Hardware Insights Forum

Because governments worldwide have proven utterly incapable of providing more than spotty enforcement of safety standards, I might as well play my part in protecting people. Here are my risk rating levels:

• Dire – horribly dangerous, even when new.
• Inevitable – might not pose immediate danger, but still pretty much guaranteed to go wrong in the end.
• Severe – not strictly a guaranteed problem, but still highly likely.
• High – not necessarily a >50% chance of catastrophe, but still too likely to leave alone.
• Medium – maybe not frightening, but definitely not negligible.
• Low – not likely to cause many practical failures, but could still be an occasional issue.
• There are also some cases where the risk level is difficult to judge.

To keep post sizes within reason, I have also sorted the products (and in a few cases, other items) into these problem categories:

1. Items that never met the relevant standards in the first place
2. Unforeseen/unanticipated problems, and items not presently covered by standards
3. Loopholes in existing standards, or standards that are just low overall
4. Issues other than physical safety
5. Other bad products
6. Places plagued by toxic phenomena

I also have two aside posts: "Close calls" and legitimate products sold with dangerously wrong descriptions.

There was another one about extremism in the name of "good intentions", but I've split that off from the main thread so that sensitive persons can avoid seeing it accidentally.

Think we can add all the horrors I reviewed, it's summarised more than enough in the reviews:

Whitenergy ATX-350W (05749)
Spire SP-ATX-350WT-PFC
Linkworld Stability Power LPJ9-25 420 W

Think that all three easily meet the Dire risk level, with several safety hazards as well as being of generally poor quality and a white-day lie to the customer in most of their parameters.

It's getting quite long already (22 items if I've counted correctly), so what do you think?

I could perhaps divide them into categories such as:

1. Non-compliant in the first place (10 or 11 so far)
2. Unforeseen/unanticipated problems; items not covered by present standards (5 or 6)
3. Loopholes or just generally low standards (2 to 4 depending how counted)
4. Issues other than physical safety (3)

Some items fit into 2 categories (possibly more in the future) at once so I'll have to think a bit about how to do them…

EDIT: Done so now (I've saved the original monolithic post to HDD just in case you don't like the new format).

This is where most of the biggest dangers lie (and have done so for as long as I can remember, and most likely well before that too).
No upgrades to the standards themselves (which lately have mostly been minor anyway) can ever fix products which make no attempt to meet them.

Countless counterfeit USB chargers (and other small off-line SMPS)
Risk level: Even Dire barely does justice to these. (Big Clive has a song about them.)
Faults: Various, but the worst being that isolation is almost non-existent in many examples (so they would hardly even qualify as Class 0 to be honest, which itself has been strictly forbidden for decades in decent countries although remains common in the Unbelievably Stupid Arseholes for small appliances).
How to avoid: Stick with the original supplies for brand-name devices where possible (as long as the original still works or replacements are reasonably priced), otherwise find a reputable third-party manufacturer and supplier.

Counterfeit mains cords, marked as “DHT Tech” (among others) (there is a real DHT, but visual similarity between it and the fakes is non-existent)
Risk level: Dire
Faults: Wires can be as small as 0.08mm² (and in copper-coated aluminium or other junk materials to boot), such that the worse examples are essentially guaranteed to melt down if loaded to the full rated 10A (or if the connected load goes short-circuit, for that matter).
Remedy: No use trying to salvage something this horrible. Just destroy it (e.g. by twisting the plug pins out-of-shape with pliers), dispose of it, and obtain a proper cord (which there aren't generally a shortage of anyway) to replace it with.

“Universal” mains sockets (and adapters)
Risk level: Dire
Faults: Attempting to accommodate so many plug types in the same position usually results in unsatisfactory contact, which can lead to overheating (and possible fire) under high load. Schuko and Belgian/French/Czech plugs will usually go unearthed (a minority of these sockets do warn not to insert those plug types, but that's scant mitigation). Even worse, however, are the many misinsertions possible; a particularly scary one being that with many plug types, it's possible to insert the earth pin into a live contact , making the whole appliance live to touch. Many common examples also have other construction flaws, as in what Big Clive calls the death-dapter; an even scarier form of adapter has multiple sets of plug pins, which are supposed to be disconnected when retracted but can often be extended simultaneously and therefore exposed and live. (A different type of multi-plug adapter has totally separate plugs, with the socket rotated or shifted to connect to one at a time; this could be made safe, but would anyhow be bulky and unwieldy to use. See examples 6 and 7 here.)

Whitenergy ATX-350W (05749)
Risk level: Dire
Faults: Falsely claims to be a Class II (double-insulated) appliance; improper inlet (may not comply with IEC 60320); inadequate input filtering; obsolete two‑transistor standby supply; sub-standard creepage distance between primary and secondary; improper earth connection; underspecified +5V rectifier; horrible capacitors; exploded at 260W load; bad voltage regulation.

Spire SP-ATX-350WT-PFC
Risk level: Dire
Faults: Uses cheap non-safety-rated ceramic capacitors instead of Y2 class; poor earth connection; obsolete two-transistor standby supply; chloroprene glue; very bad capacitors; underspecified +5V rectifier; bad soldering; overheats at 280W and shuts down at ≈300W; voltages way out-of-spec under +5V cross-loading.

Linkworld Stability Power LPJ9-25 420 W
Risk level: Dire
Faults: Uses cheap non-safety-rated ceramic capacitors instead of Y2 class; bad capacitors; underspecified +3.3V and +5V rectifiers; hot-melt glue; terrible voltage regulation; overheating PPFC inductor; can't even dream of delivering 420W.

Smoke/fog machines (mis)using an XLR connector for mains-voltage control
Risk level: Dire
Faults: Some (or all) of the XLR pins will be live, enabling electrocution (especially if pin 1 is live) and equipment destruction (if connected to audio or DMX cables). There were also some machines (that somehow got approved) which misused an IEC 60320 outlet for this purpose (abusing the earth terminal).
Examples: Lex Lighting F400 and these two eBay specials

“Child safety” covers for BS 1363 sockets
Risk levels: High to Dire, depending on just how badly the cover designer screwed up.
Problems: They're pointless in the first place, since BS 1363 sockets have had mandatory safety shutters from the beginning (in 1947!). Second, far too many covers are wrongly dimensioned or badly shaped, which can at best cause them to pop-out and at worst ruin the socket they're plugged into. Some examples are also flexible enough to misuse as a shutter opener, and the worst of all can actually enable needles (or similar) to be inserted alongside their pins .
Remedy: Just destroy the bastards (and if there's political red tape in the way of doing so, then complain loudly to those imposing it).
If you happen to have a socket with broken or malfunctioning shutters, then just get it replaced (well, if really old more stuff may have to go too).

Uni-T UT230B plug-in power meter (at least the UK version)
Risk level: Severe?
Fault: Badly designed neutral connection which is impossible to tighten properly.
While mentioning Uni-T, some of their multimeters (especially the UT71E) also qualify due to improper input protection.

