Reference Mini Part 1 (of many): Introduction
I've been thinking and wanting to do this for a long time, and I'm gonna do a series of more advanced deep dive speaker design postings, with a bit of a twist. I'm gonna largely center it around my speaker, the Reference Mini, which is a cost and complexity no object small speaker with the goal of getting as much big speaker performance from a small speaker as possible, or, simply, building the highest performance small speaker possible.
However, instead of just talking about how great my speaker is, I'm also going to talk in great detail about the numerous flaws and mistakes in design of this speaker that holds back its potential. I hope people could learn from everything I've learned from this speaker. This is going to be a series of long posts. I will try to keep it more accessible to as many people as possible, but it will generally be quite technical. It will go through everything from parts selection, speaker design, measurements, and how it tries to do well in as many aspects of speaker design as possible.
Please feel free to ask any questions on the design, or topics you'd like to see me go into more detail.
Reference Mini
4 driver, 4 amplifier, linear phase, time aligned, DSP controlled directivity *small* speaker
Frequency response: 33Hz-20000Hz, +/- 1dB
Internal volume: 3.5L, 0.13 cu ft.(Yes, there is no air inside ��)
Dimensions: 6.25" x 12" x 7.5" HWD.
Driver selection
Tweeter: Scan-Speak Illuminator beryllium D3004-604010
Midrange: Scan-Speak Discovery 10F-4424G
Woofer: 2x Wavecor WF152BD05
Passive Radiator: 2x Tang Band PR14, tuned at 35Hz
Amplifier: 2x ICEpower ICEedge 100AS2, 2x100W per amp
Audio interface: Asus Xonar U7
I've been thinking and wanting to do this for a long time, and I'm gonna do a series of more advanced deep dive speaker design postings, with a bit of a twist. I'm gonna largely center it around my speaker, the Reference Mini, which is a cost and complexity no object small speaker with the goal of getting as much big speaker performance from a small speaker as possible, or, simply, building the highest performance small speaker possible.
However, instead of just talking about how great my speaker is, I'm also going to talk in great detail about the numerous flaws and mistakes in design of this speaker that holds back its potential. I hope people could learn from everything I've learned from this speaker. This is going to be a series of long posts. I will try to keep it more accessible to as many people as possible, but it will generally be quite technical. It will go through everything from parts selection, speaker design, measurements, and how it tries to do well in as many aspects of speaker design as possible.
Please feel free to ask any questions on the design, or topics you'd like to see me go into more detail.
Reference Mini
4 driver, 4 amplifier, linear phase, time aligned, DSP controlled directivity *small* speaker
Frequency response: 33Hz-20000Hz, +/- 1dB
Internal volume: 3.5L, 0.13 cu ft.(Yes, there is no air inside ��)
Dimensions: 6.25" x 12" x 7.5" HWD.
Driver selection
Tweeter: Scan-Speak Illuminator beryllium D3004-604010
Midrange: Scan-Speak Discovery 10F-4424G
Woofer: 2x Wavecor WF152BD05
Passive Radiator: 2x Tang Band PR14, tuned at 35Hz
Amplifier: 2x ICEpower ICEedge 100AS2, 2x100W per amp
Audio interface: Asus Xonar U7
Last edited:
Reference Mini Part 2: Design goals
With the Reference Mini, I had some exceedingly ambitious goals for a tiny speaker. I essentially strived for as many attributes as I can get from a large speaker, or simply, a high quality, well designed speaker. I wanted to go through the list of my goals, whether I achieved them, and whether it was a good idea or not.
1. Maximum bass SPL capability as possible, with a secondary goal of very high SPL capability across the entire band.
- With a F3 of 31Hz, and over 100dB from 40Hz and up, this was done nearly perfectly, in the sense that even with the knowledge I have today, I could not make a higher output speaker. It was well worth the effort as the end result is astonishing.
