Jump to content

The ultimate small speaker - final design peer review thread


lowerFE

Recommended Posts

Probably not, or at least, not enough to make it worth the effort.  Part of the trouble is that the filters you need to separate the high and low contribute group delay themselves, and the summed response has some non-minimum phase group delay character as well.

 

I spent some time trying, and had all the problems you mentioned. It is extremely sensitive to the amount of delay I apply, and the resulting FR is quite different with different delays in increments of 1ms. I got something reasonable after an hour or so of playing around. The group delay around tuning was significantly reduced, but I wouldn't call it optimal. The FR curves down a bit, probably due to some cancellation from phase differences, and I'm not sure if that reduces maximum clean output. The group delay is measured on my desk, nearfield. Not sure if that's valid for measuring group delay. I also only played with IIR filters and did not attempt to use FIR filters for the crossover or any attempt at correcting phase. A quick A/B revealed no sound quality improvements, but it might be because my new room is quite bad for bass. I haven't figured out whether I have a huge 50-80Hz null or I have 2 huge 45 and 90Hz room modes. 

 

I will investigate this further at a later time as it seems like it could be promising. Right now still focused on building the speaker. The enclosure proved far more difficult and time consuming to build than expected because of accuracy issues when squeezing every bit of space possible. Looking at options to CNC the enclosure now or make it with fibreglass. 

Link to comment
Share on other sites

As I am working on tuning this speaker on the DSP and making it linear phase, I was reminded of a problem that I will have horrible group delay in the bass region because of the combination of a vented enclosure and massive shelf boost to flatten the bass response. 

 

I remember for my previous speaker, group delay exceeded 2 cycles below a certain frequency, let's say 60Hz. I don't have anywhere near enough taps to brute force correct the phase shift. But what if, say I separated the <60Hz band, and I delay the >60Hz band by the equivalent delay in ms of 2 cycles at 60Hz, which is 33ms. Would doing this improve sound quality more than it hurt it?

 

I don't see why you can't correct that with FIR.  Let's say you want to compensate down to 2 cycles at 20hz = 100 ms.  At 44.1 khz sample rate, that's 8,000-ish taps.  How many do you have available?

Link to comment
Share on other sites

I don't see why you can't correct that with FIR.  Let's say you want to compensate down to 2 cycles at 20hz = 100 ms.  At 44.1 khz sample rate, that's 8,000-ish taps.  How many do you have available?

 

I only have around 1500 taps available, and I'm running on 96KHz, not because I want to, but because that's the only option on the miniDSP 2x4 HD. 

Link to comment
Share on other sites

  • 2 weeks later...

I have bought every single component needed for this speaker. Unfortunately I learned the hard way that it must be done by a CNC machine. Just working on the CAD drawings for this speaker for CNC. Because now I'm doing this on a CNC machine, it opens up other possibilities of what could be done. Here's a rough drawing to give you an idea of what the speaker looks like and how crazy space limited it is. 

 

I always knew this speaker would be very difficult to build, but I had no idea just how insane it is until actually doing it. In my mind it didn't seem so bad, probably doable in a weekend. But in reality, every little thing is so much more difficult because of how darn space limited I am. Simple things such as just how to rear mount the drivers became another exercise because of the 1/4'' thick wood. It took nearly a full day just to cut the frames off the drivers. The incredibly small midrange and tweeter spacing causes problems in sealing the drivers, not to mention the extreme difficulty of cutting it to shape without a CNC machine. 

 

There are other things that happens when you actually build it out or model it in a 3D software. The depth of the speaker was forcibly increased by 1'' from 6'' to 7'' for several reasons such as the two woofers cannot be mounted magnet to magnet because the magnets repel each other when placed that closely and not having enough space for the larger amp. This increases internal volume, and now 2 PR's might not be enough. Should I add a third PR and lose the inert enclosure, or convert a high excursion Tang Band woofer to be a PR, increasing the baffle size and decreasing PR "efficiency", so I have to put more thought into questions like this. 

