The Obsolescence Brief Subscribe
Home TeardownsUSB-to-serial bridges

CP2102 vs CH340, FT232RL and PL2303: the USB-UART bridge crosses compared (2026)

USB-UART Bridges · Issue 08
CP2102: NRND at DigiKey / full production to Mar 2029 (SiLabs) FT232RL cross: itself EOL PL2303: counterfeit-driver history (Code 10) CP2102N: successor, per AN976

Engineers reach for a CH340 as the low-cost CP2102 replacement, but it is a different vendor's part with its own driver and USB Vendor ID / Product ID (VID/PID), so a host built around the Silicon Labs CP2102 will not recognize it without software changes. A USB-to-UART bridge is defined by its vendor's driver and VID/PID, not by its pinout, so matching the footprint does not make the part interchangeable. The driver and the package are the two things a distributor cross-reference will not tell you. This page compares the CP2102 with the crosses engineers weigh, cp2102 vs ch340, cp2102 vs ft232 and cp2102 vs pl2303, and Silicon Labs' own CP2102N. None of the parts compared is a drop-in. These are suggested candidates to validate in your own design, not drop-in equivalents.

01

Status // The part in question

Lifecycle status

The CP2102 is not end-of-life. DigiKey flags the CP2102-GMR "Not For New Designs" ("Not recommended for new design, minimums may apply"), with stock still on hand. Silicon Labs publishes no NRND or EOL word and states, verbatim, that it "expects to continue full production of the CP2102 until at least March 2029." The two records describe a distributor Not-For-New-Designs flag on a part the manufacturer still ships: the CP2102 remains in production and orderable, with no last-time-buy or EOL notice from Silicon Labs.

The flag is a design-out signal, not a supply emergency: it warrants a planned migration rather than a panic buy. Any move off the CP2102, the CP2102N successor included, is a requalification and should be budgeted as one.

Current stock, broker availability and minimum-order terms are dated data for the email, not this page.

02

Candidates // Why these

The candidate set

The comparison covers the four crosses in the table plus two near-siblings, the Silicon Labs CP2104 and the WCH CH9102, each with a specific fit-catch to validate in your own design. For reference, the CP2102 is a USB 2.0 full-speed USB-to-UART bridge with an integrated clock and Silicon Labs' VCP (virtual COM port) driver, enumerating under VID 0x10C4 in a QFN-28 package.

CH340WCH · the most-searched crossREDESIGN
Package SOP-16
Driver WCH driver, VID 0x1A86
Clock External 12MHz crystal (CH340G)

Catch: cross-vendor, cross-shopped on cost. Cross-shopped on unit cost, but a different vendor driver and USB VID/PID (WCH 0x1A86), an SOP-16 package, and the CH340G needs an external 12MHz crystal.

CP2102NSilicon Labs · successor & migration pathACTIVE
Package QFN-28 / 24 / 20
Driver Silicon Labs VCP (same family)
Migration Porting guide AN976

Catch: a different part, not a drop-in. Silicon Labs publishes AN976 "Migrating from a CP2102 to a CP2102N," whose existence signals register and hardware changes that force requalification; it ships in QFN-28/24/20, so the smaller variants change package and pinout.

FT232RLFTDI · the hub cross, itself EOLEOL
Package SSOP-28
Driver FTDI VCP/D2XX, VID 0x0403
Live part FT232RNL

Catch: cross-vendor to an EOL part. cp2102 vs ft232 searches, but the FT232RL is end-of-life (LTB of 1 June 2024 passed), so the live FTDI part is the FT232RNL. FTDI's VCP/D2XX driver and VID 0x0403 replace Silicon Labs'. Full swap on the FT232RL cross page below.

PL2303Prolific · counterfeit-driver historyCAUTION
Package Varies by version
Driver Prolific driver, VID 0x067B
Live part PL-2303GC

Catch: a Code 10 counterfeit trap. cp2102 vs pl2303 searches. Cross-vendor to Prolific's driver and VID 0x067B; Prolific's supported line is the PL-2303GC, and counterfeit PL-2303HX (Chip Rev A) chips cause a Windows Code 10 fault. See the PL2303 cross page below.

CP2104Silicon Labs · the active siblingACTIVE
GPIO Four (CP2102 has none)
Lifecycle Full production to Nov 2035
Driver Silicon Labs VCP

Catch: treat it as a redesign. Four GPIO where the CP2102 has none, and full production to at least November 2035, but a different pinout, package and feature set.

