problem on burst mode in MPC5777C EBI

[marcellobellini]

in MPC5674F i use  in EBI, the burst mode on CS0 and it is ok.

on MPC5777C not work

the 32 bit word ODD is good, but the 32 bit even is swapped

Is possible a cpu bug?

[davidtosenovjan]

Apps response:

"As burst mode works fine for the device( according to my test) and flash connections are good, I suspect something is going wrong with software.

So I would suggest 2 experiments that costumer should carry out

1. Run my sample SW and try to see the behavior.

2. Try to use SRAM on their board for burst access and see what is happening."

[marcellobellini]

are you news?

[davidtosenovjan]

Unfortunately not. I have reported your problem to the responsible engineers and I was told there is no known issue and it'll be investigate. In the week I have been informed it has been assigned to other person. Ill let you know asap.

[davidtosenovjan]

I am forwarding an answer I have obtained during christmas:

"I looked at the schematics T249-09CPU and It is for MPC5674F. The schematics looks good and it can be well applied to MPC5777C board. But except one correction. They used AD[8] and AD[16-30] to latch from ALE. But for MPC5777C,only available physical pins are AD[9-30] and AD8 is not a physical pin.
Above is also the case in schematics T345-11ME.

Further, in T249-09CPU, why are they multiplexing  AD8 pin? This seems suspicious too.

I am assuming they are either using our EVB or they are using their EVB designed for MPC5777C.

I would recommend to review the above points and share with me schematics of EVB for MPC5777C for further investigation.
Also please ask them to review "section 27.1 Chip Specific EBI information" and "27.6.6 Summary of Differences from MPC5xx"."

Please share schematics related to MPC5777C (currently shared are only partially related to MPC5777C).

[marcellobellini]

on the mpc5674f and mpc5777c there is not address A8 and then I latch data AD8 to get A8 and everything works properly.
So external memories work perfectly with A8 if you do not use the burst mode. with mpc5674 work also burst mode

EBI diffrence from mpc5674f and mpc5777c are only for new bit on ORx and BRx register, present on mpc5777c but not on mpc5674f

for example APS bit.

so now i sell  the board 345 without burst mode and is all ok.

For APS problem is other board and not work because CS2 not generate Address31 but only Data31

[davidtosenovjan]

Comment for apps:

"The latching of AD8 makes sense.

Please ask following to the customer.

1. In 345, does burst access to both S29CL032J0RFAM010 and CY7C1441-AV33-133AVX gives the wrong results?
2. Please share the MSR register configuration."

[marcellobellini]

1) i have tested only S29CL032 because is foundamental this part, but i will try also ssram next week

i set clockout to 44Mhz need in 32bit multiplexede mode

2) in MCR(MSR?) register i change  only bit 29 to 1 for some external bus part with 16 bit bus and bit 31 to 0 for 32 bit mode

But i have try to set bit 30 and other bit 29 and 31 in alternate value with jtag, but the problem not change

This 345 board is different from 397 board(other my board in production with mpc5777C) where i need of APS=1 for shift address and where bug on pin D_CS2 not set ADDRESS A31(shifted to A30 with aps=1) in thi case i use 16 bit non multiplexed bus and i not use burst mode.

[davidtosenovjan]

I have been confirmed EBI burst access works fine with MPC5777C. There is no silicon defect on the device.
Application engineer recommend to check following points in FLASH connection:
“1. BDIP connection: It seems this signal is connected through a voltage divider connection. Please ensure that it is not causing problem. Is it recommended connection from FLASH supplier? NXP design directly connects BDIP signal to ADV# pin in its schematic.

Also BDIP is connected to J4 which is NC in datasheet of FLASH. Please make sure if it is just a typo or connection error. It should be connected to J3 which is ADV#.

2. TS: TS is connected to J3 which is ADV#. Again make sure if this just a typo or connection error. Schematic says LBA# signal. there is no such signal mentioned in datasheet as well.

3. Similarly, K7, K8 and C7 is NC but all three are connected to VCOREF.

4. CLK input has also some input circuit before CLKOUT goes to K3. Please make sure if this also recommended by flash supplier or not.”

[marcellobellini]

I did other test on ebi burst, but it does not work.

the signals are good, but in read mode the first word  contains the value of the third word.

same problem of the past

Can you send me the register configuration with which you have did test on ebi burst?

[marcellobellini]

HI David

1)in spansion s29cl032 Adv signal is TS pin J3 (LBA for ST flash)

Bdip is present in ST flash and normaly is not present R346 it is not populed when i use spansion

2)Yes is correct in spansion flash

3)K7, K8 and C7 is a request in ST flash to connect vcoreref, but in spansion is NC and is not a  problem

4)no problem for fan out of clock, also in evb frescale clockout has other input

i remembar that this soluction work with MPC5674F without problem from some years and in many hundreds boards

[davidtosenovjan]

I am not aware of any erratum related to such behaviour. Could you provide further details (sheme of connection, your setting, scope measurement etc.)? Thanks

[marcellobellini]

i use same register ebi configuration from MPC5674F

OR0 ff000020

BR0 21000081

attach scheme of external flash memory

A30..A8 address bus latched from ALE

AD0..AD31 external data bus

[davidtosenovjan]

What is your device part number and mask revision?

[marcellobellini]

i use an PPC5777C mask 2N45H

but i have test also an PPC5676R on board where i use normaly the MPC5674F in burst mode

MPC5676R have a similar problem

Is possible i need set other parameters/register in addition to BR0 and OR0?

[davidtosenovjan]

I believe the only EBI difference (BR register) between MPC5674F and MPC5676R are following:

I am sure if some of that may cause mentioned behavior, I don't thinks so.

If you have further detail as measurement, logs and so on, give it to us.

[marcellobellini]

the burst reads eight word or four word, i have try both configurations

the problem is that the  data word  is correct except that the first word is wrong, its value is equal to the third word

then only one of 8 word is wrong and is always the first that takes the value of the third

[davidtosenovjan]

Hi, I am sorry for delay, I was out of office.

Could you show me how you have connected flashes with MPC5674F and MPC5676R devices? I suppose the scheme you have already attached is for MPC5777C device, is it so?

[marcellobellini]

the pattern is similar, also the OR and BR configuration is similar