Switches

Pulling switches out of a mechanical keyboard safely

Push the switch straight down until the two retention clips on its top and bottom faces clear the plate, then lift it out — pulling at an angle or yanking the keycap still attached is what breaks…

A mechanical keyboard is one where each key sits on an individual switch with its own spring and its own metal contact leaf, rather than on a rubber dome sheet with a common membrane underneath. That matters here because the switch is a discrete part. On a hot-swap board it plugs into a socket soldered to the PCB and can be removed by hand. On a soldered board it cannot — not without a desoldering iron, flux, and a willingness to accept that you may lift a pad. The single most important step in this whole job happens before you touch anything: find out which one you own.

Turn the board over and look for a small badge, a model number, or a sticker. Failing that, pull one keycap and look at the switch housing.

Read the full story →

02

New

Updated 09:14 UTC
Winding a mechanical watch without overdoing it
Watches

Winding a mechanical watch without overdoing it

"How do you wind a mechanical watch without overdoing it?"…

13 SepRead →
What separates a good mechanical keyboard from a bad one
Buying

What separates a good mechanical keyboard from a bad one

The difference between a good mechanical keyboard and a bad one comes down to five things you can check before you buy: what the case and plate are made of, whether the switches are soldered…

12 SepRead →
Getting the winding routine right on a mechanical watch
Watches

Getting the winding routine right on a mechanical watch

Wind it once a day, at roughly the same time, and stop when you feel resistance — that is the routine I would pick over the "top it up whenever I remember" approach.

11 SepRead →
How to clean a mechanical keyboard safely
Maintenance

How to clean a mechanical keyboard safely

The mistake is pulling every keycap off at once and dropping them into a bowl of water with dish soap.

10 SepRead →
Cleaning a mechanical keyboard without damage
Maintenance

Cleaning a mechanical keyboard without damage

The safe way to clean a mechanical keyboard is to remove the keycaps, clean the exposed plate and switches dry first, use isopropyl alcohol only on the keycaps and the case, and never pour or spray…

9 SepRead →
What makes a keyboard mechanical
Switches

What makes a keyboard mechanical

A mechanical keyboard is one where each key sits on its own independent physical switch, and that switch closes the circuit itself when you press it.

8 SepRead →
Mechanical keyboard switches and how they feel
Switches

Mechanical keyboard switches and how they feel

The feel of a mechanical keyboard switch comes down to four measurable things: the force curve it presents as you press, where along the travel that force peaks and drops, whether it makes a noise…

7 SepRead →
Low profile mechanical keyboards explained
Ergonomics

Low profile mechanical keyboards explained

A low-profile mechanical keyboard does not have a shallower version of a standard switch inside it. That assumption causes most of the disappointment people hit after a week of typing.

6 SepRead →
Ergonomic mechanical keyboards that ease wrist strain
Ergonomics

Ergonomic mechanical keyboards that ease wrist strain

Ergonomic mechanical keyboards ease wrist strain by letting you change the two things a flat slab of plastic fixes in place: the angle of each half and the distance between your hands, so your…

5 SepRead →
Incorporating a retro classic look into your setup
Ergonomics

Incorporating a retro classic look into your setup

If you are aiming to incorporate a retro look into a modern tech setup, I would advise adopting the controlled analog infusion approach—using mechanical devices as focal points rather than trying…

4 SepRead →
Budget options for quality mechanical keyboards
Buying

Budget options for quality mechanical keyboards

For budget mechanical keyboards, building your own kit is vastly superior to buying a pre-assembled board off the shelf; it gives you control over every single component and ensures that the overall…

3 SepRead →
Choosing a quiet mechanical keyboard for your desk
Ergonomics

Choosing a quiet mechanical keyboard for your desk

To select a quiet mechanical keyboard, you must focus almost entirely on the switch type and proper dampening techniques, rather than solely on the keycaps or case material; switches that employ…

2 SepRead →
Show all %d

Wi‑Fi 6 is the wireless standard you should buy into now, but the number printed on the box tells you almost nothing about the speed you will get: the IEEE 802.11ax amendment supports a maximum throughput of 20 gigabits per second at the MAC data service access point, while a plain 2x2 client at 80 MHz tops out at a theoretical 1.2 Gbps and a single-antenna 20 MHz device at 143 Mbps. Everything else in consumer technology — the mesh node you add, the laptop you keep for six years, the printer on the desk — follows the same pattern. The specification describes a ceiling; your configuration, your environment and your willingness to replace the weak link decide what you actually experience. The useful skill is not memorising standards. It is knowing which constraint binds first, what removing it costs, and when leaving it alone is the right call.