- For midrange and treble output, it actually exceeds most floorstanding speakers due to efficiency. Most speakers use drivers under 90dB sensitive, and all drivers are powered by a single amp, usually not much more than 100-200W, with power wasted for baffle step correction and tweeter and midrange attenuation. The Reference Mini is actually a high sensitivity design, with 90dB sensitivity for the tweeter, 91dB for the midrange, and a combined 93dB for the 2 woofers. They are individually amped with 100W per driver for 400W total per speaker, and the speaker does not give up 6dB of midrange and tweeter amplifier power for baffle step correction. All of that makes for one exceedingly SPL capable small speaker.
2. Maximum sound quality
- With some of the best drivers and extremely optimized crossover design, the sound quality is absolutely sublime. I have tried very hard to find high end speakers better sounding than the Reference Mini in order to learn from and further improve the speaker, but the Reference Mini sets the bar very high, and as a result most high end speakers are simply inferior by comparison.
- That said, the goal of maximum sound quality was not done well. My bar is exceeding high, the absolute best. This is definitely not the absolute best sounding speaker that could be made for this size, and there are a number of improvements available that can, in totality, dramatically improve sound quality, both measured and subjective metrics. I have definitely heard speakers that sound better, some much better, than the Reference Mini, either in totality, or in some certain areas. Some areas it could improve is having lower diffraction issues and narrower dispersion.
3. All in one package with internal amplifiers, power supply, and DSP
- Internal amplifiers and power supply were achieved, but DSP was not. Leaving the DSP out turned out to be a smart choice as the power and flexibility of a computer based DSP dramatically helped with learning and the capabilities of this speaker
4. Absolute minimal box size
- By the metrics of absolute minimal size, this was done very well. It wasn’t really possible to make this speaker any smaller than it is. However, the form factor is a mistake, because the speaker is meant to be placed horizontally, which made the speaker take up a lot of desk space, and made the speaker seem a lot bigger than it is. Placing it vertically had negative impact to the sound, and required a stand in order for the side passive radiators (now top and bottom) to have clearance, adding to its apparent size.
Each of the 4 goals above will have at least one separate post going into much more detail on how it was achieved. I hope you enjoy following my journey to better sounding speakers!
With the Reference Mini, I had some exceedingly ambitious goals for a tiny speaker. I essentially strived for as many attributes as I can get from a large speaker, or simply, a high quality, well designed speaker. I wanted to go through the list of my goals, whether I achieved them, and whether it was a good idea or not.
1. Maximum bass SPL capability as possible, with a secondary goal of very high SPL capability across the entire band.
- With a F3 of 31Hz, and over 100dB from 40Hz and up, this was done nearly perfectly, in the sense that even with the knowledge I have today, I could not make a higher output speaker. It was well worth the effort as the end result is astonishing.
- For midrange and treble output, it actually exceeds most floorstanding speakers due to efficiency. Most speakers use drivers under 90dB sensitive, and all drivers are powered by a single amp, usually not much more than 100-200W, with power wasted for baffle step correction and tweeter and midrange attenuation. The Reference Mini is actually a high sensitivity design, with 90dB sensitivity for the tweeter, 91dB for the midrange, and a combined 93dB for the 2 woofers. They are individually amped with 100W per driver for 400W total per speaker, and the speaker does not give up 6dB of midrange and tweeter amplifier power for baffle step correction. All of that makes for one exceedingly SPL capable small speaker.
2. Maximum sound quality
- With some of the best drivers and extremely optimized crossover design, the sound quality is absolutely sublime. I have tried very hard to find high end speakers better sounding than the Reference Mini in order to learn from and further improve the speaker, but the Reference Mini sets the bar very high, and as a result most high end speakers are simply inferior by comparison.
- That said, the goal of maximum sound quality was not done well. My bar is exceeding high, the absolute best. This is definitely not the absolute best sounding speaker that could be made for this size, and there are a number of improvements available that can, in totality, dramatically improve sound quality, both measured and subjective metrics. I have definitely heard speakers that sound better, some much better, than the Reference Mini, either in totality, or in some certain areas. Some areas it could improve is having lower diffraction issues and narrower dispersion.
