De-smarting the Marshall Uxbridge
tomscii.sig7.se277 points by fanf2 a day ago
277 points by fanf2 a day ago
If you're going through all the effort to design a PCB have you thought about driving the I2S input digitally? I skimmed through the AD85050 datasheet and it has internal DSP functionality which would have been already tuned for the drivers and box by Marshall. The reason powered boxes sound decent despite their relatively cheap hardware is because of the extensive processing they have in the background to compensate for any hardware defiencies.
As the AD85050 has a stereo I2S input there's a possibility for the actual crossover to be either done on the amp chip itself (with the same signal driving both channels) or done on the Amlogic SOC. The latter would be ugly as you would need another DSP chip on your board to do the crossover functionality, or perhaps you could program the AD85050 via I2C to add the appropiate low and high pass filters.
A two channel A/D converter would work on the front end, as you could drive both channels with a single analog input to get a stereo I2S out with duplicate channels to drive the amp. A USB input would be much messier if you want true stereo using two speakers unless you plan on doing routing on the software side. With SPDIF you probably could get away with splitting the signal and using a SPDIF to I2S converter chip in each speaker, but you would still need some way to separate out the left and right channels. The AD85050 has mixing functionality via I2C which may help with that.
And of course, all this might be more work than desigining an amp in the first place, and it really depends if you want to explore the analog or digital side of things.
> The reason powered boxes sound decent despite their relatively cheap hardware is because of the extensive processing they have in the background to compensate for any hardware defiencies.
I will not argue that that could be one ingredient, but a couple of months ago I did a toy for my kids, I bought decent speakers, placed them in a cheap plastic box, and was absolutely amazed bybthe sound quality. The amplifier is a sub 1 dollar class D bought in a Raspberry Pi shop. No processing at all. If the box is sturdy and sealed, and the speaker is good, is incredible what you can do.
Most cheap amp ICs perform well when they're outputting less than a watt, with distortion barely audible. Try connecting the same amp board to your main HiFi system if you have one and do some listening tests against the original HiFi amp. Then turn it up and it's a completely different story.
As always the speakers are the crucial part and having decent speakers will make a big difference. What a DSP can do is correct bad speakers to some degree. A typical cheap computer speaker has a muddy midrange, can't reproduce past 13 kHz or so, and has little bass due to the small driver. With DSP the manufacturer would typically low pass the amp input, smooth out the nonlinear frequency response, lift the bass a bit, and apply compression and limiting to increase perceived volume and protect the system. The results are still constrained by physics but the manufacturer is in this case able to save money on the drivers and box while getting similar sound quality.
I'm interested in which speaker and amp those were. Also, the plastic box :)
The amp: https://www.reichelt.de/de/de/shop/produkt/entwicklerboards_...
The speakers: https://www.reichelt.de/de/de/shop/produkt/breitbandlautspre...
Box: https://www.reichelt.de/de/de/shop/produkt/gehaeuse_serie_op...
Design of a friend, final product looks like this: https://hackaday.io/project/198249-untonie-antony
Not the OP, but if you have a little budget, HifiBerry's AMP2 [0] sounds great. After my dad gave his Hi-Fi stack to me (due to having no space at home), I built a small system with this and connected to a set of passive 2.1 Kenwood Hi-Fi speakers for him. They sound amazing, plus HiFiBerry OS is superb for connectivity.
I just want to note that software is built with collaboration of Bang & Olufsen. Both hardware and software oozes quality.
[0]: https://www.hifiberry.com/shop/boards/dealing-with-blocked-p...
The hifiberry documentation is a mess. Let’s say you want to get rid of your Sonos system and start at zero with a Pi (it’s not my first one)… which features could be replaced by hb? What hardware should I buy? Which licenses?
I don't have a Sonos system, but I'll try my best to run it down for you.
Platform has two starting points: Raspberry Pi + AMPs (or DACs) or Beocrate [0]
If you go with the former, you need a Raspberry Pi 4 or 5 (even 2GB models are OK), HiFiberry board of your choice, a case (available from HifBerry), and a PSU in 19-24V range.
My setup is as follows:
- Kenwood 2.1 passive Hi-Fi speakers.
- HifBerry AMP2
- Raspberry Pi 4 / 2GB
- Steel case for Raspberry Pi + Board
- 20V Meanwell power supply with a barrel jack.
If you want go all in, you can add a DSP into the mix which allows parametric eqaualization, and room compensation with optional USB Mic (similar to how B&O speakers measure room to self-optimize). Also you can design your own DSP chains and upload to the DSP.
