Many of our decisions were based on what we had driver compatibility for. Software ALWAYS took a lot longer to get right at Sonos than hardware, largely because of the huge number of internal and external compatibilities we were required to maintain.
We had a working amp + DAC signal chain with the TPA series, so we reused that. We also had working wifi drivers with the MT7613. Much of my job was convincing the RF people that they didn't need to break out all the RF chains, which would have tipped the product into unprofitable territory.
A lot of the design decisions were based on hardware cost and speed of development. We reused a lot of subcircuits we'd previously validated on other designs.
Not really - specs and sound wise, it's not any different from any other integrated class D analog input amp. Pretty run of the mill THD and SNR specs.
It also has one hell of a startup transient. It was not a fun exercise to get that passing startup click/pop requirements. I said some words about TI in anger while bench testing.
Analog Devices is making some neat digital amp chips that have an I2S digital feedback path from the chip to the digital source (processor or similar). The idea is that you can use this to digitally correct for imperfections in the transducer.
Never used them in depth, but the demo they showed us on an EVK was neat. You can trivially pick up 6dB improvements in SNR, or more with tuning and characterization.
My beef is really more centered on the fact that, if you're already at -100db of SNR, what's the point?
I frequently see efforts that use cheap, popular, globally available off the shelf hardware for more than it was meant to. I love that and it really showcases the benefits of hackable, modifiable things. Someone could make this as an argument of user-serviceable, accessible hardware instead of sealed units that only the manufacturer can open. Also, do we need to talk about "warranty void if seal broken" stickers?
~19 years ago my first real introduction to software/hardware/firmware hacking and electronics/soldering was turning the digital, one-time use "disposable" cameras that CVS etc. sold, into reusable digital cameras. Plenty of issues along the way and drawbacks with the final product, but for the time it was one of the cheapest digital cameras you could get (because the price was heavily subsidized on the fact that you would be returning it to develop the pictures, at which point it could be cleared and sold again). That definitely helped put me on the tinkering path I'm on now.
> Someone could make this as an argument of user-serviceable
I think the opposite, really. Selling overprovisioned hardware, with differentiation between models happening in software happens because they devices are locked down and aren't user serviceable or intended to be modified.
Users upgrading cheaper, over-provisioned, possibly loss-leading hardware to get more expensive functionality means no one would buy the more expensive model, which would drive up the cost of the cheaper one.
> Users upgrading cheaper, over-provisioned, possibly loss-leading hardware to get more expensive functionality means no one would buy the more expensive model, which would drive up the cost of the cheaper one.
No, not really. If you think tinkering people who would open up their IKEA Symfonisk has a nontrivial overlap with Sonos customers, you're in for a surprise. Richer "nice off-the-shelf device" people will stay that way, DIY with an IKEA speaker people will also stay that way.
What it does is it really encourages tinkering, thinking and innovation in the small fraction of people who would do that. It breeds better engineers, more engaged teenagers, a more flexible society. So really - all the positives and almost none of the negatives.
Gone are the days when a $100 processor could go toe to toe with a $360 processor. I'd say that's due to the fact that overclocking is gone and the main differentiation is number of cores, which can't be unlocked these days.
I'm all for that as well, except I have become slightly more sympathetic to the warranty aspect. Yes, we absolutely need "void if seal broken". I would love for something to be released that was fully intended to allow tinkering, but once you do so, you get no warranty service as there's no way to know what mods were done. I think that's totally acceptable
Invalidating a warranty based on such seals is in fact illegal:
Generally, tie-in sales provisions aren’t allowed.
These are provisions that state or imply that a
consumer must buy or use an item or service from a
particular company to keep their warranty coverage.
Here are some examples of prohibited tie-in sales
provisions.
“To keep your new Plenum Brand Vacuum Cleaner
warranty in effect, you must use genuine
Plenum Brand Filter Bags. Failure to have
scheduled maintenance performed, at your
expense, by the Great American Maintenance
Company, Inc., voids this warranty.”
“This limited warranty shall not apply if the
warranty seal has been broken, removed, erased,
defaced, altered, or is otherwise illegible,”
where a device cannot be repaired without such
effects.
That approach has many perverse incentives. Fortunately, we've already figured this one out: manufacturers should only be able to disavow their warranties if they can prove that a modification caused the need for a repair.
