Sunday, November 18, 2018

Logitech / Slimdevices SqueezeBox Classic (SBv3) Repair Considerations

Hi all,


this is a troubleshooting help for Squeezebox Classic (SB3) devices and the typical issues they may have or get after being in the market for 10 years or more.



This is as great little device, featuring a really nice design and good sound output quality. It adds value to any Hi-Fi set, and can be obtained at very reasonable prices nowadays.
Alas, nothing is perfect, and many units which are sold in used condition nowadays have more or less obvious flaws, or may start getting issues anytime. May this post help you if that happens :o)

I have seen the following problems (in random order):
  • 1 - display showing DAC CAN'T TALK
  • 2 - fading displays
  • 3 - issues caused by failing power supply
  • 4 - crashing instead of playback starting
  • 5 - analog audio failing whereas digital outputs are working
  • 6 - unit is not starting up at all
  • 7 - Ethernet connection not working (but Wi-Fi is okay)
Let's get into the details of each, but first there are some sanity checks you might want to try before grabbing the screwdriver. Otherwise, just skip to the section Issue Information below.

Note: Some parts of this document are still incomplete and will be finished soon.

Basic Troubleshooting

Power Supply Voltage Check

Measure the power supply output voltage. It should be +5V or a little more. The problem is that a multimeter is not enough to diagnose a power supply as the measurement is done without an appropriate load. The "idle" voltage output cannot be compared with what is going on if everything is connected, it may look perfect but as soon as the slightest load is applied, it may break down.
There are cheap adjustable battery testers at eBay that you can use to measure the voltage as the current is driven up. It is a safe and comfortable way to verify if a power supply is inside its specification or not, and definitely recommended if


At 2 Ampères, the output voltage should still be close to +5V. If it drops before reaching this point, it needs to be exchanged or repaired. See the Issue: Failing Power Supply section below for more info.

Try The Different Reset Variants

Before even opening the Squezeebox, you should try the two reset options:
  • unplug the power cord
  • on the infrared remote control, push and hold one of the following buttons:
    • + / ADD --> factory reset. This removes all settings captured in the setup process, i.e.:
      • the name you may have assigned
      • the IP address
      • the Wi-Fi settings (network SSID and password)
      • display language
    • 1 (on the numeric pad) --> Xilinx reset. This reprograms the built-in FPGA which is dedicated to controlling the display, the DAC, and other components, so it is a rather central component that in case of program errors may cause all sorts of failures
  • plug the power back in, keep pushing the selected remote control button, and keep pointing it at the Squeezebox
  • wait until the message "factory settings" or "programming Xilinx" appears
  • release the remote control button
  • wait what happens. The unit should normally reboot after a few seconds
If nothing is happening at all, the unit either does not power up far enough to read the IR command, or the display is dead and you cannot see the results.
In the best case, the unit has already recovered and can be used normally again.

Temperature Check

Place your hand on the top of your Classic while it is supposedly powered (no matter if it is in standby or playing something). A working Classic is always warmer than the ambient, thanks to the permanent display filament heating. If it feels rather cold, that indicates the circuitry may not be working. If you find the unit is not warm, check the power supply next.

Network Check


If the display is no longer showing anything, this does not mean the Squeezebox is dead yet. First check whether it is still available on the network. The basic steps to achieve this is:
  • find the device's IP address
  • ping the IP address and see whether there is a response
You may need the device MAC which is the six-byte hexadecimal code you can find on the label of every Squeezebox device. It usually starts with 00-04-, and contains four more unique segments. This is the internal address that is also shown in Logitech Media Server diagnostics, and the DHCP server, if applicable.


Finding the IP Address

The IP address is a unique address in your network that is assigned manually or automatically to your Classic in the setup phase, or during startup. It consists of four 8-bit segments (numbers between 0 and 255), separated by period characters, such as:

192.168.0.10

Depending on the network topology, you need to use different places to find the IP address:
  • If a local Logitech Media Server is used:
    • the LMS web interface will easily let you access the Squeezebox devices registered. The Information tab of the LMS settings page will give you a list of all devices currently registered. You should find the Classic in question there. If you can still control it from the LMS interface, a remote control, or an app, and audio output is still working, it's probably only the display that has failed whereas the rest of the circuitry is still fine. If the Classic is no longer listed, and does not respond to any remote command, check the "alternate" network interface: if usually running wireless, use wired Ethernet now, or vice versa, and power cycle the Classic. This is not highly reliable as you won't be able to reconfigure the Squeezebox accordingly without a working display, but it's still worth a try.