3. All in one package with internal amplifiers, power supply, and DSP
- Internal amplifiers and power supply were achieved, but DSP was not. Leaving the DSP out turned out to be a smart choice as the power and flexibility of a computer based DSP dramatically helped with learning and the capabilities of this speaker
4. Absolute minimal box size
- By the metrics of absolute minimal size, this was done very well. It wasn’t really possible to make this speaker any smaller than it is. However, the form factor is a mistake, because the speaker is meant to be placed horizontally, which made the speaker take up a lot of desk space, and made the speaker seem a lot bigger than it is. Placing it vertically had negative impact to the sound, and required a stand in order for the side passive radiators (now top and bottom) to have clearance, adding to its apparent size.
Each of the 4 goals above will have at least one separate post going into much more detail on how it was achieved. I hope you enjoy following my journey to better sounding speakers!
You packed in quite a bunch into one enclosure. My questions are...
(1) Given that you have two large woofers with their magnets close together does that affect the sound quality at all?
(2) Are the woofers in phase or out of phase and do they cancel the back wave inside the box for higher fidelity resulting in lower cabinet vibrations.
(3) How are these supposed to be set up. In a bookshelf...close to the wall...or three feet away from the wall and depending on that whats best?
(4) Given that these would cost around a $1000 a piece to make would there be a market for these given their sound quality? Are these a super high quality special use you have in mind that people would be willing to pay for?
(1) Given that you have two large woofers with their magnets close together does that affect the sound quality at all?
(2) Are the woofers in phase or out of phase and do they cancel the back wave inside the box for higher fidelity resulting in lower cabinet vibrations.
(3) How are these supposed to be set up. In a bookshelf...close to the wall...or three feet away from the wall and depending on that whats best?
(4) Given that these would cost around a $1000 a piece to make would there be a market for these given their sound quality? Are these a super high quality special use you have in mind that people would be willing to pay for?
Thank you everyone for the compliments
1) No
2) In phase. They do not cancel sound inside the box, since they're in phase
3) It can be set up any way you want, then apply the proper DSP correction for distance to front and side walls
4) No plans to make them commercial, although I'm sure there are tons of people who would pay top dollar for something like this. This sounds much nicer than the Devialet Phantom that's selling like hotcakes.
Yup, mobile computer speakers!
1) No
2) In phase. They do not cancel sound inside the box, since they're in phase
3) It can be set up any way you want, then apply the proper DSP correction for distance to front and side walls
4) No plans to make them commercial, although I'm sure there are tons of people who would pay top dollar for something like this. This sounds much nicer than the Devialet Phantom that's selling like hotcakes.
Yup, mobile computer speakers!
Reference Mini Part 3: Deep and loud bass from a small box (aka come at me Hoffman)
Perhaps the most fun part of this project is trying to extract as much deep bass out of a small box. There’s something magical about seeing a tiny speaker belt out a ton of bass. It is also going against physics, since it dictates that big bass requires a big box. While my speaker is not breaking any physics here, what I’m doing is getting as close to the limits of what physics will allow, and for most speakers, it is nowhere close to the limits.
The formula to get as much bass as possible in a small box is simple
1. Get as much displacement as possible that can fit in the box.
2. Make sure the driver(s) can handle the power needed to hit x-max
3. Make sure the amplifier can supply all the power the driver needs to hit x-max
4. Use DSP to apply Linkwitz Transform to extend the bass response to your desired F3
5. Use high pass filters and limiters to prevent bottoming/thermal damage to the drivers
6. Profit!!!
Actually doing it though, is not so simple. I will go into more detail on (1) and (2)
(1) should be pretty obvious. For a sealed box, driver displacement (surface area * excursion) directly determines the maximum SPL for any frequency. Higher displacement = Higher maximum bass SPL. For passive radiator speakers like mine it is still true, because it is either driver displacement or passive radiator displacement that is producing the output.
(2) The other major challenge The first thing to understand is that box size has no direct effect on maximum bass output. All that a smaller box size does is reduce the efficiency of the speaker at bass frequencies, as well as an earlier roll off. Therefore, in a sealed box, if one applies the correct bass equalization to extend the bass response, and assuming the driver can thermally handle the additional power, then the smaller box will perform identically to a larger box.