Regardless of the OS you run (HifBerry OS or HifiBerry OS64 (which is beta)), you get the following services out of the box:
- Roon
- AirPlay
- Spotify Deamon
- Spotify Connect
- RadioBrowser & TuneIn support
- Squeezelite for Logitech/Squeezebox
- Music (for local files)
- Bluetooth
There are also extensions, but I don't use any of them and don't know what's available.
I use the system mostly with MPD to play music from my and dad's personal collections. MPD can also connect to a Samba server.
If you have any other questions, I will try my best to answer. Hope this helps.
It's been shown that, at least for guitar speakers, the box they're in doesn't matter at all. The entirety of the sound quality or lack thereof is in speaker itself. Of course that is only one speaker , no crossover to worry about.
Jim Lill has a video on Youtube testing exactly that. It demonstrates conclusively that cabinet design for guitar speakers makes a difference:
https://www.youtube.com/watch?v=-eeC1XyZxYs
This is hardly surprising; cabinet design matters for every kind of loudspeaker. Note also that electric guitars can produce a wide range of frequencies, especially once you add distortion. Distortion generates additional tones both higher and lower in frequency than those already present.
Oh man, I've got some egg on my face for this one. I actually have a guitar cabinet that has a special geometry that disperses the sound in an area instead of directionally like guitar speakers normally do, which I bought explicitly for that purpose! Despite this, I still said what I had said, a couple posts up. How embarrassing.
That's entirely untrue, particularly at lower frequencies. This isn't audiophile cork-sniffing - the behaviour of a loudspeaker can be radically altered by the design of the enclosure. Entire textbooks have been written on the topic.
https://www.amazon.com/Introduction-Loudspeaker-Design-John-...
Marshall speakers, from my experience, has a brand sound signature, and that tuning is not very optimal for every genre of music.
Replacing the DSP with a simpler amplifier may allow to get more detailed sound from the drivers and the box themselves and may create a more pleasant listening experience.
From what I have seen, the drivers seem pretty full-size for that box, and any disturbing sound characteristic can be tuned with a simple equalizer. A more dynamic approach might create audibly weird sound profile if done wrong.
Modern DSPs are magic, but I still prefer an audio pipeline where things show their deficiencies and not hide things real-time.
I have the same thoughts about the approach, and I'm actually working (on the back burner) a similar thing. It's a harman kardon "smart" speaker with a similar design where the brains are on a separate daughterboard and that's now fried.
I've already figured out the control signals and have designed a new daugterboard with an ESP32 to drive the I2S output. I just need to figure out how to downmix the audio to mono and to DSP the L/R channels into tweeter/bass outputs, or to find some code already out there that does this. Any help/pointers here would be appreciated!
One thing you might find helpful is to prototype things with GNU Radio and a GRC flowgraph. I'm not sure that would be useful for running on the ESP32, but you could at least tinker around with signal processing tactics that you could implement on it.
I assume you mean AD85050 (rather than AD8255). And yes, the last paragraph before "Going all-in" is about the idea of driving the I2S. But the I2C config sent to the ESMT chip would have had to be reverse-engineered as well...
I have this speaker and cannot believe how annoying the smart features are. I'll be talking on the phone in my apartment and the speaker will think I'm trying to prompt it.
"SORRY. YOUR DEVICE IS NOT CONNECTED TO THE INTERNET. PLEASE CHECK YOUR BLUETOOTH SETTINGS AND TRY AGAIN." (at max volume!)
It's unbelievable. I'm not an EE, but would love to know how I can disable these incredible unsmart features.
There's an article that tells you that. I believe it's linked somewhere above this comment.
The article removes all the smart features (not just the annoying ones) and requires pretty serious knowledge of analog electronics. Probably it doesn't fit the bill for the parent comment.
> and requires pretty serious knowledge of analog electronics
If you want to understand the whole thing in depth, then yes, I guess so. However, at the end it just links to the already made project published at MIT license that you can simply replicate with barely any knowledge. It's an equivalent of self-compiling a software project after checking out its repo, which sure, may seem overwhelming if you never did that before, but ultimately it boils down to some reading comprehension and step following exercise.
>It's an equivalent of self-compiling a software project after checking out its repo
I would not assign this to someone as their first electronics project. They would have to order all the components and get the PCB etched themselves. There's a fair bit of soldering, and a usable soldering iron is not cheap. And there's no undo button in hardware: if you solder something in backwards and pop it, you get pay for a new component and wait for it to ship to you.
Why would you bother with etching such PCB yourself if you can order 10 of them for $4, and that's including shipping? Get two sets of components in case you screw up and then just solder all this exclusively through-hole and one-sided stuff with cheapest iron around.