 

Originally I aimed to have this speaker completed by about now. Unfortunately I'm looking at around late December to finish the model and have the parts cut, and another month for putting everything together and applying finishing, and another few weeks for outdoor measurements and tuning. So I'm looking at February before it's done. 

 

Believe me I want nothing more than to get this speaker done. The anticipation is killing me. I also don't want to ever setup my current mini speaker again. It is a massive wiring mess where 11 boxes (3 amps, 3 power supplies, 2 miniDSP's, 2 stereo to mono boxes, one optical cable splitter), 26 sets of cables (10 RCA, 6 banana plugs, 5 power cables, 3 optical, 2 USB) and a standalone power bar are needed to power 2 little speakers. I can't wait to cut that to just 2 speakers, 2 miniDSP's, and an optical splitter. Unfortunately I will have to set it up at least 3 more times in the next 2 months...

 

I attended the Burning Amp last week, and I heard the LXmini from Linkwitz Labs. Sounded really spacious and big (a little too big for my taste actually), but nevertheless got me curious about dipoles. The LXmini is omnidirectional to 800Hz. Mine is omnidirectional to 400Hz. Both speakers also follow the 1/4'' wavelength crossover frequency for a lobing error free radiation pattern. I had an idea on having the option to turn the speaker into a dipole by adding another set of tweeter and midrange to the rear baffle of the speaker. The ICEpower amp for the midrange and tweeter have no problem supporting the second driver, in fact it is specifically rated to be able to handle 20W 8 ohm on one channel (2 tweeter) and 82W at 2 ohm on the other channel (2 midrange). If I feel like blowing $5000 and a whole bunch of time I'll attempt this. Only reason I'm not doing this on the current speaker is that I don't have room for the rear tweeter because the woofer amp is mounted on the top baffle. And I thought a small speaker can't get more expensive than this...... (Well, I could have used some $300+ woofers, but they would have performed worse for bass than the $140 woofers I used here) 

 

But before that I want to look at moving away from a miniDSP setup and maybe to a Intel NUC setup for DSP, inspired by BassThatHz's work on a PC based DIY DSP setup. This is why the miniDSP's weren't inside the speaker, even though there is space. One feature I feel that is very important for sound quality is the equal loudness contour to maintain tonal balance at any volume. At low volume I want a more V shaped curve, at high volumes I want the bass and especially the treble to be reduced, otherwise the speaker could sound too bright at higher volumes. I'd also get much more taps for FIR filters, and can actually do more than 2 bands for the compressor without the tedious wiring solutions right now involving combining signals back into mono for the woofers. Not to mention, it would be much cheaper than 2 miniDSP's

post-1515-0-35175700-1479779638_thumb.png

Link to comment
Share on other sites

$5k! Whew that's an expensive pair of mini speakers.

 

The amount of cables, amps, power supplies, etc...Is exactly the reason I'm leaning toward active with built in amplifiers for my "someday" speaker build. More expensive but only 2 cables needed.

 

That's a cram packed little speaker. Did you think about moving the active 3 way section closer to center and mounting the PR's on the ends with perhaps an end cap on each? Wasn't sure if you thought about that arrangement or if it would even work.

Link to comment
Share on other sites

$5k! Whew that's an expensive pair of mini speakers.

 

The amount of cables, amps, power supplies, etc...Is exactly the reason I'm leaning toward active with built in amplifiers for my "someday" speaker build. More expensive but only 2 cables needed.

 

That's a cram packed little speaker. Did you think about moving the active 3 way section closer to center and mounting the PR's on the ends with perhaps an end cap on each? Wasn't sure if you thought about that arrangement or if it would even work.