Unfortunately, this is not true in the real world, because driver motors are not perfect, and higher input power to the driver causes greater motor related distortion. If you reduce the box size by half, the efficiency below the box resonance decreased by a factor of 4, and therefore you need 4x more power to achieve the same SPL compared to a box double the size. Therefore, besides having enough displacement, the other big limitation for bass from a small box is thermal limitation from power handling.
In the next post I’ll post some examples of simulations of the power levels needed for various enclosure sizes. Sneak peek: power input levels are not what it seems from voltage levels. The impedance peak makes a huge difference in reducing actual power seen by the drivers
Perhaps the most fun part of this project is trying to extract as much deep bass out of a small box. There’s something magical about seeing a tiny speaker belt out a ton of bass. It is also going against physics, since it dictates that big bass requires a big box. While my speaker is not breaking any physics here, what I’m doing is getting as close to the limits of what physics will allow, and for most speakers, it is nowhere close to the limits.
The formula to get as much bass as possible in a small box is simple
1. Get as much displacement as possible that can fit in the box.
2. Make sure the driver(s) can handle the power needed to hit x-max
3. Make sure the amplifier can supply all the power the driver needs to hit x-max
4. Use DSP to apply Linkwitz Transform to extend the bass response to your desired F3
5. Use high pass filters and limiters to prevent bottoming/thermal damage to the drivers
6. Profit!!!
Actually doing it though, is not so simple. I will go into more detail on (1) and (2)
(1) should be pretty obvious. For a sealed box, driver displacement (surface area * excursion) directly determines the maximum SPL for any frequency. Higher displacement = Higher maximum bass SPL. For passive radiator speakers like mine it is still true, because it is either driver displacement or passive radiator displacement that is producing the output.
(2) The other major challenge The first thing to understand is that box size has no direct effect on maximum bass output. All that a smaller box size does is reduce the efficiency of the speaker at bass frequencies, as well as an earlier roll off. Therefore, in a sealed box, if one applies the correct bass equalization to extend the bass response, and assuming the driver can thermally handle the additional power, then the smaller box will perform identically to a larger box.
Unfortunately, this is not true in the real world, because driver motors are not perfect, and higher input power to the driver causes greater motor related distortion. If you reduce the box size by half, the efficiency below the box resonance decreased by a factor of 4, and therefore you need 4x more power to achieve the same SPL compared to a box double the size. Therefore, besides having enough displacement, the other big limitation for bass from a small box is thermal limitation from power handling.
In the next post I’ll post some examples of simulations of the power levels needed for various enclosure sizes. Sneak peek: power input levels are not what it seems from voltage levels. The impedance peak makes a huge difference in reducing actual power seen by the drivers
This remains a really cool project. Congratulations.
Since it's rare for music to have sustained sine tones at low-frequencies (although it does happen), is there much argument for having lots of amplifier headroom?
It would let you shrink the box size further, but you might need some motional feedback to counteract the motor-borne distortion rise that you've noted earlier.
Something I'm wondering now is if it's possible to take similar principles, and apply them to the PA system world.
Cheers,
Chris
Since it's rare for music to have sustained sine tones at low-frequencies (although it does happen), is there much argument for having lots of amplifier headroom?
It would let you shrink the box size further, but you might need some motional feedback to counteract the motor-borne distortion rise that you've noted earlier.
Something I'm wondering now is if it's possible to take similar principles, and apply them to the PA system world.
Cheers,
Chris
Amplifier headroom is very important. Pop music is filled with heavy sustained bass. The headroom is much appreciated with loud bass heavy music. It doesn't have to be sustained for minutes, but as long as the amp can burst for a few seconds, then that's generally enough for most music.
Well, the most limiting part is not the power, it's the distortion as that rises to inacceptable levels before the power limitation is reached. You are using a DSP, so the first thing you should implement is a low-cut at or just below the tuning frequency. Even if it doesn't look 'hifiy-esque', it does improve the sound quality immensely. The same way it works in the bass, it also works in the midrange and highs, which translates into steeper crossover slopes or higher crossover frequencies. Or, in other words, keeping the drivers out of the range where they produce distortion.
