2019-11-06 04:17 AM
stm32h743ii
When we set the OSPEEDRy[1:0] = 11 we have glitch spike (undershoot) on clock and data. it was improved when we set OSPEEDRy[1:0] = 01.
In datasheet SPI interface characteristics page 307 said
Output speed is set to OSPEEDRy[1:0] = 11.
2019-11-06 04:57 AM
It is really a slew rate setting you can ease it back if it is too aggressive for the loading present.
2019-11-06 05:19 AM
Thank you very much.
Do you have any document share with me that I can use for reference to close the case?
or Do you have further other advice to fix the glitch spike?
2019-11-06 05:56 AM
No, i don't have any docs, the IO cells in these devices are a cascading chain of inverters with increasingly larger transistors capable of dumping more energy into the output, the speed/slew setting allows for this to be adjusted based on how aggressively you want the edges to rise/fall, and the capacitive loading presented to the pin. If the edges have negligible rise-time at the 00 setting, use that.
Your other option with short lines is to place series resistances on them, typically in the 27R to 33R sort of range to temper spikes and reflections.
2019-11-06 06:01 AM
> place series resistances on them
Using lower OSPEEDR is the equivalent of doing this at the processor side - the smaller transistors behave as resistors when switched fully on.
The DS slew rates are given for certain load capacitances, and that depends on the tracks length, layout, and number of loading pins at the other end. As long as you don't intend to drive a big number of slave devices from one pin, you are usually safer with the lower settings.
Sometimes, glitches/ringing etc. at high speed (fast edges) may indicate also inadequate return (ground) associated with given signal, i.e. improper layout. You are really looking at a transmission line at high speeds. Consider, that a 1ns slew rate edge is principially a 1GHz signal which has significant harmonics up to few tens of GHz.
JW
2019-11-06 10:26 PM
Thank you.
return (ground) is the one that I should review.