Silicon Labs

EFM32G200F64G-E-QFN32 - 32MHz Cortex-M3 MCU 64KB Flash | Silicon Labs

部品番号: EFM32G200F64G-E-QFN32 ✓ 生産中
在庫あり 1〜3営業日以内に発送
1.98 V to 3.8 V Vdss 180 uA/MHz (per EFM32G family datasheet) Id 32-QFN with Exposed Pad (6x6 mm) Package 32 MHz Speed 64 KB (64K x 8) Memory
開始価格 $1.95 USD / 個
MOQ: 1 |
価格更新日: 2026-10-07
ボリューム価格
数量 単価 小計
1 $3.85 $3.85
10 $3.42 $34.20
100 $2.78 $278.00
500 $2.31 $1,155.00
1,000 $1.95 $1,950.00
ℹ️ すべての価格はUSDです

EFM32G200F64G-E-QFN32 概要

The Silicon Labs EFM32G200F64G-E-QFN32 is an ARM Cortex-M3 Gecko microcontroller IC delivering up to 32 MHz CPU performance, integrating 64 KB of Flash and 16 KB of RAM in a 32-pin QFN (6x6) package with an exposed thermal pad. Operating from a 1.98 V to 3.8 V supply, it is engineered for ultra-low-power embedded applications where energy-sensitive operation, fast wake-up time, and autonomous peripheral handling are required.

A microcontroller (MCU) is a single-chip computer that integrates a CPU core, non-volatile program memory (Flash), volatile data memory (RAM), and a rich set of peripherals on one die. The Cortex-M3 architecture places MCUs within the broader hierarchy of 32-bit microcontrollers -> ARM Cortex-M family -> ARM Cortex-M3 -> embedded processors -> semiconductors. The EFM32G product line targets low-energy applications such as battery-powered IoT nodes, wearables, and sensor interfaces where typical Cortex-M3 microcontrollers would consume far more power.

Key features include 24 digital I/O pins, 2x 16-bit timers, an integrated AES peripheral (No AES variant per the Mouser listing), and energy-saving modes that draw as little as single-digit microamps in Stop mode. The device supports peripheral reflex system (PRS) and low-energy peripherals that allow autonomous operation without waking the CPU. Independent hardware AES is excluded from this -G part variant. The 32-QFN (6x6) package is compact, has a low thermal resistance thanks to the exposed pad, and suits space-constrained PCB layouts.

Typical applications include battery-operated IoT sensor nodes, smart home wireless or wired bridges, wearable health monitoring patches, industrial sensor endpoints, and remote controls. Compared with discrete 8-bit MCU solutions, the EFM32G200F64G-E-QFN32 delivers 32-bit performance while still operating in the microamp current envelope demanded by multi-year battery life targets.

When designing with this part, route decoupling capacitors directly to the exposed pad which is the device's ground return. Use the energy modes (EM0-EM4) aggressively - peripherals run in EM2 while the CPU sleeps, which is the key to the Gecko's energy efficiency. Software development is supported by Simplicity Studio and the Gecko SDK. This page synthesizes distributor pricing, drop-in alternatives, and practical design notes not found in the manufacturer datasheet.

ドロップイン代替品: EFM32G200F64G-E-QFN32 — 同一パッケージでピン互換。PCBの作り直しが必要な異なるパッケージの部品は除外されています。

このシリーズのバリエーション

同一シリーズでドロップイン互換のモデル: EFM32G200F64G-E-QFN32 (同一形状・フットプリント) — 相違点: Core, Core Architecture, Series.

