NUCLEO-F767ZI SPI Ringing Observed

[yuva]

Hello Team,

We have Purchased the NUCLEO-F767ZI Eval board for our board bring up activity.

We have configured the SPI with 15Mhz, i can see lot of ringing from the EVAL board which is more than 4 to 4.5V at Clock and MOSI lines can you look into it, any solution for this?

We have observed that at the output of the EVAL board no other devices connected to the Eval board.

[TDK]

Some things you can do:

• Add series resistors or termination resistors.
• Lower pin speed.
• Ignore it.

It's hard to measure overshoot accurately.

[yuva]

Hi,

We are using the SPI lines for Display and that display has the maximum core capability of 4V even for logic levels.

We have tried adding a 20E series resistor in the SPI, 2 displays have gone bad no response from the display, so need to check something can be done for that.

[TDK]

Also keep in mind that the MCU doesn't have access to voltages above VDD. Any ringing you see is likely due to trace inductance. Consider looking at other signals/pins to see if they have the same issue. Could be a scope/measurement issue and not something physical at all.

[bramble]

@yuva You can check the GPIO configuration for your SPI connections, they have the option to configure speed. You do not need, and should not enable, very high speed for your SPI at only 15 MHz. This configuration affects how hard the signals are driven and the faster you drive them will result in more ringing due to the impedance mis-match.

It's hard to do proper series termination on a dev board since you can't generally place a resistor near to the driving pin, but even a small series resistor at the receiving end (your display) will likely round-off the edges. It's a technique I've used before although it's not technically the right way to series term. Parallel term (mentioned also by @TDK ) would be perfect but I don't know if there's enough drive available to match your line impedance which is probably a few 10s of ohms.

[AScha.3]

Hi,

so at first improve your measurement skills. The cpu runs on gnd -- 3v3 , thats all that can come out.

Everything else is an artifact of your connections and measurement.

And these more recent cpu's are really fast and also fast switching, so without proper track impedance and termination you will get...ringing. Nothing new about this.

To control this basically you should switch as slow as possible, but as fast as needed. Not faster !

So for 15MHz look at ds , to check the appropriate pin speed: 

If pin at 3v3 and low cap loading, medium/low speed would match, speed 01 .

So set this as pin speed for all spi pins (clk, data...) and check again.

And check the signal on your scope with an fast signal generator first, set at 15M square wave out, low impedance.

Just to verify, what you "measure" is the real signal and not only artifacts.

And here some pics to show, what comes out - just from physics :

20MHz 3V square , 1ns rise/fall , just looked with a scope probe (5pF load ! )

same at 5ns rise/fall time :

And with a 100 ohm resistor, to reduce ringing, also 1ns rise:

So if your wires > 30mm or so, add series damping resistors on all signal lines.