With your speakers already set, it's worth trying what the crossover frequencies and crossover slopes are indeed and where are the actual limits. And that's quite a bit above your standard 'power x and excursion y' because the distortion is not linear nor is it directly dependent on the enclosure size. The main question is: If your bass is limiting the maximum spl (or in other words, dymanimcs), would it help to reduce the mid-xo frequency or rise the mid-xo freuency? If the bass does a huge excursion, you'll get high distrotion of its higher frequency reproduction. That means, if your bass driver has to play a low-mid range, that would result in a high distortion there if the bass has to do a huge low frequency 'work'. To reduce the upper frequency range (upper xo frequency) of it would also reduce a lot of the distortion in the range where it worked so far but at the same time it would increase the distortion of the midrange because that driver has to do more excursion because of the lower frequencies it has to take than it had to do before.
The main question here is now, where is it actually limiting the reproduction? If it is the woofer, then the highest ever possible improvement is to reduce the excursion of the frequencies it cannot reproduce anyway. That means, at a ported (or PR) speaker, put a low-cut (subsonic) filter in place at or shortly below the tuning frequency. If the bass driver(s) is(are) your problem, that likely improve your headroom by likely 10-12dB!!!
If your limitation is one of the other drivers, you'll have to modify a lot more of the xo settings though and the effective gain in dynamics will probably be lower than that.
Reference Mini Part 4: Woofer selection process to maximize bass and midrange sound quality in a small box
This post will go through the process of how to determine the ideal characteristics you want in a woofer, which will guide the process of what to look for in a driver. What I was looking for is:
1. Optimize for maximum bass performance in a small box
2. A close second goal is maintaining midrange sound quality and output.
To maximize bass performance in a small box actually just comes down to a few parameters. The maximum bass output is limited by
1. Woofer displacement
2. Woofer power handling capability
3. Amplifier output
Having enough amplifier output and woofer power handling determines the amount of displacement that can be utilized. However, there are practical limitations of how much amplifier power I have, especially when I want the amps to be able to fit inside the speaker. In this case, that means I have 200W available to the woofers.
Woofer power handling is largely determined by the mass of the voice coil as well as the ventilation. However, there are practical limitations to power handling as well. Ventilation is not going to do much in a speaker this this small. Having too much voice coil means lowering midrange efficiency, which is not good for midrange output. 200W from the amp is about as much as a pair of 5.25" woofers can handle as well.
So what we need is a woofer that is more power efficient. This can only really be done in one way, by increasing the motor strength of the driver. This is not the same as BL. Motor strength is BL^2/Re. Motor strength is the BL normalized for impedance. BL is a meaningless and very misleading figure by itself, because it means nothing without knowing what the DC resistance of the driver. BL can be raised to whatever you want by changing the impedance of the woofer, which is trivial to do. Motor strength can only be raised by a stronger motor.
If a driver has twice the motor strength as another driver, it is 2x more efficient, or uses half the power for the same SPL. If a driver has half the motor strength as another driver, it is half as efficient, or uses 2x power for the same SPL.
Here’s a comparison of motor strength between high end 5.25” woofers that could be contenders.
Wavecor WF152BD05 (the woofer used): 12.6
Scanspeak Revelator 15W/8530K: 8.4 (1.5x more power needed)
Scanspeak Illuminator 15WU/4741T: 11.3 (1.1x more power needed)
SB Satori MW13P: 6.3 (2x more power needed)
Seas Excel W15CH001: 4.7 (2.7x more power needed)
Audio Technology C-Quenze 15H: 6.8 (1.85x more power needed)
As one can see, out of all high end 5.25” woofers available on the market, the Wavecor has the highest motor strength of all. It is 50% more power efficient than the venerable Revelator woofer, and up to 2.7x more power efficient than the Seas Excel. This difference is even greater in the real world because the less efficient driver will suffer from thermal compression to a greater extent.
The higher motor strength of the Wavecor also gives it the highest midrange output by 2-4dB depending on the woofer being compared to. The Wavecor also has a very linear BL curve from Klippel testing with nicely symmetrical suspension linearity as well, which indicates excellent midrange performance. Because of its combination of the highest motor strength for maximum bass output and excellent midrange performance, both in sound quality and output, this is why the Wavecor WF152BD05 is used for the Reference Mini, and how a pair of tiny speakers can exceed 100dB above 35Hz and 110dB in the 100-300Hz range! (Even higher in a room) It's a lot of fun to see and hear such a small speaker play so loud!