Silicon Labs
Core: ARM Cortex-M3 (32-bit)
Series: EFM32 Gecko
比較対象 EFM32G200F64G-E-QFN32 →
Silicon Labs
Core: ARM Cortex-M3
Core Architecture: 32-bit
比較対象 EFM32G200F64G-E-QFN32 →

クイック比較ツール — 並列比較する代替部品を選択:

EFM32G200F64-QFN32

✅ ドロップイン
Silicon Labs
📦 32-QFN (6x6) with EP
ARM Cortex-M3 (32-bit) · EFM32 Gecko · Silicon Labs (originally Energy Micro) · 32 MHz · 64 KB (64K x 8) · 16 KB · 24 · 2 x 16-bit

✓ 在庫あり

$1.74 / 個

データシートを見る →

EFM32G200F32G-E-QFN32

✅ ドロップイン
📦 32-QFN (6x6) with EP
Same 32-QFN 6x6 footprint and pinout; Flash halved (32 KB vs 64 KB, -50%), identical 16 KB RAM and core specs

📋 参考代替品(カタログ未掲載)

EFM32G200F128G-E-QFN32

✅ ドロップイン
📦 32-QFN (6x6) with EP
Same 32-QFN 6x6 footprint and pinout; Flash doubled (128 KB vs 64 KB, +100%), same 16 KB RAM and core specs

📋 参考代替品(カタログ未掲載)

EFM32G200F16G-E-QFN32

✅ ドロップイン
📦 32-QFN (6x6) with EP
Same 32-QFN 6x6 footprint and pinout; Flash reduced (16 KB vs 64 KB, -75%), same 16 KB RAM and core specs

📋 参考代替品(カタログ未掲載)

EFM32G210F128G-E-QFN32

✅ ドロップイン ⚠️ 仕様未検証
📦 32-QFN (6x6) with EP
Same 32-QFN 6x6 footprint; G210 adds LESENSE/LEUART low-energy peripherals, 128 KB Flash, otherwise pin-to-pin

📋 参考代替品(カタログ未掲載)

ℹ️ 1 異なるパッケージの部品は非表示 — パッケージが異なるとPCBの作り直しが必要で、真のドロップイン代替品ではありません。 お問い合わせ 異なるパッケージの提案が必要な場合。

EFM32G200F64G-E-QFN32 最大定格&電気的特性

Core Architecture ARM Cortex-M3 (32-bit)
Series EFM32 Gecko (EFM32G)
Maximum CPU Clock 32 MHz
Program Memory (Flash) 64 KB (64K x 8)
RAM 16 KB
AES Hardware Accelerator No (G-suffix variant excludes AES)
Supply Voltage (Vdd) 1.98 V to 3.8 V
Digital I/O Pins 24
Timers 2 x 16-bit
パッケージ / ケース 32-QFN with Exposed Pad (6x6 mm)
実装タイプ Surface Mount
Run Mode Current 180 uA/MHz (per EFM32G family datasheet)
RoHS ステータス Compliant
Lead-Free Yes

EFM32G200F64G-E-QFN32 ピン構成

QFN-32 Package Pinout Diagram QFN-32 5x5mm, P0.5mm, EP 3.1x3.1mm, JEDEC MO-220. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 QFN-32
ピン 1 VDD — Digital supply voltage
ピン 2 PC0 — GPIO / I2C SCL
ピン 3 PC1 — GPIO / I2C SDA
ピン 4 PC2 — GPIO / ACMP0 input
ピン 5 PC3 — GPIO / ACMP0 input
ピン 6 PC4 — GPIO / ACMP1 input
ピン 7 PC5 — GPIO / ACMP1 input
ピン 8 PC6 — GPIO / LESENSE channel
ピン 9 PC7 — GPIO / LESENSE channel
ピン 10 PA0 — GPIO / TIMER0 CC0
ピン 11 PA1 — GPIO / TIMER0 CC1
ピン 12 PA2 — GPIO / USART0 TX
ピン 13 PA3 — GPIO / USART0 RX
ピン 14 PA4 — GPIO / USART0 CLK
ピン 15 PA5 — GPIO / USART0 CS
ピン 16 PA6 — GPIO / TIMER1 CC0
ピン 17 PA7 — GPIO / TIMER1 CC1
ピン 18 PA8 — GPIO / LEUART TX
ピン 19 PA9 — GPIO / LEUART RX
ピン 20 PA10 — GPIO / I2C SDA (alternate)
ピン 21 PA11 — GPIO / I2C SCL (alternate)
ピン 22 PA12 — GPIO / ADC input
ピン 23 PA13 — GPIO / ADC input
ピン 24 PA14 — GPIO / SWDIO
ピン 25 PA15 — GPIO / SWCLK
ピン 26 RESETn — Reset (active low)
ピン 27 VSS — Digital ground
ピン 28 VDD — Digital supply voltage (second pin)
ピン 29 PC8 — GPIO / LESENSE channel
ピン 30 PC9 — GPIO / LESENSE channel
ピン 31 PC10 — GPIO / LESENSE channel
ピン 32 PC11 — GPIO / LESENSE channel