Strong SRT5438 DVR
Risk level: High
Faults: The primary heatsink stands dangerously close to the unearthed top cover (which can be easily pressed onto the heatsink) and while not directly connected, is only “isolated” by about 1mm of pad clearance and an electrolytic capacitor sleeve (which has no specific insulation rating).
The transformer isolation, while not the worst I've seen (at least the windings aren't directly touching), is also substandard (ordinary wire with narrow margins).
Remedy: You could bodge in a safer PSU, or the connector for a suitable external unit (since it only requires +5V and +12V you could use an external HDD PSU).
In my case, however (since I didn't need another DVR) I just salvaged the HDD and a few other pieces.

Mains plugs (of any type) with a partially insulated earth pin (most common in the UK, but are now also found in Australia/New Zealand)
Risk level: Depends on what you power via them, of course, but anything with MgO-insulated heating elements will bring it well into the High category.
Problem: Given that the earth pin is deliberately connected to the chassis of a Class I appliance, there is no reason to insulate it, and this insulation can stop the pin from making contact in the socket (especially since the earth contact is often closer to the surface, or even right at the surface in many BS 1363 sockets).
Even if used only for Class II, they are prohibited (BS 1363 allows a fully plastic “earth” pin to open the shutters, but never one with partial insulation).
Remedy: Any item that has one of these plugs is counterfeit (and therefore will most likely have other violations) anyway, so should not be repaired.
For the Jaycar plug-in timer (officially recalled) its plug is integral, so can't be replaced even if you wanted to.

Euroflex Monster SC3X1 steam cleaner
Risk level: High
Faults: Steam valve may get stuck open, posing danger of scalding (while my own skin doesn't burn that easily, I can imagine it being a serious threat to small children or other people with a skin disease/disorder); thermal fuse seems to be counterfeit with an abnormally high cut-off temperature.
So yet another badly-engineered (and manufactured with parts of unknown provenance) piece of consumer junk (and not even an excusably cheap one); consumer magazines fail us yet again.
Suggested substitute: Kärcher seems to make decent models, although you must take care not to end up with the wrong plug if you order them from Amazon (to get it with an Australian plug, the order number is 1.513-142.0) and the somewhat short (4m) cord may compel you to use an extension cord with it (and this house isn't even a big one overall).

Solutions AS2601 6-outlet power-board (and no doubt other store brands made by the same manufacturer)
Risk level: High
Faults: Poorly aligned busbars which can short active to neutral when even casually bumped.
(And given that the circuit breaker has only UL approval, I don't trust it very much either for the reason mentioned in the 'low standards' section below.)
Recommended substitute: Kaifeng KF-MSD-6A aka BizLine BIZ 390527/BIZPB0005 (available at Rexel)

Jaycar “PowerTech” MS4040 (Original Model No: WH-4-3) 4-outlet power-board (and no doubt its relatives)
Risk level: High
Faults: Cord has copper-clad aluminium cores, also with poor-quality insulation; dubious method of wire termination inside (the wire ends are solder-coated, then the brass busbars are folded over them to crimp ); when overloaded, the circuit breaker arcs before snapping into the tripped state.
Recommended substitute: Kaifeng KF-MSD-4A aka BizLine BIZ 390525/BIZPB0001 (available at Rexel)

Generic fan heater, model FH-07A (sold under Celsius, Goldair, Number8 and possibly other brands; with only slightly varying pseudonyms)
Risk level: High
Faults: Violates double-insulation requirements at least once (and probably twice); thermal protection barely works (and the casing melts); damaged (and poorly terminated) internal wiring.
Remedy: Nothing that can really be done to fix this unit; even if you disable the heat altogether, it won't make a good stand-alone fan.
(You might be able to salvage some parts, at least.)
Suggested substitute: Kambrook KFH700 (hopefully little-changed from the KFH600 and KFH660); or better yet (if you can afford it), DēLonghi HVS3032

(Kambrook recently priced-up the KFH770 to AU$100, at which point you may as well go all the way to the DēLonghi instead.
Also, you may want to avoid using the oscillation motor in the KFH660/KFH770 or similar heaters, and even disable that if untrained people will use it:
I haven't personally seen the internal wiring broken by stress; but it's plausible given that at 2.5RPM, even the mere 1,000 hours Kambrook use as their endurance test will rack up 150,000 flexing cycles. Scale that up to a more sensible 10,000 hours, and you could get 1.5 million cycles. I'm not sure if oscillation is even a good idea in stand-alone fans, really, observing this example.)

Sunair 838 hair-dryer
Risk level: High
Faults: Cord is clamped only on the inner cores, not outer sheath (and squashed quite hard in there); wire ends were solder-dipped before insertion into terminal block, and cord neutral in my unit even had its insulation trapped on the underside.
Remedy: Take out the terminal block (there isn't really space for it), remove the wires originally going from terminal block to switch, cut off the pinched part of the cores, re-strip the cord, solder cord wires into the switch terminals, mount the switch and clamp the cord sheath (turning the clamp the right way up).

It also has no RFI suppression, so I suggest adding an X capacitor (as big as you can get with 10mm lead pitch) and bleed resistor.
You'll need one with long leads to mount it past the switch mounting post (even they may barely reach), and be sure to insulate its active lead.
(Decent hair-dryers also include ceramic caps across the motor and from its terminals to body, but that'd be cumbersome to add now.)

CordTech CPC4-W (see customer reviews on Bunnings' page)
Risk level: Medium–High
Fault: Socket contacts can hardly be reached by the plug pins (causing pop-out), as the socket design fails to account for the casing thickness.
Alternatives: I'd generally go with a normal power-board and separate USB supply; however if you insist on a cubic combination device, the Jackson Power Block is a reasonable option (especially if you're prepared to recap and poly-mod it) albeit with only 2 ports (2.1A total).

Pseudo-C7 cordset included with old (1970s/1980s?) Remington shavers in Australia
Risk level: Medium–High
Fault: Socket end was shortened to 8.5mm (from the 10.5mm minimum in IEC 60320), resulting in unreliable contact in proper C8 inlets.
Observing the 1.0A rating on both ends, I'm surprised the cable appears to be (then-)normal H03VH‑H2X0.75 and not tinsel-cored H03VH‑Y.
Management: If you still have the original shaver, you can continue using it there. If you don't, then destroy and dispose of the cord (or you can cut the ends off and reuse the cable for something else; though I don't think H03VH‑H/equivalent is approved here for mains anymore, now being suggested for ELV only).

This is among several then-approved items which have soured me on Australian electrical standards; for the most part, stuff got safer as we did less of our own thing and followed the European/IEC standards more (although with some newer standards, we're getting our act together a bit).
I can forgive truly ancient items (like our obsolete 7.5A appliance coupler having apertures large enough to insert the plug pins, including earth, into), but not semi‑modern stupidity like this. (What was wrong with just using the C1/C2 couple there, for crying out loud? Did any mains shaver draw more than its 0.2A?)

Another (Ronson) shaver cord I've seen has a tiny socket (much smaller than even the IEC C1), with contacts close to the surface and soft material which I can press down to the contact level with only my thumbnail ; you can count it in as a “bonus”.

