Good sound in a small package

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'll admit, this is probably the most off of the wall topic I've posted:)

It's my assessment that as of late, many manufacturers are successfully stuffing decent sound into extremely small packages. I've owned a late 2016 MacBook for about a year now. Surprisingly, I was not disappointed with the sound coming from the speakers. No it doesn't have bottom end, but the perception of a little bit of bass is there as well as imaging that beats some 2 channel systems I heard at RMAF this year! Don't flame me here, this is just my personal observation.

I've also given Sonos Play:5 and Bose Lifestyle 650 auditions and can't say they were bad sounding systems given the extremely small size and footprint. Being a DIY guy, I can't say I could make something that size with the same output. While I would never trade my DIY home theater or 2-channel LX521 systems for any of these, I'd like to try and build an ultra small 2-channel system for my own engineering edification.

I assume all of these solutions hinge on DSP which is probably why these small format systems have improved significantly in recent years. But playing with MiniDSP on my own speaker projects seems good for prototyping crossover slopes and frequencies, I'm always happier with a well designed passive x-over when working with adequately sized systems.

However, with micro systems what are the tricks that can be played to create a pleasurable listening experience? I suspect peaks and dips at precise locations attribute to imaging. So things I'm thinking about:
1. What target response is beneficial when you have a 3dB point of 200Hz?
2. Systems that play loud and extend to 50 Hz in small boxes must use either low efficiency drivers or extreme equalization of low frequencies to achieve the goal. What is a better route? Is applying EQ below driver Fs appropriate?
3. Any successful DIY micro systems out there for inspiration?

As an initial seed to this thought, I measured my MacBook Pro tonight using calibrated Earthworks M30 through RME babyface. See attached average of 3 measurements around head position with the laptop sitting on my desk. This is NOT quasi-anechoic but meant to get an idea of what I'm hearing where I'm sitting..... I can do quasi-anechoic when I setup to test another build this week if there is any interest...
 

Attachments

  • Late_2016_MacBook_Pro_Touchbar_Response.jpg
    Late_2016_MacBook_Pro_Touchbar_Response.jpg
    106.9 KB · Views: 651
  • Late_2016_MacBook_Pro_Touchbar_Distortion.jpg
    Late_2016_MacBook_Pro_Touchbar_Distortion.jpg
    127.1 KB · Views: 607
frugal-phile™
Joined 2001
Paid Member
How small… we have built a number of small speakers that sound quite good. 2-2.5 litre for FF85wk or Alpair 5.2 (exension to 80 Hz or so in the big boxes). I have a set of 2.5 litre boxes with Alpair 6.2p (not mine pictured there) shaped such that i can fit into a briefcase with amp & use an iPod/iPhone or my MacBook as a source.

Not current, but (essentially) the same small box for the FF85:
microFonken.jpg


Not as extended at either end (and pricer), but still decent and very smooth is the Scan 10F 4424 fits in 8/10 litre. Really quite small.
SS-10F-nScan-Ken.jpg


dave
 
I took the principles used by the manufacturers you've mentioned (small box, passive radiators, lots of EQ), and applied them to my Fostex FE126E drivers that've seen the full P10 treatment.

New project - Fostex FE126eN in a small PR enclosure

Photobucket has eaten the pictures after they decided they don't want to be a free image hosting site any more.

Short story is be prepared to throw plenty of EQ around, and make sure its easy to adjust the passive radiator tuning at first. Dynamic EQ would probably be useful if you've got it.

Chris
 
frugal-phile™
Joined 2001
Paid Member
Our standard milliSize box is 5.5 litre if we make them sealed, 4.7 litre is we allow for the variable size slot vent. We stuff a wide arrary of different drivers in that box. Here with a set of Alpair 6.2PeN coloured Rose — with the volume near optimum for this driver, one of the best sounding combos.