 

 

Damn it Ricci! Your comment just cost me another $5000 because you just suggested something that opened up a whole bunch of new possibilities for improvement for this speaker! Now I can't call the speaker I'm building as good as it gets for a small speaker!  :angry:

 

In all seriousness, I did consider that, but I originally brushed it off because if the speaker is used as a stereo pair, it would be placed vertically, and the PR's would sag from being mounted vertically, especially with all the added weight. Also, the speaker would need feets on the right side, so when vertically placed, the PR has a slot for the sound to escape. However, the protruding feets would not look nice when placed horizontally. 

 

However, I didn't consider all the benefits that this configuration can bring that I only realized after you mentioned this again. 

 

1. First, the speaker can be made smaller!  :). I really didn't like having to increase the speaker size from the planned 14" x 6" x 6" WHD to 14.5" x 6" x 7". I can probably drop the size to 13" x 6" x 7" with this configuration. The depth increase isn't so bad because it is not very noticeable, but decreasing baffle width from 14.5'' to 13'' is very noticeable. 

 

There is a lot more baffle space for the TMW drivers, which means the driver can be placed in the same plane, like a TMW. There are several benefits

 

2. There is a lot more baffle area around the tweeter. This opens up the possibility to make a waveguide for the tweeter. It would be a very shallow waveguide that only controls directivity above 5000-6000Hz where the tweeter stops radiating 180 degrees and starts narrowing. This would make for a smoother power response above 6000Hz, as well as reducing diffraction effects.

 

3. I can use the Audax HM100Z0 midrange. This is a slightly larger but much more (~5dB higher, 93dB @ 2.83V 8 ohms) sensitive midrange. This would allow the speaker to play even louder and more dynamic than it already is, and likely exceeding most bookshelf/floorstanding speakers, but of course below high efficiency pro speakers. The 2 woofers can do ~110dB to 400Hz, but the original Scanspeak 10F mid can only play to about 102dB in its pass band. This mid can play ~107dB across its pass band. Many have mentioned this as one of the most detailed drivers ever, and some have mentioned this midrange to be even more detailed than the Scanspeak 10F. 

 

4. I can use the Tang Band W5-1138SMF as a PR without increasing the speaker size. On the original, I would need to increase the baffle width by 0.375" if I wanted to use the TB woofer as a PR. When placed on the side, the speaker can still maintain the smaller baffle size! This woofer will provide higher excursion limits than the Peerless PR currently used. 

 

5. There is space for another tweeter and midrange on the rear baffle for dipole configuration. 

 

But what's potentially more interesting is that this configuration would make the speaker a lot more deceiving as it seems like the speaker only has 3 drivers, and only 1 woofer that is pumping out the ridiculous amount of bass rom a mini speaker. 

 

So all of this would go into the V3 of this speaker. There is way too much complexity to be adding this to the V2 speaker I'm building now. I thought the V2 will be as good as it gets, and I won't be able to improve on the physical speaker itself much more. But looks like I can! However, I need a lot more knowledge in dipoles and waveguides before I can tackle this. If you know anyone that can help with waveguides, please let me know. 

 

As to the cost, I probably only need $4000, about $900 more for extra drivers for dipole, and still not as expensive as your nuclear options! But really, even the current speaker I'm building will spank any traditional (non HE) commercial $5000 speaker for sound quality. In fact, if I do this right, it might be quite difficult to find any traditional commercial of any price that'll outperform this speaker for sound quality. It will take a special speaker, like high efficiency, horn, all out DSP speakers like JBL M2, B&O Beolab 90, Kii Audio Three, or others to outperform this. It is using some of the finest drivers available with a proper crossover implementation that takes into directivity, off axis response, power response, lobing error (where it has none), phase (linear phase), time alignment, group delay and of course flat response into proper account with no compromises to dynamics or maximum volume compared to traditional speakers. Most mega bucks high end speakers do poorly for most of those aspects of speaker performance mentioned above, which is kind of sad. 

Link to comment
Share on other sites

Didn't mean to throw the project off course but if it helped you think of some options you hadn't considered that's good.

 

Waveguides aren't so bad. In this case since everything is so compact that's likely going to be the #1 design criteria. As far as building a custom one this small it might be worth looking into having it rapid prototyped.

