Fubar Solar Power Setup

[ruku]

Thiiis one's special. Not my design, but I need to do analysis on it.



So we have a sensor that needs +/- 5 V power. Threw on a switching regulator to take care of that . Problem is, the output is bipolar, need to make it unipolar. Op Amp you say? Fshaw, silly! Lets do something else.

Split the rails, drop the potential for the AVR microcontroller.

There has GOT to be a plethora of things that are wrong with this. For one, there's no charging protection on the battery, but disregard that. It's the least of my concerns. When the sun is not there, the switching regulator no longer runs, so the microcontroller is back at normal potential.

Can anyone tell me WHY this design is bad? Innately I know this, but I can't explain WHY.

Edit: Fixed the schematic, was wrong on something.

[Jimmy]

What are you trying to acheve here

[IanB]

I'm confused. I can't make much sense of that drawing.

Does the switching regulator supply the mystery device at the lower right and also the resistor divider?

Where does that regulator get power from if the sun isn't shining?

Where do the +/- 1.65 V taps off the resistor divider go to? Is any current being drawn from them?

What exactly is the mystery device?

[ruku]

Haha. Terrible me, I didn't explain it.

TL;DR - We have a sensor (the mystery box) that provides a bipolar signal (+/- 1.65 volts). The problem is the ADC on the AVR can't read anything below 0V (or so).

Instead of properly biasing our signal, we lowered the "ground" potential for the microcontroller. That way, signals at -1.65 Volts appeared to be at 0 V, since the microcontroller's ground was at -1.65 volts.

[IanB]

When you analyze circuits, you can't only think about "voltages" (I use the quotes there deliberately). You must also think about where the currents are flowing (or need to flow) in a circuit.

If you have a truly isolated power supply, and you use that for the 0 V power connection to the microcontroller, and you use the independent battery to make the 5 V power connection to the microcontroller, how is an electrical circuit going to be completed for the current consumed by the micro?

Why on earth don't you just connect the AVR power rails to -5 V (0 V) and 0 V (+5 V) on the regulated power supply?

If your sensor output voltage is then in the wrong range for the ADC on the micro you should use a voltage level converter to bring it into the right range.

[ruku]

A - If you have a truly isolated power supply, and you use that for the 0 V power connection to the microcontroller, and you use the independent battery to make the 5 V power connection to the microcontroller, how is an electrical circuit going to be completed for the current consumed by the micro?

B - Why on earth don't you just connect the AVR power rails to -5 V (0 V) and 0 V (+5 V) on the regulated power supply?

C - If your sensor output voltage is then in the wrong range for the ADC on the micro you should use a voltage level converter to bring it into the right range.

A - Nope, not quite isolated. See how I'm connecting it to the ground through the rail splitter?
B & C - Sure woulda made a lot more sense than what the previous designer did... I agree. Though my hands are tied. I can't change the design. The only thing I can do is convince management that this is a bad design.

[IanB]

A - If you have a truly isolated power supply, and you use that for the 0 V power connection to the microcontroller, and you use the independent battery to make the 5 V power connection to the microcontroller, how is an electrical circuit going to be completed for the current consumed by the micro?

B - Why on earth don't you just connect the AVR power rails to -5 V (0 V) and 0 V (+5 V) on the regulated power supply?

C - If your sensor output voltage is then in the wrong range for the ADC on the micro you should use a voltage level converter to bring it into the right range.

A - Nope, not quite isolated. See how I'm connecting it to the ground through the rail splitter?

OK I see see now what I missed: the AVR is powered directly by the battery and then referenced to the independent supply for the sensor thingy.

B & C - Sure woulda made a lot more sense than what the previous designer did... I agree. Though my hands are tied. I can't change the design. The only thing I can do is convince management that this is a bad design.

So the solar panel is connected to the input side of the switching power supply. Where does the return current go on the input side of the power supply? Shouldn't the power supply be connected to the battery and not to the solar panel?

If the power supply were powered from the battery, it seems like the circuit would work. What is wrong with it apart from the darkness issue (and the lack of charge regulation on the battery)?

[ejeffrey]

Passive rail-splitters are a terrible idea for low power (i.e., battery/PV powered) gear.  You need an op-amp to buffer your split rails.  Once you have an op-amp, make a level shifter for the sensor output and you are done.

When there is no sun do you need the sensor or only the MCU to remain powered?  If you also need the sensor, you will need to put the switching regulator connected to the battery, and power everything from that.

[House91320]

Sticking to Dave cad nice.

[Simon]

Sticking to Dave cad nice.

DaveCad is one thing, "that" is jiberish !