DIY: Building a little BoomBox project

[NickK]

The Mrs wants a little boom box with a CD player. Emphasis on 'little' and CDP being for hassle free use. Now the maximum size of the box agreed is approx 260x220x180mm. Given this will be listening to music whilst doing other stuff, this doesn't need to blast in terms of volume but my own personal requirement is decent sound quality. The size of the box means this will not be a bass bombastic box but will get full range down to a decent level without resorting to tricks.

I've ordered the Mark Audio CHN-50 drivers - 50mm, ~4ohm and will make small vented speaker cabinets out of 12mm plywood to a proven design with these drivers. The nominal power rating for these are 7 watts. Yes seven whole watts.

I could have gone for the 20 watt Puvia 7.2HD drivers - better overall but also needing a far larger cabinet (ie each being 450mm+ high, 8 litre capacity minimum) and thus blowing the 'small' requirement.

As the box is small, but the power requirements also <20 watts, I have options for amplifications:
* Class A/B - good sound, produces heat but can drive 4ohm drivers easily.
* Class D - decent sound nowadays, can be sensitive to driver load resulting in increased distortion but would run stone cold without needing heat sinking. 4ohm is more of a challenge than A/B.

USB CD slot load -> RPi -i2s-> DAC -> opamp -> TDA chip amp -> driver.

It should be possible to use a RPI zero 2, with a small touch screen/buttons to run a media player including streaming WiFi radio. The digital out from the RPI can drive a decent DAC and a simple opamp to buffer the DAC output.

When I'd looked at the Puvia 20W speakers I came across TDA chip amps that can do A/B at 100W into 4ohm! So I could take a design around that and build an A/B amp with a lower wattage part.

I'm still considering which DAC to use - one option is a ESS 32bit DAC that would allow I2S and would work easily with the RPI.

Powering this all should then be easy enough with small regulated supply, completing the box.

Will keep you updated during the build.

 

[NickK]

Parts ordered:
* Mark Audio CHN-50 = 8W 4ohm speaker drivers
* IQAudio DigiAMP+ = 2x18W into 4ohm (also does 8ohm) Digital Amp for Raspberry Pis. This has a DAC inbuilt and Class D amplification. It also provides power regulation to the RPI. This runs on 12-24Vdc.
* Rotary encoder and IR sensor (hope this works with the apple remote)
* USB external slot load CD/DVD drive

Parts I have
* RPI Zero2W
* 13.6V x 5A (7A at 50% duty cycle) linear power supply
* Thick speaker copper wire
* 7" HDMI touch screen - I have this for another project but this is to test the power consumption as it's USB powered vs running through HDMI too.

Parts I will look ordering after:
* 63W 19V SMPS power supply brick
* 2.8" LCD HDMI capacitive touchscreen.

I'll use Volumio media player and if need patch in anything required.

The fun with this build is that the digital amp uses a number of GPIO lines, so to does the rotary encoder etc. Also the USB needs careful planning as the CD player is entirely powered through USB2.0 - this may mean I need to use a powered hub but half of this playing is trying to understand what power consumption everything is as the vendors don't really put good power requirements together (the CD drive is USB2.0 which should be 5V 500mA max according to the standard but you never know nowadays).

Once this is electrically working, I will order the birch plywood and do the dove tailed box for the speaker drivers. I will then know if the intended touch screen will work or need more power.

Will update with the photos as the build progresses.

The basic look will be similar to this:

Source: DIYAudio member's build with CHN-50s with veneer over plywood cabinet.

 

[NickK]

Just deafened myself. The Mrs was like WTF and can you turn it down I'm expecting a call.. Turns out there doesn't seem to be a volume test.. and it starts at full whack when you start a French internet radio station.



At the side- the speakers in their box, The RPI Zero 2W and DigiAMP+. A DVD R/W slot load player connected and everything powered by the DigiAMP+ power supply connected to a old bench linear power supply (13.6V).
So I got Volumio 3 to work - that's basically painless to setup but - and this is the annoying thing - they only allow CD playback on the premium payment plan.. err.. right. That's £50/year to play your Adele CD.

So the next thing is to play and find a free media player that can work headless.

