then you'd like, it might help a little. Comb filtering doesn't appear in the lateral off axis either, it's only the vertical, unless you've got horizontally arranged drivers.

I placed an order for two of those wave-guides too, at some point in the not too distant future I'll try producing a comparison to the DXT, hopefully using the same type of graphs that Geddes uses.

 

[Elbert]

Hi 5th!

A very good point, next time arround I rig things up for doing measurements, it wouldn't cost me much to do some extra measurements in smaller increments.

Like you point out, I don't expect to see any "wild" artefacts between the angles I've measured at so far, but a better resolution can perhaps be usefull to reveal some trend or tendency.. At the end of the day, I think extending the measurements out to 90 degress might prove the more interresting advance as this should show how the speaker really behaves as directivity issues really start to make them selves noticed!

Congrats on finishing your DSP! 16 channels?? wow, that should allow you to do allmost anything! Now you only need a true anechoic test-lab and calibrated Bruel & kjaer microphones to get the most out of it!

Speaking about microphones.. Just for the sake of it, I did a measurement without the calibration file I've been using so far for my DIY mic.. and that yelded a measurement curve without the dip/ artefacts I've been scratching my head over for the top octave.. Perhaps the calibration file isn't right? I'm suddenly having doubhts wether the calibration file was for the Linkwitz mod & amplifier job I have or the "plain" WM61 mic capsule.. hmm. must re-investigate..

On the topic of summing the sub input to Mono, that's an interresting one..
The main rationale for justifying this is the fact that stereo location is non existent at susch low frequencies, and that most LF info is ixed in mono anyway..

The first rationale is a matter of scientifically established physiology and psycho acoustics.

For the second one, back in the days of the analogue long-play record format, mixing LF in mono was dictated by the physical limitations of disc-cutting and playback needle tracking. I wonder what the practice actually is today nd what a non-mono LF mix would mena for stereo VS mono signal sub set up?

With the flexibility you have for future experimentation and four separate sub's, I guess we can expect some interresting answers in the future!

I actually had the honour/ pleasure of visiting SEAS a good few years back. In order to stay competitive, they did source quite a few of their components, some undoubtedly from "low-cost" countries, but everything was per SEAS design and specification, and all final assembly and quality controll was done in-house, certainly no bought-in and re-branded drivers.

Back in the "old days" the norwegian SEAS distributor would every now and then have some special deals on what was apparently OEM surpluss drivers from SEAS. regrettably, this never happens any more. I remember picking up a pair of 6,5" coax woofers with carbon fibre cones for about 35£ each! Ironically, these drives ended up in one of my cars, bu I have a feeling they will come out one day and be used in a way that will do them a bit more justice...

Have you ordered a pair of those oval 18sound waveguides??

If so, it will be very interresting to hear what your findings will be!

 

[5th element]

Hi 5th!



Congrats on finishing your DSP! 16 channels?? wow, that should allow you to do allmost anything! Now you only need a true anechoic test-lab and calibrated Bruel & kjaer microphones to get the most out of it!

There's a full anechoic chamber 15 minutes away by train at a local university. I've been inside it too I took the acoustics course there for a year, a few years ago now. When I say I've finished the DSP I think you misunderstand me, the PCB, with the DSP chip is finished, it has been for a while, I've been using it simply to pass unfiltered stereo. The DSP chip is fully functional, I've had it doing filters etc, what isn't finished is the 8 channels of D/A conversion.

The DSP can accept (per my configuration) 4 stereo I2S channels. This was done so that it can extend to do 7.1 at some point if I want to go that route. I have the ability to go up to a maximum of 24 output channels if I use a DAC that is compatible with TDM, most multi channel DACs are, but their performance suffers compared to the premium D/A converters out there. This isn't really a problem, as I'd use PCM1792s or equivalents for any critical channels, such as the main tweeter/mid/bass and perhaps a centre channel, everything else could be handled by a multichannel DAC. This is way off in the future though and will maybe happen (who am I kidding it will happen) once I have everything currently on my plate sorted out.

Speaking about microphones.. Just for the sake of it, I did a measurement without the calibration file .. Perhaps the calibration file isn't right?

I'd always be wary of any generic calibration files. Small deviations in a response would require an individual calibration per mic capsule and possibly recalibration every couple of years.

I think I got relatively lucky with my ECM8000, as it appears to measure pretty much flat, or at least flat enough to get a really decent job done. I haven't had to second guess it either, it always sounds like it measures.

For the second one, back in the days of the analogue long-play record format, mixing LF in mono was dictated by the physical limitations of disc-cutting and playback needle tracking. I wonder what the practice actually is today and what a non-mono LF mix would mena for stereo VS mono signal sub set up?

There will be some difference between the left and right channel I'd expect in some music, as to the audibility of this, I have a strong suspicion that it's probably undetectable. People tend to report back on the benefits of using stereo subs too, but then this could easily be explained by giving a more even in room bass, rather giving any real stereo perception to the bass itself.

With the flexibility you have for future experimentation and four separate sub's, I guess we can expect some interresting answers in the future!

I'll have to see how this pans out, I don't, however, expect to see anything that differs really from what's already been documented on the subject.

I actually had the honour/ pleasure of visiting SEAS a good few years back. In order to stay competitive, they did source quite a few of their components, some undoubtedly from "low-cost" countries, but everything was per SEAS design and specification, and all final assembly and quality controll was done in-house, certainly no bought-in and re-branded drivers.

Honour is the right word! I think a lot of people would love to look around SEAS/SS etc

Back in the "old days" the norwegian SEAS distributor would every now and then have some special deals on what was apparently OEM surpluss drivers from SEAS. regrettably, this never happens any more. I remember picking up a pair of 6,5" coax woofers with carbon fibre cones for about 35£ each! Ironically, these drives ended up in one of my cars, bu I have a feeling they will come out one day and be used in a way that will do them a bit more justice...

This does happen more often then you'd think, only not towards us quite as much. Madisound operates as the sole distributor for some of the Nordic loudspeaker manufactures and frequently it will have cheap deals on extra OEM stock. The last ones that I can remember were a buyout on a black coned and black phase plug variant of the W15CH001, as far as I know it's specs were otherwise the same. Also there was a buyout on some revelator 10" bass units with an aluminium cone, these were also 12 ohm if I remember right.

I do think that if there were any real performance gains to be had, by using different cones/motor combinations etc, that we'd see them in the standard line up. The only thing that I personally want is SEAS to release an excel midrange driver with a magnesium cone.

Have you ordered a pair of those oval 18sound waveguides??

If so, it will be very interresting to hear what your findings will be!

No just the monacor WG300. This is almost perfect diameter wise to mate with the W15CY I use as the upper midrange. Another benefit is it's cheap and doesn't appear to require too much modification to work with normal dome tweeters. In fact it seems that the XT25 will drop in without any modifications necessary at all to either of them. That combo will most likely be my first test, that's if I can get the faceplate off of the XT without too much hassle, Vifa have glued some of it together which could complicate things somewhat, they didn't use to do this so it's annoying that they changed that. I don't relish the thought of destroying a tweeter, but as tweeters go the XT isn't too expensive.