Fixing Bose Quietcomfort 35

Short version:
Bose Quietcomfort won't turn on - open it and check if a plastic piece sitting on the switch is intact.
Bose Quietcomfort won't enter pairing mode - it's possible that the switch on the PCB  can't reach  position responsible for entering pairing mode. Open the headphone and check if moving the switch directly helps, if so, continue reading.


Headphone won't turn on when the switch is moved.
The problem was this small plastic part which is visible from outside as a switch and which moves the actual switch soldered on the PCB. To fix it you will have to open the right headphone. Start with gently removing the ear cushion and the fabric inside, then undo three screws:

Without screws aluminium cover can be removed. The actual switch is highlighted:

From the other side of the aluminium cover you see a plastic part which should sit firmly on the switch. In this case it is broken:

Find two pieces of the broken part and glue it together, I used two compound epoxy glue.

It's not pretty but it holds well.

Bose Quietcomfort won't enter pairing mode.

The mechanical tolerances of the switch, this plastic piece and aluminium cover are so tight, that it can happen that the headphone will not enter pairing mode. You move the switch but it neither enters pairing mode nor removes paired devices. The reason is, that the oval hole in the aluminium cover is to small and the switch will not contact in the "bluetooth" position. I extended this hole. It was around 8,9mm long but I extended it by approximately 0,75mm towards the bluetooth logo. I used a round file for this:

It is barely visible that the hole is longer:

But this bigger hole allows reaching the final position of the switch.

Fixing VOX AD60 VT guitar combo

The VOX AD60 VT blown the main fuse when turned on. I received it with a comment that someone has already tried repairing it.
After opening the housing it was obvious that someone has already played with transistors in amplifier stage.

For test purpose I used two power supplies with current limitation to supply the power stage. I connected them to points marked with yellow boxes on the circuit diagram below. After turning power supplies on, I observed that without input signal both power transistor (Q13, Q16 - blue ovals) are turned on completely. My power supplies entered constant current mode. I removed both power transistors (Q13, Q16 - blue ovals) and turned supply again, this time I observed that R87 (green box) overheated. I measured voltages across R84 and R79 (blue boxes) and realized, that Q12 and Q15 (green ovals) are fully turned on, while they shouldn't. While inspecting pairs of transistors Q5-Q6 and Q7-Q8 I noticed that in each pair there are different types of transistor, while they should be actually the same. I swapped Q6 and Q8 as they should be according to the circuit diagram and after that the amplifier (still without power transistors) draw approx. 20mA. I soldered power transistors back and tested the amplifier. It works well. 

The problem was, someone was trying to fix this amplifier or to test transistors but mixed them up while putting them back in the circuit. This caused a shoot through current, the power transistors survived that but the fuse tripped. 

On the very beginning, when I observed that there is high current flowing from both positive and negative supply rail, I thought it is something with phase inverter, and not with upper or lower power stage. I did once mixed up transistor while fixing a power stage and after that I can say, I learned my lesson and this time was able to find the problem quite fast.

EMC Troubleshooting speaker relay matrix

There is relay a matrix designed to switch between active and passive speakers in a music shop. There are actually two matrices, each in a different showroom. It has been observed that one matrix switches speakers almost inaudible while the other makes a quiet but annoying click. Here is a screenshot of a signal recorded at the output of the speaker relay matrix. This is matrix switching line signal for active speakers so it was possible to record it directly:

Recorded distortion on the output of the matrix, at the beginning and ant the end there are disturbances which I believe are 4 bytes digital transmissions coupled on the audio signal.

Because of the location and surrounding equipment only one rack with a matrix can be opened, namely the one which generates audible disturbances, it looks like this:

Wiring arrangement in the matrix

The matrix consists of daisy-chained modules, each module is switching one input signal to 1-of-8 output jack sockets. A single module can be simplified to the following:

Single module of the relay matrix

In reality a single module is more complicated (I assume GND is switched too) but it describes the idea. A complete matrix is built in two rows and wired like this:

Current wiring of 10 modules in the matrix (outputs not shown because they are shielded on the other side of the panel)

See how many times digital signal cable is crossing and running parallel to analog signal wiring. I believe there is magnetic coupling between digital and analog lines. To make it worse, both are wired without shield, such arrangement is actually prone to cross-talks and requires longer cables. I suggested changing arrangement to something like this:

Improved wiring of 10 modules in the matrix

In the improved wiring concept there are shorter cables, and there is no unnecessary zigzag. I believe this will remove unwanted disturbances on the switched analog signal. Rearranging is possible because every module has its own configurable address, I am not even sure what exact type of the digital transmission is used. Maybe daisy-chain order can be changed even without readdressing modules. I will update this post when it's done.

Fixing Dell 2400MP projector with power supply PA-4331-1D-LF

Power, temp and lamp control lights are continuously on when projector is turned supplied. Nothing changes after pressing power button. In the user manual there is no description of such a combination of control lights.