Laser PW-4ADAPT travel adapter set
Risk level: Medium
Faults: No fuse in the adapter for BS 1363 outlets; earth contacts sit below L+N in the adapter for US/UK plugs to Australian outlets.
My recommended alternatives: You could:

• Search for an adapter that's actually safe (if you can find a review by a knowledgeable person, which may be a crapshoot)
  In particular, if you want to adapt to BS 1363 outlets, you'll be better off buying the adapter in the UK (where their fuse requirement is actually enforced).
• Obtain a (good-quality) plug and socket of the types required, and connect them with a suitable piece of flexible cord
  (Provided you're electrically competent, or willing to pay an electrician to do it)
• If the appliance(s) you want to import or travel with have IEC 60320 inlets, then just find the appropriate cordsets locally
• Replace the plug, if importing an appliance for permanent use in your country (provided it's compatible with the local mains supply)

Other unfused “BS 1363” adapters: Crest PWA05292 (also included in the 3-pack PWA05160), Korjo 'KA UK' (and 2-socket variant 'KA UK DA')
Jackson Industries (another Australian-owned company) provide some with fuses (not sure if all are BS 1362 though) and some without.

Go Travel 095 Schuko to AS/NZS 3112 adapter (discontinued, I don't know if the replacement model is better)
Risk level: Medium
Fault: Earth clips in the Schuko socket (CEE 7/3) come up short, so may only make contact at about the same time the power pins do (or with such a small margin that it comes down to the dimensional tolerances). Of course you can work around this by fitting it to the Schuko (CEE 7/4) or CEE 7/7 plug before plugging it into the Australian socket, but being strict about it, recalled it must be. I also got the casing to burn for 11+ minutes, once ignited.
Recommended alternatives: The same as listed in the entry for the Laser set.

Dimplex 3088T convector/fan heater (fortunately discontinued)
Risk level: Medium
Faults: Power is over the limit for Australia, due to incorrect choice of heating elements (although the nameplate said the normal 240V 2400W, it actually used an element assembly totaling 2500W at 230V which scales to 2722W at 240V (that's 11.3A instead of the legal 10A, causing 29% greater-than-normal dissipation in the conductors)). It also has quite a few other -worthy design flaws, including that the two elements are equal halves despite the switch suggesting 3 power settings . Overall, if this is at all representative of Dimplex's stuff, then they're not a reputable brand despite popular opinion.
This shouldn't have passed safety approval, at any rate.
Recommended substitute: DēLonghi HSX3324FTS is basically an equivalent (plus on/off timer and tip-over switch, and with the ability to run fanless at full power), but built properly (I'd go so far as to liken the construction to that of Japanese electronics ). Or if you're on a budget and 2000W fanless is satisfactory, then you can try the HCM2030; that model does have the usual lightweight casing (albeit with some reinforcing stampings) on modern convectors, and the foot mounting arrangement is notably insecure (with just a single screw each side which easily comes loose with movement, which once caused a foot to fall off the unit my sibling used and consequently tripped the thermal cut-out), but at least it's wired correctly.
Both of these DēLonghi models are also Class II if that's worth anything, whereas the Dimplex was Class I (which is required on the oil-filled column heaters but probably doesn't cost noticeably less to implement on these vented-box types).

Auriga “Multi-Media Speaker System”, circa early 2000s (unknown model, but black and has about 34mm midrange and 13mm tweeter drivers)
Risk level: Medium
Fault: The mains transformer (6V 0.2A) has no thermal fuse and if its output is shorted, gets too hot (about 167°C on the secondary in my test, where the official UL limit is 150°C; its working temperature class is probably no more than 130°C, and even that's a stretch observing that the secondary leads are UL AWM 1007 rated for only 80°C) to pass impedance-protection criteria (it might only qualify in an open-frame application, not inside a sealed plastic box).
Indeed the transformer bobbin is almost melting at that temperature (and things would get even worse in a hot summer).
To top it off, the speaker enclosures are not self-extinguishing.
Mitigation: Add a secondary fuse (200mA or 250mA slow-blow) if you have to keep using it. Otherwise just get some better speakers

Emporium “Beckett” table luminaire (model CWTLE1412 according to box, or C1312TLE27 on the label)
Risk level: Low–Medium
Faults: Only a single-pole cord-line switch (which is implicated in an official recall of other such lamps), so if wiring upstream is transposed or faulty, then a shock hazard exists when changing the bulb (even with the switch off).
The cord itself has copper-clad aluminium (or whatever) cores in a non-standard strand size and count (not quite 0.75mm²).
Remedy: If you're that fond of the style, you could replace the cord with a proper copper one (and double-pole switch).
Alternatives: I have not found any properly-made equivalent sold nowadays, unfortunately. Perhaps it's time to move on to native LED luminaires?
Some TaoTronics models, at least, seem half-decent (apart from the USB charging port generating TV interference when loaded) in design and construction, and they run from small plug-pack SMPS (easy to replace if failed).
E27 and B22 have long-outlived their usefulness (and “safety” level) anyway; if you insist on keeping them, just accept that you'll have to wire them properly.

Although this doesn't explain why single-pole switches seem to be normal on desk lamps…

Stylux T-869 “Trax” desk luminaire
Risk level: Low–Medium
Faults: Only borderline compliance with double-insulation requirements; plug pins are out-of-tolerance at the short end.
Remedy: Replace the plug (or even the whole cord if you're up to the task), and check the insulation on the internal wiring.
Again, the switch is single-pole; I don't see any sensible reason why the rule would differ based on the switch placement, so your guess is as good as mine.

Connected Switchgear B22 lampholders/battenholders (BH, BHC, LH01)
Risk level: Low for mains voltage, but becomes High if you use them in an extra-low voltage lighting application (given the higher current).
Fault: Although rated for 3A (typical for B22 holders), they're actually only safe up to 2A or so (the springs will overheat and anneal fairly quickly at 3A, and within seconds under the 3.75A overload test); while this will rarely be a problem with mains lamps (2A at 240V is already 480W into a resistive load, with the lampholders typically only rated to endure the heat of 250W-at-most lamps anyway), be careful if you're using ELV lighting or connecting any other (Class II, and not dependent on A/N polarity) loads to the lampholder. (There is nothing wrong with their B22 plug, model BCA; which is just as well since the Clipsal 466 was discontinued.)
Replacements (if you insist on keeping B22): Clipsal 2530SS (battenholder) and 501 (pendant lampholder)
If you have any self-respect, choose a protected battenholder (2530SS or old-stock 530SS) for mains use; unfortunately Clipsal don't make a 501SS.
(Obviously, if using for extra-low voltage it doesn't matter whether you choose the protected or unprotected variants.)

PEC PLS012 B22 lampholder (found in Westinghouse RS662V*7 side-by-side fridge/freezer and presumably related models)
Risk level: Depends on the lamp rating (40W maximum in the mentioned fridge) and physical stress; probably Low–Medium in this situation.
Fault: The springs (strangely stainless-steel) get nearly red-hot at the rated 3A, and don't seem to endure physical stress very well either (I had difficulty removing the bulb, because the B22 pins suffered severe arc damage from bad contact).
Replacement: I used a spare Clipsal 501 (which has a flange to go in place of the PLS012's tabs).
Note that there's little room behind the lampholder in this fridge, so if you use ferrules on the wire ends, their insulation collars (if any) must be cut off to fit.

