Joe's Technology Blog: 2017

Hi there!

Apparently more and more of people's Transporters are having problems lately as they are getting around 10 years old. This is to point out the issues and how to solve them.
This is a documentation of my experiences so far with my own Transporter as well as four devices that I had on the table to fix them. So this is certainly not comprehensive or complete but covers popular issues, such as:

fading displays
failed +5V switching power supply, causing:

"dead" displays no longer showing anything while the rest of the unit works
boot loop
not powering up at all
blowing 500mA mains fuse
repeatedly clicking relays
glitch noises such as intermittent chirps or squeaks
random server reconnects and Wi-Fi trouble

noisy DAC / bad analog output

Let's get into the details of each.
You will need some tools for disassembly:

a 1/16 inch hex bit for the outside screws holding the top part of the case, and to separate the back panel from the bottom part of the case (if you need to do soldering work on the mainboard)
a TX10 torx driver for the internal screws

If you plan to repair the existing PSU (+5V power supply unit), you will also need:

desoldering tool or material like solder wick
an appropriate soldering iron (I recommend one with no more than 20 Watts, set at ~320°C)
small diameter heatshrink tubing and a heat gun
new capacitors:

2x 680µF/10V
1x 22µF/400V or better 450V

a hot glue gun (optional but recommended)
some adhesive tape

Fading Displays

Just like any Squeezebox that employs a VFD-type display, the Transporter's displays are also susceptible to the burn-in problem. See here for my blog entry on the Boom that explains the VFD and its potential problems in depth.
Just to add here, the Transporter is also keeping the displays 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 Transporter is not busy otherwise. Similar to the Boom, the automatic brightness adjustment exaggerates things a little and tends to put the displays on a rather high brightness level which encourages accelerated pixel burn-in.
This is the main reason for shadows appearing "over" the actual display content. But the VFDs are also aging evenly with the years. Comparing a display that was driven rather softly to a brand-new one will also show that the newer one is overall considerably brighter.
I think it's safe to say that the displays age only while being powered. The filament wires are constantly on even in standby, and even if the displays are configured not to show anything in standby mode. I used to state that the filament can bear this easily but I might be wrong here. I could not explain the overall aging (of all pixels, also the less-used ones) otherwise.
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.
There is no blog entry on this yet but a lot of the Boom-related article applies to the Transporter as well in terms of the actual desoldering / removal process of the old displays and setup of the new ones. The housing is totally different of course. I may add a separate article on this when I have time.

Some more information about VFDs in general and how to identify the aging issues, see here.

Consequences of a Failed +5V Switching Power Supply

This is all about the yellowish small PCB between the front panel and the green mainboard in the left half:

This module is constantly powered, independently from any other part of the circuitry. It creates a constant 5V / 2.3A supply to drive the display as well as the front panel circuitry. It continues through the white flat flex to the mainboard where it powers the digital parts of the circuitry. So even if it does not seem so, most of the Transporter directly depends on this little module. This is why its failure causes so many different problems. Here's a closeup of the module in question:

In my opinion, Logitech made a questionable design choice here. While the rest of the Transporter is state of the art in terms of build quality, this looks like a Chinese 3 US-$ part, really something not up to the quality standard of the unit at all. So it's no surprise this is the most popular point of failure. Probably an afterthought. However, there are three (sometimes four) screw posts in the Transporter housing which fit the screw holes of the PCB perfectly so this power supply has always been part of the concept.

Symptom 1: Dead Displays

When you observe that your Transporter is playing music as it used to, so it is fully controllable via LMS in the browser, a Controller, or the infrared controller, but the displays stay dark (and the usual heat they radiate is missing), then this is a pretty safe indication of a failure in the +5V power supply module.
So one of the failure modes is that the displays stay off while the power supply is apparently still capable of powering the rest of the Transporter enough to make it work. In some cases, even the front panel controls (the TransNav knob for instance) are still usable.
The repair options for this are discussed further below.

Symptom 2: Boot Loop

Another failure mode stemming from this PCB is when the Transporter starts with the usual relays clicking and the Logitech / Transporter logos appearing on the displays, then restarting with the same procedure indefinitely. A usable level is never reached, and it is not possible to control the Transporter through IR controller, or LMS. It won't even become visible on the network.
In this case, the power supply board is capable of doing its thing for a few seconds until it collapses and internally restarts over and over.
The repair procedure is same as for the previous issue so continue below to find out.

Symptom 3: Transporter Not Powering Up At All

If your Transporter stays dark and does not give a single click after plugging in, it's the third failure mode of the display power supply. In this case it has virtually no output anymore, i.e. the front panel has no power. As the CPU checks rather early in its boot sequence for the presence of the front panel's circuitry, it will reset before anything else observable happens. An indication of this failure is the LED that is near the CPU board is blinking permanently. It is usually off and flashes along with infrared commands coming in.