Opening the projector:

There are silk screen descriptions on the PCB so it was easy to identify and measure voltages generated by the power supply PCB, the 5V net looked like this:

Which is mean that the power supply is not able to regulate voltage. It looks like a single cycle of the flyback is causing overvoltage and it leads to power supply shutdown and another try in couple of milliseconds when voltage drops to 4V. Visual inspection of the C206 5V output filtering capacitor reveals that the capacitor is swollen.

This capacitor has been replaced with a new low ESR capacitor but the power supply didn't start either. This particular power supply turns out to be very sensitive to correct filtering capacitors selection. After replacing the old capacitor with 3 electrolytic hybrid ultra low ESR capacitors the projector started working normally:

The C602 capacitor value is not critical (as long as it is at least 3300uF) but low ESR is.

I suppose that high temperature in the projector accelerated electrolytic capacitor's wear-out. Most of devices I repair is supplied by a flyback power supply and most of flyback power supplies suffer from issues with output filtering capacitor which usually works in high temperature, high current and high frequency.

Seeing problems with electrolytic capacitors another time I decided to buy a meter UT612 RLC meter:

I like price-quality ratio of UNI-T devices but looking at specs reveals, that the higher frequency you use for a measurement the lower capacitor values can be tested. In case of the projector with a capacitor of 3300uF I was able to measure ESR with highest  frequency of 120Hz. For a flyback power supply working with tens of kHz this measurement should be performed with at least 10kHz to reflect real conditions capacitor has to deal with.

Fixing tube amplifier Suprem Concort 80

The amplifier came with a note that it doesn't turn on. Control lamps remain dark after flipping the switch. The reason for that was damaged switch. It made a click noise but there was no connection. Establishing reliable connection requires force to applied on the lever in side direction, then the control lamps lit but there was no signal coming through the amplifier. After flipping the standby (Vollbetrieb) switch, one could hear that the power stage was working because it wasn't complete silence, it was typical quiet noise of a working tube amplifier.
For this amplifier there is no circuit diagram available, but there is one very similar (Superem Luxor 40 https://elektrotanya.com/suprem_luxor_40_sch.pdf/download.html ) which except for number of inputs and tubes reference designation is the same. I checked heater resistance of all tubes but they all appeared fine.
Keeping in mind that tube amplifiers require connected speaker I measured all supply voltages including bias and they were correct.
I observed that one of the ECC83 tubes look different, see the middle tube on the picture:

See milky deposit in the top part of the tube which lost vacuum, it should be uniform, shiny surface.

In a normal tube there is shiny surface on the inner part of the glass, above a getter. (see https://en.wikipedia.org/wiki/Getter ), the middle tube on the picture has no shiny surface but a milky deposit. This is a sign that the tube has lost vacuum, and that particular tube drives the end stage of the amplifier. This explains why the amplifier was mute. I have no vacuum tubes in my workshop so I shuffled other tubes in the way that there was no signal for vibrato but all input channels should work. And it works. It sounds pleasant with a bass guitar. I ordered replacement tube and will build it back together soon.

Fixing Allen & Heath Zed 420 mixer

The mixer was turned on 24 hours a day for around two years. After that time once its LEDs were not as bright as in the past and the signal was not getting through the analog channels. The mixer was returned from the service with a note, that power supply was damaged and that a price tag for the power supply makes it unprofitable to repair. After that, the owner brought the mixer to me. I observed that output filtering capacitors, right behind rectifying diodes are swollen. It was clear indicator that their lifespan was exceeded. See swollen capacitors, especially the one marked as (2):

 I replaced those with identical capacity and voltage, low ESR capacitors and it worked again.

Fixing acoustic research Phantom Sub 12D

In the subwoofer there is one externally available fuse and two built in fuses which protect symmetrical low voltage supply for the analog path and the amplifier.
The problem was that internal fuses blew immediately after turning power supply on.

This is a D-Class amplifier and there are some descriptions on the web that in this case output filtering capacitors C16 and C17 should be replaced what the owner already did but fuses kept on blowing. I received the Amplifier without speaker housing and took the amplifier PCB out:

Type code of key components on the PCB are removed but it turns out that this PCB is exactly the same as the one used in JBL SCS150SI, SCS160SI and SCS180.6S subwoofers.
The circuit diagram and service manual are available online.

With the circuit diagram available I noticed, that the owner replaced C16 and C17 as they were damaged (you can see corrosion on the amplifier PCB resulting from leaking capacitor's electrolyte) but there were more components requiring replacement, in this case it was Q10, Q11 and Q5. I assume that output filtering capacitor C16, failed to short circuit because of electrolyte leakage what caused Q10 and Q11 to work with a short circuit as a load, this damaged that MOSFETs. Q11 was failed open and D10 failed as a short circuit between gate and drain what caused Q5 to fail. After replacing all three mentioned transistors, the amplifier started to work.

Fixing Ebike battery, resetting OZ890

My Ebike's range dropped significantly. I took the battery apart and and found 7 pouch Lithium-Polymer cells and a BMS printed circuit board based on the OZ890. Every cell consists actually of two smaller cells connected in parallel so the battery architecture can be described as 7S2P.