Wi‑Fi 6: what the standard actually guarantees, and what it doesn't

Wi‑Fi 6 (IEEE 802.11ax) modifies the physical layer and MAC sublayer of IEEE 802.11 to support high-efficiency operation, and in its fullest mode it is capable of at least 20 Gbit/s maximum throughput at the MAC data service access point. That figure is the standard's own headroom, not a promise to your devices. Real rates scale with channel width and the number of spatial streams: 2x2 at 160 MHz reaches a theoretical 2.4 Gbps, 2x2 at 80 MHz reaches 1.2 Gbps, 2x2 at 20 MHz reaches 287 Mbps, and a 1x1 radio at 20 MHz reaches 143 Mbps. Same standard, same router, a spread of more than an order of magnitude.

That spread explains why upgrading the router alone often disappoints. If your laptop has a 2x2 radio and your apartment forces you onto narrower channels because the band is crowded, the access point's extra spatial streams sit unused. The efficiency features of 802.11ax — better behaviour when many clients share one channel — matter more in a dense household than raw peak rate, and they show up as steadier latency rather than a bigger number on a speed test. You notice the difference in a video call that stops stuttering when someone else starts a download, not in a benchmark.

The honest trade-off: Wi‑Fi 6 hardware costs more than the previous generation and delivers little on a small network with few devices and a clean spectrum. If you live somewhere with thin walls, one or two clients and no neighbour's network bleeding through, the money is better spent on a wired backhaul between two cheaper nodes. Buy Wi‑Fi 6 for density and for the years of client devices that will arrive with it, not for a headline rate you will never see.

Mesh, extenders and backhaul: the decision that determines whether your network is good

Fix the backhaul before you touch the radios. A mesh system is only as fast as the link between its nodes, and a wireless backhaul spends half its capacity repeating traffic that a cable would carry once. If you can run Ethernet between the router and the furthest node — even through a loft, a conduit or along skirting board — do it, and accept a cheaper two-piece kit over a three-piece one. The cable costs less than the third node and removes the bottleneck permanently.

Where cable is impossible, tri-band mesh with a dedicated backhaul radio is the next best choice, because the node-to-node traffic stops competing with your devices for the same channel. Dual-band mesh without a dedicated radio works in a small flat and falls apart in a long house, where the far node repeats a weak signal at half rate and you notice it as a room where downloads crawl while the rest of the house is fine. Powerline and MoCA adapters are the underrated middle option: they are unglamorous, they cap out well below the wireless peak, and they are stable in a way that no repeater is.

Extenders deserve their reputation. A repeater that halves throughput and creates a second network name is a patch, not a fix, and it is the right answer only when you need one dead room covered cheaply and nothing else. The cost of doing mesh properly is upfront money and, often, a cable run you did not plan for. The cost of doing it badly is a network that works everywhere except where you sit.

Choosing a laptop and phone on specs that survive their first year

Buy the memory and the storage, not the processor tier, because memory and storage are what you cannot change later. A machine with more RAM and a larger SSD will still feel current when a faster-clocked chip with less memory is swapping to disk. The processor generation printed on the sticker matters far less than the thermal design around it: a thin chassis that throttles under sustained load loses to a thicker one with the same chip, and you only find out when a long export or a compile takes twice as long as the review suggested.

The connectivity spec is where buyers get quietly future-proofed or quietly stranded. Check the Wi‑Fi radio's stream count and channel support, not just the "Wi‑Fi 6" badge, because a 1x1 20 MHz radio and a 2x2 160 MHz radio are both Wi‑Fi 6 and differ by a factor that shows up on every large transfer. Check the ports too. A laptop with no card reader and no full-size HDMI pushes you into a dock, and a dock is another thing to lose, power and troubleshoot.

The trade-off is honest and boring: paying for headroom you will not use for two years, versus paying again sooner. For phones the calculus tilts the other way — battery health, not compute, is what retires a device, and a battery replacement is almost always cheaper than a new handset. Replace the battery, keep the phone another cycle, and spend the difference on storage you will actually fill.

Additive manufacturing: where 3D printing earns its place and where it doesn't

Additive manufacturing is the process of joining materials to make parts from 3D model data, usually layer upon layer, as opposed to subtractive manufacturing (cutting material away) and formative methods (moulding and casting). "3D printing" is the non-technical synonym most people use, and a 3D printer is the machine that builds those geometries by successive addition of material according to the model. The distinction that matters in practice is not the vocabulary. It is that additive methods win when a part is complex, custom or low-volume, and lose when it is simple and needed in quantity.

For a hobbyist or a small workshop, the decision tree is short. If you need one bracket, a replacement knob, a jig or a prototype enclosure, printing it is faster than sourcing it and cheaper than a minimum order. If you need the same part in a structural application, printed plastic has anisotropic strength — it is weaker across the layer lines than along them — and the failure mode is a clean split along a layer you can see. Orient the part so the load runs with the layers, or design the joint so it is not relying on layer adhesion at all.

The cost people underestimate is the finishing and the failures. A print that warps off the bed, stringing between features, or a first layer that never adhered is time you do not get back, and the fix is usually mechanical — bed levelling, a slower first layer, an enclosure for materials that shrink — rather than a settings tweak. Where the part must be dimensionally tight, strong, or produced in volume, subtractive machining or forming is the right answer and printing is the wrong tool. The applications span engineering, medicine, education and architecture precisely because each of those fields needs custom geometry in small numbers; none of them needs ten thousand identical brackets.