Symptom 4: 500mA Fuse Blowing When Being Plugged Into Mains

Another repair proved the 5V supply can also be the culprit when the 500mA mains fuse blows each time the Transporter is plugged in. This can also be the case if one of the toroidal transformers is shorted, but that usually goes along with some noticeable humming noise before the fuse blows. If the 5V supply is shot, the fuse will blow instantly on mains connection.

This also applies if you find the fuse has blown, and a new fuse stays intact. This may last for some hours or even days, but keep in mind there was a reason the fuse blew, so you should not assume that putting in a new fuse fixed the actual issue.

Symptom 5: Spastic Relays

This may happen with a single incident at first which does not quickly repeat. Out of nowhere, the Transporter gets dark and restarts after some seconds. While this is very infrequent at first, the interval will get shorter over time until all that happens is that relays are clicking, maybe the device stays on long enough for a logo to appear, and then they click again, repeating the same cycle indefinitely. At some point the logo will no longer be shown because the device restarts sooner. And then it's basically the boot loop described in #2.

Symptom 6: Noises in the Playback Stream

Something I observed in my own Transporter is, besides some more of the issues listed here, that during regular playback it randomly inserts a sound similar to a little bird's chirp. I think this was pretty rare at first and got worse over time until it happened about 5 or 6 times per minute at irregular intervals.

What's interesting here is that it affects all outputs, even the digital ones. So the error is not in the DAC but way sooner.

I tried to record a section where the chirp occurs, here is a sample:

Link to original recording

Link to filtered sample

You can download and play back these samples, the issue can be heard right next to the end. For the filtered sample, I have applied Audacity's graphical EQ to drop all frequency bands to the bottom except the one around 6.3kHz which was amplified as much as possible. After that, normalized the entire sample to -1dB.

This is what the waveform in this area looks like in the filtered sample.

Looks kind of analog because it appears to fade in and out, which can also be perceived in listening, A frequency analysis of this section got me here:

Hmm, suspicious. Peak around 6553 Hz, a value that looks oddly familiar, considering that 65535 is the highest value for an unsigned 16-bit field, and here we are close to a tenth of that.

Whatever, might just be a coincidence, and I have no further technical explanation as to what is the source of this sound. But since I repaired the power supply, I never heard it again, so it could be some kind of glitching that happens when the PSU drops out and recovers just quickly enough so the Transporter won't crash or reboot, but gets out of whack somehow, and inserts this artifact along the way.

Anyway, I consider this another thing related to the power supply as fixing that also fixes this issue.

Symptom 7: Reconnects and Wi-Fi issues (probably related)

I found that my Transporter would randomly display "Connecting to server..." in the left display while it was sitting there in standby, and no other parts of my network were touched at the same time, so there is not actually a reason for losing the server connection. My Transporter is configured to run over Wi-Fi so I cannot tell if the same would happen on a wired connection. Anyway, this always ended up in the Transporter getting back to sleep after a few seconds, and being operable further on.

However, sometimes the Transporter would disconnect and stay so, and refuse to connect to the home Wi-Fi network, or not even find it. A reset, triggered by keeping the POWER button pushed for 5 seconds or so, usually resolves it, but the issue keeps comes back.

This is another one I observed in person, and found it cured by repairing the 5V PSU.

Failure Analysis

Presuming the Transporter is already opened, check the on-board LEDs first. This gives a good indication of where to continue.

There are four red LEDs, one is near the CPU (A) board in the left half, and three are in the DAC section (B).

LED A should be on shortly on initial powerup (when the mains plug is connected), and then stay off. It only lights up later when the Transporter receives commands from the infrared remote control, or (probably) from the 3.5mm remote control port in its vicinity. Whether this LED ever lights up depends on the main processor. If it stays dark all the time, the CPU is either missing a power supply, or is defective in itself.

The LEDs in group B are usually switching on at the same time. The left two indicate power for the negative and positive reference voltages of the DAC, the rightmost one indicates DAC digital supply power. If any of these LEDs is not going on, it is more likely that the mainbaord power supply (the two blue toroidal transformers) are at fault.

The main reason for the 5V PSU failure is apparently heat (besides a choice of low-quality components). Everybody who is a little more into fixing Hi-Fi equipment from the eighties or nineties knows that, most of the time at least, issues arise from aged capacitors. Electrolytic capacitors specifically because their liquid or gel electrolyte dries out over the years which lowers their capacity more and more until the circuitry that relies on them can no longer work. Another problem is that the design of consumer appliances is often focused on maximum turnover, i.e. the components used are sometimes not ideal for their place because manufacturers picked the cheapest ones.
Electrolytic caps have three values to be taken care of: maximum voltage, capacity, and maximum working temperature. Voltage and temperature are determined by the surrounding circuitry and it is always a good idea to add some headroom to these values. For instance, if you know the circuitry will use the capacitor in a place where up to 240 Volts may be present, use a capacitor that is specified up to 400 Volts because 240 Volts are converted to peaks as high as ~350 Volts during rectification, and the primary filter capacitor is typically placed right as the next component after the rectifier. But also in a pure DC area it is better to pick capacitors capable of handling considerably more voltage than they actually have to in daily life. They will hold up longer because they are never stressed by working at their designed limits.
Likewise, capacitors are often specified for 85 °C temperature where 105 °C would have been the better choice, just to be sure. In switching power supplies, there are multiple components that run pretty hot by nature. If an electrolytic cap is placed in the same region as components running red-hot, failure is guaranteed sooner or later. But using caps capable of handling higher temperatures, you can push the failure further ahead.
In this respect, PowerPAX did their homework, all electrolytic capacitors are properly rated:

C1: 22µF 400V 105°C
C2a: 680µF 10V 105°C
C2b: 680µF 10V 105°C
C3: 470µF 10V 105°C

The output filter caps (C2a+C2b, driven in parallel) are 10V rated which is double the output voltage. So far, no obvious reason for why it fails anyway, right?
Let's have a look at the temperatures:

30 Minutes later:

These images are from a working power supply where no failure ever occurred. However, remember that this unit is permanently on, i.e. the components are permanently fed and there is just the room in the Transporter's housing to distribute the heat. The casing does not have any air vent so it will circulate in there. Here is no heatsinking whatsoever. So these temperatures may go well beyond the 64.7°C I measured rather early after a cold start. The measurements from 30 minutes later show that the main switching transistor at the top got about 7 °C warmer whereas the other one accumulated 2 °C additionally. This may get worse over time. I would not be surprised if they reach around 100°C at some point. Outside temperature adds to this, too.
Let me repeat the layout photo from above for comparison:

While the switching transistor at the top is far away from the big primary filter cap (top right), the second transistor (or diode?) can freely radiate its heat towards the secondary filter caps C2a and C2b. This may weaken them over time.
You will probably find this:

Observe the dark brown region about the Zener diode ZD3, a bit left of the center. This is a clear sign of excessive heat, and it's in close proximity of C2a and C2b, also not too far away from C3 (indicated just by silk screen on the left, at 45° angle, because I desoldered it).
So the failure may have come from the Zener diode's heat instead of the transistor's. In a working PSU, this diode is not getting warm significantly at all so this might be happening only as the capacitors are dying.
While the capacitors may appear good in direct measurement, my recommendation is to replace them all. That's not the maximum ecological sustainability but these caps are probably already failed or close to failing after 10 years of constantly being powered.
You may also find one or more caps having bulged tops or having risen from the board with bulged sealings at the bottom. A capacitor is designed to fail at the top, hence the little indentations you can see there in the form of a plus sign or three indentations going from the center to the outside. The metal can is thinner along these lines to ensure that they are the defined failure point should overpressure cause a rupture. In this case, (very unhealthy) electrolyte and other stuff shoots out. On a board mounted vertically this is good because the electrolyte will spray in a direction away from the board. However, if the board is mounted horizontally as in the Transporter and many other devices, a blown-up electrolytic cap can hit a large region around it. Therefore it is best not to let this happen at all.
The rupture points also fail sometimes, for instance the bottom sealing may break sooner than the thinned metal lines which causes the electrolyte to leave the capacitor at the bottom. While both failure modes may cause damage, the more severe damage comes from seal failure because the electrolyte is very aggressive and starts damaging the PCB immediately, up to completely destroyed PCB traces in the worst case. It is much more concentrated then due to the little space it has to distribute. If you observe patches with darkened PCB traces and / or dull-looking solder points, that is a pretty safe indication of accelerated corrosion by electrolyte. A UV (ultraviolet) light also helps reveal the electrolyte.
Capacitors showing any of the said weaknesses should go as soon as possible. Again, my recommendation (specifically for this Transporter power supply) is to replace all of them in this case no matter how good they look outside. If you have proper measurement equipment like an LCR meter, you may also desolder the caps and do an extensive check, replacing only the ones having bad values. A thermal camera can help reveal that a capacitor is getting unexpectedly warm which indicates that it does not work properly anymore. They should never "produce" heat in operation.

Fixing the +5V Switching Power Supply

Depending on your skills and comfort you can fix the power supply or replace it. I like to be sustainable if possible. Actually the value of the failed parts that need to be replaced is typically no more than 1 or 2 EUR / USD.
The model is a PowerPAX SW3376 and is distributed by RapidOnline. See here for the product info / order site:

https://www.rapidonline.com/electrical-power/power-pax-uk-mini-open-frame-smpsu-eup-5vdc-2-3a-85-2251

You can see there that a single unit costs only around 12 USD. The module may be available in other countries with other resellers, too.
So many people may wonder why one should bother repairing the unit if the replacement is that cheap. It may also be a consideration that a new module has a completely new life span while replacing single components will leave the rest of the board in its aged state, so trouble may come back.
I have some arguments against that:

while a new module has a certain appeal to it, you can be sure that it will fail again within the same time span as the one you discard. The build is the same, and so are the components. The design flaws are still in there and will cause it to fail again
if you are capable of doing a component-level repair, you will be just as capable in case the unit fails again. It's way cheaper than buying the entire PSU as a replacement
let's not forget about shipping cost, taxes, customs etc. which will raise the price even if you place a larger order
it's kind of wasteful to discard a PCB with about 30 components on it just because three or four of them failed, and replacing them is so easy.