 

[NickK]

So far I've been trying a couple of things:
* Librelec - a kodi based system, no luck as there's less support for RPI. I'm not going to fight the operating system and build teams, so I switched.
* OSMC - again another kodi implementation but more RPI support. However it Kodi doesn't really support headless configuration. Hence the system can't find the wifi network nor does it seem to allow USB SSH. Lots of playing around and I'm probably closer to buying a Mini-HDMI to HDMI connector than I am to solving it.

Also it appears Kodi also uses mysql. One of the unwritten requirements is that the Mrs will happily pull the power lead out without warning. So I either put in a set of caps with a sensor to 'shutdown now' or the databases are likely to be corrupted quickly.

I did setup on the RPI4 for a bit just to bypass the headless issues. I have some additional observations:
* The DVD external drive is noisy and loves going fast - I'll probably have to set Linux to limit the rate of reading or have it buffer the entire song into memory first. I know you can use "eject -x n" to set the maximum speed of the drive.
* The default Linux media handling is dire. It takes an age to read the identify the disk as CD and then VLC spends it's time working out the best way to mess up the sound, by resetting the audio to the HDMI channels for every song.

So there's more issues to solve, and the OSMC seems to swing and not recognise the disk, the same sounds as the ALSA media player which doesn't recognise a CD file format. I'm hoping there's a easy option, but it takes time to try each one.

As you'd expect the divers sound a little harsh at the moment. Supposedly they take about 100 hours to break in from new. I also need to hook up the encoder and IR sensor.

I'll try MPD too. It looks like there's a few options with ALSA and the rotary encoder too.

Once it's working I'll have a look at the frequency response and if there's noise. I've got some large Coilcraft 470uH inductors (3.7A 400V) so with a 6000uF 50V cap and a resistor it should help reduce any power supply noise, in theory that could also be used as part of the unplug-shutdown power. LTSpice seems to show low mV ripple.

 

[NickK]

So summary so far:
* Volumio 3 - premium $, very professional and easy to setup headless without a monitor connected with an app on the phone.
* Kodi - free, no headless support, bad RPI support
* Librelec - Kodi variant, no headless support bad RPI support
* OSmc - Kodi variant, no headless support, bad RPI support
* MDP/manual setup - linux haters, don't even try. Also you end up with a very broken experience in having to work with all the commandline packages, most didn't work with the DigiAMP+ without some hardcoding or patching.
* Moode - free, very professional and easy to setup with headless support by auto starting an AP point and web page connection on phone/computer. Configure for your WiFi and away you go. Also has lots of DAC/DigiAMP+ support, support for rotational encoder, multi-room etc. AND it slows the DVD player to a normal rate using eject -x automatically. Also ssh login works off the bat. The only addition I have made in the boot/config.txt is "max_usb_current=1" which means the RPI will provide full current to the CD drive. It's not caused an issue but just in case as it's recommended.

I'm currently attempting to get Moode to play a CD, in theory it should do this automatically, however at the moment I have good vibes about Moode.

EDIT: Moode also has inbuilt support for GPIO buttons for operating playback functions.

 

[NickK]

So playing from CD using a bodge patch, the resulting FFT - 20KHz range setup and left with max marking so I get the maximum for each frequency bin:


Now the scope is upgraded to 200MHz, so the next large noise peak is at 760KHz+. The scale here is large - the entire audio frequency range is in the first division on the left.


Sorry for sitting the scope on my lap due to the mess on my desk.

So the noise floor isn't anything to write home about - the probes were attached at the speaker. As the entire system is isolated, I can use the ground clip on the scope, the entire system then floats one the oscilloscope ground reference.

Once I work out how to install the bode plot AWG frequency generator on the moode device then can it up to get a frequency response bode plot.

 

[NickK]

Have you introduced your wife to Alexa? My mum rarely listens to music because of the "faff" of CDs. Potential to integrate a home assistant into this project as well?

She's not really taken with with any of them (Google too).

Moode will do a CD rip and store - which is an option I suspect. It would be easy to add a CD storage SSD in there.

Ok, so I've seemed to improve the CD audio performance. It was stuttering and suffering from what I thought was buffer under-runs causing the music to wait until the drive started up, seeked and loaded more data.
The RPI Zero2W only has 512MB of ram so we need to be careful not to overdo this too much.

Under MDP settings (under audio settings):
Audio buffer = 4MB -> 16MB
Output buffer = 128MB (no change)
Input cache = disabled -> 128MB

Under the system tab CPU set to performance.

