0
\$\begingroup\$

Working on a new revision of a design with a new I2C LED driver connected to SAM S70 Cortex-M7, I noticed some high speed ripple on 5V rail when using a 1Ghz scope. The PCB has 2 I2C buses and both of them pollute 3V3 and 5V rails respectively. However, the p-p ripple on 5V is higher.

enter image description here

enter image description here

The I2C bus for the LED driver is pulled up to 3V3, the LED driver sits on 5V rail. The termination resistors are placed close to the MCU. MCU configures the pins to open drain when TWI is used.

The only thing I could do to reduce the ripple is to try different combinations of termination and pull-up resistor values, plus put filtering caps. The best combination so far is 220 Ohms termination resistors and 5.1K pull-ups. Interesting thing is that SDA line causes the highest ripple when it reaches 0V (controlled by the MCU). This is actually a TWI bus, so there is no slew control for this interface.

Is the anything I can do to minimize this ripple?

\$\endgroup\$
9
  • 4
    \$\begingroup\$ I’d look carefully at how your probing the signal. You shouldn’t be seeing those glitches. Make sure the probe is earthed to a solid gnd on the pcb , preferably as short as possible. High speed scopes are great, but they can show you ghosts if you’re not careful. \$\endgroup\$
    – Kartman
    Commented Jan 16, 2022 at 23:30
  • 1
    \$\begingroup\$ I suspect it is not really there - try touching the scope probe to ground clip with the scope ground clip in the normal place. The scope trace should be completely clean. I have never heard of IC2 lines interfering with anything. \$\endgroup\$
    – Kevin White
    Commented Jan 16, 2022 at 23:33
  • 1
    \$\begingroup\$ Are you using PCB that you designed yourself? Did you ground and decouple with EMC in mind? Although it's very likely that fellow commenters are right in this being some issue with the measurement, it would be good to keep in mind (edit: just saw the voltage scale, yeah that doesn't seem right) \$\endgroup\$
    – crossroad
    Commented Jan 17, 2022 at 2:54
  • \$\begingroup\$ Well, the spikes were first discovered without SDA and SCL probed. The frequency of the ripple pointed out to the I2C bus. That's why I hesitant to suspect the incorrect probing. My question should sound more like: has anyone else saw this phenomenon? \$\endgroup\$
    – Max Morgunov
    Commented Jan 17, 2022 at 3:35
  • \$\begingroup\$ Please share your schematic. For example, termination resistors aren't typically used in I2C. So where did you connect them? At the master? At the slaves? What value capacitors did you use and where did you place them? Etc. \$\endgroup\$
    – The Photon
    Commented Jan 17, 2022 at 3:47

0

Reset to default

Browse other questions tagged or ask your own question.