EDIT 2023-06-12: there is another high-quality power supply that I have tried as a replacement. Find more about this story below.

In any case, you will have to remove the PCB so let's start at that.

Opening the Transporter

unplug the Transporter from all connections, most importantly from the power line
unscrew the outside screws, 4 on each side, with the 1/16 hex bit
pull the top cover away from the front. It will slide over the back panel.

CAUTION! Be extra careful in the first few millimeters because there's a capacitor mounted on the front panel board that might be sheared away from the board if the top case drops onto it, and it does tend to do so! Be sure at all times that you push the top case evenly to the back, and support it so it does not drop down after it leaves the edge of the front panel.
This may be hard at first because there is really little tolerance in the Aluminum parts, so be sure to push/pull the top cover straight to the back and avoid tilting

Checking the Fuse etc.

there is a slow-blow glass fuse (500mA / 250V) near the mains connector on the main board. Check it for continuity. If your Transporter is completely dead, and no light at all can be seen, this could be a very simple reason. A blown fuse is cheap to replace and if the next one does not blow, consider your work done. As fuses are also parts susceptible of aging, they can blow at some point with no given reason.

The fuse might also have blown as a consequence of the 5V PSU failing. So better have more than one at hand when you try and replace the defective one. In case the next one also blows straight away, the PSU needs to be taken care of first before you can replace the fuse again.
locate the +5V PSU in the left front quarter of the housing

PSU Board Removal And Preparations

unplug the two wire connectors. The one on the edge facing the green mainboard might not have a supporting screw, so be careful when you unplug the connector there
the PSU is bolted down with three or four TX10 screws. They are not magnetic by the way so even if you have a magnetized screwdriver, the screws will just fall down
carefully lift the PSU board out, avoiding to touch its underside or any metal parts
it is good practice to short the large primary filter capacitor (labeled C1 in the picture below) because it might give you a high-voltage shock that is actually dangerous. There are multiple ways to discharge it:

A typical classic 120V or 230V light bulb is recommended if you have one at hand. With a socket and a bit of wire you can just connect the capacitor's terminals to the light bulb. It will probably not flash or glow at all but after a second or two the capacitor's charge is practically gone
You can also use a resistor that is specified for the voltage and can take some more current. Use one of the chunky ceramic ones because the flimsy standard resistors may blow up when overloaded
An emergency "poor man's" fallback method, albeit not recommended if you can help it, is to short the capacitor with a piece of metal. Just be sure it is isolated on your end. This may cause strong arcing so try to be as far away from it as you can

if you have ordered a spare PSU, you are practically done. Just put the new board in the place of the old one, and screw everything back together. Consider a quick test before you put the top cover back on.
if you chose the DIY repair path, desolder these electrolytic capacitors:

C1 and C3 are held down by Silastic material. C1 can be freed for instance with a blade that you push below it. Don't be too violent because the rectifier diodes are hidden beneath (might be a good thing to eventually remove the Silastic from them to give them a better chance of staying cool)
C3 has little room around it so once it is desoldered, all you can do is pull it out with as much force as needed to overwhelm the glue. If it does not give way at all, make sure the pins are actually free and no longer soldered, and try again
by the way, C3 is rarely at fault so you might leave it in place first and see how far you get by replacing the other three

A Word On Electrolytic Capacitors

With a few exceptions, they are unipolar, i.e. they have a negative and a positive terminal and must be placed accordingly! If you reverse them accidentally, they will be destroyed and that's the least of the problems. In high-voltage areas, they can blow up violently and cover everything with strongly alkaline electrolyte which has similar effects as acid. Luckily, the silk screen on the board indicates which is the negative terminal, and capacitors are also marked, typically with a stripe along one side of the body:

The left red arrow points to a small "+" sign on the board. This is not consistent in all places but makes it even easier to recognize the polarity. What you will always find is some sort of marking for the negative terminal which is the black filled half-circle in this case. So this is where the marked end of the capacitor (negative terminal) needs to go. You can also identify the positive wire which is longer than the negative one. See these images at Google for examples of electrolytic capacitors. They all have a clear indication of where their negative terminal is. If they don't, they are probably not capacitors, or bipolar, anyway not applicable for this repair.
For reference and to remove remaining doubts, here is the polarity map another time (red = +, blue = -)

Some More PSU Info

As we know that capacitors cannot deal well with heat, we have a good chance now to improve the situation. Instead of putting the replacement capacitors back into the same spot, my recommendation is to place them as far away from the board as possible by using the full length of the capacitor's legs. Use shrink wrap to ensure that they do not short with each other or with any other component on the board.
A result might look like this (C3 was not replaced here because it was still good):

