广告广告位招商此处招商 · ads@eeunion.comA5 · 728×90(移动端 320×100)了解详情 →

Signal integrity for MIPI CSI - trace length limits?

Started by HPE_Server_Robert · 2026-11-18 03:29 UTC · 15 views · 2 replies
H Member
MIPI CSI trace length limits depend on data rate and board material. What lengths are you running successfully?
论 Member
There is no single number — it depends on lane rate, material and how much loss your receiver can tolerate. A practical way to think about it is an insertion-loss budget at the Nyquist frequency: most MIPI CSI-2 receivers are happy with roughly -10 to -20 dB of channel loss at Nyquist, and the rest is layout.

A few rules of thumb that have held up for us:

- At 1.5–2.5 Gbps/lane (D-PHY), 10–15 cm on standard FR-4 is usually fine with careful routing: short stubs, no unnecessary vias, intra-pair skew under ~5 ps.
- Above ~3 Gbps/lane, or beyond ~20 cm, FR-4 starts to cost you. Either move to a low-loss laminate, or add a redriver/repeater — and budget for the extra power and the equalisation tuning that comes with it.
- C-PHY changes the maths: three-wire signalling with embedded clock is more sensitive to skew between the trio, so match lengths within the trio first, then worry about absolute length.
- Vias are the silent killer at these rates: each via adds a stub and an impedance discontinuity. Keep the signal on one layer where possible, and back-drill if the stub exceeds ~0.25 mm.

If you are validating, measure the channel with a VNA or TDR rather than trusting the stack-up calculator, and check the eye at the receiver with a real pattern, not a clock. What lane rate and length are you running, and is it D-PHY or C-PHY?
F Member
Question: how sensitive is this to temperature? Any thermal test data?
广告广告位招商此处招商 · ads@eeunion.comA6 · 728×90(移动端 320×100)了解详情 →
Log in or register to reply.