3
\$\begingroup\$

I am new to my current job and one design inherited from my predecessor was as follows. I am wondering if this is the best practice to make a symmetrical circuit. enter image description here

I was told that the two wires (R+ and R-) should keep the same length to make a symmetrical circuit. To do that, the designer put some turns on the layout. It looks weird to me, but wondering if this is a common practice. The frequency in the circuit is up to 1.5GHz.

\$\endgroup\$
  • 1
    \$\begingroup\$ Very common practice. It's also built into that software. Well, it looks like Altium, but more importantly it looks like not-Eagle, so it probably has advanced features like length matching. \$\endgroup\$ – Samuel Dec 8 '15 at 17:27
  • 1
    \$\begingroup\$ I was always taught that matching the length of the two conductors in a differential pair is extremely important. \$\endgroup\$ – mkeith Dec 8 '15 at 17:31
  • 2
    \$\begingroup\$ No offense intended, but if you don't understand differential pairs, what are you doing working on a high-speed design? \$\endgroup\$ – DoxyLover Dec 8 '15 at 18:28
  • \$\begingroup\$ @DoxyLover The OP doesn't necessary not know what a differential pair is. You can understand his answer in the way that he doesn't know which layouts are used to achieve length matching. Perhaps he expected one big loop instead of meanders? \$\endgroup\$ – Ariser Dec 8 '15 at 20:26
  • \$\begingroup\$ Thanks folks for the insights! Yep, me too think matching the length of a differential path is crucial, to keep the phase synchronized. The curved line just looked a bit weird... I was trying to find an example from some evaluation board layout (e.g. analog devices, variable gain amplifiers). But all the examples I reviewed all used straight lines. That was why I need some experts' insights. Thanks again to all comments! \$\endgroup\$ – Peng Peng Dec 8 '15 at 20:27
1
\$\begingroup\$

1.5 GHz has a wavelength of 20 cm and a golden rule is not to worry too much if there is a line mismatch of less than one tenth of this. That's 2 cm and maybe if the signal were a sq wave you'd be keen to reduce this by another factor of 5 or 7 to account for significant harmonics in the waveform.

So maybe mm precision is needed. You decide!

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.