This post will go through the process of how to determine the ideal characteristics you want in a woofer, which will guide the process of what to look for in a driver. What I was looking for is:
1. Optimize for maximum bass performance in a small box
2. A close second goal is maintaining midrange sound quality and output.
To maximize bass performance in a small box actually just comes down to a few parameters. The maximum bass output is limited by
1. Woofer displacement
2. Woofer power handling capability
3. Amplifier output
Having enough amplifier output and woofer power handling determines the amount of displacement that can be utilized. However, there are practical limitations of how much amplifier power I have, especially when I want the amps to be able to fit inside the speaker. In this case, that means I have 200W available to the woofers.
Woofer power handling is largely determined by the mass of the voice coil as well as the ventilation. However, there are practical limitations to power handling as well. Ventilation is not going to do much in a speaker this this small. Having too much voice coil means lowering midrange efficiency, which is not good for midrange output. 200W from the amp is about as much as a pair of 5.25" woofers can handle as well.
So what we need is a woofer that is more power efficient. This can only really be done in one way, by increasing the motor strength of the driver. This is not the same as BL. Motor strength is BL^2/Re. Motor strength is the BL normalized for impedance. BL is a meaningless and very misleading figure by itself, because it means nothing without knowing what the DC resistance of the driver. BL can be raised to whatever you want by changing the impedance of the woofer, which is trivial to do. Motor strength can only be raised by a stronger motor.
If a driver has twice the motor strength as another driver, it is 2x more efficient, or uses half the power for the same SPL. If a driver has half the motor strength as another driver, it is half as efficient, or uses 2x power for the same SPL.
Here’s a comparison of motor strength between high end 5.25” woofers that could be contenders.
Wavecor WF152BD05 (the woofer used): 12.6
Scanspeak Revelator 15W/8530K: 8.4 (1.5x more power needed)
Scanspeak Illuminator 15WU/4741T: 11.3 (1.1x more power needed)
SB Satori MW13P: 6.3 (2x more power needed)
Seas Excel W15CH001: 4.7 (2.7x more power needed)
Audio Technology C-Quenze 15H: 6.8 (1.85x more power needed)
As one can see, out of all high end 5.25” woofers available on the market, the Wavecor has the highest motor strength of all. It is 50% more power efficient than the venerable Revelator woofer, and up to 2.7x more power efficient than the Seas Excel. This difference is even greater in the real world because the less efficient driver will suffer from thermal compression to a greater extent.
The higher motor strength of the Wavecor also gives it the highest midrange output by 2-4dB depending on the woofer being compared to. The Wavecor also has a very linear BL curve from Klippel testing with nicely symmetrical suspension linearity as well, which indicates excellent midrange performance. Because of its combination of the highest motor strength for maximum bass output and excellent midrange performance, both in sound quality and output, this is why the Wavecor WF152BD05 is used for the Reference Mini, and how a pair of tiny speakers can exceed 100dB above 35Hz and 110dB in the 100-300Hz range! (Even higher in a room) It's a lot of fun to see and hear such a small speaker play so loud!
Langhub-Subwoofer 175mm Reckhorn D-165, +/- 8mm Hub!!! Mit neuartiger Abdeckung | eBay
Paramètres haut-parleur de THIELE et SMALL, sans filtre ni ampli
9,38*9,38/3,6=24,44
Xmax=+/-8mm
Xmech=+/-16mm
Paramètres haut-parleur de THIELE et SMALL, sans filtre ni ampli
9,38*9,38/3,6=24,44
Xmax=+/-8mm
Xmech=+/-16mm
Last edited:
Reference Mini Part 5: Passive radiator sag and how it affects output.
The Reference Mini has Tang Band PR14 oval passive radiators with ~100g of mass added to reach a tuning of 35Hz, mainly to reduce excursion on the 5.25" Wavecor WF152BD05 woofers and reduce the BL induced efficiency loss from large excursions. It also certainly helps getting more output from the speaker while reducing distortion.
