Nut clean

While I was removing the condensate trap from the gas central heating boiler, to check for any blockages as part of an hopeful attempt to get the boiler working again, I removed a screw that didn't need to be removed. It was in a tight spot underneath the boiler and the pipework meant that I couldn't easily get my fingers to it. During extraction this wasn't a problem, but putting it back again was a different matter.

Of course it fell off the end of the screwdriver and straight down one of the holes cut for the boiler piping. I could hear it pinging off the pipework on the way down. It seemed to have gone all the way to the floor behind the fitted units in the utility room. So I had to get down and lever off the kickboard, then poke around the detritus underneath looking for T25x5mm BZP machine screw. I couldn't find it, but came up with a brass 22 mm compression nut instead. In the end I found that it had gone down, then sideways, coming to rest in the cupboard underneath. To get it back into its' socket on the boiler I had to resort to taping it to the end of the screwdriver so it didn't fall off again before the threads had engaged with the socket.

That done I turned to the newly uncovered nut; probably it had been dropped by a plumber at some point and wasn't worth the effort for him to retrieve. It had been used, as it was caked in jointing compound and was well tarnished and quite dirty. Gemini suggested that I started by soaking it in washing up water to soften the jointing compound and the attached dirt, which I did for an hour or so. Next I applied an engineer's pick and removed most of the compound and gave the whole thing a scrub with an old toothbrush which lifted some of the dirt and more of the compound (probably a brass wire brush would have done a better job but I didn't have one to hand). Then I gave it a good covering of WD-40 and left it for 15 mins and applied the pick again, which removed all of the compound on the external surfaces and on the bevel that makes the seal with the olive and more of the stuff caught in the threads.

After that I soaked the nut for an hour in some old pickle juice (vinegar) with one teaspoon of salt to remove the tarnishing, then rinsed it off under the tap and give it another brush. Most of the jointing compound was gone but I had to sit down with the pick and give it close attention to make sure I had removed any visible remaining compound. Finally I soaked it in dishwater again for another hour and left it to dry.

At this point it should have been nice and shiny, and the threads and bevel were, but the outside, while a lot better, wasn't fully clean. It was, however, good enuf. Next time a brass brush and either some fresh vinegar or a slightly more aggressive acidic cleaner might work better/quicker.

Time taken: Half an hour, hands-on.
Cost: 1 mL WD-40, £0.02; 5 g table salt, £0.02; waste pickle juice, free; received a 22 mm brass compression nut, £2.00. Total cost, £1.96 credit.

P.S. Yes, I know I should have provided pictures, I just forgot!

Gas central heating boiler woes

Fresh from the heights of my success in replacing of one of the motorised zone valves on the central heating system, the gas boiler decided nuh-uh 😒 and promptly ceased functioning within a day. Buggrit! 😖

The boiler in question is a Vaillant ecoTEC plus 624 and was giving a fault code of F.29: Failure during operation: Re-ignition unsuccessful. According to the Installation & Maintenance Manual this can be due to:-

1. Gas supply temporarily stopped
2. Flue gas recirculation
3. Condensate duct blocked
4. Defective earthing of the boiler
5. Ignition transformer has spark failure

All very complex. However, as a first diagnostic action I reset the system by pressing the flame button three times (Manual section 13.2) to see if it was just a one off error. Unfortunately it was not. As with the zone valve replacement, I will provide no pictures to protect the incautious.

Next for some basic observations; the system water pressure was OK; I could hear both the water pump and the blower for the burner running; and I could hear the gas valve opening and closing; but could not hear any click of the igniter firing.

Looking at the fault list, there shouldn't be a reason for the appliance to suddenly have an earth fail (#4 - unless it is an internal fault). I had wired the new zone valve into the earth but that shouldn't have affecting anything; I checked anyway and it was fine. Four of the possible causes (#s 1, 2, 4 & 5) involve looking at the actual gas flow system and that is strictly beyond my pay grade (and illegal to mess around with to boot), so I could not rule them out. But I could not see any blockage in the flue gas pipe externally (#2) and I could confirm that there should be no reason for a sudden interruption in the gas supply (#1) as the gas hob in the house was working normally. The one possible cause that I could investigate myself was a blocked condensate duct (#3).