Which, maybe an appropriately placed sticker can help determine. Note I said help. Whether the sticker is enforceable or not, it's an easy way for the manufacturer to know whether the user even accessed internals that aren't required for normal operation, streamlining some warranty cases.
TL;DR - invalidating because of sticker bad, using present sticker to fast track warranty cases good.
Invalidating because of sticker - bad. So sticker saying so is also bad. OK if sticker says "Breaking seal at own responsibility" might be a measurable improvement.
Let's not pretend we haven't seen 1001 cases of "You've upgraded your laptop's RAM, so the laptop hinges that we see break in thousands of laptops is your own fault" (Dell).
Sure, I kind of assumed that when used appropriately in this way it would say something appropriate that wasn't untrue as well, such as "warranty implications if broken", which them people might at least be interested in looking up what those implications are (presumably longer turnaround time as it needs more investigation).
I did the same hack, except using two Symfonisk's and just connecting the existing speaker lines to a pair of floorstanding two-way speakers (Kef iQ70 from the 2010's).
The little amp in the Symfonisk is plenty powerful to drive these large speakers. The crossover is completely off by default, but by using Trueplay the crossover gets tuned properly and the sound is pretty damn good (albeit a tad boosted in the highs).
It's a great way to get super convenient (Airplay 2) and high quality sound in a room, just with a power cord for each speaker. And these quality Hi-Fi speakers are readily available for cheap second-hand... some pics here: https://twitter.com/jstsch/status/1529204823912075264/photo/...
So if you open up the Symfonisk, you'll find that the cables running up to the tiny tweeter and woofer speakers have a nice flat crimp connector on them.
So simply unplug those (carefully, use tweezers to squeeze the connectors and they'll pop off without friction), and connect those on some cables which you'll run to your speakers. The connectors can be found at your local electronics store.
The speakers you'll be connecting need to have a crossover that you can bypass (typically four connectors on the back versus two, with a piece of metal in between to short them). Total upgrade would be to mount all the electronics inside, but I haven't dared to cut mine open yet
Instead of modding it into a new amp like this post does they likely just spliced the speaker wires right into the speakers. The upside is it was easy the downside is you can’t just plug and play after the fact.
Indeed! Nice thing is that two Symfonisk's give you great stereo sound on external speakers. And no soldering inside the Symfonisk, just some connectors to unplug and run the cable out of the bass reflex port, so easy to restore in original condition if you'd want to. So thanks to the magic of these tiny transistor digital amplifiers, in total you'll spend ± 500 euros and have better — and super convenient — sound than quite expensively priced systems.
Yeah, the article doesn't make it at all clear what the point of this project might be. I gathered by the end that the idea is to rip out the guts of an Ikea speaker, add some binding posts and a custom case, and come away with a hacked-up, 1-channel replacement for the Sonos Amp.
Tell you what scares me though: I don't like seeing that they doubled the amount of DRAM in the new one. Hopefully they did that only because it was becoming impossible to source the 256MB module. I hate to see it because if they allow themselves to use it, eventually there will be a software release that doesn't work on the older 256MB model, and they'll deprecate it.
Both IKEA and Sonos have been decent in terms of support of older hardware, so I wouldn't be too wary of that. But, companies rarely feel incentivised to maintain older products these days.
I am pretty happy with Sonos support but they did already EOL all their original products up to everything with less than 128MB of RAM, on the basis of not being able to run the newer software. Those older platforms also had SuperH CPUs, which might have been part of the problem.
My Playbar from god knows when (5? 6? years ago) still gets updates and is on S2 now. But then yeah, there are the older devices which are EOLed.
IKEA still supports 5-year old smart switches they no longer produce even with their latest smart bulbs. But no idea if that will continue with their move to Matter (or whatever the new hotness is called).
It is almost certainly because its easier to source larger chips. The hardware product I work on has twice the RAM it needs, and an order of magnitude more eMMC than it needs, because we couldn't get smaller parts.
I love these kind of hacks. Nicely done. However I must disagree with one thing in the original article:
> Brilliant sound
I have a Symfonisk bookshelf speaker in my kitchen, and I don’t think the sound is brilliant. It is too boomy for a speaker that is supposed to be placed near a wall or in a bookshelf. The low price forces some compromises and I suspect this is one.