  • If Logitech Media Server is not used (e.g. you have registered your Classic with mysqueezebox.com and use it only for web radio streams):
    • if you are using Wi-Fi:
      • ... directly from your internet gateway / router:
        • check the router's configuration interface (typically in a web browser) to find whether the Classic is still registered, and if so, what IP address it is running on
      • ... via a Wi-Fi access point:
        • the access point may have its own configuration interface that you might use to see what is going on in your wireless network. See if you can find an entry for the Classic in the list of registered Wi-Fi clients
    • if you are using wired Ethernet:
      • if you have a hub or switch that has indicator LEDs on each port, check the port to which the Classic is connected. If no activity and connection is indicated, that probably means the Classic's CPU is not powered, for whatever reason. You can omit the other checks at this point and turn towards the power supply (see below)
      • if you have your own DHCP server, use the configuration interface of it to find whether the Classic is still registered and what IP it is running on

Pinging the IP Address

Once the IP address was acquired, you can now try to "ping" it. If you have a computer that is capable of giving you a terminal or console, all you need to do is start the terminal or console, and run the following command:

ping <IP address>

Where <IP address> needs to be replaced by the four-part IP address you found in the previous steps. The ping command is universal so it should be supported on any Windows, MacOS, or Linux platform.
The command output should read like this (typically you will get one reply per second over four seconds):

Pinging 192.168.0.10 with 32 bytes of data:
Reply from 192.168.0.10: bytes=32 time<1ms TTL=64
Reply from 192.168.
0.10: bytes=32 time<1ms TTL=64
Reply from 192.168.
0.10: bytes=32 time<1ms TTL=64
Reply from 192.168.
0.10: bytes=32 time<1ms TTL=64

Ping statistics for 192.168.0.10:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 0ms, Maximum = 0ms, Average = 0ms


If instead you are getting error messages such as the following, and / or more than one second is spent between ping attempts, it means the target does not respond to the ping, i.e. is not connected, or not working:

Pinging 192.168.0.10 with 32 bytes of data:
Reply from 192.168.0.11: Destination host unreachable.
Reply from 192.168.
0.11: Destination host unreachable.
Reply from 192.168.
0.11: Destination host unreachable.
Reply from 192.168.
0.11: Destination host unreachable.

Ping statistics for 192.168.0.10:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),


The statistics may also indicate a 100% packet loss which is just as bad.

At this point it is a bit clearer what is the problem. If network connectivity is good, and the unit can play back normally, and responds to commands from the infrared remote control and/or commands from LMS / mysqueezebox.com / the app of your choice, the unit is basically alive, but the display-related part is blown.
Otherwise, if network connectivity is not working, either the power supply is shot or another of the more complicated issues is present.

A Look Inside

CAUTION! I must warn y'all that wearing a ESD grounding strap or similar, and using ESD-safe equipment is imperative when working with the SB3. It's notoriously sensitive to even small discharge events which immediately turns the unit into a precious paperweight. Be sure to be grounded at all times when working with the innards of a SB3!

Basic Construction

Most of the circuitry is on the PCB side that you only see after extracting the PCB. These are the different boards:


The connectors board is soldered to the mainboard so it cannot be removed easily. The CPU board is connected to the mainboard via a proprietary 60-pin FPC connector. The Wi-Fi card sits in a MiniPCI connector frame on top of the CPU board.
The following image illustrates the position of the adhesive pad that is to ensure the CPU board is mechanically fixed to the mainboard:


You have to overcome this adhesive pad to be able to remove the CPU board. They really used tough material for sticking the boards together. Be aware that the other end of the CPU board just rests on that FPC connector, so you should be pretty gentle not to break or bend anything there. Adhesive remover may help soften the glue up.

Removing the Wi-Fi Board 

The Wi-Fi board is just clipped in and not glued. It is secured by two metal brackets that keep it pushed down into the white (or sometimes black) MiniPCI connector of the CPU board that is located between the large mainboard and the Wi-Fi board.
There are two half-circle shaped slots on each side of the Wi-Fi board into which the metal brackets are clipped. Bend the brackets outwards carefully:

The board will then lift out to an angle of something around 25 degrees, offering a first glance at the CPU board beneath:


You can then just pull it out of the connector so we get to see the brains of the Squeezebox:


I recommend leaving the antennas alone as the micro SMA connectors are quite delicate, and really tiny. Leaving the antennas connected causes a little more need for attention but is still better than torn-off PCB connectors, or the other side being torn off the wire. Even if the disconnection part worked, plugging the connectors back in is tedious and can easily end up in damaged contacts as well if the connectors are not perfectly aligned.

Issue Information

Issue 1: Messages Like "DAC CAN'T TALK"

If the display is giving you unexpected messages that keep popping up, and may also appear in huge letters across the entire display, a Xilinx reset may be the solution (info on how to reset can be found above). The message suggests that the processor or FPGA cannot connect to the digital/analog converter (DAC). If the software reset does not help, the problem is most likely somewhere else.

