PDSCH EVM that fails after a few-sample timing slip while DMRS SNR still looks green

Started by LunaWei · 2026-11-17 01:49 UTC · 13 views · 3 replies
L Admin
[Technical article / FAQ, rfforum Editorial Team]

A 5G NR (or LTE) conducted measurement can show healthy DMRS SNR and still fail PDSCH EVM. A timing slip of a few samples, a window that is centered on the DMRS but not the data, or a FFT start that walked after a frame drop, often explains it. The reference symbols still correlate; the data symbols do not.

1. What the timing window owns
- DMRS is a known sequence and can look fine with a slightly wrong start, especially if the equalizer is trained on those symbols alone. PDSCH then pays for residual ISI and a partial CP miss.
- UE or test-set time tracking can lock to a strong path or a delayed echo. A green DMRS SNR with a red PDSCH EVM after a cable or fading-profile change is often a window, not a PA.
- Frame / slot boundary slips after a dropout reset the window. The next capture looks "randomly worse" until you re-sync.

2. Freeze list before you open a PA or baseband ticket
- Timing offset (samples / ns) reported by the analyzer, and whether it was frozen or tracking
- Same capture: DMRS SNR, PDSCH EVM, and frequency offset together; a few-sample walk that kills only EVM is a window first
- Re-acquire / hard sync and repeat; if EVM returns, do not ECO the RF lineup
- CP fraction and window alignment in the test set; a "wide window" setting can hide or invent the fail

3. Habit
Never freeze "PDSCH EVM failed, PA linearity" from a single capture that still shows green DMRS SNR. Archive timing offset and a re-sync A/B next to the EVM line. If a re-acquire clears the fail, the amplifier was never the patient.

4. Practical gate
- Read timing offset on the failing capture
- Hard re-sync and repeat EVM
- Only then look at ACLR, IQ imbalance, or a real baseband bug

Discussion: on your last "DMRS good, PDSCH EVM bad" ticket, did the fix land in timing window / re-sync, test-set settings, or a real RF/baseband issue - and do you require the timing offset on the same sheet as EVM?

— rfforum 编辑部 / Editorial Team
L Admin
Addendum — write timing offset (samples or ns) on the same line as DMRS SNR and PDSCH EVM. A hard re-sync / re-acquire A/B comes before a PA or baseband ticket. If EVM returns after the re-acquire, the amplifier was never the patient.

Watch for a window that is centered on the DMRS but not the data, or a start that walked after a frame drop. Frequency offset belongs on that same line; a few-sample walk that kills only EVM is a window first. Check whether the test set was frozen or tracking, and whether a "wide window" setting is hiding or inventing the fail.

Open: on your last "DMRS good, PDSCH EVM bad" ticket, did the fix land in timing window / re-sync, test-set settings, or a real RF/baseband issue — and do you require the timing offset on the same sheet as EVM?

— rfforum 编辑部 / Editorial Team
O Member
A few-sample timing slip that leaves DMRS SNR green is almost always a receive window or trigger offset, not an equaliser problem. Check the timing advance tracking loop under Doppler ramp rather than at static Doppler; the loop bandwidth that passes a static test can lag a 1 kHz/s ramp. The EVM floor also moves with the phase noise of the local oscillator, so verify with the same LO on both ends before blaming the DUT.
R Member
We've tested a similar approach - decent stability, but BOM cost goes up a bit.
广告 AdRenesas 瑞萨微控制器芯片产品照片Renesas Electronics微控制器、模拟与射频/时钟方案 · MCUs, analog, RF and timing访问官网 →免费展示 · 资料来自公开信息 · 如需撤下请联系 ads@rfforum.com
广告 Ad广告位招商 / Advertise hereads@rfforum.com · 刊例 Rate cardT2 · 728×90
Log in or register to reply.