I've switched off SoX resampling and set the ALSA output mode to Direct(HW). The CPU load is about 0.94 playing a CD. No stuttering or buffering. 250MB of available memory free. I also note that Moode doesn't use any swap space which is nice but it also doesn't have ANY configured.. so it can't offload any OS stuff not in use.

It seems (as you'd expect) the CD playback needs some caching which make sense given it's slower than RAM or the SD card at access.

I think I can probably set the SOX resampling back on. It's got a noticeable effect on the sound when playing 44kHz CD.

EDIT - switched the SOX resampling and non-direct back on - load varies but no stutters.
Current set it to resample 44/48kHz (ie CD) to 192kHz and also to widen the 16bit CD to 24bit.

EDIT2
They now have about 10 hours on the clock, the original sound was harsh and very hard treble with some CDs. I've been giving a diet of pretty much every style. There is a definite difference between the start and now. I don't wince when they're playing. I suspect that this will be the biggest change and over the remaining 90hours, they will change less and less.

I've also put two 1000uF 25V "low ESR" caps I recovered from a dead SMPS into the mix - they're across the power supply into the DigitAMP+. The voltage only varies ±0.01V now even with the DVD player running. So I think the Linear regulated power supply has quite slow caps. I'm not going to swear the amp sounds better lol, I suspect adding some low ESR to decouple the IC chip supply in parallel with the 220uF that it has for each channel would make the most difference (it's like making each on/off digitally work more efficiently.

I can tell the difference between CD 44/48K 16bit and 192K 24bit, it seems clearer and darker. However that may be the production on the CDs playing into account.

One thing that surprised me about the CHN-50s is the weight and the strength of the magnets. They weigh more than you expect. Gawd knows what the Puvia with a larger magnet weigh!

I played Adele to the Mrs through them when she got home. I held the two drivers up (no cabinet) - her response was "that sounds good". They are good sounding, the cabs will be small, but will give a decent sound - not like my big transmission 1/4 wave harlechs deep, but then they're only small boxes with a port on the front.

Ported vs Sealed - Sealed is the best sound in that it doesn't have any boom or weakspots but ported gives some oomph bass at the expense of a little clarity. In the end the seal takes more power to get the same level as a ported box but would retain the clarity. For this it's going to be 'indirect' listening so a ported box works.

I also talked to her about making a solid state 'rip' of her CDs - she has them but having them stored in the box means she doesn't need to dig them out. So I think a small SSD or a larger SD card may be in order (16GB card in use atm). The install of Moode isn't large but I thought I may as well put my CDs in there to that are currently sat in the attic.

 

[NickK]

Just thinking about power this morning. The DigiAMP+ designers recommend a 19V 63W SMPS. The amp would be 36W (probably closer to 40W) and 13W for the RPI with full load that's 48W. So a 63W means it's running at 76% - all good to prevent the SMPS power factor system from dropping into the low load or varying the power too much.

The beauty of the Class D and SMPS is that it's 88%+ efficient, so we're not loosing too much power. The down side is EMI noise on the output. An option here is to look at a medical power supply blocks - in this case a Meanwell 19V 90W power supply (120mVpp) - that gives good noise filtering, doesn't get too hard on the digiamp power regulator but allows us to tap off power into a powered USB hub for the display should we need to. Only issue with going for a 19V rather than a 24V is any RC/LRC/voltage regulator then drops the line further. So we'd be left with a 4% line regulation rather than a 1% or less (in the case of the LT regulators) and no additional filtering.

Probably over thinking it to be honest, I should see if I have a 12-15V SMPS around here. Perhaps a laptop block and see what effect that causes in terms of noise through to the speaker.

 

[NickK]

Stopping of radio/operation (ie goes quiet)
Looking at the logs it appears the decoder is "too slow" and thus the system mutes to prevent under-run. Interesting. So out of nowhere it stops due to slowness - that sounds like it's CPU throttling. Previously I had this set to "performance" CPU governing but I have now gone back to the "default" which is that it scales the CPU depending on load. This seems to prevent it from overheating and throttling.

Hangs
I've noted a few unresponsive (both web and SSH) halts, requiring a reboot. I thought that may be memory related but I I think that may also be CPU heat related. One of the interesting things they did with the SoC CPU chip is actually stick the RAM on the top of the CPU like a sandwich inside the CPU casing. So heat is more of an issue. I could add a small heatsink to cool this - along with the DigiAMP IC. I've not noted a hang since the CPU governor has been set to default rather than performance.
I've also added a 1024MB swapfile and I've seen a couple of times when it's used it but not in normal operation.

