2016-04-19 08:38 AM
Hello,
We are using the STM32F439 in our application.While using the simulation with IBIS model (Hyperlynx), we have discovered undershoots on FMC pins.The voltage in undershoot is about -866mV and the current limit is about -11mA.The duration of the undershoot is 2nSec and repeatable every 50nSec.From the Data Sheet of the STM32F439, we understand that the voltage on FMC pins not allowed to fall below -0.3V, and if the voltage is lower than the -0.3V, the current has to be limit to -5mA.Our questions -1. Is the requirement in the Data Sheet of maximum current of -5mA dependent on the duration of the undershoot? (for example, for short duration the maximum current allowed is higher)2. Is an undershoot of -866mV with maximum current of -11mA for 2nSec every 50nSec on FMS pins can cause a damage to the ST32?Thank you, Waiting for your answer,Itai.2016-04-19 10:41 AM
How aggressively are you driving the outputs? ST tends to dial back the slew-rate on FMC pins. ie runs them at the 50 MHz setting, rather than 100 MHz
2016-04-19 11:16 AM
Hi az.e,
Your description is not clear enough. Can you give more details about: - Model schematic that you have simulated ? - Frequency ? - Pins - Snapshot of simulation result ? - Device package ? -Hannibal-Hello,
We are using the STM32F439 in our application. While using the simulation with IBIS model (Hyperlynx), we have discovered undershoots on FMC pins. The voltage in undershoot is about -866mV and the current limit is about -11mA. The duration of the undershoot is 2nSec and repeatable every 50nSec. From the Data Sheet of the STM32F439, we understand that the voltage on FMC pins not allowed to fall below -0.3V, and if the voltage is lower than the -0.3V, the current has to be limit to -5mA. Our questions - 1. Is the requirement in the Data Sheet of maximum current of -5mA dependent on the duration of the undershoot? (for example, for short duration the maximum current allowed is higher) 2. Is an undershoot of -866mV with maximum current of -11mA for 2nSec every 50nSec on FMS pins can cause a damage to the ST32? Thank you, Waiting for your answer, Itai.2016-04-21 01:06 AM
> While using the simulation with IBIS model (Hyperlynx), we have discovered undershoots on FMC pins.
I think you forgot about the external part - the PCB and the drain (the memory IC). Under/overshoots are reflections, which are caused by non-matching impedance. Source, transmission lines (PCB tracks) and drain should have the same impedance (complex resistance) to minimize reflections. I think you need to represent this properly in your simulation. Forum is kinda unresponsive today, again ... ;-(
2016-05-01 03:07 AM
2016-05-01 03:10 AM
The problem occurred while the FMS pins used as an inputs.
However, we simulates in 50Mhz setting but the behavior doesn't change.
From: clive1Posted: Tuesday, April 19, 2016 7:41 PMSubject: STM32F439 - Undershoot on FMC pinsHow aggressively are you driving the outputs? ST tends to dial back the slew-rate on FMC pins. ie runs them at the 50 MHz setting, rather than 100 MHz
2016-05-01 03:18 AM
Hi Hannibal,
-We used the originals models we got from the companies. The memory is MT48LC16M16A2B4-6AIT.-The frequency is 20Mhz-The pins are the data pins on FMS. All the 16 pins show the same behavior.-I will attach snapshot soon.-The package of the Microcontroller is BGA176, the memory package is BGA54.2016-05-01 03:31 AM
Hi AvaTar,
Of course, that we have simulates with the PCB , the specific memory and the microcontroller that in used.Because of limitation of board place, there is no option of adding series resistors.2016-05-01 11:12 PM
> Because of limitation of board place, there is no option of adding series resistors.
It is finally your decision, to compromise functionality and stability for cost and size. However, the very advantage of a microcontroller is having RAM/ROM and periphery on one silicon, without an expensive external bus. Have you checked other options, like other MCUs with sufficient memory, or other interfaces like Quad-SPI ?