TMS320F280049: SCI BOOT timeout

Hi PA N,

We can confirm that there is *no* timeout on "autobaud lock".  Please let us know if there are any further questions. 

Cheers!

Krishna

Hello,

Thanks for the answer , but that is not at all what i expected...

My question is related to the SCI Tx pin being set to 0 "erroneously" while DSP wakes up. and SCI boot enabled.

1. Does this mean that the DSP will hang on the "autobaud lock " forever while the TX pin is driven low?

2.Does the DSP exit and jump to FLASH as it will never receive the "Key"?

3. If this hangover is possible, How to prevent it form happening or minimize it?

Thank you,

Regards,

PA N.

Hello ALL TI,

Could someone analyse my POST and give me some feedback please?

We d like to know if during boot with SCI boot enabled if the TX pin is inadvertly driven low, the DSPis going to hang on the autobaud?

Thank you,

PA .N.

Dear PA. N.

The specific concern you have appears to relate to an unexpected short on the TX pin.  First, that would result in a faulty input configuration of the device and it is important to prevent that from happening, as you know.  

In regards to the processor waiting on an autobaud lock, there is no timeout associated with that operation.  As a result, the SCI boot flow will perpetually wait for a lock. 

Hope this answers your question.  Please let us know if we can be of further assistance. 

Cheers!

Krishna

Hello Krishna,

As the TI DSPs usually propose several BOOT methods, i m surprised that a simple short on the TX Pin can lead to an undefinite hang of the DSP.

1. You confirm that a permanent short of the Tx pin, while other BOOT is also enabled, for example FLASH BOOT, leads to loss of the DSP, though communication can never happen ? i.e. The DSP never jumps to FLASH?

2. In the case of a non permanent short to the TX pin, you confirm the problem disappears as soon as the TX erroneous value disappears...

Thank you,

Regards, PA N.

Hi PA N, 

In regards to your comment below:

>> As the TI DSPs usually propose several BOOT methods, i m surprised that a simple short on the TX Pin can lead to an undefinite hang of the DSP.

Can you please explain why you are surprised?  What is the behavior you are expecting and what is the basis and rationale for that.  Just need some clarification from you because I may be missing something. 

Now in regards to item 1, you are now asking about FLASH BOOT.  Do you have both GPIO pins (GPIO 24 and GPIO 32) configured to HIGH? 

For your second question, yes I would expect the boot process to complete as described in the TRM.  Please see the flow chart in  Figure 4-5. Overview of SCI Boot Function.

Thanks, 

Krishna 

Hello Krishna,

Sorry for the late reply...

As quoted in your response from my question:

>> As the TI DSPs usually propose several BOOT methods, i m surprised that a simple short on the TX Pin can lead to an undefinite hang of the DSP.

To clarify my thinking:

The SCI boot mode uses a check of several keywords to definitely enter boot : the DSP checks for "Valid KeyValue 0x08AA" then "Jump to Flash" if Key is not valid. I expected the same behavior if the autobaud is not locked after some "timeout".

I found this a very weak point in the SCI BOOT when jumping to Flash BOOT is also possible.

In SPI, I2C, CAN boot mode, the DSP exits if the Key is not valid and there is no risk of hanging on a erroneous value of a GPIO at start.

This was my point. Now that we are aware of this weakness, we will think our boot method differently.

 Thank you,

Regards,

PA Nicoletti.