Firstly I noticed that there is one cell with unusually fast voltage drop, when the cell voltage dropped to 3,2V the BMS turned the complete battery of to prevent damage. Removing kapton foil revealed a crack on the battery pouch:

Replacing complete battery is not an option, 6 of 7 cells are still fine. I wasn't able to find exactly the same replacement cell so I decided to cut the corners and built a cell combination which will fit into the place of the damaged one. 6 pieces of 18605 cell fit perfectly and if they are connected 2S3P then their voltage matches original li-po cells and their capacity is even higher than rest of the cells. I ordered 6 new 18605 cells already welded in the desired shape and put it instead of the damaged cell.

My strategy was to fix the battery by replacing damaged component and I believe that as long the new cell has got the right voltage and at least the same capacity (ideally a bit higher) then it has to work.

OZ890 BMS had to be reset to accept the battery. In this case one of the middle cells were replaced, it means, during the complete operation OZ890 was powered up continuously and it detected 0V on one of the cell which was replaced. I wasn't able to reset the OZ890 with original o2micro USB I2C interface, but disconnecting the BMS from the batterypack completely did the trick. After reset the BMS allowed normal charge and discharge procedure. This means, that there is no Coulomb counting in the BMS and that there is voltage supervision only. I don't expect this trick to work with more sophisticated BMSes. Fixing this battery wasn't all that difficult. It has been couple of month since I did it and the battery still serves its purpose. 

Fixing Rocktile GB-15 bass guitar combo

I bought a damaged super-cheap bass combo for practicing at home. Description provided by the previous owner mentioned that signal overdrives and that it sounds unpleasant.

After turning this on for the first time I noticed that:

• there is unwanted overdrive/distortion
• distortion changes over time
• after couple of second there is no sound

As always I checked supply voltages first - they were fine. Power stage built around TDA2030 worked too so I concentrated on the preamplifier. Replacing opamp didn't help so I had to understand how the circuit works. There was no circuit diagram available so I reverse-engineered it from the PCB:

After connecting a signal generator to the amplifier I noticed that it operates normally for a while after touching pin number 2 or 3 of the 4558 operational amplifier. Using oscilloscope I observed that voltage on the pin number 2 of the 4558 is slowly rising and when it saturates the circuit doesn't amplify anymore:

I knew that this has got to do with an input bias current of the operational amplifier. In one of my favorites books "The Art of Electronics" I read that input bias current problems become visible some time after turning the circuit on, when impacted capacitors charge, for low input bias current amplifier it can be even minutes after turning a circuit on.

According to circuit diagram every input pin of the operational amplifier has got a DC path designed for input bias current. I measured R7 resistance on its pins and it appeared to be fine but by measuring resistance between pin 3 of the 4558 operational amplifier socket I noticed that the R7 is not there. I inspected the PCB and discovered that there is a crack on the PCB between the R7 and GND. On the movie below you will see (sound is important for that movie).

After fixing that PCB crack with a drop of soldering tin the amplifier started working

One more note here is the fact, that the input jack socket shorts input to GND when there is no plug inserted. It is so to avoid noise when there is no guitar connected. But to connect a signal generator to the PCB you have to be aware about that and to plug something in to open the input socket.

Fixing DENON DCD-510AE CD player

While visiting friends I heard their complaints that they have a CD player which keeps on resetting. It wasn't able to eject the CD (there was one inside). On the display it was continuously saying "READING". From time to time this player was able to start playback but it ceased when pressed "next". I was told that the CD module was already replaced twice with no result. My first thought was that it has got to do something with supply voltage. Circuit diagram is easy to find on https://www.manualslib.com/manual/1115299/Denon-Dcd-510ae.html#manual .
I started by checking 5V and 8V net, see sample trace below. I also observed (you have to believe me because I have no waveforms for that) that 8V was dropping to lowest values when a skip track or eject button was pressed. Spinning the CD loaded the 8V and 5V net significantly too.

Blue trace i 8V and yellow is 5V.

5V and 8V are both generated by linear regulators and it is obvious that 8V is overloaded so much that 5V falls below reset threshold. I inspected C914, IC93 and C916, they appear fine. I measured current consumption on 8V (approx. 1A max) and supply transformer voltage. According to my crude calculations C914 value is way too low, and I believe that the power supply is calculated without margin. I don't have 22000uF capacitor and I didn't want to order one but I realized that I have a 230V to 5V 1A switching mode power supply and keeping in mind that the main load on the 8V net is the 5V regulator I decided to add an additional power supply to provide 5V from another supply.
I removed IC94 and connected additional power supply (see yellow rectangle) parallel to C917. This new power supply is connected in parallel to the primary winding of the main transformer of the CD player T901

It's not ideal, I could try experimenting with the original power supply and low ESR high capacity alternatives for the C914 but I had an extra switching mode power supply lying around so I used it. It works just fine.