I was originally concerned that the DVD drive was pulling too much power but the voltage doesn't drop at all from the PSU so if that is a problem then a powered hub for the DVD drive/screen would be in order. So I think power and heat are still an issue with Class D/Pi in this way, but the heat sinking is small.

The sound has improved. I've been feeding it a diet of bass heavy reggae and dubstep to free up the cone a little more.

I've also found a local woodyard that stocks BB grade 12mm baltic ply. £98/sheet For speaker enclosures you want the least resonance from the wood in addition to some dampening material. Typically people will use MDF or better HDF (high density) in cab building but MDF in practice isn't as good as birch ply as long as this has a decent number of layers. So 9 layers for 12mm or 13 layers for 18mm being about optimal. The result should be little resonance from the panels. Then I'll add dampening material to reduce the reverberation inside.

This is the design I'll use:
https://www.kjfaudio.com/wp-content/uploads/2020/08/CHN50-slot-vent-box.png

 

[NickK]

So I've wired up the encoder using the Moode rotary encoder support but it's not particularly responsive. It does work but just slow which doesn't make it a great user experience. An alternative is to use a separate rotary encoder process that controls ALSA volume directly however I don't that will be much different to the current functionality.

The RPI Zero 2W which operates about the same speed as the RPI3 is finding it difficult I think with the media player. MPD is taking 50-88% of the CPU according to top when playing a CD, the player often gets some pauses that impact even the responsiveness of an SSH session running top! The CPUs (all 4) aren't being taxed, but from my FFT work and performance testing, I know that the Zero 2W memory bus speed is slow compared to the RPI4. All that upscaled audio bus transfers seem to be causing a problem (perhaps temps too causing throttling).

Just connected up the powered USB hub. I looked under the DVD-writer drive and it states "Input 5V 1.5A" ouch! USB normal is 500mA until the later USB ports have increased that. The RPI can supply higher current but I think even with the "max_usb_current=1" setting, the DigitAMP+ 2.5A max is split across everything (it's very difficult to get this info as it's not really listed anywhere).

With the powered hub providing the power, the drive is faster and seems to be happier. I've also not noted any drop outs yet.

I've noted on the top list that the CPUs spend a lot of time waiting, so I think that's waiting on the drive - hence the lower power provided to the drive may have caused it issues. I thought that may be the issue.

RPI Zero2W - note the decoder under runs due to performance

Feb 26 15:55 : alsa_output: Decoder is too slow; playing silence to avoid xrun
Feb 26 15:55 : alsa_output: Decoder is too slow; playing silence to avoid run
Feb 26 15:55 : alsa_output: Decoder is too slow; playing silence to avoid xrun
Checking /dev/cdrom for cdrom...
        CDROM sensed: HL-DT-ST DVDRAM GP70NS50  1.01 SCSI CD-ROM
Feb 26 15:56 : player: played "cdda:///10"

RPI4 - same SD card ..

Feb 26 16:46 : player: played "http://ice1.somafm.com/beatblender-128-aac"
Feb 26 16:46 : player: played "http://ice1.somafm.com/beatblender-128-aac"
Feb 26 16:47 : player: played "http://ice1.somafm.com/beatblender-128-aac"
Checking /dev/cdrom for cdrom...
        CDROM sensed: HL-DT-ST DVDRAM GP70NS50  1.01 SCSI CD-ROM
Checking /dev/cdrom for cdrom...
        CDROM sensed: HL-DT-ST DVDRAM GP70NS50  1.01 SCSI CD-ROM
Checking /dev/cdrom for cdrom...
        CDROM sensed: HL-DT-ST DVDRAM GP70NS50  1.01 SCSI CD-ROM
Checking /dev/cdrom for cdrom...
        CDROM sensed: HL-DT-ST DVDRAM GP70NS50  1.01 SCSI CD-ROM

The CDROM messages are the patch/bodge to get CD playback - the folks at Moode have said they will never support CD playback natively. So this CD playback which is a hack from the CD Rip plugin (without the rip) is the only way. I suspect the continuous checking/reinitialising of the CD drive is causing the drive to interrupt the data transfer and thus cause an under-run.