253893d1323383828-minionken-fonken-picture-gallery-mmar-ken6p-red-comp-jpg


dave
 
I know a thing or two about this area :eek:

To answer a few of your questions

There is absolutely nothing wrong with EQ'ing below Fs if you do it right, and there is no better route. Unfortunately, doing it right is very difficult with DIY tools. The Sonos Play 5 measures flat anechoically to 30Hz. Yes, 30Hz! The beauty of the Sonos over DIY is that they developed an extremely effective limiter so that the woofers never reach the point of audible distortion. This is key. Without this you have a system that is easily overloaded. Therefore, the F3 of the system starts at 27Hz, and then the F3 gradually goes up as the volume goes up woofer's clean excursion limits are reached. The F3 of the system is always the lowest possible given the volume level and the available clean excursion.

This is extremely effective because it is the best of both worlds. The F3 is as low as possible when within the excursion limit, so with something like the Sonos Play 5, you get F3 of 30Hz at regular listening volumes, which is 90% of the use case. This is much better than non EQ'd speakers that would probably have a F3 of 60Hz at best. With most people listening to pop music, a F3 of 60Hz vs F3 of 30Hz is a HUGE difference as most bass in pop music is around 40-50Hz, and a speaker with F3 of 60Hz will completely miss the bass.

You can also have high sensitivity and deep bass, but Sonos used a very low sensitivity driver, most likely to push Fb lower to increase deep bass efficiency (not sensitivity, efficiency, they're very different, especially around Fb). The maximum bass SPL of a speaker is limited by driver displacement. With a smaller enclosure you just need a lot more power to hit the SPL's. The size of the enclosure has no effect on maximum SPL if you have enough power and the driver can take it. Power is cheap, so this is an easy choice to make.

You wouldn't believe this, but it is extremely hard to DIY something of the same size that'll beat the second gen Sonos Play 5 (which is far better than the first gen). The Sonos is far better than you would think. The Sonos didn't sound bad to you, because it legitimately sounds good, but because it is a lifestyle system you have a hard time believing it actually sounds good. Sonos did a really good job. They didn't use any psychoacoustic algorithm DSP tricks, just a really well designed speaker design with good quality drivers and full use of all the benefits of an active DSP crossover implementation.

If you want to see a DIY micro project, this one takes your idea to the extreme.

Pushing the limits of small speakers - The Reference Mini build thread
 
Last edited:
I'd like to try and build an ultra small 2-channel system for my own engineering edification.

I assume all of these solutions hinge on DSP which is probably why these small format systems have improved significantly in recent years.
1. What target response is beneficial when you have a 3dB point of 200Hz?
2. Systems that play loud and extend to 50 Hz in small boxes must use either low efficiency drivers or extreme equalization of low frequencies to achieve the goal. What is a better route? Is applying EQ below driver Fs appropriate?

A DSP compressor which limits the drive signal to never exceed the mechanical limit of the driver is how they do this, something I patented a long while ago and we implemented in DSP first in a phone with a vented speaker tuned to 100Hz
US Patent for Loudspeaker-dependent audio compression Patent (Patent # 6,201,873 issued March 13, 2001) - Justia Patents Search

The compressor is non-typical in that it restricts the input voltage so it never exceeds the mechanical limits of the driver at any frequency.
For a vented for example, the max voltage available follows a second order high pass function below the system tuning frequency.

When the drive voltage is high enough to exceed the speaker's available linear motion, the frequency response changes.

Attached is an example of how the response changes as drive level increases, capping the bass output (this device probably violates my patent LOL), taken from:
Oluv's Gadgets: Review: JBL Flip 3 - the UE Boom killer

The compressor is used to create the distortion in a controlled way that sounds acceptable to the ear, instead of just overloading the driver. It also uses less total power from the amp allowing the speaker to play louder in the region that's not excursion limited. The best way to define the max voltage vs frequency is to define an acceptable distortion threshold (vs frequency) and determine the drive voltage that achieves this. The patent explains how to build this into the compressor, its actually very simple and doesn't require any sub-banding.

As for the best target response, the response attached shows a slight rise from 600Hz as you move down to 100Hz, then a slight peak. The LS3/5A used a similar shaping to give the impression of better balance:
BBC LS3/5a loudspeaker Stirling Measurements | Stereophile.com
I know at least one well regarded kit that did the same thing, so the response is proven. Cutting it off at 200Hz sounds like overkill. If using a DSP you should be able to get down to at least 120 Hz even with a small driver.