Security and updates: the maintenance schedule nobody puts on the box

Patch on a schedule, not when something breaks, because the interval between a vulnerability being published and being exploited is short and your router, your NAS and your smart plugs are the devices most likely to go unpatched for years. Start with the router: firmware updates there fix remote-access flaws that no amount of strong Wi‑Fi passwords will cover. Then the devices that face the internet — cameras, doorbells, anything with a cloud account — and then the computers.

Two habits carry most of the value. First, change default credentials on every device that ships with an admin password, including the ones you think nobody would bother attacking. Second, put the cheap internet-connected gadgets on a separate network from the machines that hold your files, so a compromised plug cannot reach your laptop. Most consumer mesh systems offer a guest network that does this job adequately, and using it costs you nothing but a few minutes of setup.

The limit of this advice is that some devices simply stop receiving updates while they still work. A camera that has not seen firmware in years is a liability, and the honest choice is to replace it or to accept the risk with eyes open and isolate it. Multi-factor authentication on the accounts that matter — email, banking, the cloud account that controls your home — blocks more real attacks than any single piece of software you can install, and it costs you a few seconds per login.

Buying decisions in order, and what each one costs you

Sequence the spending: network first, then the devices that use it, then the peripherals, then the experiments. The network is the shared resource every other purchase depends on, and it is the cheapest to fix relative to the frustration it removes. Devices come next because they last longest and are hardest to change. Peripherals — docks, monitors, storage — are where you buy once and forget. The 3D printer and the smart-home gadgets sit last, because they are the purchases most likely to be abandoned and the least likely to affect daily work.

DecisionChoose this whenWhat it costs you
Wi‑Fi 6 router vs. older standardMany devices, dense neighbours, or a device refresh already plannedHigher price; little gain on a small, clean network
Wired backhaul vs. wireless meshAny run of cable is possibleInstallation effort; a cable you have to hide
Tri-band mesh vs. dual-bandLarge home, no cable, many clientsCost of the extra radio; a bigger footprint per node
More RAM/SSD vs. faster chipAlways, on a machine you keep for yearsMoney upfront; you notice the chip less
Battery replacement vs. new phoneDevice is otherwise healthyA repair appointment; you keep an older camera and screen
3D printing vs. buying or machiningOne-off, custom, or complex geometryFinishing time, failed prints, weaker across layers
Isolated guest network vs. flat networkAny cheap internet-connected gadgetA few minutes of setup; some devices dislike it

The pattern across every row is the same. The specification gives you a ceiling; the binding constraint — a narrow channel, a wireless backhaul, a soldered-down module, a layer line, an unpatched camera — sets the floor. Find the constraint before you spend, and spend on removing it rather than on a bigger number on a box. The trade-offs are real and none of them is free, but the ones you choose deliberately are the ones you can live with.

Why the same Wi‑Fi 6 label hides a factor-of-several gap between devices

The client radio, not the access point, usually sets your ceiling. A 2x2 Wi‑Fi 6 device at 20 MHz is capped at a theoretical 287 Mbps; the same standard in a 1x1 20 MHz configuration reaches 143 Mbps. Bump the width to 80 MHz on that 2x2 and the figure climbs to 1.2 Gbps, and at 160 MHz with 2 spatial streams it reaches 2.4 Gbps. None of those numbers requires a different router. They require a different client, or a wider channel, or both — and the client is the part you cannot upgrade.

This is why a new access point can feel like nothing changed. Open the adapter properties on the laptop and check the negotiated width and stream count before blaming the router; if it reports a single stream on a narrow channel, the 20 Gbit/s figure at the MAC data service access point is irrelevant to you. The fix is either a USB or M.2 adapter with more streams, or moving the laptop closer to the node so it negotiates the wider channel it already supports. The cost of the adapter is small; the cost of assuming the router was the problem is a returned router and the same speeds.

Reading a spec sheet without being fooled by the headline rate

Ignore the single largest number on any spec sheet and find the configuration it was measured in. The Wi‑Fi 6 peak of 20 Gbit/s at the MAC data service access point is a standard-level ceiling, achieved only with the widest channels, the most spatial streams and a clean spectrum. Every consumer box quotes a number derived from one ideal configuration; the useful question is which configuration, and whether your devices can reach it. A router advertised at the top of the range and a laptop stuck at 20 MHz will negotiate at the laptop's rate, not the router's.

The same habit applies to the 3D printer and the laptop. A printer's quoted speed assumes a layer height and material it may not print well at; a laptop's quoted clock assumes a thermal envelope it may not sustain. Ask what the number costs in real use — noise, heat, failed prints, throttling — rather than what it promises on the box. The trade-off is a few more minutes of reading before you buy, which is cheaper than the return shipping on the wrong device.