Modern condensing boilers recover heat from the exhaust of the burner by condensing water vapour out of the flue gas (CH4 + 2O2 → 2H2O + CO2 - North Sea gas plus oxygen from the air burns to give water and carbon dioxide and heat) thus giving up the latent heat of condensation. The water produced by the boiler is channelled from the burner into a condensate trap where it collects until there is enough to discharge via a plastic pipe to an external drain (in our case). Thereby avoiding the production of a slow but steady stream of water that can freeze solid in winter and block the pipe. The drain can also be blocked externally or internally by detritus from the boiler as it slowly oxidises the burner (common with aluminium burners, less so with stainless steel) and the resulting particulates are carried away in the condensate stream.

The condensate trap was accessible from underneath the boiler and was attached bayonet-style. When it was removed, there was a bit of detritus in the trap but not very much at all. There could still have been a blockage further up the system, so I took the case off the boiler and unhooked the condensate drain and confirmed that it was clear. The plastic drain pipe was not holding up any water but to check that the drain was fully open I poured some tap water down the drain to confirm that it disappeared smartly. So that ruled out a blocked condensate duct.

Just in case it had been the boiler condensate drain or the earthing of the zone valve. I restarted the boiler and it worked, once, then started failing again. So it appears to have an intermittent but persistent fault. At this point I was running out of options, so I checked some of the diagnostic codes for the igniter system:

Dx code	Description	Value
D.061	Number of ignition device cut offs	4
D.064	Average ignition time	1.7s
D.065	Maximum ignition time	4.1s
D.068	Unsuccessful ignitions at 1st attempt	7
D.069	Unsuccessful ignitions at 2nd attempt	5

According to Gemini,

An average ignition time of 1.7 seconds is normal, but a max ignition time of 4.1 seconds is unusually long (typically, a healthy ecoTEC ignites within 1–2 seconds). Coupled with high retry counts before locking out, this points directly to an internal combustion, sensing, or gas-regulation issue.

It pointed to one of the following, igniter, burner cleanliness, gas valve or spark delivery via the main PCB as the most likely causes. So it was definitely time to call in a more competent man.

The Gas Safe Engineer arrived the next day, confirmed the problem and started with a service replacing seals, cleaning everything as necessary and replacing the pressure sensor. That first step did not solve the problem. A further investigation discovered that the other zone valve on the system was faulty (stuck microswitch) so that was replaced, still no fix. The next step was to replace the gas valve that may have been overworked by the always-on zone valve.

So a couple of days later the engineer was back to replace the gas valve to the boiler and after fitting and a thorough test the boiler was up and running again.

For almost a week…

Time taken: Seven hours (at £75/hr ex VAT)
Cost: Service kit and pressure sensor, £46.74; zone valve head , £54.99; gas valve, £257.80. Parts & labour all in, £881.43.

Replacing a leaking central heating valve

Well, winter arrives and a sensible man's mind turns to the central heating system (apologies to Alfred, Lord Tennyson). As local temperatures were dropping it was time to fire up the boiler and restart the central heating (CH). It had been working properly before it was put to bed in the spring so there should have been no problems.

And getting it started was, indeed, no problem. I just repressurised the system and switched the boiler back on. What was a problem was the dripping noise coming from behind me. The zone valve controlling the hot water loop was steadily leaking from the head unit. Buggrit.

At least diagnosing the problem wasn't going to be too difficult. The CH was switched back off and a bucket placed under the offending valve, while the system water pressure bled back out via the valve. As, given the temperatures, this was a bit of a time pressured situation, I didn't take any pictures so what follows is just a quick report. Also, whilst I am comfortable working on a 240 V system, I don't want to provide too much detail just in case (ie don't do it unless you know what you are doing).

Generally, the head units of motorised valves cause the most problems and that is why they are replaceable separately from the valve bodies. However in this case it was obviously a problem with the valve body itself. But as it had leaked through the head unit I decided to replace the entire assembly.

The first job was to clear a space and get access to the valve, which took longer than it should have. Too much crap dumped in the utility room (mostly by me though, so hey-ho). Next was to isolate the CH system so that, A) it can't turn itself back on while the system is dry and B) there was no power to the junction box where the valve electrics were hooked in.

