PSM and eDRX about ubxlib HOT 14 CLOSED

Commit 9261d9c represents the implementation of 3GPP power saving and E-DRX support for SARA-R5/SARA-R5 modules. Note that to support 3GPP power saving operation you will need to make sure that PWR_ON and VINT are both connected from the module to your MCU. Also note that three new port APIs were required as part of the implementation: uPortEnterCritical()/uPortExitCritical() and uPortTaskGetHandle(); if you have your own port you will need to implement these APIs.

The best place to start in terms of finding out about the power saving APIs is probably here:

https://github.com/u-blox/ubxlib/tree/master/example/cell/power_saving

...while the APIs themselves can be found in u_cell_pwr.h.

And, lastly, power saving is a complicated topic! It has taken all of this year to make this work satisfactorily and, of course, there may still be holes. Please open an issues here if/when you find the holes :-).

from ubxlib.

A quick update on this: there's a bit of a chain of things going on:

• there is a subtle issue in the rather old SARA-R5 modules (00B version) we use on our test farm which means that we don't always get the OK response to an AT+GPS=0 command (i.e. switching GNSS off) if power saving is enabled; internal logging of the module shows it being sent but we don't always see it on the physical UART, must be something to do with the power saving sequence,
• the above needs to work when we run tests of the existing Cell Locate feature (since the tests have to switch between having the user using the GNSS chip and giving it up to be used by Cell Locate by issuing AT+GPS=0),
• hence, in order to run our tests with power saving on, we need to switch all the SARA-R5 modules on our test farm to the 01B version,
• we now have the bits needed to do that switch,
• with that done we can commit the change which enables AT+UPSV to be used when the UART flow control lines are connected between the MCU and the module; this change was completed a few weeks back, is tested locally and ready to go but couldn't be committed because then the existing Cell Locate tests would start failing.

Hopefully we will to get through all of the above next week. Then AT+CPSMS.

from ubxlib.

@eeFLis: how important is the feature for you, timescale-wise? I ask because it is now next on the list but there are other, shorter-duration-yet-lower-priority things [e.g. a recent request to see if it is possible to support Arduino-ESP32] that we could look at first.

from ubxlib.

@RobMeades we have made a quick and dirty implementation so that we can work with PSM and eDRX. But we would like to see it fully integrated into ubxlib so that all power saving features like IDLE mode also can be used. This will be especially important when the SARA-R510S-01B with NB-IoT is released. But back to your question. If it can be implemented in Q4 that would be great.

from ubxlib.

Thanks for the swift response and background information, appreciate it. Agree that this is a high priority feature and, as I'm sure you've found out, it is not always easy to accommodate (especially when, in the ubxlib case, it needs to work across our different modules). I've not got a SARA-R510S-01B myself yet [even though the code for it is being written 30 metres away :-)] but, yes, being able to power save properly is kinda essential to the NBIoT use-case.

We will aim to deliver this feature for you in Q4.

from ubxlib.

@eeFLis: an update for you - a pre-requisite to implementation of AT+CPSMS is to implement AT+UPSV. Sometimes known as "UART power saving", this allows the innards of the chip to switch off a load of clocks which saves quite a lot of power but not anything like as much as AT+CPSMS. I've had AT+UPSV working for a while now but there is a corner case, specifically to do with switching off a GNSS transceiver that is connected to the cellular module, that is causing problems and hence I can't check the code in yet. And I don't think I want to implement AT+CPSMS on top of this until I'm sure it is a good thing.

Anyway, I was wondering if you thought it would be beneficial for me to push a preview_ branch here with just the AT+UPSV bit done, assuming of course that connecting a GNSS chip via the cellular module is not part of your use-case? Just that I suspect Q4 may otherwise pass with no progress on this [that is visible to you] otherwise.

Up to you of course; if it would be a distraction, or of no particular use, to have sight of a partial implementation then let's not do it and I can only apologies for my tardiness.

from ubxlib.

@RobMeades Thanks for the update.
We do not use a GNSS chip so a preview branch would be helpful.
I hope you still keep the AT+CPSMS implementation on the ToDo list, since as you said it saves much more power.

from ubxlib.

Thanks for responding. I determined today that what I thought was something odd on the cellular module side seems actually to be a really subtle bug somewhere in my code so I need to spend more time hunting it down before I inflict it upon you even in a preview branch.

AT+CPSMS is most definitely not off the menu, it is what I was originally aiming for, just that unfortunately AT+UPSV has to be addressed first or AT+CPSMS won't actually do anything.

I'll post here when I have at least got AT+UPSV working and we can decide what to do next.

from ubxlib.

@RobMeades I would like to echo the importance and value of this feature. It would provide benefit to us as soon as you have it ready. It could be argued that most users of this chip selected it because it is low power, and software support for those low power modes could be seen as a basic requirement for a driver. This is certainly true for my use case and I am not using the GNSS chipset or Arduino.

from ubxlib.

Understood. In order to help me find the rather subtle issue in the AT+UPSV implementation, which must be completed before AT+CPSMS, I diverted to create a version of ubxlib that will run on a PC, attached to a SARA-R5 module via USB. I finished that on Friday so I'm hoping the problem will now reveal itself. I will update here when I know more.

from ubxlib.

I have now pushed the AT+UPSV part of power-saving, at least the initial implementation of it, here (see commit a971101). I say the initial implementation because, as it stands, for SARA-R5 you cannot use the CTS/RTS flow control lines while AT+UPSV is enabled (see the commit message for the details of why). It is possible to fix that but I will do it as a separate commit.

Next, AT+CPSMS...

from ubxlib.

Hi Rob.
Thanks for the update.
Good to hear that things are moving forward.
especially since the 01B version supports NB-IoT and power saving will be even more important.

from ubxlib.

The changes that allows AT+UPSV power saving to work correctly when the UART flow control lines are connected with a SARA-R5 module are now pushed here, see commit b57b7ec.

from ubxlib.

A quick note ref. implementing this feature: the code needs to know when the module has entered deep sleep. On some module types a URC is emitted when this occurs but on others, specifically the SARA-R41x-02B modules, there is no URC and so the only way we can tell that the module has entered deep sleep is if VInt has gone low.

Hence the implementation will need VInt connected through from the module to a GPIO line of the MCU to use PSM/E-DRX with SARA-R41x-02B modules.

from ubxlib.