GPRS module turns off randomly

Mark_Ennamorato 2014-06-22 17:03:04 UTC #1

My GPRS module will be running for a few minutes, sending text messages, then turn off.
I suspect running from the Tessel power is not reliable and I should use external power source?

Mark_Ennamorato 2014-06-22 20:59:27 UTC #2

Since using an external power supply it seems a LOT more stable, however having other issues at the moment..

ekolker 2014-06-23 01:38:00 UTC #3

I wrote a blog post that deals with this a while back.

The SIM900 chip is designed to run off a LiPo battery, and therefore has built-in protections, lest it damage itself or the battery. These protections include an undervoltage warning that is emitted when the chip's power supply rail falls to 3.3V, as well as an automatic power down when the rail falls below 3.2V.

Because 3.3V is all the Tessel can supply, the GPRS module includes an external power input and a separate voltage regulator which, combined, provide the SIM900 with 3.465 V. It is recommend that external power be used if possible.

When the GPRS module transmits, it uses a lot of power, so much so that it may turn itself off because it thinks its "battery" is critically low. Again, lots more about that in the blog post.

All that said, it's been a while since we've seen the module power down during our own internal testing/work. Do you have anything else plugged into the Tessel?

Mark_Ennamorato 2014-06-23 02:01:49 UTC #4

Hi thanks,

Yes i assumed it was not going to work well powered off the Tessel..
No i have nothing else plugged into the Tessel
I wish there was a screw-type connector for external power though..;-)..

Now i just have to figure out why I cannot talk to the module after the script runs..sometimes

ekolker 2014-06-23 18:33:00 UTC #5

There are two places where external power can be applied. Are you using the 0.1" male header pins?

The gray/beige connector on the GPRS module serves the same purpose as a screw terminal, but doesn't require a screwdriver. Just push down (gently) on the round indentation with a ballpoint pen to insert/remove wires.

b3ll4t0r 2014-12-02 15:03:45 UTC #6

I'm not an ee major. to that point, I read the blog post and most of it was over my head. it would be super awesome if someone would post: "I used a power supply like this (0.45A 50-60Hz Output 5.1V==2.1A") that I found here at http://awesomepowersupply.com/thisoneworkswithtessel."

Any help here?

ekolker 2014-12-02 19:18:14 UTC #7

@b3ll4t0r the punchline of that blog post is that capacitors give us a reservoir from which we can draw when the transmitter requires a lot of power. The bigger the capacitor, then, the lower the voltage drop...in theory. In practice, a combination of the power supply proper and the capacitors' characteristics govern how well the "reservoir" performs.

As it turns out, the trouble starts when the capacitors behave non-ideally, and especially when we encounter these current spikes. What we've seen in further tests here is that the ESR (effective series resistance) and ESL (effective series inductance) of the capacitors on the board are reducing the voltage the SIM900 sees at its power rail during these transmit bursts. Sometimes this effect is so pronounced that the module browns out, causing the performance issues everyone has been having.

The solution we recommend is "more of the right kind of capacitor very close to the power pins", not "use this specific power supply". Adding a 10uF ceramic capacitor to the SIM900's power pins removes any noticeable dip in the power rail during TX spikes and alleviates the brown-out problem. More details (and a picture) in this thread.

Either of these two capacitors should work for the fix:

...but any ~10uF ceramic capacitor you can get your hands on should be fine.

Mark_Ennamorato 2014-12-03 07:10:55 UTC #8

@ekolker Does this fix the sagging voltage problem 100% of the time ? I don't think many people are going to be able to solder 0805 SMT capacitors on this board, and even the leaded cap is going to be very hard... any chance you are going to redesign this to put enough of the right caps on here ?

ekolker 2014-12-03 07:26:15 UTC #9

@Mark_Ennamorato I'm wary of absolutes and superlatives...but for all intents and purposes and as far as I can tell, yes, it does fix the brown-out issue 100% of the time.

We've already done a spin of the module and made the necessary changes. I don't know off the top of my head if the new version is in the store yet, but @jia might. That said, I think the hack is well within reach for someone who has modest soldering experience. I can write up a quick guide if people are interested.

For reference/the electrically inclined: it's easy to trigger an oscilloscope on the falling edge of the SIM900's power rail (*cringes*) during transmit bursts on the first generation hardware (without that cap in place), but I was unable to trigger once the cap was there. That translates conservatively to <50mV drop on the rail, which is well within spec for the SIM900.