Once that was done it was time to drain the CH system. There is a stopcock just underneath the boiler that is below the level of the valve, so I opened it and let it run for 10 mins. Then went upstairs and opened all the bleed valves on the radiators and had lunch while the system drained out.

With the system emptied the next task was to remove the duff valve. First, I disconnected the wiring after A) confirming with a multimeter that everything in the box was dead and B) taking a picture so that I could re-wire to the same pattern (the wiring layout is generally common between these valves). With that done I was able to remove the head unit and then out came the spanners and the valve body was removed, leaving a gap for the new valve.

The replacement valve is a Drayton 27100 Motorised Valve (ZA5/679-2); as I had no time to look into what was the better option I just went down to the nearest plumber's trade counter and asked them for a replacement. Fortunately the spacing between the inlet and outlet connections for these valves is also pretty standard. So, after checking the olives on the 22 mm pipes were OK and dabbing on some new jointing compound, the new valve body was fitted into the gap whilst taking care to hold the body and not twist the existing pipework when I tightened up the connections. The new valve head was then attached, manually set to Open and was wired back into the junction box.

The final job was to cross my fingers and refill the CH system from the filling loop. Happily no more leaks. The system was then repressurised and each upstairs radiator bled in turn, repeating the process until all the air had been removed and the pressure stabilised.

Now everything was ready for a test and, switching the boiler back on, all worked as it had been.

Hurrah!

For approximately eighteen hours…

Time taken: Most of a day (including tidy-up prep and going to get new valve).
Cost: Drayton valve, £62.78; 0.5 g jointing compound, £0.03; 100L of water, £0.40 - Total cost, £63.21.

Logitech G935 Headphones Repair

Fortunately, some fixes are quick ones.

Here is a pair of Logitech G935 gaming headphones that you can't hear anything through, due to some problem with the ON/OFF switch. Their sound quality had steadily degraded and the switch had to be jiggled around to get any sort of reasonable output; but eventually even that didn't work. The ear pads are also shot, as all the pleather has peeled away from the fabric covering the foam cushions.

Logitech G935 gaming headphones
with badly worn ear cushions

A very brief web search demonstrated that this is a common problem with these headphones (and with Logitech headphones in general); with the contacts in the switch becoming dirty or oxidised over time. Gemini gave a few options for a fix. The first, toggling it on and off rapidly and wiggling the switch around to get a decent signal, I had already exhausted.

After a quick look at a couple of YouTube videos on disassembling the headphones [1/2] it looked like trying to get contact cleaner into the switch from outside would be challenging. So I decided to open up the headphones and strip down the switch as per the excellent YT vids.

G935 case removed for switch access

Six screws round the left hand side speaker opened the case, with a further two holding the button fascia and the PCB into place. All removed with a jeweller's screwdriver.

G935 button module disconnected

Once unsecured I removed the flexible flat cable to separate the left speaker assembly from the headphones for ease of access.

G935 power switch disassembled

If you squint you can see that contacts are not very bright. To clean them I applied a squirt of WD-40 Contact Cleaner and a bit of vigorous brushing with a small brush (yellow TePe size 4, 0.7mm interdental brush) that happened to fit the recess with the contacts in it perfectly. Also by happenstance it had been stored next to a cotton wool bud and the cotton wool had entangled itself with the brushes' bristles, which only helped to clean up the contacts.

A word of caution about the contacts on the top half of the switch. Clean these carefully, they aren't securely attached and are really thin (and therefore easily lost). Although not too difficult to reseat.

G935 power switch cleaning

After cleaning the contacts were visibly brighter and on reassembly everything was back to 100% performance.

The next task was to source some replacements for the ear cushions that had perished. I started with the Logitech Support Pages, which stated "There are no Spare Parts available for this Product". More superb customer service from a US company...

I did find genuine spares on iFixit at £23.00 plus another £8.40 shipping for the princely sum of £31.40. As the headset is only worth £30-40 now, this is egregiously expensive. So next I tried Amazon, which Gemini recommended, 57% cheaper at £12-15 (inc. P&P) and some had a replacement headband included. Then eBay, about £8 (75% cheaper) again with headband and finally onto AliExpress, my new favourite parts shop, £3.57 (inc. P&P from China!) or 89% cheaper than the official Logitech parts. At this price I could buy ten pairs for each genuine part. So, even if they don't last the four years that I had the originals for, if they manage more than 5 months they are more cost effective than the Logitech parts. Yes, I know that the quality may not be as good, but I am not a high end user and that doesn’t excuse Logitech from trying to rook their users. However, let's see what they look like when they arrive.

