signpost
signpost copied to clipboard
lab11
Module Time Sync
To have a hope of tight time synchronization, modules are probably going to need to wire PPS to a timer/counter (TC) pin, so that it can trigger a counter start. This might not really help, because the SAM4L doesn't have a 32bit counter that can take an external trigger anyways, but it's something to think about.
Why couldn't you connect it to a capture line on a free running timer?
- Prabal
On Saturday, March 18, 2017, Joshua Adkins [email protected] wrote:
To have a hope of tight time synchronization, modules are probably going to need to wire PPS to a timer/counter (TC) pin, so that it can trigger a counter start. This might not really help, because the SAM4L doesn't have a 32bit counter that can take an external trigger anyways, but it's something to think about.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48, or mute the thread https://github.com/notifications/unsubscribe-auth/AH6xX32r-lLeQVFRWu6WIw1NwqF5dGFwks5rm2Z7gaJpZM4MhTAW .
The only 32bit free running timer peripheral does not have any way to interface with external i/o in hardware, and its fastest clock source is 32khz. I was going to test full accuracy of time-stamping events in software using this timer today.
I expect to achieve synchronization in the 10s of uS using this method, which is probably good enough for something like localizing an audio source, but still much worse than our original expectation of synchronization given the accuracy of the pps line.
On Mar 20, 2017 7:44 AM, "prabaldutta" [email protected] wrote:
Why couldn't you connect it to a capture line on a free running timer?
- Prabal
On Saturday, March 18, 2017, Joshua Adkins [email protected] wrote:
To have a hope of tight time synchronization, modules are probably going to need to wire PPS to a timer/counter (TC) pin, so that it can trigger a counter start. This might not really help, because the SAM4L doesn't have a 32bit counter that can take an external trigger anyways, but it's something to think about.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48, or mute the thread <https://github.com/notifications/unsubscribe-auth/AH6xX32r- lLeQVFRWu6WIw1NwqF5dGFwks5rm2Z7gaJpZM4MhTAW> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48#issuecomment-287780626, or mute the thread https://github.com/notifications/unsubscribe-auth/AGIJULATo3Lyead3X38Gx3xfTzJUVTWpks5rnpC4gaJpZM4MhTAW .
Wow...that's interesting and a bit disappointing that the MCU is missing this pretty basic functionality. I'm surprised that there isn't a 32-bit timer block that can be configured to be free running and that also gives you a capture line.
- Prabal
On Monday, March 20, 2017, Joshua Adkins [email protected] wrote:
The only 32bit free running timer peripheral does not have any way to interface with external i/o in hardware, and its fastest clock source is 32khz. I was going to test full accuracy of time-stamping events in software using this timer today.
I expect to achieve synchronization in the 10s of uS using this method, which is probably good enough for something like localizing an audio source, but still much worse than our original expectation of synchronization given the accuracy of the pps line.
On Mar 20, 2017 7:44 AM, "prabaldutta" <[email protected] javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:
Why couldn't you connect it to a capture line on a free running timer?
- Prabal
On Saturday, March 18, 2017, Joshua Adkins <[email protected] javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:
To have a hope of tight time synchronization, modules are probably going to need to wire PPS to a timer/counter (TC) pin, so that it can trigger a counter start. This might not really help, because the SAM4L doesn't have a 32bit counter that can take an external trigger anyways, but it's something to think about.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48, or mute the thread <https://github.com/notifications/unsubscribe-auth/AH6xX32r- lLeQVFRWu6WIw1NwqF5dGFwks5rm2Z7gaJpZM4MhTAW> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48#issuecomment-287780626, or mute the thread <https://github.com/notifications/unsubscribe-auth/ AGIJULATo3Lyead3X38Gx3xfTzJUVTWpks5rnpC4gaJpZM4MhTAW> .
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48#issuecomment-287787169, or mute the thread https://github.com/notifications/unsubscribe-auth/AH6xX6ycpDQEDa8HUQKXsTBOtmivIdNCks5rnpU_gaJpZM4MhTAW .
Yeah, I was pretty shocked. Usually you can at least chain two 16bit timers together in hardware, and I can't even find a way to do that.
I might resort to running our synchronization evaluation code on the cc2538, but I haven't looked into it much yet. Hopefully that would at least allow us to demonstrate the capabilities of the platform, even if most of the modules we currently have can't achieve that accuracy.
On Mar 20, 2017 8:24 AM, "prabaldutta" [email protected] wrote:
Wow...that's interesting and a bit disappointing that the MCU is missing this pretty basic functionality. I'm surprised that there isn't a 32-bit timer block that can be configured to be free running and that also gives you a capture line.
- Prabal
On Monday, March 20, 2017, Joshua Adkins [email protected] wrote:
The only 32bit free running timer peripheral does not have any way to interface with external i/o in hardware, and its fastest clock source is 32khz. I was going to test full accuracy of time-stamping events in software using this timer today.
I expect to achieve synchronization in the 10s of uS using this method, which is probably good enough for something like localizing an audio source, but still much worse than our original expectation of synchronization given the accuracy of the pps line.
