8 Replies Latest reply on Apr 25, 2019 6:32 AM by tobrc_4182961

    Migrating from S25FL032P Serial NOR Flash to S25FL064L

    tobrc_4182961

      I recently migrated from the S25FL032P to the S25FL064L NOR Flash and during testing I realized that the Block Erase (64K Sector) command 0xD8 takes much longer than the original slower chip; the original flash, S25FL032P took about 6.5 seconds to erase 24 blocks 64K sectors while the newer flash, S25FL064L took over 8.5 seconds to complete the same task.

       

      The S25FL064L should complete this task much faster than the S25FL032P according to the migration document and the datasheets.

       

      Addition info:

      The sector sizes are equal for both flash chips.

      S25FL064L Block Erase Times (64KB physical sectors) –  Typical Erase Time:450 Ms,  Max Erase Time: 1150 Ms

      S25FL032P Block Erase Times (64KB physical sectors) –  Typical Erase Time:500 Ms,  Max Erase Time :2000 Ms

       

      What could cause the newer faster flash to perform slower than the older slower one?

        • 1. Re: Migrating from S25FL032P Serial NOR Flash to S25FL064L
          BushraH_91

          Hello Tony,

          Thank you for contacting Cypress Community Forum.

          Can you please send us the waveform? How many S25FL064L chip taking longer time than S25FL032P?

           

          Regards,

          Bushra

          • 2. Re: Migrating from S25FL032P Serial NOR Flash to S25FL064L
            BushraH_91

            Hello,

            1. Were the tested 24x64K sectors contain the same data in FL064L and FL032P before erasing? Data pattern has impact on the erase time.
            2. How many devices has the customer tested? The erase time may various on individual devices (even various on individual sectors on the same flash unit). From the typical erase time, in general, FL064L is 10% faster than FL032P. However, a fast FL032P could erase faster than a slow FL064L. As far as the erase time is within the spec max erase time, it is normal.
            3. Were the test flows (include how to check the erase completion) same for the test with FL064L and FL032P?
            4. Were there possible some delays in-between the sector erases?

             

            Thank you

            Regards,

            Bushra

            • 3. Re: Migrating from S25FL032P Serial NOR Flash to S25FL064L
              tobrc_4182961

              1. Yes, the sectors contain the same data.

              2. We tested 5 of the FL064L in the lab; however, we have many of them failed during production testing because of the device watchdog timer (taking too long).

              3. The test is the same because we are running the same firmware.

              4. No, because it is the same firmware.

               

              I was able to erase 24 blocks (64K blocks) in 8811 miliseconds for the FL064L  which is about 367 mS per 64K sector which is well within the spec of the FL064L.

               

              I was just surprised that it took 6447 mS to erase the same 24 blocks of 64K sectors on the older and slower FL032P. At this rate, it takes 6447/24 = 243 mS per 64K sector.

               

              Please note we received a sample of the FL064L a few months earlier that perform equally (sometimes better than) as the FL032P. The only difference is that the first batch of FL064L is automotive rated. which according to the data sheet shouldn't be any difference between the two. 

               

              • 4. Re: Migrating from S25FL032P Serial NOR Flash to S25FL064L
                BushraH_91

                Hello Tony,

                Thank you for your update. I will discuss internally and get back to you.

                 

                Regards,

                Bushra

                • 5. Re: Migrating from S25FL032P Serial NOR Flash to S25FL064L
                  BushraH_91

                  Hello Tony,

                   

                  According to the test result:

                  FL064L – 367ms  (spec typical 450ms)

                  FL032P – 243ms (spec typical 500ms)

                  Both are faster than the average (typical). The tested FL032P seems to be a “very fast” individual device. The tested FL064L seems to be fast but close to typical. The erase time may be various for different batch, different individual devices, even different sectors within one device. So your test result is reasonable.

                   

                  Regard to the timeout failures during production, what is the timeout timer value? The timeout timer should be set according to the spec maximum erase time (1150ms per sector for FL064L).

                   

                  Btw, from spec, the typical 64K erase time for FL032P and FL064L are similar (FL064L is 10% faster), the max 64K erase time FL064L is much faster (1150ms), FL032P is 2000ms.

                   

                  Thank you

                  Regards,

                  Bushra

                  • 6. Re: Migrating from S25FL032P Serial NOR Flash to S25FL064L
                    tobrc_4182961

                    Yes, it is within specification but i was concern about the difference in erase time; why would a slower Flash perform better than the faster one.

                     

                    Concerning the time-out time, we have made the adjustment to accommodate the slower erase time.  Previously in our firmware we were erasing multiple sectors without resetting the watchdog timer. 

                     

                    Thank you for the help.

                    • 7. Re: Migrating from S25FL032P Serial NOR Flash to S25FL064L
                      BushraH_91

                      Hello Tony,

                      Do you still have concern? Please let us know.

                       

                      Thank you

                      Regards,

                      Bushra