The hardest part will be ordering the components, or more specifically verifying that you got them right, as the KiCad project does not include a BOM. Other than that, it's as easy as it gets. Entry-level workshop stuff, about as complex as ArduTouch. People who never held a soldering iron in their life successfully learn on this kind of boards.
(and if you're willing to spare a few more bucks you could even get it assembled by a PCB fab and just receive ready-made boards in your mailbox - the prices aren't prohibitive there either)
I mean, it's a speaker in a box, so you could also just snip the speaker wires, ditch the circuit board, solder some extension wires, and plug it into an external audio amplifier box of your choice.
If you go that route you don't really need much EE knowledge.
(This is also only if you already have this box and want to reuse it. Otherwise I would just go to your next neighborhood garage sale and pick up some good speakers for $10)
> I'm not an EE, but would love to know how I can disable these incredible unsmart features
I am an EE, so let me tell you.
Use a hammer.
The author of the article starts out asking, "why would someone throw them out like that?"
Now we have the answer.
You can replace the innards with those of another, stupider, Bluetooth speaker.
...with the risk that you may get "THE BLUETOOTH DEVICE IS READY TO PAIR" instead, although others have figured out how to replace the prompt sounds on some series such as the JieLi SoCs.
> "THE BLUETOOTH DEVICE IS READY TO PAIR"
And now I wish HN had spoiler tags and a culture of using them along with trigger warnings...
Lately I've been wondering if there's a way to do this to Smart TVs. Personally, I like the name "stupify" better ;)
A year or so ago I bought a Hisense 65U88KM, which comes with Google TV. During the setup procedure it asked me if I wanted to enable the "smart" features, such as Google TV, the camera and microphone, or connect it to a network. I said no to all of them, and that was that.
Now it just acts as a dumb screen for my Apple TV box.
I did the same with a Sony A80L, which also runs Google TV. I even uninstalled the bundled streaming apps for good measure, although I never see the home screen.
It behaves like a monitor. I never see the TV UI unless I ask for it.
My LG could be configured to behave like this (and believe me, I am thinking about going that way).
I have a cheap FireTV which cannot be made to behave this way. If you disconnect it from the internet, it will still require you to interact with the (slow-ass clunky) OS in order to select a different input.
Your best case scenario with some of these smart TV's is the ones which run Android to replace the launcher. Possibly, this gets reset periodically, meaning you have to keep doing it.
Apparently there's a few Fire devices which can be flashed with LineageOS - I might try researching that and see if it is doable. A FireTV stick with LineageOS would be the best case scenario.
I find that most of those reset to some nonsense occasionally or whenever the power goes out. I make sure they have no internet connectuon, but I usually have to dig up the remote to get back to hdmi1 so my device interface will come back up. I accept the annoyance because I accepted the discount that they give to have all of the spyware crap on there that I am blocking. I wish they could sell something cheaper that is just a display, buy product managers will be product managers.
> I wish they could sell something cheaper that is just a display…
The "smart" part is what makes TVs cheaper, since that's what opens the door to a higher CLV. People willing to put their money where their mouth is can still buy dumb displays: https://www.bhphotovideo.com/c/browse/Multimedia-Displays-Di...
What exactly would that do? Just select any video input, attach your preferred input device and never touch the TV remote again.
There’s a project to load OpenWRT onto LinkPlay A31 [0], might be easier than basically replacing the insides..
> I liked it even though the sound reminded me of the “disco smile” (hollowed out mids), but I chalked that up to overly consumer-friendly default EQ settings.
The much simpler explanation is that it has hollowed out mids because it's a Marshall.
I'm completely unfamiliar with Marshall hi-fi equipment (that's kind of recent?), but their guitar amps are not really known for being inherently scooped. At least until the mid 70s, maybe even the JCM800 era, Marshalls were known for having more mids than their mainstream counterparts (Fender and Vox).
Practically every guitar amp's "netural" settings are scooped to an extent, though. People like that sound and it's "traditional". Even when people play direct into a mixing desk, I find that they almost always scoop the mids a bit, it's simply what you associate with electric guitar.
For those (like me) who are unfamiliar with this device: It looks like a Marshall amp, but is a 9" tall Bluetooth speaker.
Why did he replace the class D concept by a more power-hungry and heat-generating class B amplifier?
I bought one of these Marshall speakers to use as a practice amp, only to find that the dsp latency is unusably bad. Maybe this is an option…
Tangential question, I have a nice audiophile so and no speakers at the moment. I want some speakers for $low that have decent performance. I can build boxes (used to be a carpenter), but need a design and what look like expensive drivers and crossovers. Is DIY speaker construction actually a cheaper way of getting a top notch hifi, or should I just buy second hand?