On Mar 20, 2017 7:44 AM, "prabaldutta" <[email protected] javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:
Why couldn't you connect it to a capture line on a free running timer?
- Prabal
On Saturday, March 18, 2017, Joshua Adkins <[email protected] javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:
To have a hope of tight time synchronization, modules are probably going to need to wire PPS to a timer/counter (TC) pin, so that it can trigger a counter start. This might not really help, because the SAM4L doesn't have a 32bit counter that can take an external trigger anyways, but it's something to think about.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48, or mute the thread <https://github.com/notifications/unsubscribe-auth/AH6xX32r- lLeQVFRWu6WIw1NwqF5dGFwks5rm2Z7gaJpZM4MhTAW> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48#issuecomment-287780626, or mute the thread <https://github.com/notifications/unsubscribe-auth/ AGIJULATo3Lyead3X38Gx3xfTzJUVTWpks5rnpC4gaJpZM4MhTAW> .
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48#issuecomment-287787169, or mute the thread <https://github.com/notifications/unsubscribe-auth/ AH6xX6ycpDQEDa8HUQKXsTBOtmivIdNCks5rnpU_gaJpZM4MhTAW> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48#issuecomment-287794009, or mute the thread https://github.com/notifications/unsubscribe-auth/AGIJUPhqhHLyVAugYWRdKDft4pno6ajyks5rnpougaJpZM4MhTAW .
Probably not a bad idea. Our module selection kinds undercuts the basic claim.
It might be useful to just do this on some third-party module that runs its own ecosystem (e.g. mbed, Arduino, TI, etc.) to demonstrate both the timing and the modular compatibility.
- Prabal
On Monday, March 20, 2017, Joshua Adkins [email protected] wrote:
Yeah, I was pretty shocked. Usually you can at least chain two 16bit timers together in hardware, and I can't even find a way to do that.
I might resort to running our synchronization evaluation code on the cc2538, but I haven't looked into it much yet. Hopefully that would at least allow us to demonstrate the capabilities of the platform, even if most of the modules we currently have can't achieve that accuracy.
On Mar 20, 2017 8:24 AM, "prabaldutta" <[email protected] javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:
Wow...that's interesting and a bit disappointing that the MCU is missing this pretty basic functionality. I'm surprised that there isn't a 32-bit timer block that can be configured to be free running and that also gives you a capture line.
- Prabal
On Monday, March 20, 2017, Joshua Adkins <[email protected] javascript:_e(%7B%7D,'cvml','[email protected]');> wrote:
The only 32bit free running timer peripheral does not have any way to interface with external i/o in hardware, and its fastest clock source is 32khz. I was going to test full accuracy of time-stamping events in software using this timer today.
I expect to achieve synchronization in the 10s of uS using this method, which is probably good enough for something like localizing an audio source, but still much worse than our original expectation of synchronization given the accuracy of the pps line.
On Mar 20, 2017 7:44 AM, "prabaldutta" <[email protected] javascript:_e(%7B%7D,'cvml','[email protected]'); <javascript:_e(%7B%7D,'cvml','[email protected] javascript:_e(%7B%7D,'cvml','[email protected]');');>> wrote:
Why couldn't you connect it to a capture line on a free running timer?
- Prabal
On Saturday, March 18, 2017, Joshua Adkins <[email protected] javascript:_e(%7B%7D,'cvml','[email protected]'); <javascript:_e(%7B%7D,'cvml','[email protected] javascript:_e(%7B%7D,'cvml','[email protected]');');>> wrote:
To have a hope of tight time synchronization, modules are probably going to need to wire PPS to a timer/counter (TC) pin, so that it can trigger a counter start. This might not really help, because the SAM4L doesn't have a 32bit counter that can take an external trigger anyways, but it's something to think about.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48, or mute the thread <https://github.com/notifications/unsubscribe-auth/AH6xX32r- lLeQVFRWu6WIw1NwqF5dGFwks5rm2Z7gaJpZM4MhTAW> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48#issuecomment-287780626, or mute the thread <https://github.com/notifications/unsubscribe-auth/ AGIJULATo3Lyead3X38Gx3xfTzJUVTWpks5rnpC4gaJpZM4MhTAW> .
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48#issuecomment-287787169, or mute the thread <https://github.com/notifications/unsubscribe-auth/ AH6xX6ycpDQEDa8HUQKXsTBOtmivIdNCks5rnpU_gaJpZM4MhTAW> .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48#issuecomment-287794009, or mute the thread <https://github.com/notifications/unsubscribe-auth/ AGIJUPhqhHLyVAugYWRdKDft4pno6ajyks5rnpougaJpZM4MhTAW> .
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/lab11/signpost/issues/48#issuecomment-287796608, or mute the thread https://github.com/notifications/unsubscribe-auth/AH6xX8kBfN5MEDx-hAwzEipWfEKsZLo7ks5rnpwKgaJpZM4MhTAW .