I've listened to one at a IKEA shop and think its quality is mostly okay (speakers placement can change everything), but if I needed one I would certainly build it myself for much less money using cheap easily available ready made modules to convert to wireless normal amplified speakers.
This is a field in which DIY is still viable; the other day I posted a link to a data sheet of a set of modules promising quite serious performance (digital uncompressed audio with sub-millisecond latency), but sadly can't find any seller from whom I could purchase a few for testing.
I returned mine after a week, the HomePod Mini produced superior sound even in an open area for less money. The Symfonisk was capable of being loud, but the highs were very muffled.
I'm not too familiar with Sonos, but was under the impression that they are just high-end wifi speakers with some proprietary system.
And that Ikea's version is again the same idea, but at a lower cost (and less proprietary?)
So what is the actual benefit here? Is it upgrading the speakers of the Ikea model? Or is it adding an amp and wifi to unpowered speakers?
Or more specifically, creating a ceiling wifi speaker for cheap?
Because step way down from the Ikea speakers, and you could grab a few Echo Dots for (alarmingly) close to free, and line-out into any powered (or amped) speaker, which would seem to be about the same result? And at least with Spotify, multi-room works quite well.
This of course ignores privacy concerns, which keeps costs down.
I have the same question. Looking at the original post, the impetus seems to be "save hundreds on Sonos". I'm not that impressed with Sonos so this is like saying save hundreds on Bose.
"This laid all the technical groundwork to prove that the Symfonisk speaker should work as a general purpose mono speaker amp." Then he goes on a crazy mod and solder run that almost no one could or should ever do.
He wanted to power some ceiling speakers. I'd rather just buy a TPA 3116 amp with bluetooth and go from there or use echo dots like you say.
I've got the DAMGOO in my greenhouse and it sounds damn fine. $25.
It's for people who are all-in on the Sonos ecosystem, which supports stuff like multi-room audio (listen to your podcasts as you walk around the house), integration with music services etc.
If you just want a single pair of speakers in one room and are fine just casting off your phone, of course Bluetooth is fine.
I’ve got one of their small Play:1 stand-alone speakers from several years ago and it’s almost magical how room filling the sound is from a single little can.
Sonos compatibility. The $120 Ikea product is effectively a $320 Sonos Connect - which can connect to any arbitrary speaker - attached to a cheap bookshelf speaker.
This mod removes the Ikea speaker and adds the connectivity required to the Sonos component to effectively replace a Sonos Connect.
Because Sonos doesn't offer a $120 device that replicates this functionality, it effectively allows you to invest an hour or two of soldering and circuit board modding, and optionally $60 in 3D printing, to replace a $320 device.
Has anyone found a cheaper way to get stereo digital output (than the Port) to replace an old Sonos Connect? I have a nice DAC in my pre-amp, so I’d like to use that with streamed digital audio. My original Connect was very expensive and is now incompatible with newer S2 functionality. I have never understood why Sonos devices costs more without an amplifier or speaker! Presumably because they can?
Would this same approach work for you? I haven't looked, but presumably you could pull the signal before it hits the amplifier, and make it work that way...
This IKEA wi-fi/Sonos bookshelf speaker has a stereo DAC (per the article), but I would assume it's designated as a mono Sonos output device in terms of software, so only one channel (left) is probably active.
That's a good point. If it's like the other Sonos mono speakers, you can (in software) designate them as a stereo pair working together in arbitrary configurations - which would suggest that there may be a way to do it.
This is a great hack, however I really prefer the easier version of simply connecting the symfonisk board to a decent/great bi-amp speaker.
See, in their normal state these Sonos speakers are mono, yet they have this stereo amp in them. What Sonos does is use a crossover with that stereo amp to make it a mono bi-amp (in other works instead of left/right, it's outputting high/low to the tweeter and woofer respectively). If you simply connect one side of the amp to a normal speaker, you'll either get muddy bass or incomprehensible chirping, because the crossover splits the audio)
The whole soldering part of the mod here is to work around that fact, and bridge the audio back to a single mono channel. While this has some advantages - more power, and connects to 'regular' speakers, it's a lot more wrk, and a lot riskier.