So I waited for the slow boat from China.

Some time later…

Replacement G935 ear cushions

What arrived from Dongguan Qingyuan Underwear Accessories Co. Ltd. was a close-ish copy of the originals. As you can see, they were not quite the same shape and the stitching was not up to the same quality, though what can you expect for £3.57 shipped from China? I couldn't really feel much difference in the quality of the pleather.

Fitting was straightforward for the ear cushions. I simply stripped off the old ones and spent a few minutes cleaning up the exposed surfaces, and then the rest of the headphones as I had made clean spots. The new cushions slipped on fairly easily and looked and felt OK.

Replaced ear cushions

Unlike the ear cushions, the headbands were glued down and had to be peeled off. The black support attached to the back of the head band (see below) came away at the same time and was a little tricky to remove. But once done I placed it back into position at the crown of the

Headband support

The new headbands had a self-adhesive backing and were simply pressed into place. Time will tell how they stand up to wear and tear.

JOB DONE

Time: About 45 mins
Cost: 0.5 mL of cleaner (£0.01) & £3.57 for replacement parts.

Broken desk lamp

The Major arrived at my door with a desk lamp in hand stating that it was not working and pointing to a frayed cable.

The lamp was a gift for one of the tykes a few Xmases ago. It is a flexible spot LED design from Ikea (sold under the name Jansjö, but not any more), nothing particularly complex.

Ikea Jansjö flexible LED desk lamp - Non-functional

The issue was that the cable had almost completely broken off. Not sure how, one of the furry friends may have taken a liking to it or it might just have been abused by one of the tykes.

Damaged cable

In the same manner as an LED light bulb, the lamp needs the mains AC rectified into a low power DV voltage (4 V / 0.75 A in this case), which is taken care of in a wall wart. Rather than using a captive cable (cheaper) the lamp has a moulded 2-pin plug (above) . Presumably this allowed Ikea to buy a shed-load of wall warts, at a heavy discount, that they could pair with any LED lamp in their range.

I had not come across this particular type of plug before, but a little searching gave it as a DIN 41529, most commonly found on speakers of a certain age. DIN or Deutsches Institut für Normung is the German equivalent of the BSI and the 41529 standard, since withdrawn, specified connectors for loudspeakers. Fortunately they are still available for very little (~£0.50 a piece). But in this instance I raided a friends spares box for a replacement.

Should be a straightforward repair, just cut off the old plug and wire a new one on....

Repair of DIN 41259 plug

Of course it slipped my mind until after I had snipped off the old plug that the LED would be polarised and so would only work with the wires connected in one orientation, doh!. However it didn't take too long with the multimeter's diode function test to work out which side of the cabling was connected to the positive diode terminal.

A pity then that it didn't work once rewired. A bit more disassembly to check for other faults, such as a dodgy switch or another broken wire, gave nothing and try as I might I couldn't get into the LED housing (although, as I found out after the repair, it is possible - see the second picture here). So I decided to try the connection the other way round (risking it blowing I realise) and,

Ikea Jansjö flexible LED desk lamp - Functional

let there be light. After that it was just a matter of re-assembling it and marking it.

JOB DONE

Time: 10 mins (plus 40 mins headscratching)
Cost: £0.60

Avatar Controls Smart Plug - Cutting the Cloud

Initially I thought that I would be able to reflash the Avatar Controls smart plug that I had repaired with Tasmota using tuyaconvert. But they only work with Espressif ESP32 based boards and once I had access to the internals I found that the microcontroller was a Beken BK7321TQN32 module.

Beken BK7231T sub board

Packed into this tiny 15x20 mm board (marked WB3) is a 32-bit CPU with 2 MB of flash ROM and both Wi-Fi and Bluetooth LE wireless connectivity. Which is all well and good, but how do I get access to and change the firmware of this SoC?

