TPS2546: 60s recognition

Hi Fred,

What is the current and previous setup with the TPS2546? Can you provide block diagrams? 

Does this same case occur with different mouse? 

Thanks and Regards,

Raymond Lin

Hi Raymond,

I think it's the same setting , since they only change CPU,

they change from 1111 (S0)to 01111 (S3),

this 60s thing , does it have anything to do with CPU?

or just between TPS2546 and mouse?

Hi Fred,

It's possible the CPU could be doing something to the TPS2546 but I would like to confirm first if this is a device compatibility case. Does this issue occur with all types of mouse or just this particular one?

Thanks and Regards,

Raymond Lin

Hi Raymond,

they only test this kind of mouse, they haven't tried other kinds.

but I think the problem is not on the device , 

since one need to wait 60s, the other doesn't.  

is there any chance that mouse change from full speed device to low speed device?

or would you tell what could be the potential difference that cause this?

Hi Fred,

Let me ask internally regarding your question as I'm not too familiar with this device. I'll get back to you by the end of the week!

Thanks and Regards,
Raymond Lin

Hi Raymond,

they tried three different mouses , the outcome are all the same,

the status sequence is 1111 to 0111 and then wait 60s to wake up back to 1111,

same for last generation but except for  the 60s window,

please discuss internally and feedback to me, thanks

Hi Fred,

Thank you for the clarification, I'll provide a response by tomorrow!

Thanks and Regards,

Raymond Lin

Hi Raymond,

any update from BU?

they're in a little urgent.

Regards,

Fred

Hi Fred, 

Apologies for the delay, let me get back to you by tomorrow.

Thanks and Regards,

Raymond Lin

Hi Raymond,

we will be design out if we can't solve this, 

please help , thanks

Hi Fred,

Looping in Chris from my team to help.

Thanks and Regards,

Raymond LIn

Hi Fred,

Apologies for the delay.

I'm a little unsure about the current situation and want to clarify some things to make sure we are on the same page. In the "Main Issue" section, I will describe the issue as I have interpreted it from the information you and Raymond have shared.

Main issue:

Fred Cheng said:

my customer plug in the mouse , and he have to wait 60s so it can wake it up in sleep mode (S3).

While the computer is asleep in S3 mode a mouse is initially connected to a USB port controlled by the TPS2546 , it takes 60s for the mouse to wake the computer up.

In a previous version of this design, in the exact conditions as above, the mouse did not need to wait 60s.

Constants

• Mouse
• TPS2546
• General Layout
• Software

Differences

• CPU

Fred Cheng said:

the status sequence is 1111 to 0111 and then wait 60s to wake up back to 1111,

Timing of events

1. Mouse is unconnected
2. Computer is awake in S0, status is 1111
3. Computer falls asleep in S3, status is 0111
4. Mouse is connected
5. After 60s pass, computer wakes up in S0, status is 1111

Thoughts:

Fred Cheng said:

the status sequence is 1111 to 0111 and then wait 60s to wake up back to 1111,

same for last generation but except for  the 60s window,

The "status" (CTL 1,2,3 and ILIM_SEL pins) is all externally controlled by the CPU(?) which means any timing related to the status sequences comes from the CPU. Initial thoughts are that the routine is being handled at the CPU differently between versions. Because both generations use the same part and same status sequence, there only seems to be a difference in the timings which would point to the CPU.

[Q1] Can you describe what the customer expects the CPU to see from the TPS2546 in order to change the status. When will the CPU go from 0111 to 1111. Is there an internal timer that is waiting 60s, is it waiting for a signal from the TPS2546?

[Q2] Is this handled differently in the previous version?

.

Could you monitor the fault line, status line, and DM/DP out lines for the old and new versions, and see if they differ in value at all during these events. If you do no see any differences here then and there

Fred Cheng said:

but the strange things is they try it on the last platform which has same design same TPS2546 but with different CPU, the mouse doesn't have to wait 60s to wake up from sleep.

The TPS2546 part does not seem to have any recent updates so if the design and part is the same, the only thing that could be causing a difference is the CPU.

Because we are starting in status 0111, the data lines are not connected and load detect is active. A requirement for Wake on USB is that the data lines are connected so this may be related to the delay in wake up.