This idea sounds appealing, but wouldn't that also (generally) require standard power (115v/60hz stateside) to run the amps? Not having to worry about having an electrician run the appropriate 'high voltage' (in a residential wiring context) power to ceiling/wall speakers is appealing to me. I'd love any additional resources you have about this method though.
Oh, I understand. For some reason I was thinking self-powered (internally amplified) speakers that were being driven by a line-level source off the Sonos. Not sure why that's where my mind went, but appreciate the clarification!
Nowadays, amplifier modules often have built-in DSP capabilities, eg to psycho-acoustically improve bass or better match the speaker. Has it been verified that this is not the case? Otherwise you'd get a signal that has been altered and is likely a mismatch to any other speaker other than the Symfonisk.
These do use such a DSP, but it's not a fixed response. Sonos speakers use the phone amp with a process they call TruePlay to tune their speakers, and TruePlay still works with this after the mod, so you can retune the DSP for pretty much any kind of speaker and room combo.
The benefit of doing this is the Sonos compatibility. If you like the ecosystem but don't like the packaged speakers or the cost of their standalone amplifiers, this is a cheaper path to using your own speakers in the Sonos world.
I'm not sure that it will be fully compatible. Sonos has a feature to tune/calibrate your speaker setup for your environment. That's to reduce room gain and in general to get a better flatter response for your environment. And I'm not sure that after switching the amplifier into bridge mode that functionality will work correctly
There are a bunch of mod articles and YouTube videos about doing mods with the Gen1 and bi-amp speakers - or some even turning it into a subwoofer amp - and trueplay apparently still worked.
That said, removing the bi-amp capability may technically break trueplay, but I don’t see how. If it worked in single channel mode as a sub, it should work bridged like this.
[edit] just looked at his gen1 guide that does the same mono-bridging, and it claims trueplay still works. Which makes sense - trueplay is a feedback loop. With the speaker playing specific frequencies at specific volumes, and your phone listens to it and adjusts the DSP to make those frequency/volume pairs match what it expects.
The Sonos Amp + two built-in speakers with equivalent functionality is like $1,200. This lets you use a $100 speaker as that amp (or two of these setups for stereo sound) and then you can use much more reasonably priced built-ins that are still Sonos compatible. It's only useful if you're in the Sonos ecosystem, but it's pretty useful if that's the case.
Not all amps are equal. The Sonos Amp can competently drive 4 beefy in-wall/ceiling speakers. The Sonos Amp isn't just the guts of this product marked up 5x.
Either way, it's silly that Sonos doesn't have a simple un-amplified receiver for similar setups. Perhaps they are afraid consumers would get confused about non-amplified products given all their other products are amplified.
Either way, it's silly that Sonos doesn't have a simple un-amplified receiver for similar setups.
They do have that, the Sonos Port, but it's extremely overpriced at $449. They very gracefully do allow you to get a full two channels of audio (stereo) though.
I used the same mod to turn a few older, passive hi-fi speakers into Sonos-enabled wireless speakers. They sound (and look) a lot better than the Symfonisk, and putting the DAC and amp right into the speaker enclosure is more convenient than an external amp and wiring.
If you use them to bi-amp speakers with two drivers, the TruePlay calibration works great.
For good and easy multiroom sound obviously. It's a killer feature of Sonos. And in this particular case it's a ceiling speaker.
Stereo is not a problem here since a single Symfonisk is a mono speaker and you need a second one to get stereo. That won't change after the mod.
Seems to be most important that this amp is small and sound okayish. Still better spend your money on old serviced piece of kit that will likely survive several decades. I haven't had any new amplifiers that sounded as good as my old 70s Sony TA over-engineered amp.
Question: Is the Sonos thing that you can walk around your house streaming playing whatever on your phone and the speakers know where you are and automagically turn on, connect, and play the sound?
Start listening to something in the bedroom, head downstairs for breakfast, bedroom speakers switch off as kitchen speakers take over?
Hmm, hacking in a stereo line level out seems more worthwhile. Chip amps can be found super cheap, and will likely give better results (as would a traditional amp)
But there isn't a traditional stereo signal in there - there's a mono signal fed to a crossover which splits the high/low audio and feeds it in to the amp as 2 separate channels. The (normally stereo) amp is then used to feed the tweeter and woofer instead of L/R.
Nice work. Warm fuzzies.