If you are seeing "MAC NOT SET", or horizontal lines either drawn additionally or missing entirely, this indicates a more difficult problem.
MAC NOT SET means that the device cannot see the MAC address that was assigned by Logitech to the unit. It's basically something like the serial number, and Logitech uses it like that. On the sticker on the bottom side of the unit, you can see a six-part hexadecimal string, typically starting with 00-04-20. This is a configurable value that is stored in the EEPROM Flash chip, and for some reason, the processor cannot read the appropriate data from there.
If you go to the settings menu and attempt to set the MAC address back to what it is supposed to be, you may find that you cannot set the hexadecimal values correctly in all columns. That is a side-effect of the issue. It's no use trying, the settings will not be stored eventually.
The message suggests that the problem might have to do with networking, which leads many people to believe it has to be the Wi-Fi card. That is not the case unfortunately. It's neither the Wi-Fi board nor the on-board ethernet controller chip but a communication issue between the processor and the rest of the circuitry.
I had this issue with a Squeezebox Boom once and was never able to fix it. Another Boom that came in later could be fixed temporarily by resoldering the SDRAM chip, but the problem soon returned so I think the resoldering job did not actuallyl fix anything, it was rather just a coincidence that the unit apparently recovered, anyway the problem still persists so it must have been something else. This is still waiting for closer inspection.
So unfortunately, if you are seeing MAC NOT SET, that's still an unknown thing.

Issue 2: Fading Display

Just like any Squeezebox that employs a VFD-type display, the Classic's displays are also susceptible to the burn-in problem. See here for my blog entry on how to open a Classic and replace the display.
Just to add here, the Classic is also keeping its display on at all times except when it is unplugged. Many people use the "screensaver" feature to have clock time, date, weather info, or RSS feeds displayed while the device is not busy otherwise.
I think it's safe to say that the displays age only while being powered. The filament wires are constantly on even in standby, or if nothing is shown at all. I used to state that the filament (heater) wires can bear this easily but I might be wrong here. It might explain the overall aging (of all pixels, also the less-used ones) that appears after a time.
Anyway, after 10 years of constant use the displays will show some form of aging for sure. Enthusiasts will enjoy the result of a display replacement.

Issue 3: Failing Power Supply

I would claim that the external power supply is the Achilles heel in the overall Squeezebox Classic design. Even the rather high-quality ones that Logitech delivers along with a Classic will fail sooner or later, causing all sorts of trouble.
The power supply needs to deliver 5 Volts at up to 2 Ampères peak, and usually around 800mA are drawn from the power supply constantly, regardless of the mode the Squeezebox is in. This is a considerable permanent burden for a power supply so the switching PSUs will all age and fail sooner or later. Linear power supplies (easily identified by its weight because most of it is a massive transformer) typically last much longer but they are rather hard to find nowadays, for these reasons, among others:
  • transformers contain a good amount of valuable copper wire which makes them more expensive. Most of the weight is metal
  • the transformers also make the power supplies bulkier and heavier
  • energy efficiency in a good switching PSU with power factor correction (PFC) is a lot better than in a linear power supply
  • transformers may emit humming noise at mains frequency (50 or 60Hz) if not appropriately shielded / decoupled
  • transformers cannot auto-adjust to any mains voltage. Some switching PSUs can be used world-wide with no need to configure anything. In the best case, a transformer has multiple primary taps and circuitry on that side to adjust manually to some fixed voltages like 100, 120, 220, 240 Volts.
A big advantage that classic linear power supplies have over switching power supplies is that they do not feed much noise back into the mains network. By nature, switching power supplies are causing the AC sine wave to become more and more distorted, depending on the number of switching power supplies used in the same mains section. This is because they do not run in sync with mains frequency like transformers, instead they chop off the power they need at a rather high frequency, causing little spikes and dents, or just noise, in the sine wave, which affects other devices in the same power section as it lowers their input power quality.

If you own (or can find) a linear power supply that covers the Squeezebox requirements, it's probably a good choice to use.

To diagnose a power supply, an adjustable DC load is ideal. Nowadays these are available for next to nothing at eBay, Alibaba, Amazon etc.
Something like this will do nicely:



These modules are sold as DC loads or are also called battery testers. The one here is good because a bottom threshold voltage can be defined. If the PSU drops below this threshold, it triggers a visual and audible alarm.
Most DC loads will need their own power supply, in this case it's 12V and the Squeezebox Boom power supply fits just nicely. You will also need a DC jack for connecting the power supply under test to the green screw post at the right of the tester's PCB. A bit of DIY is needed here because wired 5.5 / 2.1 mm jacks will be rather hard to find.

