Rise Time GPIO Pads for QSPI

[User20592]

Hello,
I'm working on a TC2x7 Tricore board.
I'm want to implement QSPI with the highest possible clock frequency (50MHz). Therefore I useLVDS pads (possible, e.g. for QSPI2).
Data sheet says, the rise/fall time of LVDSM pads (MTSR, SCLK) is 2.5ns.
I measured it with an oscilloscope and achieved only 10ns rise/fall time for MTSR and SCLK. With that rise time, 50MHz are a difficult goal..
Does anything on the Evalboard influence the rise time of the pads or am I overseeing something? I'm measuring with the extension board on the 80 pin socket.
I don't know what else I can do, but somehow it must work 😄

Thanks for your help 🙂

PS: LVDS function enabled in GPIO ports, speed grade configuration has no effect, 5V or 3.3V supply makes no change, cable used: CAT5e twisted pair, 100 Ohm termination resistor

[MoD]

How you check the rise/fall time? You use the differential mode? You check on MTSR_P, MTSR_N, SCLK_P and SCLK_N?

[User20592]

Hello, I check it with measuring for example MTSR_P on one channel against ground and MTSR_N on the second channel against ground.
Finally I use the math function of the oscilloscope and subtract the positive signal from the negative signal. This subtracted signal gives me the LVDS signal and the rise time.

[User20639]

Stupid question, does your oscilloscope have enough bandwidth? You probably need 500Mhz+ to get a proper reading.

[User20592]

Yes, it has a bandwidth of 100MHz and a rise time of 3.5ns.. So, is the mistake of the oscilloscope too big? Or isn't it possible to get closer to the 3.5ns rise time of the oscilloscope? Because I'm only measuring 10ns.

[MoD]

With 100MHz bandwidth you have max. resolution of 10ns. Therefore you measure 10ns.

[User20639]

MoD wrote:
With 100MHz bandwidth you have max. resolution of 10ns. Therefore you measure 10ns.

Bandwidth != sample rate. Even a cheap Rigol scope with 50Mhz bandwidth has a 1GSa/s sample rate.

As Paulemeister said, with 100Mhz bandwidth you should be able to see a 3.5ns rise time, so I'm not sure why you can only see 10ns. Have you tried another output, see if you're seeing the same thing?

[User20592]

I can try it with QSPI3, it uses a different output pin.. can tell you more on tuesday!