However, I often have my speakers vertically, so the passive radiators are in an up and down configuration, which is bad for the suspension of the passive radiator due to the added mass. This eventually causes the surrounds of the passive radiator to sag over time.
The graphs below show outdoor ground plane measurements of my speaker. Green is taken 1.5 years ago, red is taken recently. As you can see, below 50Hz, the red curve is tilting down about 2dB vs the green curve, suggesting that over 1.5 years the passive radiator has sagged enough to reduce its output contribution by 2dB.
Lesson: Don't use up or down firing passive radiators! Not unless you like to replace the radiators every year or so, or lose output.
The Reference Mini has Tang Band PR14 oval passive radiators with ~100g of mass added to reach a tuning of 35Hz, mainly to reduce excursion on the 5.25" Wavecor WF152BD05 woofers and reduce the BL induced efficiency loss from large excursions. It also certainly helps getting more output from the speaker while reducing distortion.
However, I often have my speakers vertically, so the passive radiators are in an up and down configuration, which is bad for the suspension of the passive radiator due to the added mass. This eventually causes the surrounds of the passive radiator to sag over time.
The graphs below show outdoor ground plane measurements of my speaker. Green is taken 1.5 years ago, red is taken recently. As you can see, below 50Hz, the red curve is tilting down about 2dB vs the green curve, suggesting that over 1.5 years the passive radiator has sagged enough to reduce its output contribution by 2dB.
Lesson: Don't use up or down firing passive radiators! Not unless you like to replace the radiators every year or so, or lose output.
Actually, its bad, but it gets worse. The main negative effect is the increased distortion because of the asymmetrical suspense, lowering not only the spl but also the maximum dynamic by a lot. When I see a sag of 2dB for low output (while measuring), it's pretty safe to expect ~8-10dB lower maximum dynamic because the distortion and maximum excursion limit of the passive radiator. That also applies to subwoofer drivers and that's the main thing why you should really avoid horizontally mounted bass and especally sub drivers with high moving mass. I've told that many times but it's ignored or being laughed at for equally many times.
You can prevent it from becoming a that big issue though. It's enough to flip the passive radiator every half year so that evens it out. If the basket does not allow the reverse mount position, put it into a frame you can flip.
In general, it's ofcourse the much simpler solution to avoid such a mounting position at all and instead put the PR onto the back side. For drivers and PMs with very high Mms (vertically mounted) it's still advised to rotate it 180° (from 6h to 12h (clock)) every now and then too, as that reduces the tumbling movement too.
Reference Mini Part 6: Nuances in baffle designs – Driver location
The location of the drivers on a baffle matters considerably when it comes to performance. Take a look at the Reference Mini’s baffle design. Looks pretty normal. Many studio monitors have similar baffles. Not bad right?
Wrong
The location of the drivers on a baffle matters considerably when it comes to performance. Let’s start with baffle step loss, which is fairly well understood. Frequencies with a half wavelength larger than the width of the baffle will wrap around the baffle and radiate beyond half space. This has the assumption that the driver is located at the center of a circular baffle. In which case, the distance from the driver to any point on the edge is equal.
However, what happens if the distance to the edges aren’t equal, like in all rectangular baffles? Instead of talking about the problem for the woofer, let’s use the tweeter on the Reference Mini’s as the example, because normally one does not think of tweeters needing baffle step correction. Look at the location of the tweeter for the Reference Mini. The tweeter is about 1” away from the top edge of the baffle, so the baffle step loss starts at around 6750Hz from the top side. Doing it for all 4 sides:
+--------+------------------+--------------+
| | Distance to edge | Start of BSL |
+--------+------------------+--------------+
| Top | 1" | 6750Hz |
+--------+------------------+--------------+
| Bottom | 5" | 1350Hz |
+--------+------------------+--------------+
| Left | 2" | 3375Hz |
+--------+------------------+--------------+
| Right | 9" | 750Hz |
+--------+------------------+--------------+
As you can see, even if there is no diffraction, the tweeter’s off axis response is wildly different depending on the angle because of the wildly different distances to the edge of the baffle. Please take a look at the graph below where I averaged the raw polar measurements of the 4 sides on the Reference Mini's tweeter. You can clearly see the differences in output depending on the side. There is no way to compensate this baffle step loss accurately because it is only accurate for one angle! Therefore, the off axis response of the speaker has a different tonal balance depending on the angle, and this is harmful for sound quality as research shows consistent off axis response is more optimal for subjective sound quality.