More About Power Supplies

It's just a theory but a power supply offering more than the 2A specified might be less challenged in peak situations, and may live longer if the parts it consists of are a bit overdimensioned. From workbench measurements I can say that the SBv3 rarely needs more than 1A. Power-up spikes might reach higher though which makes 2A a good margin.
Another reason for picking a higher margin (and thereby, Amperage), is the fact that the SBv3 power supply usually dies after 10 years of permanent service. This may be proof it was actually working above its comfort zone. A more capable one might last longer.

More things to consider:
  • AC or DC: the output must be a rectified direct current, or DC. Do not try AC!
  • polarity: the inside of the barrel connector is the positive terminal, the outside is negative or ground. Please never connect a power supply with reverse polarity, it will destroy your Squeezebox in the blink of an eye
  • barrel dimensions:
    • outer diameter is 5.5mm
    • inner diameter is ~2.1mm (this might not be the entire truth though, see further below)
    • length of the metal sleeve outside is around 10 to 11mm but I think that shorter ones as well as longer ones would also work
  • size: power supplies nowadays come in almost any shape or form. I prefer the larger ones that look like a notebook charger, assuming that they have more space for heat dissipation and probably a better circuitry. Let me repeat that this is all theory and I have no proof. Bigger might be better but that's no law
  • type: discussed above
  • price: the Chinese market pushes out 5V power supplies for less than a Dollar if you want. Needless to say I would never ever connect my precious SBv3 to one of them. You will have to rely on some common sense here. Personally I find power supplies for less than 10 US-$ a bit suspicious. Even considering mass production, there are too many compromises made to reach such a low price. You can spend >500 US-$ for a high-end linear power supply assembled by naked virgins in a full moon night if you want. I would not claim that this makes any audible difference eventually but it's certainly not bad either.

Now for the barrel connectors. I must admit here I am still not entirely sure I have understood this "inside diameter" matter completely. When you take a look at the jack in the SBv3, you will see a center pin. Whenever I measured them, I got about 1.66mm diameter of the pin itself. In the jack cavity, you will also see a large metal spring contact near the bottom which makes contact to the outside metal sleeve of the plug.

The question is why the plug's inside diameter is specified so much higher than the 1.66mm the pin actually has.

To make matters worse, there are at least two types of internal contact in the plugs:

    
Metal cylinder    Spring contact

The spring contact offers a bit more flexibility as it is able to adjust for the pin's diameter. Hence the small slots outside of the springs where they can move. However, springs may wear out and lose their ability to push inwards. Sometimes it helps to bend them back.

The metal cylinder variant (in my experience this is what mostly shipped with the SBv3) theoretically offers the entire inside surface of the plug to make contact to the pin, but in practice, only a very small area will actually touch. Due to its incapability to adjust (no springs), I think this is the less fault-tolerant one. Interestingly, it still works out quite well.

There are many more variants of these connectors. These two are the most popular ones. Unfortunately, when you go about ordering a power supply, you will rarely see a high-definition closeup of the connector to help you choose.

Unfortunately, I cannot say something definitive about the inside diameter of the plug. There are choices of 2mm, 2.1mm, 2.35mm, 2.5mm, and some more. Any of them might fit but just as well might not. The spring variant may offer some more margin for error. I'm wondering why they do not have a minimum and maximum diameter, considering the spring's ability to move accordingly.

To complicate things even more, it's possible that Slimdevices and/or Logitech changed the connector types during production. So all I can say is "it depends...."

You're on your own here, sorry. If you have a shop to go to, I recommend taking the SBv3 with you and just checking out the available power supplies. If you can find one where the connection can be easily made, and is also stable if you move the cable or slightly pull it, that's a fit. Inside diameter of 2.1mm might always be good, but don't take my word for it.

 

A Word On Electrolytic Capacitors (Regarding The Power Supply)