Deta (Arlec) DET190 downlight
Risk level: Low–Medium
Fault: Ineffective cord grip, on top of generally poor quality (par for the course from Arlec).

These are products which complied at the time of manufacture (and some still do), but only because the authorities didn't think of the problems.

Antec SmartPower, TruePower, NeoPower and Phantom series (earlier generations, manufactured by Channel Well Technology)
Risk level: Inevitable (damage to powered hardware)/Medium (fire)
Faults: Almost 100% rates of failed output capacitors (Fuhjyyu, Mucon), even during the “expected” lifespan of a PC; imagine how bad they'll be now!
Remedy: If you are capable of repairing SMPS without getting electrocuted (or creating hazardous faults yourself), then you can replace the caps with good‑quality equivalents. (It might also be a good idea to modify the fan controller, to spin the fan(s) faster and reduce the temperature rise.)

If you aren't equipped to repair it, then you will have to sabotage it in such a way as to render it safely inoperable.
(For example, by cutting out the AC input wires; just make sure that their stubs can't short onto anything. The earth wires must be kept intact, though.)
Replacement (if not repaired): For +12V heavy configurations, you can use any decent modern PSU with suitable ratings.
Vintage +5V heavy PCs will be trickier; modern DC—DC units (of a modest rating) may work for modest rigs, but if you have a real power hog (e.g. dual Athlons) then repairing a +5V heavy unit (or finding one which has good capacitors in the first place) will be about the only way to do it properly.

Bestec (a.k.a. “Worstec”) ATX-250-12E (doesn't affect the -12Z and later models)
Risk level: Inevitable (mainboard damage)/High (damage to other hardware)
Faults: Two-transistor self-oscillating standby supply with a very bad capacitor in the critical location, which often causes the +5VSB (nominal) to skyrocket (as high as +18V has been reported!) The other caps aren't much good either, of course.
Remedy: While it doesn't fix the fundamental flaw, you can get more life out of the unit by replacing the critical capacitor (and others given the age of the unit) with a high-quality one. More adventurous repairers may even be able to replace the two-transistor circuit with a modern IC.
Replacement (if not repaired): Much the same as for the Antecs above.

RIFA PME/PMZ/PZB, WIMA MP-3 series and AEE Miniprint metallized paper capacitors (there may also be similar types from other manufacturers)
Risk level: Inevitable
Faults: Brittle encapsulation that cracks easily, letting moisture in causing dielectric breakdown; Class X and Y units often burn up as a result (with the added “bonus” of leakage to earth where Class Y versions are present). The RIFA (and AEE) models are most infamous for this; the WIMA types may be slightly better.
Remedy: If you own equipment containing them (they were also embedded in many Schaffner IEC inlet filters), they will need to be replaced with non-defective types (which will be polypropylene, polyester, or ceramic for Class Y up to 10nF). (As with all safety-critical components, be sure to obtain Class X and Y capacitors from reputable suppliers.)
If you decide not to repair an affected item, then sabotage it like suggested for the Antec PSUs above (in such a way that power can't get to any of these faulty capacitors, of course).

I'm a bit surprised these capacitors are still being made; the claimed advantage is better self-healing than plastic film, not that it's that much help when the case cracks regardless (indeed, they don't even need to be used to fail). (I don't know if new production has been improved, so I'll assume the worst there.)

“Amztoy” (and other brands of the same design) self-cleaning litter boxes
Risk level: Severe (even after applying the firmware update, I would still rate it as High)
Faults: Between buggy firmware and lack of hardware failsafes, during a cleaning cycle the boxes can trap and kill a cat (and have in several cases).
Alternatives: See the reviews on One Man Five Cats.

Molded-on SATA power plugs (not all are affected)
Risk level: High
Faults: Some designs develop electrolytic corrosion internally, creating a conductive path which can eventually heat up enough to cause ignition.
This seems to occur in those with soldered connections; those stated to use crimp connections are probably fine (unless that text itself is faked )
How to identify: Mark Furneaux has a good video.
Remedy: Adapter cables of this bad design should be destroyed, and swapped for a safe version. (If you get them wired onto a PSU, it may be trickier…)

Convector heaters (based on bare resistance wires or ribbons) with plastic top grilles (mostly older models)
Risk level: High
Faults: The plastic degrades from the heat (especially when operated at full power), becoming brittle, and eventually breaks; opening up large gaping holes, and possibly being melted/burned (if it falls into the heating elements).
This is especially bad, compared against the near-indefinite lifespan of the electrical parts (if competently manufactured).
Management: If you still have such heaters, avoid operating them unattended, and check the condition of the grilles regularly; when they break (or become severely discolored at any point), stop using the heater. (You may be able to then salvage the OK parts; if you don't, then amputate the cord.)

Thankfully, every newer convector heater I've seen has a metal grille (I suppose the relevant standards must have been revised).

CordTech 6WS34-TGH-PR (customer review was on Bunnings' page but that's gone; I hope it really is discontinued)
Risk level: High (tentative rating)
Faults: Weakly clipped-in socket faces which can pull out with the plug, exposing live contacts. Here's the customer review photo:
Replacement: I'd use a regular power-board and plug in a USB supply (to avoid creating extra waste when the supply fails).

The only exception to my separate-supply preference might be when travelling, to save space in the suitcase.
Fortunately, both the Jackson Power Block and original Allocacoc PowerCubes have secure socket faces (with sandwiched-in flanges, no clips).

HPM power-boards (and outlets) with rotary socket covers (see customer reviews on Bunnings' page)
Risk level: Medium in normal use, but becomes Severe if you have small children trying to poke things in (which is the whole supposed point!)
Fault: The covers break off easily, and the spring can then short between pins (the straight part also looks long enough to reach into a socket contact).

Current HPM power-boards in general earn a dishonorable mention, for stooping to the USAmericans' level by saying they're “for residential use only” (even their overpriced “heavy duty” models, which of course contain the same flimsy socket contacts as the basic type )

In any case, children would probably figure out to rotate the covers fairly quickly (so they're not secure compared to internal shutters).
So far the only modern Australian power-board I've seen with decent shutter mechanisms is the variously-rebranded “EL01084”.

However if you don't actually need the shutters, the Kaifeng power-boards remain my recommendation for quality.

“Swivel” (rotatable cord entry) mains plugs (some might also fit under Low Standards)
Risk level: High
Faults: In Arlec's design (Australia), the wires break easily after a few turns due to a lack of internal strain-relief:
(There was an official recall of a certain run albeit for a different reason, but in reality all of them should have been.)
Another type found on Belkin power-strips in North America has a different construction (sliding contacts inside) but is also infamous (maybe more so!) for burning; here's an example report (among many).
Possible remedy: For the Arlec version (if you have some that haven't broken yet), you could glue them permanently into the orientation you use them in.
(Even then, I'm not so sure I'd trust them that much…)
For the Belkin version, I don't see any remedy other than replacing the plug (or the whole unit).