There are 4 graphs in this picture, one for each direction, left, right, top, bottom. Each graph is the power response for each direction, which is the summed average of the off axis angles' frequency response.
As you can see, there are large differences in the low end response between the 4 directions. Let's ignore the 3500-5000Hz part which is diffraction related.
The blue line is the right side, and it is clear that it has the most low end output since it has the most baffle to the right of the tweeter that extends the start of the baffle step loss.
The red line is the bottom side.
The black line is the left side.
And the green line is the top, with the least amount of low end output.
As you can see, the midrange and woofer both suffer from the same problem, and so does essentially most speakers. What is the solution to this problem? One way is spherical baffles, then the distances to all edges are the same. However, that’s not exactly practical. The more practical solution (for tweeters) is a waveguide. The waveguide limits the dispersion so that it greatly reduces the interaction with the baffle. If the waveguide is circular, then it should be pretty obvious why it will mitigate this problem!
The location of the drivers on a baffle matters considerably when it comes to performance. Take a look at the Reference Mini’s baffle design. Looks pretty normal. Many studio monitors have similar baffles. Not bad right?
Wrong
The location of the drivers on a baffle matters considerably when it comes to performance. Let’s start with baffle step loss, which is fairly well understood. Frequencies with a half wavelength larger than the width of the baffle will wrap around the baffle and radiate beyond half space. This has the assumption that the driver is located at the center of a circular baffle. In which case, the distance from the driver to any point on the edge is equal.
However, what happens if the distance to the edges aren’t equal, like in all rectangular baffles? Instead of talking about the problem for the woofer, let’s use the tweeter on the Reference Mini’s as the example, because normally one does not think of tweeters needing baffle step correction. Look at the location of the tweeter for the Reference Mini. The tweeter is about 1” away from the top edge of the baffle, so the baffle step loss starts at around 6750Hz from the top side. Doing it for all 4 sides:
+--------+------------------+--------------+
| | Distance to edge | Start of BSL |
+--------+------------------+--------------+
| Top | 1" | 6750Hz |
+--------+------------------+--------------+
| Bottom | 5" | 1350Hz |
+--------+------------------+--------------+
| Left | 2" | 3375Hz |
+--------+------------------+--------------+
| Right | 9" | 750Hz |
+--------+------------------+--------------+
As you can see, even if there is no diffraction, the tweeter’s off axis response is wildly different depending on the angle because of the wildly different distances to the edge of the baffle. Please take a look at the graph below where I averaged the raw polar measurements of the 4 sides on the Reference Mini's tweeter. You can clearly see the differences in output depending on the side. There is no way to compensate this baffle step loss accurately because it is only accurate for one angle! Therefore, the off axis response of the speaker has a different tonal balance depending on the angle, and this is harmful for sound quality as research shows consistent off axis response is more optimal for subjective sound quality.
There are 4 graphs in this picture, one for each direction, left, right, top, bottom. Each graph is the power response for each direction, which is the summed average of the off axis angles' frequency response.
As you can see, there are large differences in the low end response between the 4 directions. Let's ignore the 3500-5000Hz part which is diffraction related.
The blue line is the right side, and it is clear that it has the most low end output since it has the most baffle to the right of the tweeter that extends the start of the baffle step loss.
The red line is the bottom side.
The black line is the left side.
And the green line is the top, with the least amount of low end output.
As you can see, the midrange and woofer both suffer from the same problem, and so does essentially most speakers. What is the solution to this problem? One way is spherical baffles, then the distances to all edges are the same. However, that’s not exactly practical. The more practical solution (for tweeters) is a waveguide. The waveguide limits the dispersion so that it greatly reduces the interaction with the baffle. If the waveguide is circular, then it should be pretty obvious why it will mitigate this problem!