Watchdog reset on purpose


#1

Hello,
i have some problems about my fastract xtend modem with R7.43 firmware on the area. Because of the problems (we havent found the reason), our customers have to power off/power on the modem. After powering off/on the modem, it starts working again.
We are very fed up trying to find the reason causes modem to act like that so we want to do something powering off/on the modem automaticly. i read the documantation and it says there is a watchdog mechanism on the modems. So i thought i can achive it by triggering watchdog reset manually.
I searched the forum and i see “adl_errHalt()” function causes watchdog reset. I will use this function on my open at software. So this function will reset the modem as if someone powering off/on the modem? And will it mean the same of powering off/on the modem?
Thank you.


#2

Yes, adl_errHalt() cause watchdog reset but it may different from hard reset (“power off/power on the modem” as mentioned).
You may refer to sample application “Bug” for detailed explanation of different type of exceptions.

Still, you may describe more the problem you faced so probably we may able to give better advice.

Hope it helps.
L


#3

We have a sim card holder problem. Module can not detect SIM card. Even AT+CPIN returns ERROR. Our software resets the modem by AT+CFUN but it cant help. We may have also different problems which we are not aware of. What we know is power cycling the modem solves the problem. So our quick solution can be power cycling the modem or performing a hardware reset within the open AT application.

You mean that adl_errHalt is not what we are looking for. Do you know any other way to perform a hardware reset within the open AT application?
Thanks.


#4

The only way to really get a true hardware reset is to use real hardware!

This is not unique to Open-AT!


#5

Yes, i think performing a watchdog reset doesnt mean the same as power cycling nor hardreset. So i guess we are going to design a microcontroller for powering off/on the modem externally.


#6

Have a look at the TPS22922 from TI (http://www.ti.com/product/tps22922). It is a low on resistance load switch which works very well controlling power to the modem. Use a microcontroller to drive this, and you can completely control power to the modem.


#7

Hi tomridl,
Thank you for the product suggestion but our hardware designer is going to take care of it. i think he will use some kind of relay for power switching. And a microcontroller will control the relay.


#8

Unless it is also driving other things, and those things specifically require a relay, I think using a relay would be daft!

You can’t drive a relay direct from a microcontroller port pin; you will need some kind of semiconductor driver anyhow, so you might as well use a semiconductor device to switch the modem!

See: https://forum.sierrawireless.com/t/q2687-cfun-1-vs-reset-pin-vs-power-cycle/5357/2 - which also indicates that even the Reset line is not necessarily the same as a power-cycle!


#9

I really dont know about hardware. He is designing a circuit which has relay and semiconductor too. i dont know why he is doing in this way. Also i can not be questioning him because i have no idea about the hardware design. So i think he has some reasons to apply it in this way.

Yes , i read that topic and i have come to know that reset line is not the same as power-cycle. That is why we have decided to make something that performs the power-cycling externally. That topic gave useful ideas.
Thank you for mentioning it.


#10

Perhaps you should get him to join in on this discussion… :question:


#11

So has Sierra/Wavecom has designed a module that cannot be brought from undefined state to defined state under certain conditions, which may happen frequently?

Even microcontrollers bridge their watchdog pins with reset pins and that can be reliably reset.

I work with Fastrack Xtend modules, so I’m afraid I cannot tuck a relay on top of that.

Before I decided to use them, I had modules custom-designed with a LPC2138 CPU and Quectel M10 GSM module. The LPC2138 was self-resetable always, and it drove a MOSFET to control power to M10.
Unfortunately, Quectel’s modules failed terribly after 3hour-1month after power-up :frowning:


#12

In my stuation, yes i have undefined problems which can be solved by only power cycling. However, this does not mean that all the modules are problemmatic exactly.

i was trying to perform firmware upgrate from R7.43 to R7.46 over the air. But i could not do it because module were reseting itself in a loop. That shows there can be one more problem too. So i asked my customer to send it back to me so i may able to investigate the problem better.
As soon as i get the modem, i will have a look at its internal logs, try to test it here and discover the problem. Then i will send a new FXT with R7.46 firmware to my customer. He will put it in the same field so we will see the problem will happen again or not.

There are suggestions previous posts of this topic. tomridl’s and awneil’s posts are pointing some other companent insead of a relay. Besides, to me if your modem gets stuck in an undefined state and you can solve it by only power reset, first you should try to find the reason which puts it in an undefined state.


#13

An electronic component can be put into undefined state for example by an EMP burst modifying CPU registers or cosmic radiation flipping bits in RAM or . It should bear the capability to resume proper working after that - especially by executing reset.

In my case the modules have been using watchdog about 4 times per minute for 8 hours. Some did came back online, some didn’t. I just wonder why is that so.

Edit:
An interesting issue is whether the module power-cycles SIM card, as it has a huge bug potential.


#14

There’s a practical danger to repeatedly resetting the device via watchdog.

By experiment I’ve determined that letting the device work for ~20s and them resetting it by watchdog, if done repeatedly, will eventually cause OpenAT OS to refuse to run the application. Repeated resets by AT+CFUN=1 work just fine.

After doing some more research I’ll probably contact SiWi about that - because it shouldn’t be that way. FXT009 exhibits it on latest firmware - 7.46.


#15

i think you’ll find it should - it’s to stop a bad application getting into an infinite reset loop.

Called the “Recovery Mechanism”

https://forum.sierrawireless.com/t/multiple-resets-stop-application/4319/2

https://forum.sierrawireless.com/t/bug-in-sierra-wireless-software-suite-recovery-mechanism/4542/6


#16

i investigated one issue. i created a new topic about it. It can be about recovery mechanism or ny protection mode if xtends have.
take a look pls [url]https://forum.sierrawireless.com/t/modem-waits-in-protection-mode/5419/1]