I'll post some photos on Flickr when I get a chance, but thought I would share some of my travels through the inner workings of the SoundBridge Radio power supply. I know this has been a constant discussion topic, so I hope some of you find some help here.

A friend of mine brought me his dead SoundBridge radio, along with links to this forum and the PSU problems plaguing SoundBridge Radio users. I have a bit of time on my hands, so I thought it would be fun to tear the thing open and fix the PSU. Little did I know what I was in store for.

Disclaimer: You should not even attempt to try this unless you have a solid understanding of electronic circuits and all the safety precautions needed. There are dangerous amounts of power being manipulated by this PSU. Before you try to open this sucker up and play with it, ask yourself this: Do I know enough to not accidentally die from an electrical shock?

Opening the Case
This requires three things:
1 - A T20 tamper proof star driver (the one with a hole in the driver for a center pin)
2 - A paint scraper or other implement for prying through the copious amounts of glue
3 - A well ingrained sense of patience

There are four tamper proof screws that hold the back of the unit onto the main electronics housing. You can probably find the tamper-proof driver you need at a local hardware store, but you might need to order it online. The one you need is a T-20 tamper proof star driver.

Once the screws are out, the two halfs of the unit are connected by two wire harnesses that have simple clips holding them together. There is no need to open the main electronics housing (with the LCD display), so just set it aside. The back housing (triangular) is where the PSU lives. To get this open, you need to remove four recessed screws as well as 8 small surface screws, some of which may be hidden beneath rubber gasket strips. Thankfully all those screws are standard phillips heads.

Once the screws are removed, the front panel needs to be carefully pried off, as it is held in place with about three tubes of silicone glue. An exacto knife and a thin-bladed putty knife did the trick for me, although it did take about 30 minutes to complete this without breaking the plastic housing.

Once inside, you will find three sets of wires that prevent you from opening the case very far. Two of those wire bundles have chokes wrapped around them (rectangular blocks). The chokes have small clips on the side that allow you to open them, freeing the wires so you have more working room. The third set of wires runs into a small circuit board (I'm guessing a crossover circuit) and eventually over to the speaker in the top of the cabinet. There's isn't much you can do about that one unless you want to cut the wires and re-splice them later.

What you need is the PSU which is contained in a plastic box wedged into the case. It is also glued, so a bit of prying might be needed to get it loose. The PSU is encased in a thin plastic container wrapped with copper tape for shielding. Peel back the copper tape, and the plastic pieces should separate, although you will need to feed the wires through the plastic holes to get the PSU boards out of this little box so you can work on them.

Do I need to mention that you should have unplugged the radio from the wall before doing any of this? Really? Do I?

A Look at the PSU Board
Once you are able to get the PSU board free from the case, what you will find is two boards glued together with enough adhesive to assemble a small commercial jet. There will probably also be giant blobs of adhesive strewn across the components and board in a pattern that makes you think it might have been bring-your-child-to-work-day at Roku industries.

On the unit I dissected, as well as from many on these forums, the most visible problem is a blown capacitor on the board. The one you should look for is a small black electrolytic cap very near to the two huge caps that are adjacent to each other. More than likely you will find the top of that capcitor to be bulging or even burst. There is probably a bunch of blackened goo nearby as well.

It would be easy to say "replace that cap with a fresh 220uF 35V aluminum electrolytic capacitor", but there's a bit more trick to it than that. To be able to do so, you first need to do some more prying.

Separating the two PSU circuit boards
The main board with the PSU components has a secondary PCB underneath it, which appears to do nothing more than provide a ground plane and noise shield. To reach the underside of the component board, you need to pry these two boards apart. Start by removing the several solder connections that penetrate through to the lower board. Once those are desoldered, you will need about thirty minutes with an exacto knife and several thousand words of profanity to pry the two boards apart without snapping one of them. It was also interesting to find the two MOV Varistors tied to the ground plane in a way that tells me they weren't part of the original design. (What? The UL Inspector says we need over-voltage protection?)

Replacing the Capacitor
Unsolder the old one. Solder in a new one. Put it all back together and call it done.

Except It's Not Always That Easy
So during my dissection, replacing the failed capacitor did bring the unit back to life . . . . . . . For about 15 minutes until it failed again. Think opening that case and then closing it all back up is fun? Wait until you get to do it twice.

