2026-01-08 10:19 AM
2026-01-08 11:47 PM
Hello @RAdam.1
If you refer to the latest External Memory Manager Middleware,
STM32Cube_FW_N6_V1.3.0\Middlewares\ST\STM32_ExtMem_Manager\Release_Notes.html
You will find a list of NOR Flash SFDP from Winbond and Gigadevice manufacturer which are compatible with STM32N6.
Actually, concerning Hexa PSRAM/HyperRAM, only the APMemory APS256XXN-OBR-BG was validated on STM32N6.
After checking, it seems that the APS256XXN is available at Mouser.com.
APS256XXN-OBR-BG AP Memory - Distributors and Price Comparison | Octopart component search
Best regards,
Romain
To give better visibility on the answered topics, please click on Accept as Solution on the reply which solved your issue or answered your question.
2026-01-13 10:25 AM
It sounds like I'm in the same position as the original poster.
The recommended PSRAM APS256XXN-OBR-BG is showing as nearing the end of its lifecycle and not recommended for new designs. The manufacturer recommends APS256XXN-OB9-BG as a replacement.
Can anyone comment on the compatibility of APS256XXN-OB9-BG with the N6?
2026-01-15 4:06 AM - edited 2026-01-15 4:07 AM
I reviewed the respective datasheets:
APS256XXN-OBRx – APM Octal PSRAM Datasheet, Rev. 1.00
APS256XXN-OBx9 – APM OPI/HPI Xccela PSRAM Datasheet, Rev. 1.2
The memory devices support the following serial interfaces with the Xcella support on OBx9 part number:
APS256XXN-OBRx: DDR Octal SPI
APS256XXN-OBx9: OPI/HPI and also supports Xccela™ mode for PSRAM
The main differences concern operating frequency and data rate:
APS256XXN-OBRx: Up to 200 MHz (400 MB/s in X8, 800 MB/s in X16)
APS256XXN-OBx9: Up to 250 MHz (500 MB/s in X8, 1000 MB/s in X16)
All other specifications listed below appear to be fully compatible:
Both part numbers are pin-to-pin compatible and share the same package dimensions.
Power-up initialization uses the same method and timing.
Address space, burst type and length, and command/address latching are identical.
The command truth table is the same.
Read operations use the same address/data transfer sequence and timing.
Write operations also appear to be aligned.
In my opinion, the APS256XXN-OBx9 should be compatible with STM32N6 XSPI without any modifications but limited to 200MHz with DQS.
Best regards.
Romain,
To give better visibility on the answered topics, please click on Accept as Solution on the reply which solved your issue or answered your question.
2026-02-04 2:58 AM
Hi,
Just to add few comment: APS256XXN-OB9-BG - 256Mb is recommended latest device to use (APS512XXN-OB9-BG - 512Mb). This is latest version in production, fully compatible with STM32N6 (and other STM32 MCU supporting OPI). (APS256XXN-OBR-BG is the lower speed grade, not recommended for new design)
Thanks
Alex
2026-02-04 5:06 AM
Thank you to have confirmed the point.
Best regards,
Romain,
To give better visibility on the answered topics, please click on Accept as Solution on the reply which solved your issue or answered your question.
2026-05-28 1:54 AM
Hi Romain,
do you know of any OctoSPI PSRAM with 128MB (1024Mb) that is compatible with the STMN6?
Thanks in advance
2026-05-28 3:08 AM
Hi,
You can use 2 pces of 512Mb (ASP512XXN-OB9-BG) and 2CE to make 1Gb on same memory controller
Method 1 : Optimize performance, need 2 controller and more pin (two PSRAMs can be treated as independently controlled devices)
Method 2 : Double density without controller / pin count impact. Note: due to path loading, you might need tu use IBIS simulation to confirm the data window is good enough.
Regards
Alex
2026-05-28 4:20 AM
Thanks for the fast response!
Which method would you suggest? How would the memory be accessed when using method 1? Will the hardware handle everything and I can assume it being continuous RAM when programming the firmware? If I understood the Datasheet correctly the FMC has two ports, so with Method 1 I wont be able to add another OctoSPI flash for the firmware to the MCU, right?
Best Regards
2026-05-28 5:19 AM
if you have 3 memory devices (two PSRAM and one NOR), with only 2 xSPI/OPI memory controller, you will be in method 2 anyway. One controller with 2 device, one controller with one device.
You can decide to have
a) the two PSRAM on one controller and NOR alone
b) one PSRAM and NOR on one conttroler, and PSRAM alone
This may be driven by your SW architecture, and whether the PSRAM usecase can be splitted or not, and if yes, which usecase need more performance. No general recommendation, just as one example, you may decide to do frame buffering on the xSPI, and share NOR / PSRAM for code / data.
Alex
We’re moving the ST Community to a new platform to give you a better and more reliable community experience.