There are a few things to consider for this solution:

the image shows that the front panel flat flex cable is running across where the capacitors are placed. The original Kapton sticky tape is placed approximately where the large black capacitor is now so the tape needs to be moved, or better, be replaced and put in a more convenient region
below the PCBs there is a sheet of strong plastic material, forming an isolation layer between the PCBs and the bottom metal casing. It is stuck down full length along the left edge but not anywhere else, if at all. Depends on the date of manufacture, I have also seen units where this sheet is only held in place by screw posts. Anyway, beneath the 5V PSU it can move between the bottom metal shell and the PCB which is mounted on ~5mm posts. As you will probably put the capacitors on top of this sheet, you should ensure that nothing can move much. Otherwise, the sheet along with the capacitors stuck to it might wear out the thin capacitor legs so they get brittle and lose contact. Run some lines of hot glue along the edges of the plastic sheet to stick it firmly to the metal bottom shell as shown in the photo in three places in the bottom right corner. This is at least worth considering if the Transporter is not yours and you have to send it via a parcel service. You will have to undo this whenever you want to service the board next time so before you do this, better run a test to see if it all works as desired.

The image also shows that I did not do a complete replacement. C3 was not replaced at all.
If I had replaced C3, I would have isolated and run its legs towards the left side, and stuck C3 down against the plastic shield there, just like I did with the other capacitors.
Another deviation: for C2a and C2b which are connected parallel anyway, I picked a single replacement capacitor. I think it's 2200µF/16V instead of the original combined 1360µF/10V (2x 680µF). A higher capacity is not actually necessary but I did not have any closer values at hand at the time of repair and it is not too far off the original value. It may be intriguing to pick even higher capacities, assuming that the capacitor will then do a better job of filtering the output and stabilize it. However, a much higher capacity is actually counterproductive here due to the nature of capacitors. A discharged capacitor has a very low resistance while being charged. The higher the capacity, the heavier the load for the circuitry feeding it. This is close to a short which may damage other components again. While 2200µF is considerably more than 1360µF, we are in a low-voltage environment in this part of the circuitry and 2200µF is not an extreme capacity, so the components feeding the capacitor should be capable of bearing this. Still, if you need to purchase the parts anyway, better pick 680µF / 10V to be as close as possible to the original design.

A High-Quality Replacement PSU

In the quest to find better power supplies for the Transporter, I have come across this one for around 46 EUR:

https://de.farnell.com/en-DE/tracopower/tmp-15105c/converter-ac-dc-5v-3a-15w-module/dp/1672159

It's a PSU for embedded and telecommunications environments where the demand is high for reliability and output quality. 3A is a bit more than the 2.3A the PowerPax PSU is specified for, it's far more than the Transporter needs. Actually the digital circuitry of the Transporter, including the front panel and displays, rarely exceeds 800mA but having some reserve can't do any harm.

So I ordered one to give it a try and it turned out to be ideal. It is a rather flat package, and it's encapsulated which may mean less EMI emissions and noise being sent to other components in the Transporter.

As the measurements are totally different from the original PSU, the screw posts that it was mounted on need to go. They are pressed into the case from the bottom, and can be removed by hammering on top of them until they look like here (this is the top side where the PCB used to be).

To prevent damage to any other parts of the Transporter, I suggest to remove the front panel, and the back panel (along with the main PCB) so what you are working on is just the bottom plate.

The vertical scratches you see here are from production and were not caused by me. This is the normal state of things at least in the silver/aluminum finish Transporters.

The screw posts will now protrude on the bottom side like so, and can be pulled out with a pair of pliers:

End result - in the pictures the aluminum appears to be warped but actually that is not the case.

Now it is time to determine where the new PSU should go. As a measure of care, I decided to put it as far away from the main PCB as possible, just giving enough room towards the front panel that the power connection can be screwed in:

Mark the screw holes with a thin pencil:

The front panel is still missing in these photos. If you plan the PSU placement, be sure to leave enough room for the screw terminals to be operated.

The pluggable connections that were used with the old PSU need to be clipped off, and the wires should be terminated with crimped end caps. Sorry, I forgot to take pictures of this. Here is what it looks once finished:

Here you can see the clearance required near the front panel between the screw terminals and the wire endcaps and the front panel itself. Don't get too close there

Plenty of room on the primary side

I have used 20mm M3 countersunk screws, drilled 3mm holes and gave them a funnel shape with a larger metal drill. Then used plastic washers on the inside and two nuts as standoffs. Once the power supply is in place, another plastic washer and nut complete the mounting. I have put some loctite on each of the nuts (not visible in the pictures) to ensure they won't go anywhere.