Original Microsoft Xbox internal PSUs (Foxlink or Delta made)
Risk level: High (Foxlink), Medium (Delta)
Fault: The C8 mains inlet has inadequate mounting (just a pair of flimsy plastic clips in the Foxlink; the Delta is a bit better, but evidently not trouble-free either), so plugging/unplugging the cord enough times will crack the solder joints (which will then arc and burn the board).
Microsoft infamously tried to deflect the blame to the included C7 cords (which had no problem) and provided replacement cords with an AFCI/AFDD, which went down as well as you'd imagine (given that users borrowed C7 cords from elsewhere, which made no difference).
Management: While you can resolder the joints (if they haven't already arced to the point of burning the PCB), this doesn't fix the underlying weakness.
You could add a strong glue to hold the inlet still; otherwise leave the cord plugged into the Xbox and avoid disturbing it.
Possible replacements: If you can obtain a Minebea PSU (compatible with your Xbox revision), it should be fine.
Otherwise you can adapt an ATX PSU to work (rewiring, regulating the +5VSB down to +3.3VSB, inverting the Power On and level-shifting the Power Good signal).

Since the affected Xbox revisions also had a leaky clock supercapacitor (and other electrolytics may be on their way out too), deal with that while you're in there.

Non-induction (resistance heated) cook-tops with touch (or otherwise electronic) controls
Risk level: Medium (if otherwise competently engineered) to High (if less competent)
Faults: If the electronics fail, the heating elements can be switched on unattended, igniting any combustibles on top.
(This has indeed been implicated in a few official recalls.)
Mitigation: If you're stuck with such a model, then make sure anything combustible is kept well away from the hot-plates.
If you're electronically inclined, you can examine the control circuitry (and possibly make improvements).
Alternatives: Look for models using good-old mechanical switches for the resistance elements.
(Touch controls are fine for induction zones, given their inherent safety advantage.)

The makers of these would do well to take a page from microwave ovens:
Those come with at least 3 interlock switches as standard, one of which is arranged to deliberately short the mains (blowing the internal HRC input fuse) as a last‑resort if all others fail (or more likely in practice, some idiot tries bypassing the interlocks to attempt room heating with the unit; which you shouldn't try with any form of oven anyway, just get a convector or fan heater if that's what you want and the room doesn't have a heating installation).
With each switch being even remotely reliable, you could easily sell trillions of microwave ovens before ever encountering a total interlock failure…

You can see that the four 'technical' examples so far are of U.S. origin (although two remain extant in several other countries in some form); the reason (or at least a major factor), in a nutshell, is near the end of the post.

Baby shoes (and any shoes with significant heel elevation and/or pointed toes, of course)
Risk level: Basically Inevitable if fitted at all tightly. Granted, this isn't a “catastrophic” outcome (and most people aren't even aware of “slight”/“moderate” deformation, lacking any useful frame of reference) but once it occurs, it can't really be fixed.
Problem: Almost no modern shoes are shaped sensibly enough to fully avoid constriction (and hence deformation during growth), but baby shoes are the most detrimental of all (while simultaneously serving no functional purpose). As for “aesthetics”, I find healthy feet (however rare they are in the “developed” world) far more attractive than any shoes (between millions-of-years of evolution and the transient whims of fashion, which do you expect to win?)

I don't generally give parenting advice (certainly not in-person) but beyond avoiding baby shoes outright, don't impose shoe-wearing at any age (or even encourage it apart from genuine dangers; broken glass only counts as such if it's pointing up, and I don't think that's frequent enough to justify the detrimental effects of shoes). If other people try to impose shoes, make every effort to question their arguments. Much more information can be found here (among other sites).
When you think about it, isn't it quite telling that athlete's foot and ingrown nails (as just two examples) don't often (if ever) affect hands?

Vintage U.S. sound equipment with their 'death capacitors'
Risk level: Severe
Problem: These capacitors were connected between the mains input and unearthed chassis, yet aren't even rated for reliable operation across mains.
This isn't a forgivable mistake the way items in the above post are/were; even in the 1950s electrical engineering was developed enough to understand why using such a low-rated capacitor there was a horrible idea. Rather it's an indication of just how shady U.S. electrical practices have been (and still are) compared to saner countries.
Remedy: As mentioned in that article, the safest option is to rewire the equipment for Class I (earthed) operation; but at the very least, remove the death cap or replace it with a Class Y capacitor (obtained from a reputable supplier, again).

Any and all plug-in (or outlet-integrated) “Mode 2”/“3-way” “surge protectors” (i.e. those that divert surges to earth/ground)
Risk level: Severe (equipment damage)/Medium–High (electrocution)
Problems: As mentioned in various places on Zero Surge's site (and also this article by a former president of theirs), shunting surge currents to earth creates voltage pulses along the earth conductor, causing damage to interconnected devices (and in smaller surge events, signal corruption/interference).
If the A–E (or N–E with reversed polarity) MOV is on the brink of failure, it can also pass residual current to earth at the normal voltage.

So far I've checked the MOV connections in these Australian power-boards:
A to N only

• Allocacoc PowerCubes (although they're a close call for other reasons)
• Arlec SWP-LS series4 (also a close call)
• Arlec PB12PP and Crest PW4PBS10 (both manufactured by Kaifeng, their suppressor parts are almost identical)
• Belkin F9V004auC6 (goes inline with the C5/C6 and a phone lead; interestingly it contains both a permanent 470V 14mm MOV for 200–240V mains, and a switched 200V 20mm MOV for 100–127V; so is indeed suitable for both ranges, although not in-between values like the label rating implies)
• Crest PWA04984, PWA04985 and (probably) their other close relatives
• Some HPM unit I had, but I don't remember the model anymore (although the neon indicator was blackened)
• Jackson PT4113 (no failsafe though) and other current models according to their website (however, I'd still recommend opening up to verify)

A and/or N to E

• Australian Protective Electronics 390-1 (on this occasion they also proceeded to use X2 caps from A+N to E, instead of Y2)
  (Not to be confused with their similar-looking but unrelated Power Guard 370B, which is an undervoltage/overvoltage cut-out device; although it does have its own fault) Since this isn't even a power-board (just a single-socket pass-through), there was no point in modifying it.
• CABAC PB80 (discarded now)
• Jackson PT0888 (I cut out the MOV earth wire; and since no failsafe was included, also severed their active track)
• Kambrook KD4SG (I removed the MOV PCB and connected the wires through, since it's otherwise well-built as Kambrook power-boards were; I also managed to erase the “Surge Guard” markings with acetone, although you must avoid prolonged contact between acetone and the plastic)
  Some later models (I saw one in the thrift-store alongside the KD4SG) were plastic-welded and therefore a lost cause.
• Total Power Systems POD-1 (cannot be salvaged unless you have Tamperproof Torx Plus screwdrivers, and also has a bad circuit breaker)
• Triangle P-320 (molded plug found on a C13 cordset)

Unfortunately, they are installed inside UPSes and not just power-strips/boards; so if you need the UPS, avoiding them may be a bit tricky.
PSUs themselves don't generally contain this arrangement (I've not seen it from the well-known OEMs), although the Enlight Sniper Power 450W and Infinity IN08-450 (both based on the same platform) do appear to (although the old pictures are broken, I still have a local copy), not that they're very good units anyway.