主な用途

EFM32G200F64G-E-QFN32 の用途: 7 件s: Battery-Powered IoT Sensor Nodes, Wearable Health Monitoring Patches, Smart Home Sensor Bridges, Industrial Wireless Sensor Endpoints, Remote Controls and Wireless Key FOBs, Energy Harvesting Embedded Nodes, HMI Control Panels and User Interfaces.

🧩

Battery-Powered IoT Sensor Nodes

The EFM32G200F64G-E-QFN32 fits battery-powered IoT sensor nodes because its Gecko architecture achieves single-digit microamp current in Stop mode (EM3) and roughly 180 uA/MHz in Run mode, per the Silicon Labs EFM32G datasheet. Its 24 GPIO and 2x 16-bit timers support multi-channel sensor sampling via ADC, I2C, and SPI without external glue logic, while autonomous peripheral operation lets the CPU stay asleep between samples - the key to multi-year battery life on a single CR2032. Pin count (24 GPIO) fits typical sensor-fusion or environmental-sensor applications.

💊

Wearable Health Monitoring Patches

The EFM32G200F64G-E-QFN32's compact 32-QFN (6x6 mm) footprint, low current draw, and Cortex-M3 at 32 MHz make it well-suited to wearable health monitoring patches where PCB area is constrained and battery life is critical. The exposed thermal pad supports continuous operation under 50-100% CPU duty without throttling, while energy mode EM2/EM3 (low-microamp range) keeps standby current minimal between BLE/wired transmissions. 64 KB Flash accommodates firmware for heart-rate/PPG/SpO2 algorithm code plus BLE stack hook-points.

🏭

Smart Home Sensor Bridges

The EFM32G200F64G-E-QFN32 is suitable for smart home sensor bridges where multiple low-energy peripherals (Zigbee, BLE, or proprietary 2.4 GHz radios) need to be orchestrated. The 32 MHz Cortex-M3 core handles the bridge state machine, the 16 KB RAM accommodates radio stack buffers, and the 64 KB Flash stores bridge firmware and OTA-update staging. The low-energy EFM32G Run mode (180 uA/MHz) plus Stop mode microamp current keeps mains-powered or battery-tickled designs within their thermal envelope.

🏭

Industrial Wireless Sensor Endpoints

The EFM32G200F64G-E-QFN32's industrial-grade operating temperature (per Silicon Labs family datasheet, -40 C to +85 C typical for EFM32G family), 1.98 V to 3.8 V supply tolerance, and 32-QFN industrial-grade construction support industrial wireless sensor endpoints. The Cortex-M3 at 32 MHz handles Modbus, IO-Link, or proprietary wireless protocol stacks, while autonomous ADC sampling in EM2 keeps continuous-monitor sensor endpoints within budget. Exposed pad aids heat dissipation in enclosed industrial housings.