The first decision to make was - which open source firmware system to use on the reflashed device? Broadly speaking there appearred to be two main options, the previously mentioned Tasmota and ESPHome (there are other OS projects, eg EspEasy or ESPurna, but they seem to be somewhat deprecated). Tasmota is based on the MQTT protocol, which is a venerable and widely used communication protocol that should work with any home automation system. ESPHome is a newer project and tailored to integrate with Home Assistant. As I am planning on using the latter, ESPHome seemed the more reasonable choice (though I freely admit that this choice wasn't based on any deep understanding of their relative merits).

The next decision was - how to get the new firmware onto the chip? There is a no soldering option for reflashing Beken chips using tuya-cloudcutter, which uses an over-the-air exploit that has since been patched by Tuya. However the plug was probably old enough that it might have not been patched. The second option required a small amount of soldering - which I was not adverse to - and the use of a USB-to-TTL adapter, but had the advantage of being certain that it would work. However, depending on where the soldering points were this could get a little awkward.

Much rummaging later I couldn't find a Raspberry Pi (the OTA exploit is injected via a networked Raspberry Pi). But after taking another look at the access for soldering (I would have needed to unsolder the whole Beken board) I bit the bullet and ordered a second-hand Pi 4B from eBay along with a case, a 64 GB micro-SD card and a HMDI-to-micro HDMI cable (ouch!).

The instructions from Digiblur's site are given at pace with limited explanation but I found that I could follow them for most of the journey (appended below are some notes on the process). The Avatar plug wasn't on the list of known Manufacturer/Device names so I had to install the Tuya app on a smartphone (fortunately you can use it as a guest) to get the firmware information, which was v1.0.8 for both main and MCU modules. Armed with this info I went down the list of BK7231T 1.0.8 versions until one stuck - 1.0.8 - BK7231T / oem_bk7231s_rnd_switch. Then opened ltchiptool to get details of the active pins on the device, which gave;

I: UPK: Found BK7231T config!

I: UPK: Switch/plug config

I: UPK: - relay 1: pin P7

I: UPK: - button 1: pin P8

I: UPK: Status LED: pin P1,inverted True

So, one relay, one button and one LED. Ltchiptool will output a starter YAML file that will allow for basic control of the device.

Setting up ESPHome as an Add-on to Home Assistant was straightforward but I did have a hiccough on selecting the device type as, for BK72xx boards, there were two options, WB3S and WB3L. I couldn’t see any easy way of identifying which board I had (it was only marked as WB3). So I had to compare the pinouts for the two chips on LibreTiny and make an educated guess based on the proximity of the SCK and CEN attachment points and plumped for WB3L. I did have to combine the ESPHome and ltchiptool YAML into a single piece of code in order to get it to work (with a bit of assistance from Gemini on correct formatting of the YAML output). But work it did, in the end.

JOB DONE

Time: 1 day (combined over several)

Cost: ~£0.50 materials (£73.96 tools - an investment)

Digiblur - How To Guide - Tuya CloudCutter with ESPHome LibreTiny - No soldering

Step 1 - Building the Raspberry Pi Image

Rather than use a standard Raspberry Pi OS distro the protocol uses a headless Lite version that is stripped down and accessed remotely via SSH. I have no idea whether you could just use the standard Desktop Raspberry Pi OS but I'm guessing that you probably could. However in the absence of any real understanding I just followed the instructions…

The command line instructions are:-

• sudo raspi-config - opens the system settings menu to reset the country localisation

• sudo apt update && sudo apt install network-manager - updates the Pi with a list of relevant upgrades and installs the most recent NetworkManager package to manage local network connections

• sudo nano /etc/dhcpcd.conf alterations - tells your Raspberry Pi to stop managing the wireless network interface, giving the system a static IP address

• sudo nano /etc/NetworkManager/NetworkManager.conf alterations - opens the config file for NetworkManager and takes control of the systems network interfaces

Once rebooted Docker is installed, which is a "container" that tuya-cloudcutter will run in (basically a way to ensure that a program will run on all hardware environments).

Step 2 - Install / Update Tuya-CloudCutter

Set up tuya-cloudcutter for use.