CH9102WCH · the CP2102-alike substituteREDESIGN
Marketing "CP2102 compatible"
Driver WCH driver, own VID/PID
Sold on Modules as a substitute

Catch: marketing is not a guarantee. Sold on modules as a CP2102 substitute, but different silicon, a different WCH driver and USB VID/PID, its own package. "CP2102 compatible" marketing is not a form-fit-function guarantee.

03

Identity // The driver is the tell

Same function, not the same driver

Pinout is not the axis here. No two of these parts share one: CP2102 QFN-28, CH340G SOP-16, FT232RL SSOP-28, PL2303 varies by version, CP2102N QFN-28/24/20. Every cross is a board change.

A parametric search cannot return what decides the cross. A distributor cross-reference filters by function (USB-to-UART) and data rate and hands back all of these parts, but the host driver and the USB VID/PID are not filterable columns, and neither is package fit against your layout. Anything on the host keyed to Silicon Labs' VID 0x10C4 or its VCP driver stops matching the moment the silicon changes vendor: CH340 under WCH's VID, the FTDI parts under 0x0403, Prolific under 0x067B. Only the CP2102N stays in Silicon Labs' driver family, and even that is a requalification per AN976.

CP2102 QFN-28 reference · only the CP2102N stays in the Silicon Labs driver family · every cross-vendor part changes the driver and USB VID/PID

CP2102Silicon Labs · subject
NRND
Package
QFN-28
Driver
SiLabs VCP
USB VID
0x10C4
Lifecycle
NRND / in prod.
CP2102NSilicon Labs
ACTIVE
Package
QFN-28/24/20
Driver
SiLabs VCP (same)
USB VID
0x10C4 (own PID)
Lifecycle
Active
CH340GWCH
ACTIVE
Package
SOP-16
Driver
WCH
USB VID
0x1A86
Lifecycle
Active
FT232RLFTDI
EOL
Package
SSOP-28
Driver
FTDI VCP/D2XX
USB VID
0x0403
Lifecycle
EOL → RNL
PL2303Prolific
CAUTION
Package
Varies
Driver
Prolific
USB VID
0x067B
Lifecycle
HX EOL → GC
CP2104Silicon Labs
ACTIVE
Package
Own pinout
Driver
SiLabs VCP
USB VID
0x10C4
Lifecycle
To Nov 2035
Green stays in the Silicon Labs driver family Amber driver and USB identifiers change Red package / pinout changes (board redesign)
ONE STAYS IN THE FAMILYOnly the CP2102N keeps Silicon Labs' VCP driver and VID 0x10C4, and even that is a requalification per AN976, not a like-for-like part.
DRIVER CHANGES ON EVERY CROSS-VENDOR PARTCH340 under WCH's VID, the FTDI parts under 0x0403, Prolific under 0x067B, anything keyed to 0x10C4 stops matching the moment the silicon changes vendor.
NO TWO SHARE A PINOUTCP2102 QFN-28, CH340G SOP-16, FT232RL SSOP-28, PL2303 varies, CP2102N QFN-28/24/20, every cross is a board change.

Same function, not the same driver. The host driver and USB VID/PID change on every cross-vendor part, suggested candidates to validate in your own design, not drop-in equivalents.

04

Parametrics // Side by side

Comparison table

CP2102 (NRND at DigiKey)CH340GFT232RLPL2303 (EOL versions)CP2102N
VendorSilicon LabsWCHFTDIProlificSilicon Labs
LifecycleFull prod. to ≥Mar 2029 (SiLabs); NFND (DigiKey)ActiveEOL, LTB 1 Jun 2024 passedGC recommended; counterfeit HX (Rev A) → Code 10Active
Host driver / USB VID:PIDSiLabs VCP, 0x10C4WCH, 0x1A86FTDI VCP/D2XX, 0x0403Prolific, 0x067BSiLabs VCP, 0x10C4 (own PID)
PackageQFN-28SOP-16SSOP-28Varies by versionQFN-28 / 24 / 20
Internal clockIntegratedExternal 12MHz on CH340GIntegratedVaries by versionIntegrated
Same-vendor migration notesubjectnone (cross-vendor)FT232RNL is live partPL-2303GC is live partAN976 CP2102 → CP2102N

The load-bearing rows are the host driver / USB VID:PID and the package: no cross keeps both. The two trap rows are the FT232RL (itself end-of-life, so cross to the FT232RNL) and the PL2303 (cross to Prolific's supported PL-2303GC; counterfeit PL-2303HX chips cause a Code 10 fault). Lifecycle and electrical figures are vendor/distributor-reported, not independently audited.