I did get a buffer underrun on the RPI4 but far less, also with 2GB there's more memory that can be assigned and the memory runs far faster. Add the USB3 and you have a far more capable system. The CD drive seems happier too, so to is the speed of the web page.

EDIT: After 3 hours of radio play, no buffer underruns unlike on the Zero on radio.

 

[NickK]

So V1 I've settled on the following:
* CHN50+AmpDigi+
* RPI4 - the zero2W just didn't have the processing power required it seemed due to the limited memory bandwidth. Powered from the DigiAMP.
* Meanwell 19V 90W SMPS block
* A 19V to 5V switching step down power supply to power an old belkin powered USB hub - the hub connects to the RPI4.
* 2.8" capacitive touch screen display connected HDMI and taking power from the powered USB hub. (requires 5V so let's assume 1.5A)
* LG GP70NS50 DVD-RAM drive powered through the powered USB. (requires 5V 1.5A) - I may make this an external drive to be disconnected as it's noisy.. or at least with a button to disconnect it.
* Volume knob via rotary encoder, I can make the click a mute.
* IR sensor for an old apple remote.

I'm going to use USB-MicroSD adaptors and then use two 64GB MicroSD cards plugged into the RPI4 USB2.0 ports to then store music. Takes up little space and little power compared to full 120GB SSD. The performance requirements for streaming music should be simple enough with an SD card. I can get my CDs on there too

I have a 7" screen already and I know that stops the RPI4 from booting due to power draw, hence needing a powered hub.

[RPI]
GPIO -> DigiAMP
GPIO -> volume encoder
GPIO -> IR receiver
MicroSD card slot -> OS/Moode SD drive (16GB MicroSD card)
USB2.0#1 -> 64GB music MicroSD card
USB2.0#2 -> 64GB music MicroSD card
USB3.0#1 -> Powered USB hub -> Touch screen touch control USB
USB3.0#1 -> Powered USB hub -> External USB port -> External USB drive (also allows a RPI keyboard and it's USB port for a mouse to be connected).

[Power]
230Vac Mains -> 90W Meanwell SMPS 19Vdc block
SMPS 19V -> DigiAMP
SMPS 19V -> SMPS 19V to 5V step down -> Powered USB hub power input
Powered USB Hub -> Touchscreen power input

Some random ideas for V2
* power supply and isolation of the digital power improvements
* digital re-clocking to reduce jitter
* SMPS noise reduction.
* battery pack
* digital DAB radio

 

[NickK]

Hello and congratulations on your project.
It's always nice to discover the ingenuity of others and read their journeys.

Personally, I feel like there's an easier way than using an rpi for this.
I discovered the Arylic brand (www.arylic.com) over a year ago and so far it's exactly what I'm looking for.
Let me explain. I like to create boomboxes, music boxes and more. My current project is based on multiroom.
I bought two different cards including a 2.1 amplifier and a multiroom receiver only. These cards can be modified, among other things, with respect to the GPIO outputs (which can also be modified, but for this you need an ADC Workbench program which unfortunately only works on windows for the moment).
The power is up to 2X50W and 100W for the subwoofer which is already very good and more powerful than what you are looking for for your project. But what can do more can do less.

I have, like you, the project to add a cd player (recovered from an old pc) to this amp which will probably be powered by the usb of the card, I have not yet tested if it delivers 500ma or more.

The advantage over your project is that everything can be controlled from a smartphone, without having to code or use linux or add an lcd screen or a multimedia player. A very easy-to-use application allows you to control everything, to directly launch an internet radio or spotify/deezer or other. You can plug in a usb key containing up to 1000 songs, have an aux input and even spdif in addition to bluetooth and wifi. There's even an alarm clock option.
Which for my lady is much easier to use than raspberry pi based projects.
I tinker, assemble, modify and then she enjoys it quietly.

The multiroom side is very nice and the value for money really impressive. It becomes very interesting to equip several rooms in your house rather than with brands like So.oS.
And the phone should not be left connected. If I start a web radio, I can leave my home and the sound continues since it is the amplifier itself which is connected to the internet and my telephone used as a remote control. Unless I'm using bluetooth, of course.

You can suddenly use active speakers that you already have or make your box according to your desires. You can also put passive speakers of 4 or 8 ohms.

What is the total cost price of your creation?