Link to comment
Share on other sites

Didn't mean to throw the project off course but if it helped you think of some options you hadn't considered that's good.

 

Waveguides aren't so bad. In this case since everything is so compact that's likely going to be the #1 design criteria. As far as building a custom one this small it might be worth looking into having it rapid prototyped.

 

It's all good. I hope you got the sarcasm in the previous post. I do appreciate *anything* that helps me find a better design for the speaker. Otherwise I'm left feeling, that's it? There's nothing else I can do to improve this? I had this thought for nearly 2 months, and I was thinking I might be a little sad after it gets built since I'll be "done". This should keep the fun for at least another year or two. 

 

Will look into rapid prototyping. 

 

How would a decent NUC plus audio interface be cheaper than 2 minidsp's?

 

I don't know the exact cost, but 2 miniDSP's are $440 alone, plus another $60 for the 2 summing box. I really doubt a NUC setup is gonna cost more than $500. Don't need a high powered setup. This is only for audio. 

Link to comment
Share on other sites

Question: I'm actually thinking about moving the PR's to the side in order to reduce the width from 14.5" to 13". But if I do that, does it still cancel out vibrations? There is a lot of stuff in between the 2 PR's.

 

Remember you need a 2 in/6 out (?) audio interface as well. Is it intended to be the source or just the processor?

 

This is intended to be the source and processor. I did not consider the cost of the audio interface. This is a very early stage idea that needs a LOT of research. Thanks for bringing up these things to me so I can do more research. 

 

The cost isn't a huge issue. If it ends up being more expensive than the miniDSP setup, so be it unless it starts costing a ridiculous amount (like $1000+). 

Link to comment
Share on other sites

If it's the only source then your life is much simpler as you no longer need an audio interface with inputs and you don't need a dsp engine that can sit there spinning on those inputs.

 

If you're using windows then EqualiserAPO is one option, jriver is another. The former sits in the windows audio system and does not use ASIO (or WASAPI exclusive), the latter is more flexible as it has both a loopback interface for rendering system sounds and is a player with a highly capable dsp engine as well. jriver is probably the simplest option as well as one of the most feature rich from a DSP point of view. 

 

If you prefer linux then jriver is still an option, though lacks the loopback interface, brutefir, for a convolution option, and some sort of alsa/ecasound combination. 

 

An audio interface can cost as much as you want to spend, from a few hundred up into the stratosphere :). My preference is to use the same interface for measurement as for playback so I like devices that have at least 1 mic pre and at least a few inputs. To that end, I've had a focusrite saffire pro 24 and an rme ff800 in my system though I currently use a motu 1248. A motu ultralite mk4 is probably one of the most compact units you can get with sufficient output channels, has solid drivers and should be a nice, clean interface. On the other hand, if you go for something that is purely a DAC then a cheap option is something like the minidsp u-dac8. Ultimately the right choice for you depends on how much you believe in the impact of electronics on audible SQ, certainly lots of options out there anyway.

Link to comment
Share on other sites

Looks like this speaker is getting delayed again... I wish I could have thought of this earlier (or Josh could have suggested this earlier  ;) ). There was a 2 month period between the time I made the last improvement and the time I started actually building the speaker. In that agonizing 2 months I couldn't build the speaker, but also couldn't think of any other improvements. But better late than never! Way better than finishing the speaker, then bang my head realizing I could have made all these improvements. So here they are: 

 