05

FAQ // Frequently asked

Frequently asked questions

Is the CP2102 obsolete or still in production?

The CP2102 is still in production, not obsolete. Silicon Labs states it expects full production of the CP2102 until at least March 2029, while DigiKey separately flags the CP2102-GMR as "Not For New Designs." It remains buyable but carries that distributor flag, with no Silicon Labs EOL notice published for the part, which makes it a candidate to design out on a planned schedule.

What is the difference between the CP2102 and the CH340?

Both do the same serial job; what differs is the vendor driver and the USB VID/PID. The CP2102 uses Silicon Labs' VCP driver and Vendor ID 0x10C4, while the CH340 uses WCH's driver and Vendor ID 0x1A86, so host software keyed to Silicon Labs will not enumerate a CH340 without changes. The CH340G also needs an external 12MHz crystal where the CP2102 integrates its clock, and the CH340 is generally cross-shopped as the lower-cost, maker-grade option.

CP2102 vs FT232 vs PL2303 for USB-serial, which one?

All three are USB-to-UART bridges with the same serial function, so the choice comes down to the host driver, the USB VID/PID and each part's lifecycle status. Two of the three carry a lifecycle trap: the FT232RL is end-of-life (last-time-buy 1 June 2024 passed, cross to the FT232RNL) and the legacy PL-2303HX chips have a counterfeit-driver problem (Code 10), so Prolific points to the PL-2303GC, while the CP2102 itself is in production but flagged Not For New Designs at DigiKey. The deciding factors are the driver ecosystem and each part's lifecycle, with the package and host enumeration to validate in your own design.

Is the CP2102N a drop-in replacement for the CP2102?

No, the CP2102N is a different part that requires requalification, not a drop-in. Silicon Labs publishes AN976 "Migrating from a CP2102 to a CP2102N," and a dedicated porting guide signals register and hardware differences. The CP2102N also ships in QFN-28, QFN-24 and QFN-20 packages, so the smaller variants change the pinout and footprint versus the QFN-28 CP2102. It keeps Silicon Labs' driver family, which is its main advantage, but it enters the design as a new part.

What is the replacement or alternative for the CP2102 now that it is Not For New Designs?

There is no single drop-in; the realistic paths are the Silicon Labs CP2102N successor (per porting guide AN976) or the sibling CP2104, plus the cross-vendor bridges engineers compare, the WCH CH340, the FTDI FT232RNL and the Prolific PL2303GC. Each changes at least the package or the driver and USB VID/PID, so each is a suggested candidate to validate in your own system, not a guaranteed drop-in.

CH340 or CP2102 on ESP32 boards, which USB bridge?

On an ESP32 board the serial link is identical either way; what changes is the driver and the cost. The CH340 is the lower-cost bridge and the CP2102 is the Silicon Labs part, and each needs its own vendor VCP driver installed on the host. For a new design, the CP2102 is flagged Not For New Designs at DigiKey while Silicon Labs still ships it, so the CP2102N or the CH340 are the options for a part without that distributor flag.

What is the difference between the CP2102 and FTDI?

Each runs its own vendor stack: the CP2102 uses Silicon Labs' VCP driver and Vendor ID 0x10C4, while the FTDI parts use FTDI's VCP/D2XX driver and Vendor ID 0x0403, so host software keyed to one will not enumerate the other without changes. The FT232RL that many readers name is itself end-of-life, so the live FTDI cross is the FT232RNL.

Editorially independent and from public sources only; lifecycle, package, driver and rating data is vendor- or distributor-reported, not independently audited. Alternates are candidates to validate in your own design. Confirm every detail with the manufacturer before committing.

About the author

Hermann Njike is a systems obsolescence manager who leads the cross-functional response when critical components go end-of-life, and is the single point of contact for thousands of production parts.

Editorially independent and from public sources only; lifecycle, pin and rating data is vendor-reported, not independently audited. Alternates are candidates to validate in your own design. Confirm every detail with the manufacturer before committing.

Hermann Njike

Hermann Njike is a systems obsolescence manager who leads the cross-functional response when critical components go end-of-life, and is the single point of contact for thousands of production parts.

Related briefs
The Brief, weekly

One obsolete part, worked out in full.

Read a sample issue →