The command line instructions are:-

• sudo apt install git - installs Git, which is a versioning software, but will also allow automatic downloading and installation of tuya-cloudcutter from a repository such as GitHub

• git clone https://github.com/tuya-cloudcutter/tuya-cloudcutter - downloads and installs tuya-cloudcutter

Step 3 - Install ESPHome Kickstarter firmware on the Device

Once tuya-cloudcutter is running it will need to know the current firmware version on the device to be cut. I located this via the Tuya app and stepped through the choices for my firmware and Beken chip versions until one stuck. Once tuya-cloudcutter has done its thing, you should see the reflashed device as an open network with the SSID of kickstart-bk7231. Connecting to it and opening the 192.168.4.1 address in a browser will pop up the device control page where you can give it the details of your network for it to connect. Once on your network you will be able to see its local IP number.

Step 4 - LT Chip Tool to configure the device

The next step is to install ltchiptool, which will interrogate the chip and give you a description of which pins are connected to output devices. It will also supply a basic YAML file for the device.

Step 5 - Install ESPHome Add-On or Docker Container

This will allow the move from the basic kickstart firmware to ESPHome firmware. As I am running Home Assistant adding ESPHome as an Add-on was the simplest option for me.

Step 6 - Install ESPHome Firmware and Integrate into Home Assistant

At this point things drifted a little from the protocol. Once the kickstart device is located, you can generate an .uf2 file with the ESPHome firmware and use the device's control page to upload it. After some time the device will reappear in Home Assitant with the new name you have given it. For me the process didn't result in a opportunity to include the YAML file supplied by ltchiptool, but with a bit of judicious copy and paste I could include it from the ESPHome Add-on by editing the device's YAML file there and installing it via the Add-on.

The end point of the protocol should be a reflashed device with basic control over its functions via a page on the Device & services Settings page of Home Assistant.

Avatar Controls Smart Plug - Into the Belly of the Beast

Having convinced myself that the smart plug was not working, it was time for a look inside.

But before that, an external inspection found no obvious charring or melted plastic that might have implied some sort of catastrophic failure.

Avatar Controls AWP14H 10 A UK Smart Plug

On the back of the device were manufacturer and model details and some basic electrical performance data (220-240 V input voltage and 2.3 kW / 10 A resistive load). There was also an IEC 60417 - 5016 symbol for a fuse described as a T10AL240V - which is a T[timed(slow blow)]10A[10 amps]L[low overcurrent rating]240V[mains voltage]. So somewhere inside is a 10 A fuse, which appeared to be a reasonable place to start.

The initial hurdle, then, was to open the case. A job easier said than done, as it is designed not to come apart for obvious reasons. Bigclivedotcom has published a video showing a fix for a similar plug in which he says that the cases are glued or possibly ultrasonically welded [timestamp 12:19]. Some vigorous poking with a blade and jeweller's screwdriver didn't make much progress. But a second YouTuber came up trumps; a bit of asymmetric squeezing from a quick-grip to generate a bit of radial stress resulted in a couple of cracking noises. After which the plug parted quite easily revealing the circuitry for inspection.

Avatar Controls AWP14H Smart Plug - Internal Layout

Again, no obviously burnt out components, but one of the electrolytic caps (arrowed) was suspiciously domed compared to the other two. If you watch Bigclivedotcom's video and another from the Buy it Fix it channel it appears that these caps are particularly susceptible to failing.

While it was open I had a look at some of the other components. From what I could glean I think the plug is organised as follows;

1. The T10AL250V fuse is in a board mounted package with a fusible resistor and a zinc oxide varistor for circuit protection and a single diode to rectify the AC supply into a half-wave DC supply.
2. The mains voltage is then converted to a rough 5 V DC supply by a small buck converter and a couple more electrolytic caps.
3. Our questionable 470 μF 16 V electrolytic cap smooths the output of the converter further.
4. Then a 3.3V voltage regulator supplies the control board on a riser.
5. This is a Beken BK7321 system-on-chip.
6. Finally this controls a 10 A relay that switches the plug.

The screen print on the PCB gives the board's details as HYS-01-127_V1.2 2020-07-17, meaing that the maximum age of the plug is 5 years.

I can check the fused components once the board is free from the plug base, but it looks like the low voltage cap is likely going to be the culprit. So out with the soldering iron it is then.