Thanks to Ricci's suggestion, I am able to reduce the size of the speaker by placing the passive radiators on the side. By doing this, I am able to reduce the width of the speaker from 14.5" (very tight, most likely have to be 14.75 or even 15") to 13" safe. I'm trying to get it down to 12.5". The bottleneck to reduce width further is actually the larger ICEpower amp being too long at 6.3"... If the amp wasn't a problem the width could be reduced to 11.5". Grrr. Wish the amp is something like 3.9" x 5" instead of 3.1" x 6.3". If I'm crazy I'll somehow mount the amp diagonally to reduce width, but that's probably not happening. Nevertheless, I'm happy that I can reduce the width, which makes me feel less bad about the 1" depth increase. In fact, better because a reduction in width is a lot more noticeable than an increase in depth. Another cool side benefit is that this configuration makes the speaker seem much more deceptively small, not that the original wasn't already tiny. In the original you see 2 bass drivers on the front baffle. But, this one you only see one woofer on the front baffle. The other woofer is hidden in the rear and the PR's are on top/bottom. It'll be even more ridiculous to make people think, wait, all that from just one little woofer???

 

Because the width requirement isn't so stressed anymore, the midrange enclosure is changed to a trapezoidal shape in order to have non parallel side walls to reduce standing waves. 

 

I'll be using a different passive radiator from Tang Band. It is an oval shaped passive radiator with a special ribbed surround to prevent cone rocking. It seems to be extremely high excursion because Tang Band recommends just one of these for their 5.5" woofer with 9mm Xmax when a minimum of two PR's are needed for the Sd that PR has and how much excursion the woofer has. The driver I'm using has the same Sd, but half the Xmax. Also, in their own simulation they modeled with 50W for the woofer, and had 8mm of excursion on the PR at Fb. I can assume Xmax is at least 8mm? For some reason Tang Band is making it ridiculously hard and expensive to get these. First, only one distributor caries it, and they're in Germany. I have to pay $60 in shipping to have them shipped to the US. Second, and the much more ridiculous problem: the only way to buy these PR's is to buy them in a bundle with a woofer, and Tang Band specifically says they won't separate them. They ignored the distributor's plea to separate them as well. Therefore, I have to buy a pair of their woofers just to buy these PR's, and a $60 shipping charge on top of that...... These better perform well compared to the $12.75 Peerless PR's.

 

http://www.tb-speaker.com/products/pr14-c

 

Question: The Tang Band PR seems to be a much better PR in terms of surround and suspension compared to the Peerless. This is apparent on the design and the specs, much higher Cms and Mms. However, one glaring thing is the low 3.1 Qms compared to the Peerless's Qms of 10. This is causing a 0.5dB loss in output in my simulation. Does Qms actually affect PR efficiency that much? 

 

One other concern is the Xmax/Xmech on this PR. Tang Band does not list this, but one of the big reasons for using this is because I'm under the assumption that it has a lot of Xmax. The biggest problem with the Peerless PR is the limited 6mm Xmax at only half of what the simulation says I need. I'm inclined to think an oval shaped spider/surround would be capable of less excursion than circular shaped ones?

 

On other things, I mentioned other improvements that could be done with Ricci's suggestion of side firing PR's, but after more careful thinking, they can't be done. 

 

1. The waveguide is a no go on the tweeter. I do not have the baffle space to move the tweeter to the same axis as the mid and woofer, so the current orientation stays, and there is no space for a waveguide. 

 

2. Full Dipole/Bipole setup can't be done either. I don't actually have space to put the rear firing drivers due to the amplifier placements. Maybe if I increased depth, but I don't want to increase size for something like this.

 

3. Bigger midrange can't be done for the same reason as #1. There is no baffle space, and the original configuration does not fit the bigger mid. Not a huge deal since the speaker will never be pushed to the point where the midrange runs out of steam. The bass would long run out of steam before the mid will. Maybe if I used subwoofers, then that'll come up.

Link to comment
Share on other sites

I know this is a DIY project and most of the satisfaction and value of the speakers is in the designing and building process. But for 5000$ for a pair of square speakers with a limited output and connections why not go to the market's best bookshelf speakers ever made: Devialet Phantom Gold?

Fully digital,with wireless transmission,optical input, fancy remote, very very stable connection.

The bass drivers are low impedance with a tech similar to IPal, the amps and power source are much beafier the the Ice modules, you get 103 dB at 45 hz from a pair in room at 2 meters, it looks really nice. The drivers have about 13 mm xmax and 16 mm xvar.the mids and tweeters are coaxial, proprietary design.