On the second failure, I found a zener diode (CORRECTION: Schottky Diode SB260, NOT Zener) that had failed, located right next to the capacitor I just replaced. The solder job was strange on this diode, and I soon found out why. The leads on the diode didn't fit through the holes on the board. Apparently whoever designed the board forgot to check the lead diameter and laid out drill holes that were too small. So the assembler had simply snipped off the leads, laid the diode on top of the board and then looped some small bits of wire through the holes to connect it to the pads on the bottom of the board. Nice job, guys. High quality. Nobody at the assembly plant owns a drill? Really?

Before I started replacing more components, I decided to learn more about how the PSU was built. I ran several of the part numbers and found that the basis of this PSU design is the TOP247 switcher. In fact when I pulled the datasheet for that part, one of their example circuits is pretty much part for part the PSU that powers this radio. It's not a bad design at all, so don't blame the circuit designers from Roku. The problem seems to be purely in low quality components used in assembly, as well as some crazy assembly practices.

The problem with this PSU design is that all the pieces play together very tightly to ensure the correct output. So when one component fails, it impacts voltage and current all over the board. There is a reasonably complex feedback circuit to control the switcher, that works through a set of sense resistors and an opto-isolator back to the TOP247. If a component fails, the feedback profile changes, thus causing generalized weirdness all over the board. In the unit I was working with, the failed diode caused some unsafe current loads through the feedback circuit and fried the opto-isolator as well as several of the resistors in the circuit. It also seems to have damaged the voltage reference IC on the board, which pretty much borks the whole thing.

The PSU should have two output voltages (19VDC and 9.5VDC), with the damaged feedback circuit, I am getting 22VDC on the 9.5V rail, and a constantly changing voltage reading on the 19V rail. I can only pray that the Roku people put some regulators inside the main electronics enclosure, otherwise those voltages could have fried stuff in there as well.

In the end I have decided to just replace most of the post-transformer components with new reliable parts. I'm building a list of components available through Mouser Electronics for anyone else who wants to give this a try. Nothing pre-transformer appears to be borked, but I might replace that giant mongo 100uF 400V cap just for good measure.

I'll work on getting some pictures together of the operation, as well as a parts list of replacement components and post those to the forum later this week.

Oh and one last thing. The datasheet for the TOP247 highly recommends the use of an NTC Thermistor to limit inrush current charging the capacitors at first power-on. There doesn't appear to be one of these on the PSU board for the SoundBridge Radio. There is a component that could possibly be an NTC Thermistor, but looks more like a fuse pot. (UPDATE: It's a fuse, no inrush protection on the board). Without that Thermistor, the inrush current is what is probably blowing people's power supplies like fireworks on bonfire night. I make that guess since so many postings state their radio died after they disconnected it from power and then reconnected it.

I'm gonna see if I can determine if that mystery component is actually a thermistor (UPDATE: It's not). If not, I'll try to see if I can hack a way to insert one into the board and provide that protection.

Hey spiffy_chimp,

Thanks for this very informative write up. I hope members of this forum find it useful.

Jeff

A couple of corrections to be made, thanks to PM from another member.

1 - The diode I mentioned is not actually a zener diode, but is in fact a schottky barrier diode of the SB260 variety. Anyone thinking of replacing that diode should be aware of this. 2A forward current, 60V reverse voltage.

2 - The component I thought might either be an NTC thermistor or a fuse, is actually a fuse. That means there isn't any inrush protection, which could be a cause of some of the PSU failures the community has experienced.

Right now, I'm just waiting on a parts order from Mouser to handle the replacements. I'll post again once the parts arrive to let everyone know how it is going.

I'm cautiously hopeful that a complete fix will be achieved.

spiffy_chimp,

Nice job! You also get extra credit points for getting that sucker open too.

This is just my opinion, so take it with a grain of salt...
If someone needs that much info, the probably don't have the skill set to be working on the unit in the first place. I have repaired three R1000's already and know what a pain they can be (as you pointed out) I also like the "three tubes of silicone" joke...true, true. Speaking of which, I don't think it's silicone since it still stays gooey after the normal cure time for silicone (at least the ones I worked on where still gooey (like string cheese)).

Some things I want to point out from my tinkering...

On the 8 small screws, two are longer then the other 6 (of 8) well at least on two of the three units I fixed)). These two go below the woofer. Since devices are reved, not all untis will have the two longer screws I assume.