Here is a very crude drawing of a cross-section of it. This is not to scale but I think it makes clear what I wanted to achieve:

Blue: 20mm M3 screw

Black hathched pattern near bottom: case plate

Red: M3 plastic washers

Green: M3 nuts

Orange hatched pattern: power supply case

Purple: Loctite

The two stacked nuts at the bottom plus the plastic washer could be replaced by a 5mm standoff, I just didn't have them at hand.

This setup provides some airflow on all sides of the PSU case to ensure some cooling. What also helps is the four holes remaining from the screw posts formerly removed.

Here is a look at it from the bottom (main region of interest is the top left corner):

The four smaller holes are from the removed screw posts. I used a drill for chamfering as some sharp edges may remain from the removal process. While these holes no longer serve a real purpose, they help cooling the new PSU with a bit of airflow.

I noticed rather late that I had come very close to the area of the front-left foot with one of the new screws. Had not considered that really, but you might do better and place the AC/DC converter a bit further towards the middle of the Transporter, or even further towards the outside so one screw is hidden beneath the foot eventually.

Terminal Polarities

You do not need to consider the polarity on the AC side, basically it does not matter. It's totally different on the secondary (DC) side though.

Here are some pictures illustrating the correct polarity for the 5V DC connection:

Negative terminals (both top pins) at the top, positive terminals are the two other pins

As the connector is only going in one way thanks to keying, you can use the wire imprint as a help. The negative wire is the one having the long white dashes printed on its insulation

Other than the Logitech PCB, the PSU has markings near its output terminals that cannot be misunderstood so you should be able to figure out which wire needs to go where.

I recommend using color to help identify the polarities. I used red and black sharpies to mark the terminals on the front panel connector as well as the crimped wire going towards it.

Putting Everything Back Together

The last steps of the repair process are:

ALWAYS ensure the mains power cord is unplugged until you are done and everything is safely connected and bolted down
do not bolt the PSU board down just yet
plug in the primary connector (from mainboard to PSU board)
put the PSU board back in and fix it with the three (or four) TX10 screws. Do not use hot glue yet!
There are two different models out there, one having four screw posts to properly fix the PSU, most only have three screws so one corner is left unsupported. Maybe Logitech considered this fourth screw too close to the mains input pins, and decided to leave it away for better isolation distance in later generations, but that is total speculation.
we can now test the output voltage of the board. If you connect the power cord now, there should not be any obvious reaction as the brain of the Transporter is not yet powered. Use a multimeter to measure the output of the front-facing connector on the PSU. It should be 5 Volts or a little more. If your meter reads something far below 5V, like 4.8V or less, or more than 5.5V, that's not a good sign. You should not connect the front panel power connector to it in this case. A new PSU might be the better option after all as there is something else wrong and it's really hard and dangerous (at least for me as a non-expert in this field) to find out what else is wrong in a switching power supply. So many things can go bad that I would not recommend to go deeper into this
if you have an electronic load that can be configured for 5V / 2.3A, then this would be an excellent time to see if the PSU can hold up to these values. With fresh capacitors, it definitely should
assuming the output voltage looks good, disconnect the Transporter from power again and connect the front panel connector. It is polarity-safe by a "key" in the plug. If you have trouble inserting it, make sure it is the right way around. Don't force the connector in violently because reverse polarity will definitely destroy your Transporter beyond repair.
In case you picked a different 5V PSU model, be sure to check polarity multiple times before you eventually connect everything!
it's another test round now. Connect power again. The boot sequence should be as usual now, you should hear the relays clicking and within 5 seconds, the Logitech/Transporter logos should appear on the front displays. You should be able to observe that the LEDs are lighting up as described further above. If possible, try to measure the PSU voltage output during this phase and see if you can still measure a constant 5V output. If this is not happening, or you notice a smell, smoke, any unusual noise or any other irregularity, disconnect the Transporter from power immediately! It's certainly a good idea to have a switchable power cord or extension that you can quickly disconnect with just the push of a button or switch, rather than having to pull the plug. Which is also particularly dangerous as you will have your fingers close to the mains fuse during that moment. I really hope you won't be in this situation because it is hard to say what went wrong and how catastrophic the damage actually is. If you followed the instructions closely up to this point, there should be no reason for this happening. Some advice if the odds are against you anyway:

are you sure you took the polarity of all capacitors into consideration? They might fail spectacularly, especially the primary filter cap getting fed in close to 400 volts, if accidentally reversed
where did the unusual activity (smoke, fire, noise) come from? If it is the PSU, using a new PSU might still save your day if nothing else was damaged. If anything else blew up, this may mean irreversible damage to your Transporter
did you notice any other unusual things about the Transporter in the past? Maybe that is an indication of damage other than just the PSU

hopefully your Transporter is now displaying stuff and connected to the source of your choice. If you use WiFi, at least one antenna may have to be connected first for successful network access. Check the front panel controls for functionality, everything should be as usual now
put the top cover back on. I suggest disconnecting the device completely again and putting the Transporter face down on your thighs, then sliding the top cover over the back plate. This gives you best control to push it down evenly.
Especially near the end, make sure the top case slides correctly over the front faceplate. It may hit the capacitor in the top center of the front panel PCB and damage it or shear it off. Use care and control while you slide the top case on.
note that the top cover can only be placed correctly one way as the screw holes are not the same distance to the edge on both ends. At the front side, the screw holes are a bit closer to the edge then on the back side. If you find that the screw holes do not line up with the threadings beneath, that's a pretty safe indication of the top cover being put on the wrong way
put the eight outside screws back in. Don't tighten too fast in order to keep the threads and the 1/16" insert intact.
Congratulations!