In Australia, even Clipsal are guilty of participating in this scam, with their 30SFM (a module that snaps into the normal aperture used by their 30M switch mechanisms); it is also available assembled into twin outlets, catalog numbers 25SF, 2025SF or C2025SF. (At least it has thermal fuses, so shouldn't burn the building down.) Here it is, at AU$168 RRP for just the module
(I mean, these abominations are a rip-off at any price, but bloody hell!! No doubt, they've been quite a goldmine for their makers.)
PDL's 600SM (in New Zealand) is much the same; I can't quite tell if this was also the case for the user-replaceable 600SPM (installed in the 695SP), but if you have one, check the number of pins on it (if 3, it's probably the same ordeal; if only 2, it's OK to leave in).

The North American electrical system in general, on that note
Risk levels: Typically Medium, but can extend to High or Severe for some items.
Faults: Range from flimsy ultra-cheap construction in many “residential grade” fittings (particularly bad given that the home is where fires are most likely to kill people), to the unending use of Class 0 on many small appliances (even those containing MgO-insulated tubular heating elements with their bad habit of insulation breakdown when old), to arrangements that in places seem self-servingly designed to go wrong at the slightest deviation from the “rules”.
What do you expect when you entrust a group of insurance companies (it's right there in the name Underwriters Laboratories) with product safety
Mitigation: If you are unfortunate enough to live there, then look for equipment that compares well to the norms in Europe and Australia/New Zealand.
For Class 0 appliances you can't avoid using, make sure to power them through a working GFCI. Otherwise just stay safely out of those countries.

Ionization smoke (non)alarms (in most cases)
Risk level: Severe (when a fire occurs)
Problem: Irritatingly prone to nuisance alarms from microscopic particles (e.g. burning toast); unresponsive to the larger smoke particles prevalent in real fires (especially of synthetic materials).
Solution: Unless you have a niche case which actually requires fast response to microscopic particles, use photoelectric alarms instead.
As can be found at the linked site, the continued approval of ionization units was also another of the UL's 'antics' (apparently they're finally taking action in 2022).

Doorknobs with plastic parts in the latch mechanism (the originals in my home being from Kwikset)
Risk level: Medium overall, I guess
Problem: The plastic parts will crack if appreciably stressed, and eventually break to a point where the mechanism jams and you can't open the door.
Replacement: If you're here, then I doubt you'll need instructions to change a doorknob (but even if you do, they're hopefully included anyway).
I haven't looked far enough to be sure of the entire market, but certainly the brands I've seen in Bunnings have all-metal mechanisms (Kwikset not among them, but even their current latches appear to be metal looking at their site).
Unfortunately those Kwikset knobs have a crescent-shaped axle (unlike the prevalent square axles, although Kwikset have also adopted those), so if you can't obtain a replacement latch from Kwikset, you'll have to replace the entire knobset (not that I minded much given the worn finish).

While these items don't necessarily pose any direct physical danger, they nonetheless have bad enough flaws that the consequences can cause serious financial, legal, or emotional harm to their victims; so I take them just as seriously overall.

“Promotional” USB flash drives (as given away at conventions and the like)
Risk level: Severe
Faults: Extremely cheap and low-quality (or old recycled) flash memory, which can lead to data corruption.
Mitigation (if you're that much of a penny-pincher): Don't use for anything remotely important, don't allow into an environment where they might be used for anything important, and destroy as soon as data corruption appears. That said, most are extremely slow in the first place, so if your time is worth anything you wouldn't want to use them anyway.

Conventional schooling systems (with rigid adult-directed curricula, and compulsory testing)
Risk level: High to Severe, getting worse the more “special” the student is.
Faults: Most of them well-explained in Peter Gray's blog.
Mitigation: If you are stuck with sending your child to a conventional school, then be ultra-vigilant about their mental health, and don't delay responding in an emergency. If you are a student in this situation yourself, then you will have to figure out the most viable solution for your circumstances.
At least my Mum had a strong enough will to pull me out before I suffered too much damage…

Kingston SSDNow V+200 series (and other SandForce-based SSDs?)
Faults: When worn-out (or possibly even before), they may silently corrupt data and cause insidious problems (programs will crash of course, and Windows will slow down as it tries to limp on despite being mortally wounded); even their S.M.A.R.T. reports may be oblivious to the failure. The failure of an SVP200S3120G has cost my Mum (a well-paid accountant) an easy AU$10,000 in lost productivity, and there may be other cases of the same problem. (Observing the reports, I'm wary of SandForce-based SSDs in general, now; unreliable and difficult-to-recover, so I read. Perhaps fitting with their ownership by Seagate )

Here are some items which aren't quite bad enough for an outright recall, but have come too close for a great deal of comfort:

Computing hardware and other high-tech

• AMD Athlon Thunderbird/Duron Spitfire: Lacks any built-in thermal protection, so if it loses its heatsink, it will burn out in seconds. By the way, since the 1.4GHz flagship Thunderbird dissipated 72W and the die area was 120mm² (1.2cm²), its surface heat density was 60W/cm²!
  (Typical fan heaters operate their elements at around 10W/cm², which in the event of fan failure is already enough to get red‑hot before the thermal fuse blows; and they've had compulsory 2-stage thermal protection since the 1980s if not earlier. 60W/cm² even rivals the 'suicide showers' common in South America, which to be fair lack any thermal protection themselves.)
  At least there's fire retardant in the PCBs (and most Thunderbird/Spitfire CPUs had a ceramic carrier), without which things could have been really bad; and AMD added a thermal diode in the Palomino/Morgan cores. (Although they still depended on the mainboard responding and PSU shutting off in time.)
  And to be fair, the chip probably didn't continue dissipating the full 72W after reaching its failure temperature anyway, so didn't quite get red-hot.
• Bush BHAS01UR digital TV set-top box: Got it because the display window came unstuck from the front panel (easy enough to stick back on with new double-sided tape). Where its mains cord wires connect to the switch, the heatshrink doesn't fully cover the bare ends; isolation from the casing (this is supposed to be Class II) therefore comes down to the distance from there to the cover (seems about 5mm to 6mm, but could be reduced under physical stress). Otherwise, the PSU seems more-or-less OK (capacitors aren't quality, but are probably within their ripple current ratings and measure alright for now). The RF modulator is rather bodged-in though (intended for PCB mounting, instead a multi-way connector is fitted over the 6 small pins and the ground wire just soldered onto a tab), and the logic board had a bit of chloroprene glue for reasons inexplicable.
• Topower SilentEZ 350W: As shown in our review, that goofy fan button is bodge-wired onto the underside of the PCB using two single-insulated wires fed through the creepage slot between the optocouplers. It's also full of nasty chloroprene glue, performs poorly and is surely long past its design life.
• TSA8 PSU for DEC HiNote VP500 series: Well-made electronically (with Japanese electrolytics and even an OS-CON on the output), but the plastic welds on the casing are a bit weak for comfort (it only took a spudger to tear open). (Two reinforcing pillars are actually provided with screw points hidden behind the label, but their inner diameter is larger than practical for screwing into; anyway they're quite close to the primary circuitry, so if you do add screws then make sure to insulate over their heads.)