Most of the kit I already had; soldering iron, solder and soldering bit cleaner; and clamps to hold it steady. What I didn't have (or couldn't find) was any desoldering braid, flux or IPA. So these required a bit of shopping.

The IPA (isopropyl alcohol) for cleaning the board after soldering/desoldering came from Amazon for the low, low price of £6.59. I did have some of this before, but the advent of Covid-19 and the great IPA shortage of 2020 emptied my stocks. Obviously I don't want to be sloshing IPA out from a 1 L bottle onto the workpiece, so I poured a bit into an old screw-cap perfume bottle one of the tykes had discarded. Amazon had no SDS available but the supplier, Trade Chem, were quickly able to send one by e-mail (a sign of a good company). The only issue was the cap liner coming apart but that is what you get for cheap-as-chips.

Next was flux and desoldering wick. I hadn't used an electrical soldering flux before but as this job would rely on desoldering I decided not to try and bodge it. So had a look at a couple of videos on flux (What is this flux stuff anyway? and Which one should I buy?) and decided that the whole topic was too niche to get into for a small job. In the end I bought what I felt was one of the more general purpose no-clean fluxes, MG Chemicals 8341 No Clean Flux Paste, at £7.58 on Amazon. An SDS was sourced from CPC Farnell. Along with a hilarious instruction video showing exactly the wrong way to attach a Luer lock needle.

Needlestick injury incoming

Hint: Use the needle cover to twist the needle onto the syringe, then take the cover off. That way if you slip whilst putting the needle on you don’t end up injecting yourself with rosin (yes I know it is a blunt needle but you can stick yourself with it just the same).

Onto desoldering wick and at this point I was running out of steam so I just bought some MG Chemicals Super Wick No-Clean Fine Desoldering Braid, 2.5 mm x 1.5 m 426-NS for £3.22 from CPC Farnell. Good, bad or indifferent it will have to do.

Now I had the tools and materials to hand it was onto desoldering the board from the 3-pin socket assembly. This wasn't quite as straight forward as I had hoped. As the board's main input pins were connected back to the plug socket pins they sucked heat away really fast so even with the soldering iron up at 400 ^oC and added fresh 60:40 solder it was a bit of a struggle. The wick worked quite well but I couldn't easily part the board from the base and when it did come away it took the tag off the neutral pin. There was also some naughtiness with the 10 A fuse.

Avatar Controls Smart Plug - Board dismounted (left), mains tag issue (top right) and unintended fuse damage (bottom left).

Fortunately the fuse was still OK, just a bit melted, as was the fusible resistor next to it meaning that a blown fuse was not the issue. But now I had to repair the tag before I could get back to the original repair. The tag that had come astray looked like it hadn't been punched on properly as I didn't find any pieces of brass that had come away. I did try to punch it back in but it didn't work, so in the end I flooded it with solder. This might not be a great fix but, as regluing the case will make the plug weaker as well, I am not planning on moving it around a lot. So it will have to do.

The final job was to remove the suspect capacitor, which went without issue.

Failed electrolytic capacitor

As well as the slight doming on the top, the bottom was also deformed. When I brought out the multimeter to check its value I found that its max range was 40 μF meaning the replacement cap I had bought would be over range; however measuring the old cap gave a value of 12 μF which is quite a long way from the 470 μF that it should have been. Definitely the prime suspect.

The faulty cap was from a Chinese manufacturer, Dongguan Zhuo Wang Electronics Technology. The full extent of the specs that I was able to obtain was what was written on the outside of the component's packaging. A search for an identical replacement or a similar 470 μF 16 V cap in the same size format yielded nothing so I plumped for the closest that I could find which was a Panasonic EEUFR1C471. These FR- series caps from Panasonic have low ESR (electrical series resistance) and are designed for long service lives with high ripple currents. So I replaced the duff cap with a new one and resoldered the board back onto the base.

Repair Avatar Controls Smart Plug

Once reassembled and plugged in a blue LED came to life and the socket could be switched on and off at the power button. The plug was also visible on Bluetooth as a Tuya LE device.

So it seems to be functional again. The next step is to deal with the firmware…

P.S.

I did contact both Farnell and MG Chemicals to let them know about the H&S flaw in their video, both of which replied that they "would pass it on to the relevant Dept." But it is still up. So if you have stuck yourself, they have already been warned.