In my opinion, a repair that covers these aspects is a better outcome than just a new PSU. This solution will last much longer because there is no heat-related stress for the capacitors, and they are what fails most of the time. Depending on the choice of parts, they may last almost forever. Just replacing the old board with a new one will give you another ~10 years of Transporter lifetime until it fails again. The construction is still basically the same and it suffers the same design flaws. So even if you lost all confidence in your old PSU board and ordered a new one, it may still be worth while to modify the new board by exchanging the original SAMXON caps with parts of better quality and modifying the PSU to keep the caps away from the board as shown above. This will of course void any warranty you have on the new PSU but if you know what you are doing it should not be a concern. Eventually the loss is around 12 USD if the mod ends catastrophically.

Noisy DAC / Bad Analog Output

The issue described now has nothing to do with the +5V power supply for a change. If your Transporter's analog outputs are dead, or strongly muted and/or distorted, the Digital-to-Analog Converter (DAC) might not be supplied with the required voltages. The LEDs around the DAC may serve to indicate this kind of trouble.
The Transporter uses two potted toroidal transformers (the blue modules on the far left edge of the mainboard). One generates 2x 18V AC which is converted to +15V DC and -15V DC to give the DAC its reference voltages. If this transformer fails, the DAC cannot work properly. The other is probably used to power the remaining analog circuitry. None of both should ever get too hot to touch, and the top surface should be flat and not bulged.
I found in one case that out of the two primary windings of the transformer, one had gone high resistance; it was not completely open but clearly out of order. The high resistance caused the output voltage of the transformer to drop down to almost nothing.
The relays are part of the automatic voltage selection. Depending on the input voltage, one or both primary windings of each transformer are switched in series. I found that in 240V environments, only the relay that is closer to the back will engage. It is likely that the second relay is dedicated to loop in the second primary windings. The determination of the voltage and relay constellation is done on the CPU board.
What is safe to say is that either way, only half of the transformer failing causes it all to fail. The unemployed primary winding won't save the day.
A safe indication of this is when the three LEDs in the analog area of the mainboard (right side, opposite of the transformers) are not lit, just dim, or flickering (might also apply if just one or two out of three are affected). They should be reasonably bright. What helps a lot here are the two measurement points for the +15 and -15V reference voltages. Just measure these and if they are not spot-on 15V, there's something wrong in the power rail. I was a little surprised to find that the failure is at such an unusual point as the transformer. But it's actually something that makes the repair way easier than diagnosing tons of SMD components.

I have seen one of these transformers blow up due to a malfunction of the CPU board. This Transporter is probably not recoverable. I guess the transformers themselves are of stellar quality and won't fail out of nothing. It's the driving part that kills them.

Just to be sure, the following instructions apply only if the DAC reference voltages, or one of them, are actually missing. The next steps will only fix this specific issue.

The part you need is a Talema 70014K transformer module. It can be found at RS Components for example (here's a link to the German site, you can switch to your language from there):

https://de.rs-online.com/web/p/ringkern-transformatoren/2239092

The part is rather cheap, currently (2017-12-16) it's 10.73 EUR or ~12.60 USD, and it is identical to the original part. Always good to know that original parts are still made.
You can find it on the left side of the following picture:

The right (2x 9V) transformer is probably dedicated to supplying the DAC digital control voltages, or analog circuitry around the DAC. Just a guess though.

Here's the repair approach:

follow opening instructions as shown above (in the +5V PSU chapter)
the mainboard is the large green one filling the back half of the case
to get it out, disconnect the white flat flex cable that goes to the front panel PCB. Use a lot of caution with the PCB connector as it easily breaks! Try to lift up the left and right side the same amount.