The graphs are measurements in room at 1 meter made by a guy on the devialet forum.

I have sold a bunch of these. I had Beolab 90 in our shop and on moderate levels one could hear no big difference.

Again, I love DIY but it has to offer a much bigger performance/value to worth the headache

post-3306-0-42652900-1480317892_thumb.jpg

post-3306-0-66982900-1480317916_thumb.png

post-3306-0-95620300-1480317929_thumb.png

Link to comment
Share on other sites

I know this is a DIY project and most of the satisfaction and value of the speakers is in the designing and building process. But for 5000$ for a pair of square speakers with a limited output and connections why not go to the market's best bookshelf speakers ever made: Devialet Phantom Gold?

Fully digital,with wireless transmission,optical input, fancy remote, very very stable connection.

The bass drivers are low impedance with a tech similar to IPal, the amps and power source are much beafier the the Ice modules, you get 103 dB at 45 hz from a pair in room at 2 meters, it looks really nice. The drivers have about 13 mm xmax and 16 mm xvar.the mids and tweeters are coaxial, proprietary design.

The graphs are measurements in room at 1 meter made by a guy on the devialet forum.

I have sold a bunch of these. I had Beolab 90 in our shop and on moderate levels one could hear no big difference.

Again, I love DIY but it has to offer a much bigger performance/value to worth the headache

 

So here's my thinking. 

 

First, I have heard of a pair of the Devialet Phantom Silver. The sound quality is very impressive, easily worth its asking price when compared to normal commercial speakers. However, the bass fell way short for me. It hit low, but didn't seem to hit low with any power, even when the volume is turned up. Maybe it was the room, but I made an effort to walk around the room to make sure I wasn't in a null, and it was the same everywhere in the tiny room. 

 

Now to my project. One, it was initially expected to cost around $1000, significantly less than $6000 for a pair of Phantoms. The cost did balloon to around $3000 as I discovered better components that would deliver better results, which led to an all out mentality where money was truly no object in the chase to achieve the absolute best. Even then, it's still half the price of the Phantom. That said, this project was never about saving money, it was about building something that I can't buy and to see how far I can push small speaker performance. Yes, I might have spent around 500 hours on this speaker by the time its built, but it was FUN! 

 

Two, the Phantom is impractical as a travel speaker. It is not what I consider a small speaker. It is 2.2x the size of the original design I proposed, and I have a new idea that'll shrink the size of my design to the point the Phantom is ~2.9x larger. But that's not even the biggest issue. The Phantom is 22 pounds each. I can't even put 2 of them inside a luggage bag without exceeding the 50 pound luggage weight limit (bag is 7 pounds alone). My design is around 9 pounds each. 

 

Next, performance. I'm confident I can outperform the Phantom in sound quality. If I can't do that with high end Scanspeak beryllium tweeters and neodymium mids, both considered to be one of the best in their class, then I'm doing something wrong. My design will match the benefits of coaxial characteristics and the benefits of a 2 way (no driver compromises) without the problems of coaxial drivers. The only area mine won't outperform the Phantom is in diffraction control. For bass, if we assume 13mm Xmax, and the drivers can handle 1000W, plus another 500W or more for thermal compression, both unrealistic for 6.5'' woofers, then yes the Phantom will have a 4-7dB advantage compared to mine limited at 150W and 5mm Xmax. If we assume Devialet did an incredible job and was able to get the 6.5'' woofers to move 10mm cleanly, then it still only holds a ~3dB advantage. If we assume something more realistic, like 7mm, which is reasonable as any further distortion should become noticeable and thermal compression should set in fast for such small woofers, then it doesn't really hold any advantages for bass. Remember, I'm limiting my woofers at 150W and 5mm excursion. I can push my woofers harder if I wanted to. I have more than double the amp power on tap for another 3dB output. 

