TL;DR: on LQFP48, the CH32V203C8T6 uses the same pin names in the same positions as the STM32F103C8T6, so a board can often take either. It is not the same chip. The core is RISC-V instead of Arm Cortex-M3, so firmware must be rebuilt and a different debug probe is needed. The minimum supply is 2.4 V instead of 2.0 V. And the WCH part adds peripherals the ST part lacks. Check the five points at the end before you swap.
Sources: STMicroelectronics DS5319 Rev 20 (July 2025), the STM32F103x8/xB datasheet, and the WCH CH32V203 Datasheet V3.0. Both are the manufacturers' own PDFs. We compared the two LQFP48 pin tables row by row: ST Table 5 (pages 28 to 33) and WCH Table 3-1-1 (pages 21 to 24).
Side by side
| STM32F103C8T6 | CH32V203C8T6 | |
|---|---|---|
| CPU core | Arm Cortex-M3 | QingKe V4B, RISC-V (IMAC) |
| Max CPU clock | 72 MHz | 144 MHz |
| Flash / SRAM (C8) | 64 KB / 20 KB | 64 KB zero-wait / 20 KB |
| Operating VDD | 2.0 to 3.6 V | 2.4 to 3.6 V |
| Absolute max VDD | 4.0 V | 4.0 V |
| USB needs VDD | 3.0 to 3.6 V | 3.0 to 3.6 V |
| Ambient, suffix 6 | -40 to 85 °C | -40 to 85 °C |
| GPIOs (48-pin) | 37 | 37 |
| USART | 3 | 4 |
| SPI / I2C / CAN | 2 / 2 / 1 | 2 / 2 / 1 |
| USB (full speed) | 1 device | USBD plus a host/device USBFS (on PB6/PB7) |
| ADC | 2 x 12-bit, 10 channels | 12-bit, 10 channels, 2 units |
| Op-amps | none | 2 |
| Debug | SWD and JTAG | 2-wire SDI on SWDIO (PA13) / SWCLK (PA14) |
ST DS5319 Rev 20: Table 2 (page 10), Table 6 (page 37), Table 9 (page 38), Table 45 (USB, page 73). WCH CH32V203 V3.0: Table 2-1 (page 5), Table 4-1 (pages 33-34), Table 4-2 (page 34), Section 2.5.23 (page 18). WCH notes that the C8's 64 KB is its zero-wait flash area and that a further non-zero-wait area exists (Table 2-1, note 1).
The pin map: the same, with two exceptions
Going through all 48 pins, every position carries the same name on both parts: VBAT on pin 1, OSC_IN/OSC_OUT on 5 and 6, NRST on 7, VSSA/VDDA on 8 and 9, PA0 to PA15 and PB0 to PB15 in the same places, BOOT0 on 44, and the VSS/VDD pairs on 23/24, 35/36 and 47/48. PB2 is BOOT1 on both. Two details differ:
- Pins 24 and 48 are
VDD_1andVDD_3on ST, andVDD_IO_1andVDD_IO_3on WCH. WCH's block diagram gives the I/O domain its own 2.4 to 3.6 V range. On a board that ties all supply pins to one 3.3 V rail, this makes no practical difference. - PB5 (pin 41) is marked 5 V-tolerant (FT) on WCH and not on ST. The other FT pins match: PB2, PB10 to PB15, PA8 to PA15, PB3, PB4 and PB6 to PB9 on both.
A matching pin map is what lets the same footprint and routing work. It does not make the alternate functions identical. WCH lists remaps the ST part does not have (for example its fourth USART and the USBFS signals on PB6 and PB7). Check each peripheral you use against the WCH pin table, not just the pin name.
Firmware: a rebuild, not a re-flash
The STM32F103 runs an Arm Cortex-M3; the CH32V203 runs WCH's QingKe V4B RISC-V core, which the datasheet says supports the IMAC subset of the RISC-V instruction set. A binary built for one will not run on the other. The code has to be compiled for the other architecture, with the other vendor's startup files and peripheral library. WCH's peripheral registers are documented in its separate reference manual (CH32FV2x_V3xRM).
Debugging changes too. ST offers SWD and JTAG. WCH uses a 2-wire serial debug interface (SDI) on the same two pins, SWDIO on PA13 and SWCLK on PA14, enabled by default after power-on or reset. It is a different protocol from Arm SWD, so plan on WCH's debug tools rather than an ST-LINK. (Our reading: the datasheet names the interface but does not list compatible probes.)
Five checks before you swap
- Supply below 2.4 V. ST runs from 2.0 V; WCH's operating range starts at 2.4 V (Table 4-2). A design on a sagging battery between 2.0 and 2.4 V works on one and not the other. Both need 3.0 to 3.6 V for USB.
- Firmware and toolchain. Budget for a port, not a drop-in: different core, startup code, libraries and debug probe.
- Peripherals you rely on. USART, SPI, I2C, CAN and ADC exist on both in the counts above, but register maps and some timings differ. Re-check every driver against the WCH reference manual.
- The 5 V signals. The FT pins match except PB5. A 5 V signal routed to PB5 is within limits on WCH (FT input up to 5.5 V absolute maximum, Table 4-1) but not on ST, so do not move such a design back to ST without checking.
- Clock and speed assumptions. The WCH part runs up to 144 MHz with its own clock tree; ST tops out at 72 MHz. Timing code that counts cycles, and any clock-derived baud rates, must be recalculated. (Our reading of the two clock specifications.)
When the swap makes sense
If your board already carries an STM32F103C8T6 footprint, the datasheets support reusing the layout for a CH32V203C8T6. You gain clock speed, a fourth USART, two op-amps and a second USB port. In exchange you take on a firmware port, a new debug tool and a 2.4 V minimum supply. It is a design change with a known footprint, not a part substitution.
Both parts are in the DatasheetVend component library with their manufacturers' datasheets. The pinout extraction returns each pin's functions with the datasheet page they came from, so two pin maps can be compared in code rather than by eye.
FAQ
Is the CH32V203C8T6 pin-compatible with the STM32F103C8T6?
On LQFP48 the pin names and positions match on all 48 pins; pins 24 and 48 are named VDD_IO on WCH, and PB5 is 5 V-tolerant only on WCH. Electrically and in firmware they differ.
Can I flash STM32F103 firmware onto a CH32V203?
No. One is Arm Cortex-M3 and the other RISC-V; the firmware must be ported and rebuilt.
Can I debug a CH32V203 with an ST-LINK?
The CH32V203 uses WCH's 2-wire SDI interface on PA13/PA14, not Arm SWD. Use WCH's debug tools.
Compare pin maps from the datasheets, not from memory.
A new account gets five free passes. Each pass runs up to two modules, pinout and footprint included, on a PDF up to 25 MB.