Thanks and Regards,

Chris 

[Q1] Can you describe what the customer expects the CPU to see from the TPS2546 in order to change the status. When will the CPU go from 0111 to 1111. Is there an internal timer that is waiting 60s, is it waiting for a signal from the TPS2546?

FYI, the mouse is connecting the whole time.  It's just in raptor lake CPU, somehow it needs 60s to wake from sleep, but alder lake doesn't..

but 60s is indeed in our IC, so I'm still not sure if this is CPU's problem.

as for the transition from 0111 to 1111, I think it's manually done. 

[Q2] Is this handled differently in the previous version?

I don't think so,  may I know what signal  behavior or action will DP and Dn do when waking up?

if we all see the same waking up signal at same timing , then I don't think it's not our problem, right?

Fred Cheng said:

FYI, the mouse is connecting the whole time. 

Good to know, wasn't sure about the connection as we might be worried about load detection.

Fred Cheng said:

I don't think so,  may I know what signal  behavior or action will DP and Dn do when waking up?

if we all see the same waking up signal at same timing , then I don't think it's not our problem, right?

I'm not sure what the specific signal behavior is as I'm not too familiar with "Wake On USB". I'm assuming that during the sleep state there will be no data coming over the lines, and when the mouse is moved, we will see data signals on DP and DN in lines (mouse side). The 0111 status says that the data lines will be disconnected (see Table 3 above), so I am not expecting to see any data line activity while in the 0111 status on the DP and DN out lines(CPU side).

It seems like the only signals we send back to the CPU would be either the Status or the Fault signals, so as long as all signal timings between CPU and the TPS2546 remain the same between versions, I do not think it is our problem.

Hi Chris,

so customer said in  this generation, the duration before enter S3 has shrink to 20s, i think that's why,

1. please help check with system engineer , is this 60s defined by USB-IF VESA or only by ourselves TI side?  why 60s?

if it's the latter, do we have other device support less than 60s?

2. look like 60s has nothing to do with CPU, it's completely out of TPS2546's internal counter, please confirm this with internal expert 

thanks

these answers would affect the most of the TPS2546 share on this gen and next generation notebook market, so must help on this one.

Hi Fred,

Fred Cheng said:

2. look like 60s has nothing to do with CPU, it's completely out of TPS2546's internal counter, please confirm this with internal expert 

thanks

I thought the previous version of this design did not have the 60s problem but was using the same part. Is this true, or did they just not care about the 60s in the previous version? I think you are on the same page here, but I'm confused why they are running into the 60s issue between versions when they are using our hardware for both version and the only thing changing is the CPU.

Let me try reaching out to the System engineers to see who covers this part.

Thanks and Regards,

Chris

Hi Chris,

by nothing to do with CPU, I mean this 60s is completely out of our IC internal counter.

the reason last generation can work is because the duration of suspending to sleep mode was more than 60s, so they didn't spot this issue, but now, with new CPU the duration became 20s. While our IC is still set and fixed at 60s.

please make sure you answer above two question, especially why choose 60s as spec before?

this device maybe around 70% used by our customers, me and my manager need this answer, thanks

Hi  Chris,

thanks a lot, it helps

but my question will be why design team chose 60s, since it looks like the 60s is not a USB2.0 spec.

I mean it can be 40s or 50s or any other?

Hi Fred,

I was able to talk to a member of systems and another apps engineer.

The system engineer who was working on this is not in the group anymore, and the document I sent earlier is what he provided when he left. None of the systems engineers who are on the team were around during the bring up of this part, and there is no roadmap for supporting updates in the future.

From the note, it looks like 60s is what was decided on to safely account for the "10s of seconds for a transition to S4/S5". I am not sure if it can be lowered to 40s or 50s, but from the notes it seems unlikely that we would lower it below 20s like you customer wanted, even if we could.

Thanks and Regards,

Chris

Hi Deric,

please let me know the feedback on why choose 60s as spec and is 60s recognition  including the time moving mouse or keyboard?

Hi Fred,

Deric is not on this E2E thread and I would recommend reaching out to him using email.

Thanks and Regards,

Chris