The parts that fail most of the time in switching power supplies are the electrolytic capacitors used to flatten the voltages in the input and output stages. The power supply units are usually rather small, so there is not much space around the components to dissipate heat, and the stock power supplies are also sealed air-tight which prevents airflow cooling. Electrolytic caps cannot stand exposure to heat. They have a temperature rating of typically 85°C or 105°C but even if the ambient temperature is lower but constantly above a certain level, it will still cause the electrolyte to dry out over time. Once that happens, the internal resistance of the capacitor will rise, as will the internal pressure, and it will heat up itself more and more, possibly even to the point of blowing up. Most capacitors fail rather silently though. Some start leaking which is pretty bad for a PCB as the electrolyte is chemically aggressive and corrodes other parts around it, as well as the PCB traces. While this may not happen rapidly, leaking caps may still work good enough so the owner does not notice anything. In unsealed power supplies, the tell-tale smell of electrolyte would quickly notify anybody around. That's unfortunately not working alike in the Squeezebox power supplies as they are all ultrasonically sealed all around. This type of failure may go unnoticed for months or even years.
A capacitor's metal can has deliberate weak spots, in most cases at the top, which will fail by design if pressure gets too high. This is to ensure that electrolyte will evaporate in a defined place, instead of all around as the can rips apart. Shooting out at the top (away from the PCB) is better because the overall contamination may be less extreme. Anyway, it's usually quite a mess when this happens.
A switching power supply employs capacitors in the input stage to filter the input high voltage after it was rectified. The voltage may spike up to around 380 Volts (in 230 Volt environments). Therefore, these capacitors are rather large, and specified to take up to 400 Volts. If they fail, the rest of the circuitry will have to deal with a lot of ripple and dropouts in the mains frequency (50 or 60Hz). The outcome cannot be good.
As switching power supplies create their output voltage by turning the output on and off in a high frequency (I think that's around 600Hz?), the output needs a lot of filtering to come as close as possible to a constant target voltage, and remove all of the on/off ripple of the previous stage.


The capacitors mostly fail in the this (low voltage) stage. A Classic power supply usually has two 680µF / 10V rated electrolytic capacitors in parallel to the output. The output is +5V so a 10V rating should give a good amount of headroom, still there is the heat they have to withstand constantly. One could call this a design flaw but it's the same design in every switching power supply, and there isn't much designers could do to prevent this except using capacitors of higher quality. Even those may fail sooner or later. It's the constant demanding duty that kills them.

Effects Of A Failing Power Supply On The Squeezebox Classic

You may notice issues related to bad power supplies by one or more of these behaviors:
  • weak / flashing / flickering light in the TOSLINK (digital optical) connector --> See Issue 6 below
  • Squeezebox not turning on at all --> See Issue 6 below 
  • crashing on playback --> See Issue 4 below
  • Wi-Fi failing whereas LAN / Ethernet connection is still working
  • Flickery display
  • Boot loop (Logitech logo appears, screen blanks, then it repeats after a few seconds)

Issue 4: Crashing Instead Of Playback Starting

If your Classic appears to be starting up normally, shows the boot logo, registers with the network etc., but restarts as soon as anything should be played back, that's a good indication of the digital/analog converter (DAC) being in trouble. The DAC is only used to convert digital audio to analog to supply the RCA outputs as well as the headphones connector. Even if you are using the digital RCA, or the optical TOSLINK output, the DAC is still used. So you cannot configure the Squeezebox to ignore DAC errors.
The DAC is in close communication with the FPGA on the mainboard. If the DAC does not respond appropriately, the FPGA will tell the CPU to restart in an attempt to recover from the malfunction. However, if the reason for the malfunction is a hardware defect, there is no way for the device to recover by itself.
Again, it may be failing or failed electrolytic capacitors that cause the DAC power supply and signals to exceed the defined margins. Another reason might be a power supply that is just strong enough to power the Squeezebox to the point where its playback circuitry engages, and then fails.

Please follow this link to the Capacitors article: https://joes-tech-blog.blogspot.com/2019/12/logitech-slimdevices-squeezebox-classic.html

Issue 5: Bad Analog Output, But Digital Is Working

If you have analog as well as digital equipment, and can verify that the digital outputs are working while the analog outputs are dead, muted, or distorted, this means that the DAC is still powered far enough not to fail in communicating with the FPGA, but its analog part is no longer supplied stable voltages. The first step here is to replace the electrolytic capacitors just like in the Crashing issue above.

Issue 6: Unit Not Powering Up At All

If you have verified that the power supply is okay, but the Squeezebox is dead, check the digital optical output first because it gives a good indication of how bad things are looking.
This illustrates how to find the TOSLINK port:


The port has a little protective cover that can be pushed inside on one side of the connector. Do this carefully to get a clear view of the LED emitter in the middle of the connector. It's not Laser so don't worry about your eyesight.
Usually, the TOSLINK LED should give you a solid red light, not blindingly bright, just like a typical red LED. The darker your surrounding is, the better the chance to see whether any light is emitted at all.
This LED is always turned on as long as the Squeezebox is connected to power. It should never switch off, flash, or flicker. If you perceive a reduced brightness, that's probably just a flickering on a frequency too high for the eye to notice. A slower flashing (still multiple times a second) indicates that circuitry in the power supply, or in the Squeezebox, is rebooting constantly because it detects something is wrong.
If the TOSLINK LED does not appear right, this usually goes along with the Squeezebox no longer starting up at all. You won't see anything on the display, and it won't register on the network, neither via Wi-Fi nor via the Ethernet port.
Besides checking the capacitors, also try the steps listed in Other Troubleshooting Advice below. 
It may be necessary to re-flash the Flash EEPROM because in a lot of cases it has become corrupt.

