SK-AM62: How to find PTP PPS pin on the EVK

Hello again Nick,

Thank you very much for the highly detailed explanation. I have taken my time to try to understand all of it, but I still can't find my PPS out signal anywhere.

I wonder if I'm doing something wrong, or if I'm just trying to use a feature that is still not functional.

Let me describe what I have done so far, in case you can point out something I'm interpreting the wrong way.

First, in the timesync_router that can be found in the dts file you link to, I see  this:

/* pps [cpsw cpts genf1] in17 -> out12 [cpsw cpts hw3_push] */

I still cannot locate hw3_push  anywhere on the am625 documentation, so I decided to look for SYNC_OUT pins, since you mention those in your explanation.

Luckily, I was able to locate those in the AM62X Reference Manual (AM625, AM625-Q1, AM623, AM620-Q1 SPRSP58B – JUNE 2022 – REVISED JUNE 2023). If we go to page 56, 6.3.2.1 MAIN Domain, Table 6-6. CPTS Signal Descriptions, we see:

If I'm not mistaken, I think the SK-AM62 rev E3A uses the ALW type AM625. If we cross reference these pins with the board schematics, we get these test points:

I have my EVK directly connected via eth1 to a  PTP server running the automotive-master ptp profile using ptp4l 3.1.1.

On my EVK, I am running the following commands on different consoles:

am62xx-evm:~# ptp4l -m -f /usr/share/doc/linuxptp/configs/automotive-slave.cfg -i eth1 -p /dev/ptp0

am62xx-evm:/usr/kernel-selftest/ptp# ./testptp /dev/ptp0 -P 1
pps for system time request okay

am62xx-evm:/usr/kernel-selftest/ptp# ppstest /dev/pps0
trying PPS source "/dev/pps0"
found PPS source "/dev/pps0"
ok, found 1 source(s), now start fetching data...
source 0 - assert 1294277998.000080399, sequence: 4231 - clear  0.000000000, sequence: 0
source 0 - assert 1294277999.000080400, sequence: 4232 - clear  0.000000000, sequence: 0
source 0 - assert 1294278000.000080399, sequence: 4233 - clear  0.000000000, sequence: 0
source 0 - assert 1294278001.000080399, sequence: 4234 - clear  0.000000000, sequence: 0
source 0 - assert 1294278002.000080398, sequence: 4235 - clear  0.000000000, sequence: 0

But I still can't see anything resembling a PPS on any of the five pins I list above. Do you think I might be doing something wrong, or is there something not fully implemented yet?

Hello Xabier,

I ran out of time to look through the June 2023 revision of the AM62x TRM today to see if we added a table with the outputs of the time sync router to that latest revision. The first thing I would check is to see what the time sync output 12 is.

Regards,

Nick

Ah, so do I need to wait for the June release of the TRM to be available?

I have checked the november 2022 release  just in case, and Figure 10-8 (Page 7966 - Time Synchronization Support), and Table 10-8 (Page 7968 - Push event for CPSW) seem somehow related to all of this, but I still can't figure out how to map that to outputs, since the columns for this table are "Input Index" and "Interrupt Source".

Hello Xabier,

Hmm, ok there has not been an updated TRM release yet. I am checking with the hardware team on when we can expect that, and on whether we can get some more information about the time sync router connections. Please ping the thread if I have not responded by Friday.

I will check to see how the Linux developer tested the time sync on their board.

Regards,

Nick

Hello Xabier,

Still following up on the hardware signal routing question. However, the Linux developer has provided a way to test PPS from the terminal in case this is helpful:

"

1. Request PPS signal generation by running:

echo 1 > /sys/class/ptp/ptp0/pps_enable 

2. Sample the PPS signal at 1 second interval and compare the timestamps to check if they truly are 1 second apart, by running the following commands:

# Get first timestamp
cat /sys/class/pps/pps0/assert
# Wait for 1 second
sleep 1
# Get second timestamp
cat /sys/class/pps/pps0/assert

If the two timestamps are 1 second apart, that confirms the PPS signal.

"

Regards,

Nick

Hello Nick,

Thank you very much for all your help.

I have tested these last commands, and it looks to me like everything looks fine:

am62xx-evm:~# echo 1 > /sys/class/ptp/ptp0/pps_enable
am62xx-evm:~# cat /sys/class/pps/pps0/assert ; sleep 1 ; cat /sys/class/pps/pps0/assert
1294507007.000000015#55
1294507008.000000014#56

Hello Xabier,

I have added information about the TSR output signals in this new FAQ. Hope it helps:

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1243869/faq-am62-what-is-the-time-sync-router-for-how-do-i-use-it

Regards,

Nick

Hello Xabier,

One last update: We would actually NOT expect to see the PPS signal on one of the SYNCx_OUT pins, and that makes sense when we review the time sync router outputs on the FAQ I linked above.

Time Sync Router output 12 goes to one of the CPSW CPTS inputs, called cpts_hw3_push. I am not sure why the developer chose this as their example mapping. I am following up with them separately.

If you want the PPS signal to show up on a processor pin, you need to send it to a SYNCx_OUT signal (one of outputs 20-23), and make sure the pinmux settings are appropriately set for that pin.

I do not have time to look at the AM62x SK EVM schematics today to see which SYNCx_OUT signals are routed to headers on the board that could be connected to an oscilloscope. However, if you run more tests and have any additional feedback, let me know and I will pass it along to the developer. Your timesync router mapping might make its way into our default EVM dts file.

Regards,

Nick

Hello Nick,

Thank you very much for your help! I was actually working on a status update. It is now quite late for me and I haven't had the chance to test my progress today enough, but I have definitely been able to get some kind of output that resembles a PPS signal just before I had to leave the office.

I'm not sure I will be able to post my process today, but I will give it a shot. If there's no new reply for me in the next few minutes, I hope to have something tomorrow.

It all appears to be working correctly for now.

I've had the system running for a litlle bit more than 3 hours at this point, and I have been collecting data from the generated signal. Using linuxptp v4.0 configured with the automotive profile on the SK-AM62 and a PC with an intel i210, this is what the syncrhonization looks like:

Hello Xabier,

Awesome, glad that you got it working! Also, thank you for sharing your changes to the Linux devicetree. I have passed your example along to the developer, and we will add it to the default AM62x SK devicetree file.

If you have any additional questions, feel free to make a new e2e thread and we will help you out there.

Regards,

Nick