 

So in summary, I have a speaker that is less than half the size and weight of the Phantom and still perform better for treble, mid, and almost on par with bass. However, the completeness of the Phantom with nice wireless connectivity, app and remote is very nice, and something I wish I can have with DIY.

Link to comment
Share on other sites

I figured out a way to significantly reduce the width of my speaker. Originally the width is stuck at 13" because the larger 125ASX2 ICEpower amp is too long at 6.3". I realized instead of using the 125ASX2 to power the 2 woofers, I can use 2 of the smaller 50ASX2 amps, one for each woofer, and have both channels bridged for much more power. Since the ICEpower amps are 2 ohm capable, a bridged 50ASX2 can output 170W at 4 ohm, more than the 125W on the 125ASX2. The 50ASX2 is 2" shorter than the 125ASX2, so this allows the speaker's width to be reduced from 13" to 11". This is a huge 32% decrease in width compared to the original 14.5" width, and will make the speaker look much smaller. Volume wise it is not too much smaller than the original because the depth is 1" deeper than planned, which makes up about half of the volume lost from the 3.5" width reduction. 

 

The big problem is that the 50ASX2 is not manually bridgeable. You must buy the preconfigured BTL version of the 50ASX2 if you want to use it bridged. The problem? No one sells them, and ICEpower only sell to businesses. Anyone know somebody that can give me a hand here in acquiring 4 of these 50ASX2 BTL?

Link to comment
Share on other sites

Houston, we got a big problem, and this time I'm afraid there is no DIY solution to this problem. 

 

One thing that nagged me about my current speakers is that they have the tendency to sound thin at high volumes. Originally I thought it is just the 4'' woofer running out of steam. But then after some measurement, it shows the SPL levels I'm playing at for 100-300Hz should be well within the limits for the woofer. 

 

Then I realized, the woofer is not compressing because it ran out of steam, it is compressing because of thermal compression from the huge up to 13dB EQ boost on the woofer. Even 1-2dB of thermal compression will be easily audible, since the lower midrange is now 1-2dB mismatched to the midrange and tweeter. 

 

The new speaker will require even more bass boost, and the airspace per woofer is even smaller. This double whammy will cause even more severe thermal compression. One thing I was very proud about this design was how loud and dynamic this speaker can play due to the very high sensitivity (for a non pro driver) of the woofer and the high sensitivity midrange that still maintains 90dB sensitivity even after a 3dB baffle step compensation. Because of that I was thinking it will match or exceed most floorstanding speakers in dynamic and maximum volume above 100Hz. However this all comes crashing down when the woofer suffers from thermal compression, and this will notably affect high volume playback when the woofer (which also does lower midrange) is 2-3dB lower in level against the mid and tweeter after 30 minutes of blasting. 

 

The solution is thermal compression compensation through a feedback system with a temperature sensor on the magnet, then a predetermined model is used to infer voice coil temperature from magnet temperature of that particular woofer, and then use DSP to apply the appropriate boost to compensate for the loss in sensitivity depending on the frequency and temperature. This is of course, way WAY beyond what can be done on DIY, and where I bow down to commercial DSP implementations (and curse miniDSP for having such limited functionalities for such an expensive DSP). I've known about this issue, but never gave much thought about it and thought it was something done to perfect the speaker a little more, but doesn't really have noticeable improvements. Who cares if I lose a little bit of output in the bass at very high volumes. It is already stupid loud! However I didn't realize the effect that it would have on the low midrange, and realized this is a crucial feature for a speaker that uses huge bass boost. The only thing that can be done through DIY is to have a setting that reverses the effects of thermal compression for worst case scenario and manually switch to this setting if I plan on extended high volume playback, which is not really a nice solution.

 

Looks like the high volume playback performance is just going to suffer. 

Link to comment
Share on other sites

  • 1 month later...
  • 1 month later...