Other electrical items

• Allocacoc PowerCubes: In a better time I might have given them an actual recall; but since there's much worse available in this market (as mentioned above) and relatively few good options (Kaifeng KF‑MSD‑4A being the best I've seen) anyway, I'm pragmatically letting them go for now. I still strongly recommend that you either add a thermal fuse to the MOV or just remove the MOV altogether, though (better to end up with a dead PSU than a house fire).
  In Australia/NZ there's now a better-made local clone, Jackson Industries' Power Block (though I advise the non-USB version)… Actually, now that one of my non-USB Power Blocks had a failed solder joint on the end socket earth, I could technically count them here too; but since they're otherwise a winner, this is easier to fix than a broken cord grip, and the other two I bought pass, I'll again spare them the indignity of a recall.
• Arlec SWP-LS series4 power-board: (and presumably related models such as SWP-S, FWP-S)
  Perhaps the first item here where a close call is an improvement from my expectation; on the back it warns to perform insulation testing at only 250V, which suggests MOVs from L to E and N to E (see the Loopholes/Low Standards post regarding the perils of those arrangements). On opening it up (with the needed exotic-but-obtainable triangular bit), however, it just has the one MOV from L to N; and there's no provision for others as an option (or even for a PCB big enough to hold them). It's still cheaply made though (as seems typical of Arlec products), with thin busbars (though still sufficient to pass 10A) and only a crude wire fuse for the MOV fail-safe, and the LED series resistor gets quite hot (not uncommon in these).
• Blaupunkt BCH500B: A zombie-branded product which generally sits right on the edge of the standard. I only even considered buying it to have a lower‑power model in the heater assortment, although there are surely at least a few better-built models out there (e.g. the DēLonghi HCC500 convector).
• Click C6PB: You can defeat the circuit breaker on it simply by holding the reset button (the normal breaker design can still trip although will auto-reset in that case). Choose the CLKPB6 (and a separate extension cord if you were considering the bundle) instead (see my post to identify the preferred version) – or better yet, the native Kaifeng KF-MSD-6A aka BizLine BIZ 390527/BIZPB0005 (available at Rexel).
  If you already have it, then make sure not to obstruct the button from popping out (or better yet, don't overload it in the first place )
• Metal covers for Clipsal Classic (C2000), Slimline (SC2000) and Eclipse (SL2000) outlets: While they maintain (just) enough clearance from the outlet slots, when a Clipsal 439(S) or similar rewireable plug is inserted, the cord wires (especially the neutral) can get very close to the outlet cover where looped over the plug's strain‑relief hooks. Since these clip‑on covers (unlike the Metal Plate Range's integral plates) have no earth connection, they really need to be double‑insulated from all live parts. If you're going to use metal covers on single‑pole Classic outlets, make sure they're wired correctly; for double‑pole outlets (other than the C2015D4 in an ordinary home/office installation) I would definitely stick with the plastic covers (and given the choice, the 2000 Series).
  When using those rewireable plugs in a metal-covered outlet, double-check that the core insulation is intact. When fitting those plugs, you may be able to double‑insulate the cores with heatshrink or similar (preferably clear to allow continued inspection of the inner insulation); although it might not fit if using upsized cords (1.5mm² in 10A or 2.5mm² in 15A), especially if also heavy‑duty. However heavy‑duty cords and 2.5mm² ordinary‑duty (along with H05RR‑F 1.5mm²) have thick enough inner insulation (0.8mm, as thick as the lone insulation layer on H03VH‑H and H03VH‑Y) to remain relatively safe anyway, and H05VV‑F 1.5mm² has 0.7mm (and the heatshrink may still fit). Clipsal 418(S) and similar side-entry plugs are unaffected.
  The light switch covers have no such issues, but you should always ensure the mounting screws are at a safe depth (>3mm below the grid plate).
  It is likely that metal covers for the competing HPM Excel range pose a similar concern, but I never cared for Excel anyway so haven't paid attention…
• DēLonghi DL2401TF: For the most part it's built fine (as expected from DēLonghi), but (due to some loophole in the standards) its little 500W fan heater (tacked on below the controls in a half-hearted attempt to mitigate the slow warm-up of oil-filled heaters) has no thermal fuse, just the resettable cut-out. At least that's a self-holding Klixon TH11CA081, mounted above the element so the heat will rise to it if the fan seizes; it does respond when covered, although takes over 9 seconds so the element has time to start glowing. The shroud around the element is also a thermoset plastic (phenol-formaldehyde or similar) if that helps. (In case you're wondering, I didn't buy it; it was left disused in Grandma's shed, then I took it.)
  This probably also applies to models DēLonghi sell in other countries with the same basic arrangement (although many of those add the fan heater's 500W to the total, rather than having to subtract it from the power of the main oil-filled heater); if you own any (or, for that matter, any competitors' heaters with equivalent arrangements; a discarded Dimplex model I saw uses a PTC element with no other thermal cut-out there), then please verify this for me.
• DēLonghi HS25F: DēLonghi's engineers must have had a bad day with this model, as in the unit I found the cord neutral wire gets uncomfortably close to the rear panel (like most DēLonghi non‑oil heaters it's meant to be Class II) and fan wires aren't much further from it. (The heater wires are double‑insulated, they just didn't pay enough attention at the cord and fan wires.) Also, the bottom grille slots are quite wide at 10mm (technically within the 12.5mm maximum for IP2X, but still wide enough to fit even my adult pinky fingers through) so it's not exactly child-safe (this is also one of the Dumbplex 3088T's flaws).
  It's also configured a bit oddly – the fan is connected in parallel with the low (1000W) element, not the medium (1400W) element as you'd logically expect with the twin rocker switch (and no separate fan switch); while semi-excused by the 1400W element being above the 1000W (so wire restraint and the fan wires' 105°C temperature rating rule against connecting the fan up top), you'd think they could have just put the 1400W element below (which the newer HSX3324FTS does, despite not even connecting the fan across an element in this manner).
  Another peculiarity is that the 1000W element has a 1mm² connecting wire, but the 1400W gets only 0.75mm² (still sufficient, but a strange combination).
  Fortunately, the current budget-model HCM2030 doesn't share these flaws (it has a nice tight bottom grille, and minimal slack in the cord wires).
• IEC 60320 C15 cordsets with PVC cable: The original IEC standard requires rubber cord for the C15 (up to 120°C at pins), but (certainly in Australia) I've seen plenty of molded C15 cordsets using PVC flex (H05V2V2‑F if you're marginally-less unlucky, but I've even seen them with plain H05VV‑F).
  So make sure not to use those with any appliance which actually gets hot enough to melt PVC…
  (They're OK on a kettle, but not on cooking/grilling appliances with exposed hot surfaces.) The only upside I can see to their existence is that there are far fewer counterfeit C15 than C13 cords, giving appliance makers an easy way to “key” them out (should still equip them with rubber cord though).
  If you have a melting-hot appliance, then you can equip it with a C16A (≤155°C at pins) to evade the problem.
• Jackson PTAMULTI2: Imprecise internal alignment puts the UK earth contact about 1mm behind the active and neutral.
  To be fair, it's no worse overall than the other adapters I've dealt with so far; but if you can wire up your own adapter, I still recommend that over this.
• Kambrook KFH20: I bought it for AU$10 at a Good Sammy expecting to just clean the lint out and be good to go, but it turned out that the heater frame was very brittle so I had to do away with that; at the cost of dropping by 400W, I transplanted in the TEFH211B's element assembly (below).
• Kambrook KFH200: This revised model fixed the heater frame (using the normal glass-reinforced polyamide rather than the previous brittle plastic), but when tearing it down I found that the internal “cup” connectors had aluminium crimp sleeves (at least the KFH20's one was copper).
  (Most of the KFH200s got rid of themselves due to their thermal cut-out tripping in the first few minutes of use, but beware if you have a working one.)
  Kambrook models in the years leading up to it may also be affected, and I also found such dubious connectors in the Sunbeam EC1300, cover earth connection of the Smeg TSF01 toaster (presumably also affecting the 4‑slice 2‑slot TSF02 and 4‑slot TSF03), Target Essentials TEFH211B (below) and Vornado 633 (although it's not such a big deal there given the low current) so there's clearly a spate of them around…
  Like many Kambrook heaters (and probably other appliances) from the 2000s (maybe even late '90s) until 2013 or so, the cord is also rather stiff and could crack when bent while cold. Subsequent Kambrook heaters (KFH6x0 and presumably KFH7x0) added fiberglass sleeves around the cup connectors (if my KFH660 ever breaks, I'll dissect it and update accordingly).
• Sansai DB-793 nightlight: In the one I found, the active pin's mounting post broke allowing the pin to push into the PCB; luckily it only touches pads it's connected to anyway. They also omitted the bleed resistor for the 220nF X2 capacitor, the rectifier is rather strange (4×1N4007 but not in a true bridge, such that the negative half-cycle is shunted out and they could as well have used 2 diodes), the transistor has its collector and emitter reversed, and the soldering is quite bad (with a cold joint on the LED anode). Seems par for Sansai's course though…
• Sunbeam EM6910 coffee machine: Got it to examine since the steamer's thermal fuse is open from age (even its epoxy end seal went brittle, too bad Sunbeam haven't used an ageless thermal fuse like in Black & Decker's clothing iron 15 years older). There are also three very hot resistors on the control PCBs (typical of what happens when n00bs select resistors by their rated power alone, not knowing that's just for the resistor itself in free air).
  And the fuse to the electronics input (which neither heater is powered through by the way) is a 10A slow-blow type! (The service manual actually said 3A so another n00b must have changed it in the meantime; I'll change it to a more sensible 2A slow-blow, HRC again of course.)
  The service manual itself isn't great (not that we expect much nowadays), and what's shady is that it stated the two heating elements as 1300W each (would be 2600W total, over the Australian 2400W limit); fortunately that seems to be just a typo going by their measured resistances (either that, or they fixed it in the meantime), otherwise I would recall the poor beast.
  Although obtaining the replacement thermal fuse (Microtemp G4A series; set for 240°C going by the service manual, since the original's markings were gone) was easy enough, I didn't have an easy way to connect it properly (soldering is right out) and the operation using this machine went under in the meantime; I don't drink coffee myself, so lost the motivation to continue the repair. The house is too crowded anyway, so off to e-waste it shall go…
  None of the parts I did buy for it are overly-specialized, so they can be used elsewhere as desired.
• Target Essentials TEFH211B fan heater: Got to take home from the local Rexel (the clerk said it “didn't work” but in reality it was just inappropriate for the large open space; they've since installed pendant radiators above the point-of-sale desk); it's not outright horrible (I'd grade this one a D), and even has a neat design feature or two (a shroud for airflow isolation between the heating elements and fan motor, and a 3-blade impeller which is better for harmonic stability than the 2 or other even numbers in many other units) although the fan rattled a lot due to low oil (easy enough to top up though). But what's a tad disturbing is that part of the fan shroud touches the feed terminal to the front heating element (at least the terminal itself doesn't seem to get dangerously hot). Otherwise the thermal protection responds OK (even though the bimetal switch and thermal fuse are positionally transposed compared to where reputable manufacturers put them, and the rear instead of front element is given priority), although the relatively low 112°C setting on the thermal fuse means it could age-out if used in a room 2000W can actually heat. As mentioned above, its neutral connector also has an aluminium sleeve.