Issue 7: No Ethernet

One unit that was sent here for repair actually did not connect via Ethernet whereas a connection via Wi-Fi was possible without issues. Other than that, the unit had the typical reboot issue on playback.
To be more specific, my Ethernet switch would indicate a connection all right but the SB3 would not successfully get an IP address from DHCP, and even assigning a specific IP address, subnet mask, and gateway IP would not have it connect to anything in my infrastructure.
Surprisingly, this was also fixed by the capacitor replacement!

Other Troubleshooting Advice

Note: all of the following steps require opening the case and extracting the board so you have full access to all parts of it. See here for a guide how that is best done.
You will need just a TX10 torx driver. The housing is held together by two TX10 screws at the back, and inside, there are four more TX10 screws that hold the board in place.

The unit should be completely disconnected from everything, especially the power supply, for all steps illustrated! If you do any of this to a powered unit, it may cause irreversible damage beyond repair.


Reseating the Wi-Fi board

The forums are full of threads where users describe that removing and plugging the Wi-Fi board back in cures all sorts of problems.

Once the Wi-Fi board is out, clean the gold-plated contacts at the card edge connector with IPA, and remove any dust from the connector area of the CPU board.
Now is an excellent time to see how the Classic behaves without the Wi-Fi board plugged in (only the CPU board). You may find that it powers up when it formerly didn't which might mean the Wi-Fi board is blown and needs to be replaced. See Replacing the Wi-Fi Board below for more info on this.
Before connecting power, please ensure that the Wi-Fi board is out of the way and isolated far enough that it won't short any contacts elsewhere.

To push the Wi-Fi board back in (unless you plan to reseat the CPU board as well as shown in the next section), use the same angle that the board had before you removed it. Push it firmly into its end position and make sure it's flush in the connector on both sides and not askew by any means. If in the correct position, you can push down the raised edge of the board until both metal brackets click back into place. It should not be necessary to touch the brackets for this, and should happen with no violence involved. Compare with the photo above to verify the board is in the same spot is was before. If it does not fit, pull the Wi-Fi card out and reseat it another time. Be sure not to bend the metal brackets more than necessary in any phase.
When everything is back in place, connect power and see whether things have improved.



Reseating the CPU board


It is not recommended to use any tools against the adhesive pad. Any hard or sharp tool might shear off or damage components on the CPU board. If you want to be totally sure, use adhesive remover, or IPA, to weaken it, and give the agent time to sink in. Drown it in IPA multiple times. IPA does not hurt any electronic component of the Squeezebox but you should ensure that the adhesive pads behind the VFD display stay intact (unless you plan to replace it).
The board should not be pushed hard or torn at because that puts a lot of strain to the rather delicate board-to-board connector. Gently pull the CPU board straight away from the mainboard, e.g. holding it between two fingers and pushing the fingers beneath the board, the same amount on both sides. Do not twist or skew the board. The adhesive pad will give way sooner or later, just be patient.
Once the CPU board is out, you can give the connector a cleanup. All connections are gold-plated so there should not be any corrosion, but fine dust particles could still have made their way in there.
There isn't much we can do about the CPU board. There are no electrolytics, thankfully, and everything else is more or less non-serviceable. Just clean it up if needed, and place it back on its connector.
The adhesive pad may have to be replaced, especially if it was soaked in IPA. The CPU board plus the inserted Wi-Fi board together create a weight that the connector alone cannot support in the long run, and unfortunately, Logitech did not make use of the little hole in the CPU board that could have been used to bolt the board down to the mainboard with a screw post, just like in the Transporter. The Classic mainboard does not feature a matching screw hole so it is advisable to ensure that a reliable adhesive pad can be used to hold the boards in place.


Replacing the Wi-Fi board

If you observed that the Classic works without the Wi-Fi board, and fails when it is set, it's clearly the culprit and should be replaced. See List Of Spare Parts And Sources below to find out how to acquire a new card.