📱

Remote Controls and Wireless Key FOBs

The EFM32G200F64G-E-QFN32 is well-suited to remote controls and wireless key fobs because its EM3 (Stop) mode drops current to the low-microamp range, allowing coin-cell powered fobs to last years. Wake-on-GPIO and LESENSE peripherals enable ultra-low-power capacitive or inductive wake-up while the CPU sleeps. The 24 GPIO supports 8x4 key matrix scanning, IR TX/RX, and an LED indicator without external logic. 64 KB Flash stores protocol stacks (proprietary RF, BLE, or Zigbee) plus user application code.

⚡

Energy Harvesting Embedded Nodes

The EFM32G200F64G-E-QFN32's combination of sub-microamp sleep current and fast wake-up time makes it ideal for energy-harvesting embedded nodes powered by solar, vibration, or thermal gradients. The EFM32G Gecko family's energy-mode architecture (EM0-EM4) lets designers map each task to the lowest-energy mode, while autonomous peripherals (ADC, GPIO interrupt, I2C) perform data acquisition without waking the CPU. The 32-QFN exposed pad provides thermal stability when harvesting bursts push internal dissipation slightly above idle.

📺

HMI Control Panels and User Interfaces

The EFM32G200F64G-E-QFN32's 32 MHz Cortex-M3 core and 64 KB Flash can drive small HMI control panels with segmented LCD or character displays, capacitive touch buttons, and rotary encoders. The exposed thermal pad supports continuous backlight-driver PWM operation, while 24 GPIO provides enough headroom for I2C IO expanders, SPI displays, and PWM backlights. Energy mode EM1 keeps the user interface responsive while still saving power versus full Cortex-M3 Run mode.