Another approach is to just use a driver that can take the beating. I don't like using vents in small cabinets with small drivers because it severally limits driver power handling and the mids get veiled because there isn't enough box absorption. I built and use in my office daily a 3.3L sealed system around the Tang band W4-1720. Its hooked up to a Nuprime IDA-8 (100W) without difficulty. I spent a long while finding a driver that would simultaneously play down to 80Hz in a tiny box to a decent level, and it does a good job. The in-box near field roll off response is attached. No DSP, just horsepower. it takes a beating without complaint.
 

Attachments

  • entirevolumerange.jpg
    entirevolumerange.jpg
    68.1 KB · Views: 86
  • Nap.JPG
    Nap.JPG
    33.9 KB · Views: 99
  • NapNF.JPG
    NapNF.JPG
    37.1 KB · Views: 81
TangBand W3S-871 is a good fullrange. This is in a one liter closed box and a notch filter. However the distortion in low end gets critical. Placement near a wall and eq helps.

A Tannoy Mercury FR is a 1.5 liter BR 2-way speaker, much lower distortion but no bass below 100Hz!

Creative T20 was surprisingly good until it's potentiometers stopped working
 

Attachments

  • tb w3 spl disto.png
    tb w3 spl disto.png
    125.6 KB · Views: 72
  • Tannoy Mercury FR-horz.jpg
    Tannoy Mercury FR-horz.jpg
    147.5 KB · Views: 89
  • Creative T20-horz.jpg
    Creative T20-horz.jpg
    146.4 KB · Views: 106
Last edited:
A DSP compressor which limits the drive signal to never exceed the mechanical limit of the driver is how they do this, something I patented a long while ago and we implemented in DSP first in a phone with a vented speaker tuned to 100Hz

Cool!

A couple of questions, if I may.
Firstly, is it possible to emulate this with, say, a Behringer DEQ2496 (it has dynamic EQ, which will boost or cut a band of parametric EQ according to signal level)?
Secondly, any idea where the diagrams went from your linked patent? I can't seem to see them on Firefox, and can't find a PDF to download, either.

Thanks,
Chris
 
Good for cheap but kinda coloured. They were one of the 1st decent small FR and gained a lot of traction before they became NLA.

dave

First woofer tried in the little box was the W4-657 but its a dirty sounding little thing. Can't fully escape the effect of that gunshot peak even with traps etc.

The 1720 has a real "paper" sound to it (warm, not a great amount of detail), but nice for office listening and clear even when cranked up. Will cross reasonably high if using a notch.
 
Cool!

A couple of questions, if I may.
Firstly, is it possible to emulate this with, say, a Behringer DEQ2496 (it has dynamic EQ, which will boost or cut a band of parametric EQ according to signal level)?
Secondly, any idea where the diagrams went from your linked patent? I can't seem to see them on Firefox, and can't find a PDF to download, either.

Thanks,
Chris

From googling the patent #:
Patent US6201873 - Loudspeaker-dependent audio compression - Google Patents

BTW, ignore the "adaptive filter". That was an echo canceler used in this particular application but not needed for just a speaker.

Never used the 2496, sorry can't help you there. If you find its possible, please post a follow up!
 
From googling the patent #:
Patent US6201873 - Loudspeaker-dependent audio compression - Google Patents

BTW, ignore the "adaptive filter". That was an echo canceler used in this particular application but not needed for just a speaker.

Never used the 2496, sorry can't help you there. If you find its possible, please post a follow up!

Thanks for the link. Should probably have searched that myself.

If it can work with the features offered by that unit, there are a few similar ones that would open up a lot of opportunities.

Chris
 
I love my TB W4-654 as a mid range with an SDX7 and ribbon tweeters from CSS. No longer ‘small’ though!
But I do plan a 2.8 litre vented box for some 4” SB Acoustic coaxial drivers (SB12PFC25-4) from Madisound.

Did you ever get around trying the SB12PFC25-4? If so would like to get your feedback, and also if yes, did you build the crossover as suggested by SB Acoustics?

Thanks!

Steve
 
If you want to go small then you tune low as possible (long port) and add power with equalization above the fb. Then add a high pass filter around the tuning frequency for peace of mind. You might want a high wattage driver with a low fs to help control harmonic distortion, also a low qts is generally preferred. This is easy with equalizer apo.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.