SShortSingh.
Back to feed

Why Factory Dimmer Specs Say Silicon Control and Tungsten, and What Buyers Should Check

0
·4 views

A factory dimmer datasheet phrase, "silicon control dimmable lamps, or tungsten," is a literal translation of phase-cut dimming terminology that many Western buyers find confusing. Phase-cut dimmers reduce power by chopping part of each AC cycle, using thyristor or TRIAC semiconductors, a principle long used in North American wall dimmers. Leading-edge designs suit resistive incandescent and halogen loads, while trailing-edge designs handle LED drivers more gently. Most dimmer complaints such as flicker, buzzing, or failure to dim stem from mismatches between the dimmer and the lamp rather than from defects in the dimmer itself. Buyers should check maximum load, minimum load, dimming range, and whether the unit is leading- or trailing-edge before purchase.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Matías Galiano describe cómo configura nodos ARM y fault domains en un cluster OKE de Oracle Cloud

En la cuarta entrega de su serie sobre Kubernetes en OCI a bajo costo, Matías Galiano, arquitecto de plataforma, explica la configuración del cluster OKE de producción de Tiendis. El texto detalla el cluster de tipo Basic, que no cobra por el control plane aunque no incluye SLA, frente al Enhanced de USD 0,10 por hora. También describe el uso de FLANNEL_OVERLAY como red de pods y la elección de Oracle Linux 9 para nodos ARM. Por último, el autor advierte sobre un problema con imágenes de nombre corto en cri-o, que rechaza referencias como nats:2.12 y exige nombres completos.

0
ProgrammingDEV Community ·

Cluster OKE en Oracle Cloud: red con una VCN, dos subnets y gateways sin cargo

Esta tercera entrega de la serie sobre Kubernetes en OCI al menor costo explica cómo diseñar la red de un cluster OKE. Se utiliza una única VCN con dos subnets: una pública que aloja solo el balanceador de entrada y el endpoint de la API, y una privada donde viven los nodos sin IP pública. El Internet Gateway gestiona el tráfico de la subnet pública, mientras que el NAT Gateway permite a los nodos salir a internet sin exponerse. Según el autor, el Service Gateway resultó el componente más valioso, ya que el tráfico hacia servicios de Oracle viaja por la red interna sin pasar por el NAT, evitando costos por hora y por volumen de datos.

0
ProgrammingDEV Community ·

Fixing virt-install on ARM64: Security Violation, dnsmasq Port 53, and the qemu:///session Trap

The author describes problems encountered when creating a first virtual machine on an ARM64 Raspberry Pi, despite using a correct virt-install command. The failures stemmed from libvirt, Debian, and shell defaults rather than the command itself, and each produced misleading errors. A key issue was that virsh defaulted to the per-user qemu:///session connection, which has no system networks or shared storage, instead of the system-wide qemu:///system. The article explains the difference between the two URIs and shows how to check and set the default. It warns that a VM created under the wrong URI will be invisible to the system libvirt and can appear to vanish, so explicitly specifying --connect qemu:///system is recommended.

0
ProgrammingDEV Community ·

A Timeout Does Not Prove an AI Agent Failed, Says Developer Guide

An AI assistant that submits a request and receives no reply cannot tell whether the underlying action succeeded, so simply retrying may duplicate an order, booking, or submission. The article advises separating the conversation from the action, identifying whether an operation changes state and whether a status query exists, and checking permissions before consequential steps. It recommends idempotency so that repeated attempts are recognised as the same intended action, with explicit rules for reused keys carrying different payloads. Testing should cover lost responses after successful processing, concurrent retries, genuine failures, and unavailable status services, with clear limits on when to stop or escalate. The author also suggests telling users honestly that completion cannot yet be confirmed, backed by a bounded retry policy and a reviewable trace.