What I like with these cards is that you can add things or move certain elements without necessarily having to take the soldering iron or desoldering braid or having to buy only their accessories

Cool - I just had a look at the acrylic site. I see they do streaming, without storage - you need to stream from their box I assume. That makes sense for a multi-room.

There's some hidden things in Moode - if the Moode doesn't detect a known wifi network in 90 seconds it creates it's own access point - you can simply log onto that using the phone and then just type "moode.local" into then web browser. It then works just like a app and this is how setup moode in a headless configuration. Moode doesn't need a screen unlike some of the others - in this way it resembles Volumio without the operational cost of App Store developer certificate signing. Moode also has multi-room streaming available so it would be possible to stream or connect remote NAS to the server and then stream to small Zeros for example. The sound processing (CamillaDSP or SOX) is what takes the processing power - CamillaDSP provides room correction and SOX provides resampling up or down with digital filters. With pure HW pass through the CPU hit is far less and a Zero is likely to be possible.

CD drives take more than 500mA, the LG external USB drive doesn't have anything in the packaging or data sheet but on the bottom of the drive it states "Input 5..0V 1.5A" which it is expecting to pull from the USB power socket. It sort of works with RPI set with max_usb_current=1 but it's not happy. Using separate powered USB hub you could hear the seeks being faster, less retries and it seemed to stream better.
I have a number of dead MacBookPros, extracting the DVD drives - one CD drive turned out to a Apple style IDE connection the other was a mini-SATA. Looking at powered connector adaptors they're basically 3/4 the price of a new external drive. Also once out of the case they don't have front to the slot so both open to dust and laser problems.

You're right that the price is probably higher than one of the mass produced streamers:
* 24 for a DigiAMP (2 to 8ohm) the TI chip has a Burr Brown 192kHz 24bit DAC with a class D power section. 18W/ch into 4ohms (or 2x30W max according to TI).
* 55 for a RPI 4 (15 for a zero2) or pick up a RPI4 second hand for less - I'm using one from a project that is delayed
* 35 for a 90W SMPS power brick
* 20 for MicroSD cards with adaptors
* 35 for a 2.8" display
* 40 for a pair of CHN-50s
* 98 for a large sheet of ply (case)
* 40 for the DVD/RAM external drive
* 5-10 for a 19V-5V step down supply
* 5 for various switches/rotary encoder etc
* powered USB hub I have a broken case donor that I'll just use the circuit board direct.

So currently at 362 with everything in a self contained box. However everyone knows DIY isn't less expensive, just more interesting..

 

[NickK]

I just ordered the medical 90W 19V SMPS power block, and by tomorrow afternoon I should have the space in the garage and I'll get a 1220x2440x12mm ply sheet from a local spot. That way I have bits I can make this weekend - fun with wood bits & routers

It's just occurred to me I'll need to get a set of black anodised wood screws

I'll then add the other pieces (screen, step-down power supply for powering the USB hub etc). I have a external chassis mount USB connector around somewhere so that will solve that.

 

[NickK]

So the power supply brick arrived - smaller than I was expecting for 90W!

I'm impressed - no problem with a short circuit (accidental!) and the noise level is as good as my old Maplin regulated supply in the audio region although you can see much further up 200kHz and 700kHz there is more noise at about -30dB. At the speaker the noise level is around -86dBV. That will do for a little amp.

Model is a Meanwell GSM90A19-P1M which can do 19V 4.7A.

Turns out I have a R/C Matek UBEC which does upto 27V into 5V at 4A which is a small SMPS DC-DC step-down converter. Previously I thought it was 12V max. Result.

I have the 7" screensetup and I think it works better than a 2.8" would given the Moode layout:


The screen is sat in the box I built for travel at xmas. The new SMPS brick.

Currently playing Adele 30 off the CD as a test. I should get the wood tomorrow.

 

[NickK]

I have my 12mm baltic plywood I've created a set of templates out of 1.6mm ply so I know the bits fit together. I've decided to leave dovetailing until another day - this will glued.

On the front panel I will have to:
a) drill a hole for the driver,
b) inset the face plate of the driver
c) chamber the rear of the hole for air flow

The plan indicates lagging inside but explicitly mentions not to use foam (foam isn't great from research). So instead I'm going to get some thick felt and line the inside.

