I'm trying to get power supplied to a 3g cellular board. The board manufacturer, Adafruit, changed the requirements on me after I started putting the board together, to say that it requires a li-po battery, but this is going into a barn where it won't be attended. Those batteries are a known fire hazard, so I'm trying to do a non-li-po supply. I have about 600Ah of solar-powered 12v batteries available, which I have to step down to the 4v the 3g chip wants.

The chip manufacturer says you should be able to do this with a LM2596 switching regulator, but recommends a 100µf tantalum capacitor or greater. Not having any tantalum caps handy, but having several 1.5f super capacitors around, I tried to use that, but the chip is behaving as though the voltage is dropping too low, and restarting.

This 3g chip has a slightly odd power requirement:



Diagrams are from the SIM hardware design document for this chip: http://www.adafruit.com/datasheets/SIM5320_Hardware%20Design_V1.07.pdf My LM2596 matches this diagram, except I don't currently have the TVS diode and fuse on the input side, and don't have the FB101 on the output. Instead I have a thermistor then the supercap.

When I plug this into a calculator for capacitor discharge here, I get values that say I should get a pretty minuscule voltage drop, but instead the board is resetting from voltage drop. Is this ripple? Is it the ESR of the supercap? Is this why SIM recommends tantalum caps? I'm assuming the switching supply produces more or less clean enough voltage, the thermistor is properly getting out of the way after the supercap initially charges up (seems to in my testing), and the supercap is more than large enough to deal with this surge current. So where am I going wrong. Also, which tools would I use to diagnose this? Thanks for any thoughts!

 

Check the battery by itself. Does it experience a voltage drop greater than 300mV when loaded? A larger battery, or maybe just a fresher one might be the only thing you need.

 

Check the battery by itself. Does it experience a voltage drop greater than 300mV when loaded? A larger battery, or maybe just a fresher one might be the only thing you need.

Kermit2 - there's not really a battery in this. Think of it as a very large 12v battery bank, a 12v->4v buck switching converter (supposedly rated to up to 3A), a thermistor, and a 1.5f capacitor.

Actually, here's my test rig:

 

I think that there is missing some information.
What are the data on the thermistor?
More data / type of supercap?
What's beneath the black tape?

Some years ago I developed some commercial GSM / SMS equipment.
It was something like two low ESR electrolytic capacitors on 1000uF after a switch mode power supply.
The first prototype was fitted with standard electrolytes and I saw how big the difference is on the capacitors.

 

If it is a cap ESR problem, then more capacity is the wrong approach. Parallel several smaller caps like your 100 uf to achieve an overall lower ESR for the group in parallel.
Also be sure this voltage drop isn't comin from a section of very small pcb trace or a bad solder joint. Any source of significant resistance in any portion of the conductor pathway could be dropping that voltage.

 

I think that there is missing some information.
What are the data on the thermistor?
More data / type of supercap?
What's beneath the black tape?

Some years ago I developed some commercial GSM / SMS equipment.
It was something like two low ESR electrolytic capacitors on 1000uF after a switch mode power supply.
The first prototype was fitted with standard electrolytes and I saw how big the difference is on the capacitors.

This is the thermistor, they say it's "NTC 5D-9 (Same as Ametherm SL08 5R003) Equivalent: Surge Gard SG413 & Ametherm SL12 5R003". 3A max current, "Initial Resistance is 5.0 Ohm +/- 20%, Reduces to 0.18 Ohm at 3 Amp Current", and it does seem to perform that way. Haven't taken a resistance reading when it's warmed up, but it reads 6Ω cold, and the current and voltage rise and fall like I expect them to when charging the capacitor.

Super cap is a Kamcap 5.5v 1.5f. I don't read enough Chinese (as in, none!) to see the full specs. No idea what the ESR is on that.

The black tape is covering the solder joints from my 12v supply wires to the ground wire and thermistor lead. The tape is really just strain relief for the wiring. I had the output side taped the same way, but took it back off to get voltage readings. Basically looks the same as the left-most side of that protoboard.

 

If it is a cap ESR problem, then more capacity is the wrong approach. Parallel several smaller caps like your 100 uf to achieve an overall lower ESR for the group in parallel.
Also be sure this voltage drop isn't comin from a section of very small pcb trace or a bad solder joint. Any source of significant resistance in any portion of the conductor pathway could be dropping that voltage.

That makes some sense. I get almost no resistance anywhere else. Haven't had to deal with ESR before this, and problems are usually the unknowns/misunderstood parts of the problem. That's probably also why SIM specified tantalum caps for this.

This is the sort of thing I'd need an oscilloscope to debug, yes? Most of the stuff I work with is DC and logic. I always think of oscilloscopes as an AC-world tool. But I can't think of another way (except maybe setting up an arduino to monitor the voltage for a few seconds) to watch the voltage changing here.

 

If you read section 9.2.2.2.4 on page 28 of this data sheet you will see the limits on maximum output capacitance.
The ESR of the capacitor must also be withing certain limits.
Using such a large capacitor and adding a resistor in series with the cap can cause the controller to behave erratically.
You can't arbitrarily change the component values on a switching regulator from the recommended values and still expect it to work properly.
That's not good engineering practice.

As far as tools, you need an oscilloscope.
Without that you are pretty well blind as to what's happening.

 

If you read section 9.2.2.2.4 on page 28 of this data sheet you will see the limits on maximum output capacitance.
The ESR of the capacitor must also be withing certain limits.
Using such a large capacitor and adding a resistor in series with the cap can cause the controller to behave erratically.
You can't arbitrarily change the component values on a switching regulator from the recommended values and still expect it to work properly.
That's not good engineering practice.

As far as tools, you need an oscilloscope.
Without that you are pretty well blind as to what's happening.

Look at the picture. The buck regulator assembly looks to be as specified in TIs datasheet. The thermistor and supercap are downstream of the completed regulator circuit and are essentially load for the buck regulator.