Skip to main content
Associate II
Nov 20, 2026 6:54 AM UTC
Question

STM32H5 FMC Interface Read Mode - Undershoot and Overshoot Risk Observation

Hello STM team,

I am using STM32H5 series MCU interfaced with ISS SDRAM IS42S16400N-6TLI.

I am observing 3.9V peak on Data lines when controller reads. Undershoot of -500mV is observed. All the data lines are interfaced on FT pins. 

Overshoot and undershoot peaks are observed for period of 1.2nS

We have evaluated the change in overshoot and undershoot by changing series termination from 22 Ohm to 47 Ohm. No significant improvement observed.

With same PCB layout and design with Winbond W9864G6KH memory overshoot and undershoots are within limits.

Even with 72 Ohm, I am observing same behavior with 100mV reduction in overshoot. 

I have following queries regarding STM32H5 MCU.

1. For FT data lines, Is it okay to go ahead with 3.9V peak.
2. Is it okay to go ahead with -500V undershoot.

 

Thank you.

Regards,

Rajesh Matere

9 replies

mƎALLEm
ST Technical Moderator
Nov 20, 2026 8:16 AM UTC

Hello ​@RajeshM,

1- 

Hello STM team

Please be aware this is a public community and not a peer to peer channel support.

2- 

I am observing 3.9V peak on Data lines when controller reads. Undershoot of -500mV is observed.

You didn’t share any screen short of your scope observation. Could you please do it? so others can help you.

3- 

Is it okay to go ahead with -500V undershoot.

I think it’s a typo and it’s -500mV according to your first statement.

To give better visibility on the answered topics, please click "Best answer" on the reply which solved your issue or answered your question.
RajeshMAuthor
Associate II
Nov 20, 2026 12:59 PM UTC

Hi,

 

Sorry, correction my end for undershoot observed is -700mV.

 

Thank you.

Regards,

Rajesh Matere

RajeshMAuthor
Associate II
Nov 20, 2026 12:54 PM UTC

Hello All,

 

Thanks for your response. Please find attached Overshoot and Undershot waveform.

Just to add more inputs. I have checked with 125MHz, 83MHz and 45MHz clock. No significant improvement observed after changing clock frequency. Mostly overshoots are observed in read mode.

 

Thank you.

Regards,

Rajesh Matere

mƎALLEm
ST Technical Moderator
Nov 20, 2026 1:29 PM UTC

Thank you for the sharing.

What is the level of the overshoot? same as undershoot i.e. 700mv ?

And please provide the exact part number (not only STM32H5).

To give better visibility on the answered topics, please click "Best answer" on the reply which solved your issue or answered your question.
RajeshMAuthor
Associate II
Nov 20, 2026 1:34 PM UTC

Hi,

Overshoot is about 700mV .

Undershoot is about -700mV.

Additionally attaching the STM32H5 general operating conditions specification. 

Does this mean, Overshoots are oaky for this much time period.

Thank you.

Regards,

Rajesh Matere

RajeshMAuthor
Associate II
Nov 20, 2026 2:34 PM UTC

Hi,

Part number is STM32H563.

Thank you,

Regards,

Rajesh Matere

mƎALLEm
ST Technical Moderator
Nov 20, 2026 2:38 PM UTC

Need to check internally..

You said: “Overshoot and undershoot peaks are observed for period of 1.2nS”

Need to be clear here: what 1.2ns corresponds to? is it the duration of the undershoot / overshoot?

To give better visibility on the answered topics, please click "Best answer" on the reply which solved your issue or answered your question.
RajeshMAuthor
Associate II
Nov 20, 2026 2:50 PM UTC

Hi,

Please find the attached explanation related to 1.2ns .

Thank you.

Regards,

Rajesh Matere

mƎALLEm
ST Technical Moderator
Nov 20, 2026 3:28 PM UTC

Thank you. This is more clear now.

I’ve forwarded your request internally over this ticket: CDM0067683 (not accessible by the community users). I’ll get back to you as soon as I have any feedback or question.

In mean time, I invite you to read these references:

AN4803 “High-speed SI simulations using IBIS and board-level simulations using HyperLynx® SI on STM32 MCUs and MPUs”

 

To give better visibility on the answered topics, please click "Best answer" on the reply which solved your issue or answered your question.