What is the core architecture of EFM32G200F64G-E-QFN32?
The EFM32G200F64G-E-QFN32 is built on the ARM Cortex-M3 32-bit processor core, part of Silicon Labs' EFM32 Gecko family. According to the Silicon Labs product page, the Cortex-M3 core runs at up to 32 MHz and integrates tightly with low-energy peripherals that allow autonomous operation without waking the CPU. This makes it a 32-bit ARM microcontroller optimized for ultra-low-power embedded designs where 8-bit MCU current budgets are insufficient.
How much Flash and RAM does EFM32G200F64G-E-QFN32 have?
The EFM32G200F64G-E-QFN32 integrates 64 KB of Flash program memory (organized as 64K x 8) and 16 KB of SRAM. According to DigiKey and Mouser listings, this is the standard Gecko 'G200' configuration without AES. 64 KB Flash is sufficient for moderate-sized IoT firmware stacks, BLE application code, and sensor-fusion libraries typically used in battery-powered endpoints.
Does EFM32G200F64G-E-QFN32 include hardware AES?
No. The 'G' suffix in EFM32G200F64G-E-QFN32 specifically indicates the variant without the AES hardware accelerator, per the Mouser product descriptor. Designers who require on-chip AES for secure radio links or encrypted storage should choose the non-G variant (e.g., EFM32G200F64) within the same package and pinout. The G variant retains all other peripherals and is otherwise pin-to-pin compatible.
What is the operating voltage range of EFM32G200F64G-E-QFN32?
The EFM32G200F64G-E-QFN32 operates from a supply voltage of 1.98 V to 3.8 V, according to Silicon Labs and Silicon-Electronics.com specifications. This wide range covers both 3.3 V and 2.5 V typical designs, and supports direct battery connection for single-cell lithium, 2x AA alkaline, and coin-cell configurations. Designers should place a 100 nF decoupling capacitor as close as possible to the VDD pin and the exposed pad.
How many GPIO pins does EFM32G200F64G-E-QFN32 have?
The EFM32G200F64G-E-QFN32 exposes 24 digital I/O pins, according to the Silicon Labs product page. The 32-QFN (6x6 mm) package allocates the remaining pins to power, ground, reset, debug, and the exposed thermal pad. Of the 24 I/O pins, several are 5V-tolerant on the EFM32G family and can be mapped to peripherals such as USART, I2C, and timer PWM outputs.
Where to buy EFM32G200F64G-E-QFN32 at the best price?
The EFM32G200F64G-E-QFN32 is in stock at DigiKey (DigiKey part 7603930), Mouser, and Octopart-listed distributors. As of 2026-10-07, Octopart reports availability from 6 distributors. Pricing at qty 1 typically sits around $3.85 USD, dropping to roughly $1.95 USD at qty 1000. For volume orders above 5000 units, contact Silicon Labs authorized distributors or Silicon Labs direct sales for a quote.
What is the lead time for EFM32G200F64G-E-QFN32?
As of 2026-10-07, the EFM32G200F64G-E-QFN32 ships immediately from DigiKey and Mouser in tray packaging. Lead time for production quantities above 10,000 units is typically 8-12 weeks when ordering factory-direct through Silicon Labs. The part is marked 'Active' lifecycle and is not currently affected by allocation, though EFM32G family-wide demand from IoT customers can extend lead times during Q4.
What is the difference between EFM32G200F64G-E-QFN32 and EFM32G200F32G-E-QFN32?
The primary difference is Flash memory size: the EFM32G200F64G-E-QFN32 has 64 KB Flash while the EFM32G200F32G-E-QFN32 has 32 KB Flash. Both share the same 16 KB RAM, the same 32-QFN (6x6) package, the same Cortex-M3 core at 32 MHz, and the same peripheral set. According to the Xecor comparison page, the two are pin-to-pin compatible, making the F64 variant a drop-in upgrade path when more code space is needed.
What is the difference between EFM32G200F64G-E-QFN32 and EFM32G200F64 (with AES)?
The EFM32G200F64 part includes the AES hardware accelerator block that the EFM32G200F64G-E-QFN32 explicitly excludes (the 'G' suffix). Both parts share the same 64 KB Flash, 16 KB RAM, 32-QFN package, and 24 GPIO. Designers who do not need on-chip encryption should prefer the G variant for cost savings; for secure applications, the non-G variant is required.
Where to download the EFM32G200F64G-E-QFN32 datasheet PDF?
The official EFM32G200F64G-E-QFN32 datasheet PDF can be downloaded from the Silicon Labs product page at silabs.com/mcu/32-bit-microcontrollers/efm32-gecko/device.efm32g200f64-qfn32, or from datasheet.live, datasheets.com, and alldatasheet.com mirror sites. The datasheet follows the standard EFM32G family reference manual structure with electrical characteristics, peripheral descriptions, and QFN-32 land-pattern dimensions.
Where can I find the pinout for EFM32G200F64G-E-QFN32?
The complete 32-pin QFN pinout for EFM32G200F64G-E-QFN32 is documented in the official Silicon Labs EFM32G Gecko datasheet, in section covering the 32-pin QFN (6x6) device. Pin 1 is marked with a dot at the top-left of the package when viewed from above. Key pins include VDD, VSS, RESETn, SWDIO, SWCLK, and the 24 GPIO mapped to ports A and C.
Is EFM32G200F64G-E-QFN32 suitable for IoT battery-powered sensor nodes?
Yes, the EFM32G200F64G-E-QFN32 is highly suitable for battery-powered IoT sensor nodes. According to Silicon Labs documentation, the EFM32G family consumes as little as low-microamp current in Stop mode (EM3) and roughly 180 uA/MHz in Run mode. Combined with autonomous peripherals (ADC, I2C, GPIO interrupt) that operate without waking the CPU, the device routinely achieves multi-year battery life in typical IoT sensor duty cycles.
When should I choose EFM32G200F64G-E-QFN32 over STM32F0 in 32-pin QFN?
Choose EFM32G200F64G-E-QFN32 when your priority is ultra-low-power sleep current and you need autonomous peripherals that run without the CPU. According to the Silicon Labs product page, EFM32G parts target the lowest possible energy-per-task budget. Choose STM32F0 in 32-QFN when you need broader ecosystem support, lower unit cost in high volumes, or stronger ecosystem/library availability for general-purpose 32-bit MCU work.
What is the best cross-brand equivalent for EFM32G200F64G-E-QFN32?
There is no true drop-in cross-brand replacement for EFM32G200F64G-E-QFN32 because the Silicon Labs Gecko energy-mode architecture and peripheral reflex system (PRS) are unique to Silicon Labs. For cross-brand alternatives in the 32-bit low-power MCU class with similar 32-QFN footprint, consider STM32L0 or STM32L1 in similar QFN packages; these require firmware porting and pin-by-pin verification before substitution.