Still Unsolved: Sudden Death Syndrome

(it's what I call it, I am not aware there is a more professional expression for it)

The Boom has some reputation for failing from one day to another without warning. When being powered on (plugging in the power supply), it just activates the button backlight on the front panel, or cycles quickly through the backlight LEDs in the front panel, and that's it. No Logitech logo appears, network connectivity is off, the unit does not even get warm as the filament wire heating is not even turned on in this state. It's brain dead, so to speak.
The Classic is similar in construction and choice of components, and unfortunately I have seen cases of the same symptom in the Classic, i.e. the unit does not respond to anything, at best the TOSLINK LED emits some light but for sure not the usual brightness.
I think I have isolated this error to the CPU board as all other parts of the affected Classic are working when a different CPU board is in use. However, just as with the Boom, it's mostly unknown what happened and how it can be fixed. I have extracted the Flash EEPROM already and verified that it still contains the correct data, which it typically does. So this is not the result of a Firmware flashing process that went bad or was interrupted. The only remaining components on the CPU board are the CPU itself and a DRAM chip.
One Boom came back to life when I refreshed the soldering on the DRAM chip but soon after reverted to the non-working pattern, so I was close but did not actually fix anything by resoldering.
Dealing with the CPU is really hard as it has a ball-grid array (BGA), i.e. all solder joints are hidden beneath the CPU chip. To remove the CPU, all these solder balls need to be heated up, and putting a new CPU is chanceless without the right tools because there is no way of knowing whether the soldering is complete for all BGA balls. Just one ball with a bad or broken connection will render the entire device useless.
It's well known from gaming consoles that BGA soldering suffers a lot from excessive heat generated by their CPUs and GPUs, but the Ubicom IP3023 processor used in the Squeezebox Classic and Boom does not produce a big amount of heat. It rarely warms up beyond 40°C which is way below the point where lead-free solder begins to degrade. So this is still a mystery, and I hope you did not get to this point and found exactly this set of symptoms. Because that means that I cannot help at this point. I will keep investigating, there might be a solution some day.
 
EDIT 2023-08-19: finally there is something to say about this. It's really good news actually. I have figured out that many SB3s are not powering up because part of their Flash EEPROM has become corrupt. This causes the CPU to crash and restart in the bootup process. The power consumption oscillates between 0.5 and 0.8A, and in some cases it even gets far enough to show the Logitech or even the "squeezebox" logos but eventually restarts all over. It is not fully clear yet why it happens at all but there is a solution. The EEPROM can be desoldered, reprogrammed, and soldered back in, and in all cases so far this was a success!
So what I claimed to be a dead CPU actually never was, and I apologize for this misunderstanding. It,. or its soldering, may be part of the problem but the success rate of EEPROM reflashing proves it wrong, mostly.
I am speculating here about the reasons why the Flash EEPROM becomes corrupted. It could be what hackers call "glitching", or a brown-out condition where the CPU does erratic things because it doesn't get enough power for a short period of time. As the Squeezebox firmware is capable of overwriting itself with a newer update from the LMS, and also stores some config data such as the Wi-Fi access parameters and its MAC address in Flash and both can be dynamically changed, there are some "write to Flash" mechanisms in use.  The Squeezebox firmware can be understood as one large program where all routines and static content such as graphics, fonts, etc. are in memory at the same time. Having the CPU jump from a block of code that is dedicated to playing music to a completely different part that deals with writing the Flash memory is just a matter of a program counter jump to a different memory address. This might be what happens when the power supply is on its way out and causes the first little dropouts, or the capacitors failing to stabilize the supply inside the SB3, or a combination thereof. There is no temporary write protection in place that might intercept these random instructions. From all the defective EEPROM images I took from the SB3s that could be repaired by reflashing a working image, I cannot say for sure if they have anything in common. Actually it looks like every image is different from the working one in different locations, but no two are affected the same way. It's also surprising that this random overwriting of the Flash memory always causes a boot loop, but never rather harmless things like corrupt fonts or graphics.
I know too little about how the IP3023 and the Flash EEPROM interoperate, and how the CPU processes what it finds in the EEPROM so it's really hard to say why it fails to boot. If anybody knows more about debugging here, I'm ready and willing to hand over the binary images that I have captured so far.
A deeper investigation of the area around the EEPROM gained me another insight: the test points on the back of the SB3 CPU board connect directly to almost all of the 48 pins of the chip! This means that with a "bed of nails" board the EEPROM might be reprogrammable without the need to extract it first and solder it back in later. I have tried to design such a PCB in KiCad but made a lot of mistakes. Firstly, no matter what pogo pins I tried, the drill holes were too small in diameter. Pogo pins are not known in the stock KiCad and I didn't know how to design what I need to get the PCB created as intended. I didn't leave enough solder eye around the holes so enlarging them with my own drill would not leave anything to solder to. The drill would be a bit too large also so the pogo pins would not stand exactly upright. Moreover, I have no idea yet how to mechanically connect the SB3 board in a way that it is consistently pushed down firmly enough for all pins to make good contact. Probably a 3D printer job but I have absolutely no expertise in this. It would need to include some means of mounting the frame to the adapter board stack (one signal PCB and another one to guide the pogo pins) and having some reusable clamp to hold the SB3 board down.
Then there is the question how the other side of this construction (towards the chip programmer) would look like. Certainly not a TSOP48-style connector to emulate a chip.
The typical USB programmers have Textool sockets but in order to read/write a TSOP48 chip, a more or less sophisticated adapter is needed, and it must be compatible to the programmer, so it's hard to give a generic solution here.
I have an older TL886 model here, and the adapter (bought from somebody in China) is a little PCB stack where the bottom PCB has some signal conditioning on it and at least one chip with the markings sanded off (bastards!), and on top of that there's the TSOP48 socket on its own PCB,  connected to the one below with two 24 (12x2) pin rows, the ubiquitous type we also know from Raspberry Pi, Arduino, floppy drives, older hard drives and so on. My adapter replaces this TSOP48 socket and has the same sockets towards the bottom, but needs the signal conditioning board with the unknown chip nevertheless.
However, the older TL886 is now being replaced by newer TL886-II programmers, some of which have an extra ISP connector and a totally different construction, with the TSOP48 socket on the top PCB along with an unmarked SOP8 chip, and the need to run an additional 16-pin cable from the programmer to the adapter.
I think, in both cases these unrecognizable chips provide some kind of key for the programmer to know what it is currently dealing with, but seems to be proprietary. Might be some counterfeit protection also. No idea if we could create an open board to connect the SB3 to one of these newer programmers.
 