Take care to seal the flower out of it when you are done; a lot of work and design (from what I recall RokuMike saying) went into this system and the woofer needs to have this (including the nice fancy stuffing that looks like it’s from a pillow; this is acoustic speaker stuffing the makes the woofer think it’s in a bigger enclosure as well as absorb reflection waves) and you also want to keep stuff from rattling around. So speaking of which; make sure you glue the chokes back in place too. I am using silicone untill I find something that is like the OEM sealant.

Take care when you pry apart the bottom board (which I assume is part of a Faraday Cage to cut down on “noiseâ€

Matt,

Good points there all around.

To anyone who suddenly feels empowered to open up their radio and start mucking around inside:

The power supply discussed carries full voltage from the mains in your house. IT CAN HURT YOU VERY BADLY AND POSSIBLY KILL YOU. And as Matt points out, placing components incorrectly can cause some rather spectacular displays of electrical carnage. If you aren't entirely sure you know your way around an electrical circuit along with all the necessary safety precautions, don't try this at home. If you have even the slightest doubt about performing a repair on this power supply, either contact Roku about a replacement or take it to a qualified repair technician. Seriously, this is not a hobbyist repair job.

That being said, I'm glad to hear I'm not the only one silly enough to crack this sucker open and have a go at a repair. It's too nice of a unit to just let it die because of a few cheap components.

Given that RokuLabs have a "Director of Hardware Engineering" and that the R1000 radio has a proven history of known PSU failures, is it not appropriate for a Technical Repair Bulletin, complete with drawings & components listing, to be made available to those PROFESSIONALS who have the necessary skills/training ???

The problem appears to be that Roku simply have shown that they have no interest in getting an acceptable worldwide solution (at low cost to the end-user). If the law was that ALL shipping-costs were at their expense (re faulty products) things would be very different. The ongoing situation is nothing short of a disgrace.

To some of the others who have torn one of these apart and brought it back to life, I'm having a bit of trouble with a few of the resistor values. Several of the resistors on my PSU board for the SB1000 are possibly damaged due to a component failure. A few of them are scorched to a point where I can't read the color codes accurately, and obviously I no longer trust readings from my multi-meter if the component was scorched. Some of the variations are obviously component tolerance, but others have me worried.

Anyone have an SB1000 open, or have documented the values that can help me out?

Below are the resistors in question. where possible I've listed the color code value, as well as a real reading from a multi-meter. Can anyone confirm, deny or correct any of these values before I start?

R905 - Color code: 10k, MM reading: 975 (way too far off to be tolerance)
R906 - Color code: 610, MM reading: 615 (within tolerance)
R907 - Color code: 1M, MM reading: 975k (within tolerance)
R908 - Color code: 470, MM reading: 4.7k (color code is partially scorched)
R910 - Color code: 32.4k, MM reading: 32.35k (within tolerance)

Thanks in advance if someone has those values to share.

Hi

R905 - Color code: 10k, MM reading: 975 (way too far off to be tolerance)
R906 - Color code: 610, MM reading: 615 (within tolerance)
R907 - Color code: 1M, MM reading: 975k (within tolerance)
R908 - Color code: 470, MM reading: 4.7k (color code is partially scorched)
R910 - Color code: 32.4k, MM reading: 32.35k (within tolerance)

In the units I have repaired the values are as follows

R905 is 1K
R906 is 620R
R907 is 1M
R908 is 4K7
R910 is 32.4K

I think you find that the resistors themselves are not burnt.
What has happened is that the yellow glue securing them to the board has gone black with heat, giving the appearance of scorch marks.

This was mentioned in the other long thread about the 75R resistor which another forum member had a query on.

Pete

Thanks for the values Pete. You are correct that five of the six resistors in the feedback circuit are simply discolored and not fried. One of them however, actually fell into bits when I removed it from the board.

So my plan had been to replace most everything post-transformer on the board. Not necessarily because all of the components were toasted, but because of some anal-retentive need for order and a love of shiny new parts. (it's a character flaw I am told)

Anyway, a status of how things are going: The FedEx man delivered my Mouser order on Thursday afternoon, so I replaced the caps, inductors, voltage reference IC, diodes and opto-isolator that evening. With the resistor values posted, I replaced those on Friday. I powered on the board and almost nothing had changed. The 9.5V rail was measuring at 22.5VDC while the 19V rail was fluctuating between 0-34V rapidly.