 

[NickK]

Slow progress at the moment due to priorities. I'm making a circular saw table for the smaller cuts and now the garage is physically connected to the electrics rather than an extension cord, it should be a bit more organised.

However some more O-scopery.

First the noise of my simple linear regulated power supply:


Next the noise of my SMPS power supply:


And finally the noise (not muted) at the speaker:


So the more -dB the quieter the noise floor is. As you can see the SMPS switching means you get quite a bit of noise, however a Digital amp (ie Class D) also performs switching at high speed so the output of the amp has a filter on to both combine the digital signal into a analogue wave (like an integrator) but this also provides a level of "noise reduction" in that the peaks appear to disappear but the average noise is higher.

The main criticism of digital amps/SMPS is it reduces the quality of audio - specifically "sss" sounds due to the interference, I would conclude that this could be apparent in this build, although I may look at adding filtration post SMPS - the system doesn't need the full 19V as it functions happily on 13.6V so I could add some supply filtering.

 

[NickK]

So I've now spent a little time dealing with the garage:
1. electrified the garage and replaced the strip light with an LED light.
2. built a circular saw table + shuttle with clamp out of a circular saw and some old 18mm floor particle board.
3. built a router table out of a 1/4" router and 18mm floor particle board that's been routed out to give about 12mm of routing depth

Version one of the front panel I used a 74mm holesaw and attempted to hole saw the 100mm wide 3mm recess using a drill press. Nope. The detachable hole saw is way to inaccurate.

Version two sees the use of a 6mm drill hole through the wood, and in the router board to give a diameter sweep with the 6mm router bit from underneath. This means I can make circles smaller than the router bit. In this case I was guessing on the diameters a little after realising that I needed to add 1/2 the drill bit - the result is a little larger than was expected but better result:


A 45 deg router bit on the inside means the back side airflow is better.

The router bit kicked sending the piece across and plunged into the side wall. Well this was a good second version of the front piece. What is good is the process is very repeatable and accurate.

 

[NickK]

Well "first voice" in that this box is held together with framing kit (ie corner and cord) that you can't see. The port piece is missing but it gives an idea of the sound:




The Mrs wants to vinyl decorate the outside anyway so the fact there some glitches etc is not a problem. I'll get a 90degree vice tomorrow then I can start glueing the things together. I need to order some wool felt too then I can finish the speakers and the little equipment section that will sit on top of it.

How do they sound?

Decent. I'll know more when the things are fully built with the port piece in place and inside sealed.

They can play quietly but with a good sound, they have some balanced bass but as you'd expect it doesn't extend down like a large pair of quarter wave floor standers would.

 

[NickK]

So the initial rough glueing is ready for the wool felt, once that's done I can glue up the side and sand it. There will be a cable hole up the top of the unit to connect to the amp & screen area above it. Oh yes - managed to cut my finger - but that will disappear when I give it a final sand.



The port tuning (the little bit of ply that faces back into the enclosure that the cable skirts around really helps the sounds of the speaker. Previously I was a little worried about sounding a little in a room but that has worked really well. The final wool inside will reduce transmission of waves and the final side glued on should help the the ridigity further.

I'm contemplating giving it a expoxy coat inside. It will take a week to go rock hard. However for this first attempt I thought it would be a good idea to simply keep things as simple as possible.

PVA wood glue with the ply keeps the entire thing rigid so it probably won't benefit from any internal bracing given the small cabinet size. I have some small offcuts I could make a brace from.

 

[NickK]

Took the setup and showed the mrs with the speaker drivers now screwed in (although still open on one side - the felt should arrive today!). She listed a musical that she'd not listened to for 20 years and was singing along so I think it got the thumbs up.

I think it surprised her that it can play airplay and be controllable from her iPad too. In a larger room, which has painted walls and hard floor, the sound was loud enough and with a decent bass. There's a touch of 'enclosed speaker sound' on the odd occasion which I'm hoping will disappear with the felt.

I'll strip them back down and give the boxes a flat sanding and fit the felt, but that has been a good test. Next steps are to embed the screen, the secondary power supply and USB hub, add the volume controls and IR sensor into a small box that will sit above the speakers.

EDIT: Once I have everything sanded and done, I'll take photos then she can do some vinyl work and see the difference. I has occurred to me that given the bathroom requirement, some varnish/epoxy sealant would actually be a good idea - it will help the rigidity but also it will protect the ply from absorbing any moisture and expanding.