# How critical is it for SN65HVD1780DR A/B Differential signal lines to have matched trace length?

How critical is it for the A/B differential signal lines of SN65HVD1780DR to have matched trace lengths on PCB?

• Does it say anything in the datasheet? There should be a "Layout Specifications" section. – KingDuken Jun 5 at 15:20
• What speed are you driving the RS485 bus at? – Voltage Spike Jun 5 at 15:45
• Compute skew vs % jitter of bit @ C/√Er – Sunnyskyguy EE75 Jun 5 at 15:55
• 9600 bps is the speed of data on the bus – be-ee Jun 5 at 15:56
• not an issue.... consider <60 ps / cm – Sunnyskyguy EE75 Jun 5 at 15:57

9600 bps equates to about 0.104 milliseconds per bit (1/9600).

The signal speed on a PCB microstrip (external trace) is calculated by:

where Vs is the velocity of the signal, c is the speed of light (11.8 inches per nanosecond), and Ereff is the effective dielectric constant (about 2.92 for FR-4).

The signal speed on a PCB stripline (internal trace) is calculated by:

where again, Vs is the velocity of the signal and c is the speed of light. Er is the dielectric constant of FR-4 which is usually around 4.

If you do the math out, the signal velocity on a microstrip is approximately 6.9 inches per nanosecond and the signal velocity on a stripline is approximately 5.9 inches per nanosecond. If your bit rate is such that each bit takes 0.104 milliseconds (104000 nanoseconds), then you would need a length difference of roughly 717,600 inches (11.3 miles) for a microstrip or 613,600 inches (9.7 miles) for a stripline to present enough delay in one of the differential lines for the receiver to miss an entire bit. You really don't need to worry about length matching on a PCB when dealing with only 9600 bps.

NOTE: Feel free to check my math, I rushed through this on my lunch break.