So I did some measurements, and it seems like the speaker's dispersion isn't as wide as I expected. I expected the speaker to have essentially perfect off axis response up to 4000Hz or so. Basically, the 90 degree curve should be essentially identical to the on axis curve up to 4000Hz or so. But for some reason the measurements show the 1000-6000Hz response drop as much as 4dB as the angle increases to 67.5. I expected the speaker to display essentially no response drop in that region because the drivers should be essentially forward omni in that range. However the response, very surprisingly, looks suspiciously similar to a constant directivity speaker with 135 degree pattern held to 950Hz where each successively steeper off axis angle produces a response that is reduced in level and very close in shape of the on axis response all the way out to 10KHz and a bit beyond. For some reason the tweeter isn't omnidirectional at and near the crossover frequency, but somehow matches the directivity of the midrange. I don't know how this is possible.

 

This is a very pleasant surprise, since one of the things about this design that bothers me is that the supposedly near perfect 180 degree dispersion pattern to 4KHz is detrimental for most in-room performances, even though theoretically this property makes it a more accurate speaker. Now the goal has shifted to making a more controlled directivity speaker. Since there is a rear woofer, I'm going to use that and turn the speaker into a cardioid pattern below 400Hz down to about 100Hz and minimize output to the rear of the speaker and significantly reduce reflections from the front wall. This requires throwing away up to 12dB of output since cardioids drop off at 6dB/oct from Fequal (which is at ~400Hz), but good thing this speaker was originally designed to have very high output in the 100-400Hz range, so I can afford to sacrifice up to 12dB of output for better directivity control and in room performance. If I need more output than that I'll just go back to bipole mode and get the headroom back. Will need to experiment to see if the cardioid pattern can be held above 400Hz by having the rear woofer canceling out the sound from the midrange driver too. 

 

Now I need to explore PC DSP solutions. I need 4 miniDSP's for this speaker now, and that's unwieldy and expensive (~$1000 for 4). The Intel NUC solution is much neater and more powerful (way more FIR taps for one), but requires more work to get it set up. Having trouble just installing Ubuntu on my spare MacBook Pro to try PC DSP out before I buy, but for some reason I keep getting an error message during installation. Will explore more when I can squeeze some time here and there.

Link to comment
Share on other sites

  • 3 weeks later...

For some reason the tweeter isn't omnidirectional at and near the crossover frequency, but somehow matches the directivity of the midrange. I don't know how this is possible.

 

 

This type of thing is why measurements are so important. In a nutshell most of our modeling or simulations, or common knowledge are simplifications of what are much more complex issues in reality. They say bass is omni too but that's not entirely correct either.

 

Any other news? Are you listening to these and working on dialing them in?

Link to comment
Share on other sites

This type of thing is why measurements are so important. In a nutshell most of our modeling or simulations, or common knowledge are simplifications of what are much more complex issues in reality. They say bass is omni too but that's not entirely correct either.

 

Any other news? Are you listening to these and working on dialing them in?

 

 

Yeah I agree. However, getting a proper environment to do really accurate measurements is quite challenging. 

 

Unfortunately my friend who's helping me build this cabinet is sick and can't work on them for a few weeks. This is partially why this speaker has been delayed for so long. I expect it'll be finished in a month, and then it is off to tuning. In the meantime I have been learning proper speaker measuring techniques using my current speakers (they have the same midrange and tweeter, except it doesn't have a Be diaphragm, everything else is the same) while waiting for them to be finished. I originally planned on learning this after I get them so I don't do things twice, but I'm glad I'm doing it now. It will reduce the tuning time for the new speaker significantly as now I know what I'm doing.

 

In other news, I'm shifting away from a PC DSP and decided to dive into the world of hardware and make my own DSP from DSP chips from Analog Devices or another company. It leads to the cleanest solution and have the greatest access to various DSP algorithms. However, the learning curve is STEEP. This is like the hurdle to learn about proper speaker design again, but with even greater complexity and fewer resources. But really, a PC DSP solution is not nice. While powerful, it is limited in DSP algorithms, and is at best a temporary solution. 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...