AM263P4: Segger J-link support with AM263 and Uniflash

Hi Ian,

The "Could not connect to target" error you're seeing is a known issue, and there are multiple contributing factors. Despite J-Link appearing as a selectable probe in Uniflash, the AM263Px SDK documentation, target configurations, and all flashing examples exclusively reference the XDS110 debug probe [1][2]. This strongly suggests J-Link is not fully supported for Uniflash flashing operations on this device, even though the UI allows selection.

Additionally, there are hardware-revision-specific problems that compound the issue:

1. Uniflash + AM263Px-CC E2 Revision: Known Limitation

If your TMDSCNCD263P is the E2 revision, Uniflash has a documented limitation where it does not properly support this board revision. The SDK documentation explicitly states this under the Uniflash tool's "Limitations" section [3]. The SBL fails to read the board ID from eFuse on E2 hardware, causing the bootloader to hang and preventing any JTAG-based tool from establishing a connection [4].

2. Outdated Flasher Binary in Uniflash

A separate known issue involves an outdated flasher binary bundled with Uniflash, which causes DAP access failures ("Error -2131" or "Error -1170") even with the XDS110 probe. This was fixed in Uniflash 9.1.0+ [5], and since you're on 9.5.0 you should have the fix — but if the E2 eFuse issue is the root cause, the version won't matter.

Recommended Workarounds

Option A: Use the onboard XDS110 (recommended path)

The TMDSCNCD263P has an integrated XDS110 debugger accessible via the micro-USB connector. All SDK flashing workflows are validated against this probe [6][2]. If you need an external probe, the TMDSEMU110-U (XDS110) is the supported external option [6].

Option B: Use sbl_jtag_uniflash via CCS instead of Uniflash GUI

Bypass Uniflash entirely and use the sbl_jtag_uniflash example project from the MCU+ SDK (examples/drivers/boot/sbl_jtag_uniflash). This works in both DevBoot and OSPI boot modes [7]:

• DevBoot mode: Build and load the project directly via CCS — GEL files handle initialization.
• OSPI boot mode: Create a target configuration without GEL files, run the load_sbl.js script from tools/ccs_load/am263px/ in the CCS Scripting Console first, then load the uniflash utility [7][8].

Option C: Fix the E2 eFuse issue (if applicable)

If your board is REV E2, modify bootloader_soc.c in the SBL source — replace the EFuse_read() call with status = SystemP_SUCCESS;, then rebuild and flash the patched SBL. TI indicated this fix was targeted for the March 2025 SDK release [4].

Option D: Use UART-based flashing

The standard production flashing workflow uses UART boot mode with the uart_uniflash.py script, which doesn't involve J-Link or JTAG at all [9]. Set boot mode switches to UART (SOP3=0, SOP2=0, SOP1=0, SOP0=1), then use the Python script to transfer images.

Key Takeaway

J-Link appearing in Uniflash's probe list does not mean it's validated for AM263Px. All official documentation, examples, and target configurations point exclusively to XDS110 [1][2][6]. The onboard XDS110 or UART-based flashing are your most reliable paths forward.

To help refine this further, it would be helpful to know:

• Which hardware revision of the TMDSCNCD263P you have (check the silkscreen — E1 vs E2)
• Whether the same "Could not connect" error occurs when using the onboard XDS110 debugger (this would isolate whether the issue is J-Link-specific or board-level)
• Whether you've tried the sbl_jtag_uniflash CCS-based method as an alternative
• Whether TI has provided any official response in the locked forum post you referenced regarding J-Link support timelines

1. AM263Px CCS Setup - SDK Documentation
2. AM263x JTAG Uniflash Example
3. AM263P4 Uniflash Error on CC Board - E2E
4. TMDSCNCD263P AM263P4 eFuse Issue - E2E
5. LP-AM263P Uniflash Fix - E2E
6. LP-AM263 Debugger Tool - E2E
7. AM263P4 Flash Verification via sbl_jtag_uniflash - E2E
8. AM263Px Additional Details - SDK Documentation
9. AM263Px Flash Tools - SDK Documentation

Best Regards,

Zackary Fleenor

Hi Zackary, I would appreciate a human response rather than what is clearly an AI generated comment that does not adequately understand the relatively simple question that I'm asking. Telling me to use the XDS110 is not a valid response to a question about whether the promised segger support is available.

I am able to use Uniflash with the XDS110. Similarly, I am able to use a j-link to debug the board through gdb or segger o-zone. There are no hardware compatibility issues here.

Further testing shows that we can flash S variant boards (external flash) but cannot flash SIP (F variant boards) with the flash in package

Hi Ian,

We need to check on SIP variant support with segger, since it came later in the timeline.

But from my understanding, if the flasher_jtag.out is updated for the SIP varinat then it should be able to support teh SIP package support as well.

Please see this on migrating SDK to SIP compatible: software-dl.ti.com/.../SOC_MIGRATION.html