0
\$\begingroup\$

I am new to the forum and the EE world so forgive me if technical terminology is a bit off.

I may be thinking about this too much, but what is the procedure for calculating SPI clock speed for a bus that needs to transmit large amounts of data (28Mb/s) in real time? This is for a large number of slave whos' output data rate is 200Hz each.

\$\endgroup\$
  • \$\begingroup\$ As others have already mentioned, the clock rate will be equal to the data rate in Mbit/s, or greater. How much greater? That depends on the "overhead" that slave devices impose, and communication schedule which you are planning. Post these details please. \$\endgroup\$ – Nick Alexeev Feb 21 '16 at 1:52
  • 1
    \$\begingroup\$ For example, are you doing burst reads from the slaves, or single byte reads? The latter may achieve less than 50% throughput (send 8 clock cycles to instruct to read, then read on the next 8 cycles). \$\endgroup\$ – Tom Carpenter Feb 21 '16 at 2:37
  • \$\begingroup\$ Thank you all for your responses! @Nick Alexeev - in regard to communication scheduling, could you elaborate please? is it not possible to read all sensors simultaneously? \$\endgroup\$ – ledhead1191 Feb 21 '16 at 6:56
1
\$\begingroup\$

If your aggregate data is 28Mbps, then your SPI clock needs to be at least that, but likely higher to account for addressing each device, any latency in addressing the various devices' chip-enable, and latency before they're ready to respond, and so on.

\$\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.