Starting to get annoyed, I pulled everything post transformer off the board and built the circuit on a breadboard, connecting it back to the transformer outputs on the original board. Now I have stable voltages, although far from the expected values. The 9.5V rail is measuring at 4.4V while the 19V rail hovers around 7.9V. Interestingly, the output voltages are cycling every second or so. I scoped the outputs and they spike once a second then decay (like a capacitor drain). The spike each second is accompanied by an audible click from the original PSU board.

So what have I learned? Well first of all, I need to find where I created a solder thread or short of some kind on the PSU board. Secondly, I need to find what has failed pre-transformer (or god forbid is miswired on my breadboard circuit). I'm taking a wild guess that the spike-and-drain output profile I am seeing indicates the switcher itself may be failing, possibly due to my own dirty solder job that left a short somewhere. The big horking load cap seems to be powering the circuit, but only being recharged every second or so. I'll try putting a dummy load against the outputs to see if it drains faster under load.

I'm near to considering just chucking the old board, etching a new one and rebuilding the entire thing with brand new components. I know, it's extreme, but I have this intolerance for being defeated in a repair job.

Hi spiffy_chimp,

Your bread-boarded circuit not working may simply be due to the fact that it's bread-boarded. Switching supplies can be very layout sensitive due to the high current spikes involved. However, it is also possible that there is another fault or wiring error. Were you measuring the outputs with or without any loading? Some power supply designs require a minimum loading before they will operate correctly.

Is the amplifier part using the 19V?
What about taking a 19V notebook power supply (here in Germany at ebay 13 EUROs) and use a 7809 to create the 9,5V?

That is certainly one method which has been suggested.

A small 9V regulator inside, and feed the rest from an old laptop supply.
Most use an 18 - 19V output.

Just make sure a decent heatsink is attached to the regulator to keep it nice and cool.

I bought one of the first R1000's in Feb '06 when I first read of them. Radio never came and I think I talked to someone who told me they weren't satisfied with the quality of their first products received. that they wouldn't ship until they were, and my credit card would not be charged until the unit was shipped.

It arrived in June '06 and worked until August this year. I bought one for my daughter and her husband in Sept '06, and that one failed soon after in warranty and was replaced and is still working - knock on wood!

I had already taken the PSU apart and renewed my interest through the forum. I saw the affected capacitor referred to by mr.jr and in your flickr photos. The top of mine is expanded like a can of tomato soup with botulism and the dielectric melted and burned on the outside.

Thanks very much for the photos and detailed insight into the PSU - gives me courage to take on the rest of the job. Unfortunately, the retail parts houses are not as accessible as in the past - hope I can easily find replacements.

Hello Friends,

I am writing from Germany, found this thread when google for roku SB radio repair hints. You guys really did a great job. And shame on the responsible manager at Roku for this crap quality assembly of PSU. I wonder why this device got certified with the CE label. Isn't there any fire hazard regarding the blown caps, overheated diode, and the sound proofing damping material? What was the root cause for the failure? Did the defect cap blow up the diode? Did the soldering/contact/via issue with the diode blow up the cap?

Nevertheless: Would it be an option to remove the built-in PSU completely, and provide an external PSU with matching voltages and currents? Possibly a notebook PSU?

Another issue: This device has a stand-by power consumption of 9W, which is almost the same when in operating mode. Is this high stand-by caused by the PSU itself? Is there any control wire from the display unit back to the PSU for standby, or is the PSU always running at full power?

Thanks!

I am now closer to the source of failure:

I measured the supply currents of 9.5 and 19V via an external high-end lab power supply.

9.5V: after booting the radio, measured 0.54A, regardless of turned on or in standby. This supply voltage is for the display unit's microcomputer.

19V: about 0.15A when turned on, zero in standby. This supply voltage is for the audio amplifier

The built-in PSU (Shenzhen Tailing Technology Co. LTD) is specified as:

Model TL203D-A9
Input: 100-240V, 70VA, 1.3A max
Output 19V, 2.3A (I think this is over-engineered for the built-in audio amp and speakers)
Output 9.5V, 0.37A

The difference from 0.54A to 0.37A corresponds to a permanent overload of almost 50%. This is definitely a system design bug, not a manufacturing issue.

Next I will open the display unit to check why it consumes constantly more than 0.5A, even if in standby. This bothers me the most.