I will add more information and photos about this over time.

List Of Spare Parts And Sources

Power Supply

  • Germany/EU: look for "VHBW" at eBay, they are selling replacement PSUs explicitly specified to work with given Squeezebox models. Be sure though to pick the Squeezebox Classic, not the Radio or something else
  • USA: Mouser offers some power supplies that look like they could do the job. Others like DigiKey, Farnell etc. certainly are all offering stuff that works out for you. Search for "DC 5V", then select Wall Mount AC Adapters, pick 5VDC output voltage again if needed, and 2A-3A current. The inside barrel diameter should also be selectable. As pointed out above, 2.1mm or 2.5mm are most popular but I can't say for sure which one is better.

Wi-Fi

It is still surprisingly easy to find replacement cards which fit the Squeezebox Classic thanks to the MiniPCI standard interface that is used. You mainly need to focus on the Atheros AR2413 chip that the card needs to be built upon. Any MiniPCI card having the AR2413 chip will most probably do. However, pay attention to the placement of the antenna connectors, and the physical size should be the same also. There are cards that have just one antenna port which would at least reduce, but might also disable the Wi-Fi reception of the Classic.
For buying a new card, just check the usual suspects in the web for

Atheros AR2413 MiniPCI

Buy one that is closest to the card you extracted from the Squeezebox in look and size, and where the seller mentions the AR2413 is built in. In the ideal case, there are photos in the offering where the chip is visible.

I have two examples here. The green one is what you will find in Squeezeboxes most of the time. I found the blue one on eBay a few years ago for around 3 EUR a piece. The photo where the metal shields were removed show that these two boards are mostly identical regarding the chip placement and the components surrounding it. The main difference is the power supply which I think is more efficient in the blue card. On the green card, the power regulator that can be seen in the top left corner gets really hot which makes me think it's wasting a lot of energy there. The blue card does not have hot spots like this.




There are half-size cards and MiniPCI Express cards which you should avoid. They won't work in the Classic. They are way smaller and look like this (notice the different card edge connector):

 
So if you get results like MiniPCI-e, MiniPCI-x, half size, half length or anything similar, ignore them. Not suitable for the Squeezebox.


Display

The display is a Noritake MN32032A which is still manufactured and sold today. Depending on the sales office, you may be lucky and purchase one directly from them.

I found a data sheet on Google, find the closest sales office in the bottom right corner. If all else fails, I can supply these displays, however, outside EU the shipping fees are out of control, thanks to Covid-19.

Capacitors

see here

 
<< TODO: Power Jack>>

<< TODO: Battery tester>>


Final Words

Legal Disclaimer

Some legal stuff because you never know: please bear in mind that I am writing this as a hobbyist, not a professional. I describe personal ideas here which is only one of many ways such a repair can be achieved. I cannot guarantee that following this guide will lead to a good result, and cannot be held liable for any personal, physical, or monetary damage anybody suffers by following this guide.
I am open to advice if anything described here is wrong or can be done better. Please let me know in the comments if you find there is anything left to be desired.

Repair Service

In case you would like me to do any of these repairs (including display replacement), just drop me a mail at johannesfranke74@gmail.com. I have good experience with Squeezebox devices and own several Classics myself which helps a lot in troubleshooting and comparison. Repairs are done at a fair rate and with warranty on parts and work done.

Thank you for reading!