The flat flex cable is fixed in its end position by the brownish top part of the connector. If it is pushed down, it will press against the flex cable, ensuring it cannot slip out. This piece needs to be lifted by about 1mm on each side. The small plastic tabs on both sides look promising but be very careful as they easily break off! This will make handling the flat flex cable miserable. I recommend using a (very) flat screwdriver, a blade, or a spudger, and push it under the tab. Attempt to pull the brown part up from as far "inside" as possible instead of using the plastic tab for it. The following picture is to illustrate, the spudger being the red outline shape, hopefully it makes things a bit clearer:

with the spudger or whatever you are using is in this position, carefully lift it with as little effort as possible. It is safer to lift it just a fraction of a millimeter on one side, then doing the same on the other side. This will take some patience but believe me, a broken connector is practically impossible to replace as there are so many different and incompatible forms around. Awesome soldering skills are also required to do this without any damage to the board. If you broke it and consider yourself adept enough to replace it, this is probably (!) a good replacement: https://www.mouser.de/productdetail/omron-electronics/xf2j-3024-11a?qs=sGAEpiMZZMs7i6cT6ogu456eCSW9zXHpB8HNXgh4rJ8%3D - please be aware that I never ordered one of these so I cannot guarantee that this is actually a good fit. It just looks similar and has similar dimensions, considering the pin count and pin pitch. I cannot share any experience about it though so don't hold me liable
once the flex cable is out of the way, you will see why this was necessary. One of the 13 TX10 screws that hold the board in place was hidden beneath. Bending the cable out of the way is not a good idea. Disconnecting it will also make the following work way easier
the plan is to get the mainboard out and leave the back plate connected to it. Otherwise you would have to unscrew a lot more which is not really necessarsy here
remove three 1/16 hex screws at the bottom edge of the back plate to free it from the bottom shell
undo all 13 screws on the top side of the mainboard
the next bit may be tricky. You may find that the mainboard and back plate are still hanging somewhere so you cannot easily lift them out. The reason is the plastic sheet isolation that has a hole which surrounds a metal bolt (one of the threadings for the screws that connect the back plate and bottom shell) between mainboard and back plate in the region of the big mains connector. Unfortunately, I have never taken any pictures of this location. Anyway, you will notice that the plastic sheet keeps you from taking the mainboard out because it is stuck down firmly along the left edge of the bottom shell. It takes some wiggling and right placement of the plastic sheet to get it free eventually. Especially older Transporter models do not have this plastic sheet.
once the board is out, the transformer we need to desolder is illustrated here - notice that the mains connector is in the bottom left corner of the picture. The transformer that this is about is the one further away from this connector:

so that's eight pins to be desoldered. The transformer will probably just drop out once you desoldered all pins because (thankfully) the pin are straight and were not bent in any direction during production of the board
if you are curious, you can now find out which transformer winding gave up. Each pair of pins is connected to one of the coils. All should have a rather low resistance (primary resistance is typically higher than on the secondary side). One is probably just open or has a considerably higher resistance
solder in the new transformer. I recommend soldering one pin and then another across. Then check that the transformer is flush on the board with no gap in between. This is corrected more easily if only two pins are fixed and may have to be re-heated to push the transformer further down. Just heat up the respective soldering point and push the transformer so it is flush with the board, then continue with the remaining soldering work
to check now whether the repair is a success, you will have to put the Transporter back together mostly because the mainboard won't power up without the front panel PCB being connected, and also the 5V power supply needs to be in service
it is too soon to reassemble it all completely so we will do the minimum amount of work now so the unit gets ready for testing
put the mainboard back into its place, taking care of the isolation sheet which must be in the same position as it was before disassembly, otherwise it will be hard or impossible to put it back in its position. Do not use force here or bolt it down if you feel there is any physical resistance
align the board to its screw posts and fix two or three of the TX10 screws so it doesn't all fall apart. One of these screws should be the position that is later covered by the flex cable. This makes final assembly easier if you find that everything is working as desired. Then, you just need to fix all the other screws and it's finished. Otherwise you may have to disconnect the flat flex another time with the risk of breaking the board-to-flex connector like before
push the flat flex cable back into its connector, ensuring that the brown plastic lock stays up and open while you insert the cable. Make sure the flat flex is in its end position and cannot be moved more deeply into the connector. Eventually push the plastic lock down on both sides simultaneously, try to avoid too much pressure on the outward-facing tabes because they are so unstable. Better apply equal pressure across the entire part. Ensure at all times that the flat flex does not move, goes straight down, and is not askew
reconnect the 5V power supply as described further above
carefully plug the mains power connector in but stay ready to unplug it in case anything does not go as expected. Keep an eye on the mains fuse and the LEDs
relays should click, and within ~5 seconds, the displays should show the boot logos
observe the three LEDs in the DAC area. They should all light up now permanently:

the fourth LED not shown in the picture (near the small mainboard stack) should stay off unless you send a command from the infrared remote control
to check it completely, connect Ethernet or one of the WiFi antennas and have it play a bit of music. The sound should be crisp and clear as usual now
put the top cover back on. I suggest disconnecting the device completely again and putting the Transporter face down on your thighs, then sliding the top cover over the back plate from above. This gives you best control to push it down evenly. Note that the top cover is not identical on both ends as the screw holes are not the same distance on each end. At the front side, the screw holes are a bit closer to the edge than on the back side. If you find that the screw holes do not line up with the threadings beneath, that's a pretty safe indication of the top cover being put the wrong way
put the eight outside screws back in
Enjoy :o)

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 a Transporter myself that 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!