For a nice tower or woofer box, building your own is usually way cheaper than buying new but rarely significantly cheaper than scavenging or second-hand purchases. There are tons of designs, but in general just build based off the specs of your drivers and the area you'll be placing them and you'll be alright. For specialized types like electrostats, building your own is pretty much the only way to go unless you're ready to drop 10-20k on commercial offerings.
This website is a good starting point: https://www.audiosciencereview.com/forum/index.php?pages/Rev...
Click on the magnifying glass with an "A" to get to the advanced search to be able to filter by price.
What do you consider low price? For sheer quality, it's pretty damn hard to beat the price of Kali studio monitors.
I'm really thinking of 'low enough that my wife doesn't kill me', which is a variable amount!
Thanks for the Kali tip. They are a little more expensive in Europe, but still a possibility.
I guess re DIY I'm looking for that mythical thing where it is still cheaper to do something yourself rather than buy it!
I'm curious about price - sure, the speakers were free ($240 value), but I don't think printing up a PCB is cheap, and those are some pretty big capacitors.
JLCPCB have changed the game. Five 2-layer PCBs of up to 100x100mm cost just $3.50 including global shipping. Things get more expensive if you stray from their standard specs, but you're still looking at just a few dollars per board.
The biggest electrolytic caps in this circuit cost $3.29 each in qty 1, but they're fancy "audio-grade" Nichicon caps; a standard-grade capacitor of that size would cost you $1.68 if you want a Japanese brand, or as little as $0.36 if you can settle for a Chinese brand.
And you can even order all or the SMD components soldered for less money that you can buy solder and other consumables for soldering never mind the time.
I recently got 20 reasonably complex 4 layer gold plated pcbs assembled for about $10 a piece by JLCPCB. Maybe 20 items on the BOM, 50 parts total.
It’s insanely cheap. 5 years ago when I was last regularly getting PCBs built it would cost 10x that. And it would be a really manual process - loads of emails back and forth. PCBway have managed to automate basically the whole process.
I think the biggest cost is labor. Dumpster diving for speakers and then spending dozens of hours in high skilled labor to replace the insides is actually hilarious. I wonder how much he'd charge a client for a project this size? Probably many thousands of dollars.
But I can't knock a man for having a hobby. Clearly they're optimizing for fun and nerd cred, not cost.
Article suggests that complicated tools are annoying and not right to be that complicated. But one can argue, that if such complexity is actually needed to tackle tasks (like the multiplatform compilation of the Linux Kernel for example), then it is pretty obvious that in the optimal case the learning curve is at least in a linear relationship with the complexity. If given complexity is high enough to surpass the related learning time threshold of the given person then he/she/whatever is going to be annoyed for sure.
There is no escape from this.
>I was delighted, and immediately impressed by the sound quality. The low end was punchy and deep, much deeper than the modest dimensions led me to believe. The high end was crisp and detailed.
Its amazing what a brand name does to ones ears, all of a sudden a small plastic box with SBC Bluetooth codec sounds great.
This looks an EE‘s approach who hasn’t had a lot of exposure to speaker design. You need to consider Thiele/Small parameters of the chassis, enclosure volume, baffle design and a million other factors to design a proper crossover. You can’t just ltspice your crossover and call it a day. VituixCad would be a more appropriate solution. And then you actually have to measure!
Replacing and amp and „smart“ crap is easy if there is an analogue crossover you can reverse-engineer, if it was just some DSP things get difficult quickly (unless its just a single broadband chassis, but even then…).
And no, you can’t use pre-built „standard“ crossovers or some calculator on a website either.
But other than that, nice that he saved some hardware.
Do not underestimate audio circuit tuning based on listening tests. Good ears and patience can substitute a lot of lab equipment dollars, especially for a hobbyist.
I would say its the precise inverse: very few people can do it well by ear, and it takes a lot of practise and experimentation. At the most basic, try matching the volume level of a subwoofer to the main speakers - even this is already very hard to do by ear, for reasons of human auditory perception and psychoacoustics.
You don't even need a lot of "lab equipment dollars", measuring the basics can be done with ~100€ calibrated USB mic. As a rule of thumb, you cannot develop a good speaker without measuring, unless you have advanced modelling tools and experience to use them correctly, in which case the measurements will mostly match the simulations.
The best acoustic devices in the world have been made and tuned by hand for thousands of years. "you cannot develop a good speaker without measuring, [modelling, or simulations]." Is fundamentally untrue despite what the cybernetic totalists may want to believe.