Non-electrical items

• Cosmo PU dining chairs: These aren't here due to some structural defect, but because they often have a lot of welding slag rattling around in the frame, and it can scatter out of the holes when tilted (with the risk of shorting a nearby electrical device). If you have them, then shake out as much slag as you can in a safe location, and pick it up afterward (since it's steel you can use a magnet, but a commercial vacuum cleaner should also work safely enough).

This is where I put stuff that doesn't clearly fit into any other category.

Future Motion Onewheel family
Risk level: Severe
Faults: The Pint X and XR have pinched battery management harnesses due to bad design, and the negative battery lead tends to break off.
The GT is prone to “ghosting” (where you get off, but it runs away at full speed and can injure somebody else).

Australian Protective Electronics Power Guard 370B (it ought to be Non-Compliant, but who knows?)
Risk level: High
Fault: The dropper capacitor is only rated for 250VDC. It may survive 240VAC for a short time (given that the proof voltage is typically 1.6× the working voltage), but would eventually fail for sure. (As removed from my unit it was only slightly below-tolerance, but I have no way of telling how long it was used)
If it shorts out, the only thing (apart from the powered circuitry) in series is a 56Ω carbon film resistor (which tend to burst into flames when grossly overloaded).
Remedy: Replace with a proper X1 or X2 capacitor.

Dyson DC05 Motorhead version (not sure whether this better-fits the Unforeseen/Unanticipated or Low Standards section)
Faults: Besides the official recall due to the handle breaking open, it uses a slide switch which is a bad choice for a high-power, partially inductive load.
Remedy: As Cool386 did, it can be rewired with a standard push-button switch, as per the non-Motorhead version.

I can also vindicate the Dyson AM04/AM05 recall, as the controller (and fan) PSU has (or had at the time?) no fuse .
(Yet they did include HRC fuses in the connections to both heater columns, which is not a standard requirement although it may be a good idea in PTC heaters.
If there's a good thing to say about it, it's one of the few if any PTC fan heaters to have a half-decent airflow volume, albeit entirely laminar.)

Now, I won't go into obvious examples (cults, hate sites, tyrannical or war-ridden countries etc.) here; I'll even spare those forums with power-tripping staff.
This is for places which seem like they might have a chance (or at least present a dangerous temptation), but don't.

DeviantArt (and I'd presume other art sites to be much the same unless there's a good reason otherwise): Look at the art if it suits you, but don't expect anyone there to be compassionate when it matters, and don't take the “stamps” seriously. Even if a member was decent to start with, they will very likely rot with the community (and probably already have if they've been active for an appreciable time). It's basically all superficial and the “compassion” is counterfeit, holding up no better than do the counterfeit products mentioned above. If there's someone there you feel sorry for, they're most likely a lost cause (yes, I've tried).

I think the world already has plenty of art anyway, so the loss isn't that great; I prefer putting my effort into niche pursuits.

The “Reception Family” Wikis:
What could be a better way to harden groupthink? I don't even care that their views on a lot (or even a majority) of media are agreeable; the agreeable views are pretty much worthless anyway (as you can find most of them anywhere else), so the disagreeable views make it a strong net negative.