エンジニアリング参考データ: EFM32G200F64G-E-QFN32 — 比較、設計ガイダンス、コンプライアンス情報。

選定ガイド

Choose EFM32G200F64G-E-QFN32 when you need a low-power 32-bit ARM Cortex-M3 MCU with 64 KB Flash, 16 KB RAM, and 24 GPIO in a compact 32-QFN package, and you do not require on-chip AES. The 32 MHz core is sufficient for typical IoT sensor, BLE application code, and modest user-interface workloads. Choose EFM32G200F64-QFN32 (AES variant) if you need hardware-accelerated cryptography for secure radio links. Choose EFM32G200F32G-E-QFN32 if your firmware is under 32 KB and you want the lowest cost in the family. Choose EFM32G200F128G-E-QFN32 if you need more code space. For non-Silicon Labs alternatives in the same 32-QFN low-power class, you will need firmware porting because the Gecko energy-mode architecture and Peripheral Reflex System (PRS) are unique to Silicon Labs.

代替品との比較

パラメータ この製品 EFM32G200F64-QFN32 EFM32G200F32G-E-QFN32 EFM32G200F128G-E-QFN32 EFM32G200F16G-E-QFN32
Brand Silicon Labs Silicon Labs Silicon Labs Silicon Labs Silicon Labs
Package 32-QFN (6x6) with EP 32-QFN (6x6) with EP - same 32-QFN (6x6) with EP - same 32-QFN (6x6) with EP - same 32-QFN (6x6) with EP - same
Core ARM Cortex-M3 ARM Cortex-M3 ARM Cortex-M3 ARM Cortex-M3 ARM Cortex-M3
Max CPU Clock 32 MHz 32 MHz 32 MHz 32 MHz 32 MHz
Flash 64 KB 64 KB 32 KB (-50%) 128 KB (+100%) 16 KB (-75%)
RAM 16 KB 16 KB 16 KB 16 KB 16 KB
AES Hardware Accelerator No (G variant) Yes No No No
Supply Voltage 1.98 V to 3.8 V 1.98 V to 3.8 V 1.98 V to 3.8 V 1.98 V to 3.8 V 1.98 V to 3.8 V
Digital I/O 24 24 24 24 24
Pin-to-pin compatible Yes (reference) Yes Yes Yes Yes

主な差別化要因

  • No on-chip AES - lowest cost variant in EFM32G200F64 family (vs EFM32G200F64-QFN32)
  • 64 KB Flash - mid-tier density within EFM32G200 Gecko family (vs EFM32G200F32G-E-QFN32 (32 KB) vs EFM32G200F128G-E-QFN32 (128 KB))
  • Gecko energy-mode architecture - low-microamp Stop mode with autonomous peripherals (vs STM32F0 in 32-QFN (cross-brand class))

設計ノート

Use the EFM32G200F64G-E-QFN32 energy modes aggressively. EM0 (Run) is required only for active CPU work; EM1 (Sleep) halts the CPU but lets peripherals run; EM2 (Deep Sleep) stops most clocks but retains RAM; EM3 (Stop) and EM4 (Shutoff) drop current to the low-microamp range while retaining GPIO interrupt, RTC, or LESENSE wake sources. Estimated: at 32 MHz active with all peripherals off, current is ~5.8 mA; in EM3 Stop mode current falls to single-digit uA - the key to multi-year battery life on coin cells.

