I have this eternal struggle going. It seems to be an endless game. The mission is pretty simple, but it hasnt been possible to pull it off yet.
Mission: Make a Gadgeteer device stand on a field powered by a 12V car battery, and report back to a server over HTTP on a daily basis.
I have given up on the NETMF networking stack over cellular (using PPP), so I have turned to basic usage of the SIM900 firmware, which support HTTP communication. I have made it work a few times, but its about as unstable as the PPP solution.
After browsing lots of pages with other user of the SIM900 module it hits that the Seeed schematics might not be completely following the requirements. I have seem Seeed modules before that was working, but not in any way well designed (sorry Seeed).
The “big” capacitors, C1, C3, and C13, are all in the schematic in area A-to-B, and 4-to-6 (top right-ish area). Of the big 470uf tantalums, C1 and C3, one is on the input to the regulation circuit, and one on the output; the big electrolytic supercap C13 is a supplemental power source for the module, and goes through a diode D3 to the VBAT net. C2 is a small 100pf (?) that’s also on the output.
As Andre says, adding capacitance to smooth out noise is usually done as close as possible to the power hungry device. In this case, the ideal location would be as physically close to the VBAT input pins, so if you could find pins 54 and 55 of the SIM900 (or 57 & 58 for that matter) you could connect to those with your additional capacitance. But there’s a reasonable amount of capacitance there, and I’d be interested to see if there was actually noise on that line before assuming extra capacitance might help.
They may not help with the power transients that the module objects to. The supercap is HUGE, at 0.22F and 5.5V, but that’s not going to protect against all fluctuations, even though it is massive.
You mention your target state is huge 12V battery. I don’t really expect a power source like that to fluctuate much. What power conversion do you have to power the module and mainboard ? Do you have the same issues when not powering from a 12v battery, but say running from a 12v mains power supply through the same power module ?
Having said all that, sometimes you just have to add more capacitance - it’s usually the ESD value that is more important than the capacity, in high drain scenarios.
If you have it, check the cellular module datasheet. SIMCOM are very good on this in regards to the power supply requirements. The thing to watch with big capacitors is the ESR as this will dictate the ability to respond to fast transients. Also watch as big caps also appear as an initial short on the supply on power up and it can slow the rise time of the supply as the cap charges up. Make sure the reset remains active until the supply is at the correct level.
For a SIMCOM modem board I designed, I used 100uF low ESR TANT’s on each power input with a 0.1uF CERAMIC X7R in parallel and two 330uF on the power supply output. These then provide the required transient response from the modem during transmission bursts. Good thick tracks on the power rails also help. I used a 4 layer design with 1 internal layer for the power to the modem.
And still through your DP module ? Perhaps it will be a little more responsive (a car battery can certainly deliver more than the 2.5A power pack you’re using) but the DP is potentially still the bottleneck. You still have to remember that the power conversion for the module is done on-module, which is where the power concerns will really be.
lol. No I just meant your comment that the design isn’t that polished. Maybe they’ve revised other modules (non Gadgeteer) as they learnt from the original implementations, but the Gadgeteer modules from Seeed always went unloved. Even as far as drivers; it is pretty easy to see their business model was churn out relevant modules and hope to make money on them with no reinvestment in improving or even moving the drivers into newer frameworks.
here’s how Seeed modified their SIM900 Arduino module v2 - check out the oscilloscope diagrams towards the bottom of the post. GPRS Shield V2.0 - Seeed Wiki That can’t be affecting your module as you only see instability after some time, but still they made some reasonably significant power changes,…
@ njbuch - I can confirm that I’ve also experienced these power issues whit HTTP calls. For simple SMS, power up, registering with the network, getting IP address and all that it works.
To me the issue was that I could never get an HTTP response. The request was sent OK but the module always reset before receiving the response. I ended up connecting a bench power supply to get it working.