>The best acoustic devices in the world have been made and tuned by hand for thousands of years.
Says who? You should really acquaint yourself with current research, starting with Toole's infamous "Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms".
> Says who?
Anyone who has read about or experienced instruments and music halls that were made before the commodification of electricity?
That book sincerely answers that sort of confusion. Audio production (here music creation) is very different from audio reproduction. There is no real way of judging the former, except for room acoustic which has been scientifically guided for a long time.
Yeah but there's a pretty important caveat here. The person who's actually doing the listening is the only person's opinion who matters about the quality. This is very much of a case of "if it sounds good it is good" That's why you're doing here because that is literally the only person who needs to be satisfied with this arrangement.
Thats valid.
My point was more along the lines of „a layman reading this might draw the conclusion that this is the proper way to solve this problem, because the text is written with an engineering mindset and demonstrates a certain degree of sophistication“, and I wanted to caution people that this is not the case.
> You can’t just ltspice your crossover and call it a day
Turns out you can, they just did and are happy with the results.
Truly. When will audiophiles get over their depression and realize we solved audio like 30 years ago? Sorry, silicon got too powerful and too good, audio was just no match.
This is kind of true, yet it is not the case that those competently designed speakers dominate the market. It is dominated by awful Bluetooth boxes, terrible headphones and expensive snake oil.
Which means most people can’t tell the difference. This is mildly upsetting to the people who can.
Do you know of a way to convert a measured speaker model into a spice model that can be simulated with the rest of the circuit?
You seem to be very knowledgeable about this subject. I have some Google Nest Audio speakers that sound fantastic, but have the same problems as bluetooth speakers, and lack a 3.5mm input jack to convert into normal speakers. Do you have any recommendations on how to do this, for someone with minimal audio knowledge and some basic EE?
The Google Nest Audio speakers are kind of a special case. They only sound good because they use a sealed, extremely rigid cast aluminum sealed enclosure with a high excursion driver. The performance of these speakers with a regular crossover and amp will be poor, due to the low efficiency of the enclosure/small driver.
To get around this, Google put in the TI TAS5825M smart audio amp. By measuring the speaker parameters through V/I measurement and a model, it drives the speaker in a closed loop way with far more power than it would actually be able to handle nornally to compensate for the resistance from the enclosure air pressure, and throttles to maintain the coil at a safe temperature. The chip also does DSP to compress the audio signal, cutting the peaks off the bass as needed when the volume is turned up so volume is maintained at the cost of bass.
One way to explore could be to just feed I2S audio from an I2S ADC i.e. PCM1808 to the digital input of the amplifier. The processing is internal to the amp so theoretically you won't lose the tuning. However this may turn out to be a relatively annoying reverse engineering project with fine magnet wire involved.
Note: I2S is different from I2C - the amp will likely have both. You will likely need to keep the original system around to program the amp over I2C (or capture the transaction and replay it) - otherwise you will likely get no audio.
The "raw" audio performance of this device (just an amplifier connected directly to the internal speaker and dsp on the computer) is impressive, kicking out bass down to 40Hz. It will, however, not last long like that. Reports online are that these blow speakers easily even when used with the default amplifier.
I would recommend that if 3.5mm input is desired, to replace them altogether with the IK Multimedia iLoud Micro Monitors. These have sound quality just as good as the Google at similar size, with the same DSP tricks, but have regular inputs and no smart features.
>You can’t just ltspice your crossover and call it a day.
I'm sure that's true, but how important is that really for a set of crappy plastic speakers?
At the end of the day, if you're happy and it sounds ok: thats fine, most people can't tell the difference anyway or they get used to however it sounds. But it will inevitably leave a lot of potential performance on the table.
[dead]
[dead]
what exactly is the problem with smart speakers? Most people just want to stream Spotify
"We are delighted to announce an update to our smart speaker line: we are sunsetting the online services for this model. Your devices will be disabled in March 2025. They will no longer work, and please bring them to an electronics recycling center near you.
You are eligible for a voucher to get a discount when you upgrade to a newer model."
This is pretty much word for word what has already happened at least once.
Sonos at least reversed their decision to disable the devices when they sunset S1. They just made them incompatible with their current system but allowed people to download the old apps.
Unfortunately, devices that were already marked for trade-in before they made the decision are still completely bricked.
I bought a Yamaha CD-N301, came with a web radio, where you could setup a user, and make an index of URIs with webradios. After that you could select the station from a menu in the front. 3 months after purchase, the service started costing 3 dollars per month. I only use the CD function now.
I contacted Yamaha, the answer was “it still works, we never promised it would be forever free” PoS