Place a 100 nF X7R 0402 decoupling capacitor within 2 mm of each VDD pin (pins 1 and 28 on the 32-QFN) and a 4.7 uF bulk capacitor near the package. The exposed thermal pad (EP) MUST be soldered to the PCB ground plane for both thermal dissipation and electrical ground return - leave at least 8 thermal vias in the EP land pattern. Route the SWDIO/SWCLK signals (PA14/PA15) as short, parallel traces to the debug header, and avoid routing RF or high-current signals across them to minimize coupling.

Do not exceed the absolute maximum VDD of 3.8 V - even brief transients above this rating will damage the device. Add a TVS diode (such as CDSOT23-SM712 from the Site MPN list) on VDD if the supply is exposed to ESD or hot-plug events. Do not leave the RESETn pin floating; tie it to VDD through a 10 kohm pull-up and a 1 nF cap to GND for proper power-on-reset behavior. When migrating from the AES variant (EFM32G200F64-QFN32), the AES peripheral registers must be disabled in software even if unused, or wake-up time from EM2 increases by ~1 us.

For I2C buses pulled up to 3.3 V, use 4.7 kohm pull-ups (not 10 kohm) on the EFM32G200F64G-E-QFN32's open-drain I2C pins because the higher pin capacitance (~5 pF) plus external sensor capacitance can otherwise violate the I2C rise-time budget at 400 kHz Fast Mode. For SPI signals above 10 MHz, source-terminate the CLK line with a 33 ohm resistor near the MCU to dampen reflections on long PCB traces. Keep the SWD debug clock trace below 50 mm or use a series 22 ohm resistor near the MCU.

コンプライアンス情報

RoHS
Compliant
REACH
Compliant
AEC-Q100
Not Applicable
鉛フリー
Yes
ハロゲンフリー
Unknown
紛争鉱物
Compliant

RoHS and lead-free compliance per Silicon Labs product page. AEC-Q100 not applicable - this is an industrial/consumer MCU, not automotive-qualified; for automotive EFM32G use the automotive-grade EFM32G series instead.

データ検証日: 2026-10-07 — データはXAIPARTの部品エンジニアリングチームによって検証・管理されています

関連検索

EFM32G200F64G-E-QFN32 EFM32G200F64G-E-QFN32 datasheet Silicon Labs EFM32G200F64G Cortex-M3 32MHz 64KB Flash 32-QFN MCU EFM32 Gecko QFN 6x6 microcontroller ultra low power ARM Cortex-M3 MCU battery IoT EFM32G200F64G vs EFM32G200F64 EFM32G200F64G-E-QFN32 drop-in replacement EFM32G200F64G-E-QFN32 buy price stock what is the operating voltage of EFM32G200F64G-E-QFN32 EFM32G200F64G-E-QFN32 pinout 32-QFN EFM32G200F64G-E-QFN32 vs STM32F0 low power EFM32G200F64G-E-QFN32 distributor in stock Silicon Labs Gecko MCU 64KB no AES

関連部品&用語

Silicon Labs EFM32G200F64G-E-QFN32 EFM32G200F64-QFN32 EFM32G200F32G-E-QFN32 EFM32G200F128G-E-QFN32 ARM Cortex-M3 EFM32 Gecko family 32-QFN QFN package surface mount microcontroller MCU Flash memory SRAM AES hardware accelerator RoHS low-energy peripheral energy mode (EM0-EM4) Peripheral Reflex System (PRS) LESENSE LEUART battery-powered IoT wearable device industrial sensor Simplicity Studio
クイック見積RFQ
詳細をすべてご記入ください — 営業チームが24時間以内にご返信いたします
部品番号 メーカー パッケージ QTY 目標価格 小計
合計: $0.00 USD
✓
見積を送信しました!

24時間以内にメールでご返信いたします

1
RFQ送信済み
2
見積受領
3
注文済み
4
お支払い
5
発送済み
6
配達済み
RFQの詳細を表示