Comments on: Monitoring Battery Voltage

By: Jolly Logic

Jolly Logic — Wed, 07 Sep 2011 16:30:45 +0000

Or this, three small resistors, two pins, and no residual drain:
1. VBATT–>5K ohm–>(Pt A)–>5K ohm–>(Pt B)–>100K ohm–>VBATT
2. Connect (Pt A) to your A2D port
3. Connect (Pt B) to an open drain output (hopefully you have one) of your microcontroller.
If you drive (Pt B) low with your open drain output, you can read 1/2 VBATT at (Pt A). When (Pt B) is high impedance, no current flows (both ends of the chain are at VBATT.

By: Jolly Logic

Jolly Logic — Wed, 07 Sep 2011 08:10:55 +0000

Or rather, less is wasted (your micro’s high isn’t as high as the battery’s)…

By: Jolly Logic

Jolly Logic — Wed, 07 Sep 2011 07:18:49 +0000

Came to this late, but wanted to post one nugget now that I found it.
Use the voltage divider approach, but to minimize current when you’re not measuring, ground the bottom of the divider using one of your micro pins (held low). When you’re not using the divider, hold that pin high, and no current is wasted when not measuring. Voila!

By: MoJo

MoJo — Tue, 17 May 2011 22:22:05 +0000

I agree with Andrés, you should get adequate results with your system. I did some basic tests with a similar set up (lead acid battery, charged by solar) and it was accurate enough to give a rough estimate of available time left before shutdown.

Most cheap electronics use something similar but built into an IC. Might be worth looking for some battery monitoring chips.

You don’t say what uproc you are using but AVRs have a couple of useful features. The first is protection diodes on the pins which will prevent damage to the AVR if the voltage is too high (assuming low enough current!) which can be handy when you want to charge the battery in circuit. Many also have an analogue comparator that you can use to measure voltage drop over a resistor to calculator current draw. If cost isn’t an issue then a hall effect sensor and op-amp to scale it is a better option though.

By: Andrés

Andrés — Tue, 17 May 2011 20:15:45 +0000

In my opinion (not a EE) your circuit will work. If you consumption is repetitive and you can draw your discharge curve (easy with a mc: http://www.google.es/search?q=battery+discharge+curve&tbm=isch), you only need to set the Idle-mode point in a voltage before the “knee” of the curve.
Regards.

By: John

John — Sat, 26 Feb 2011 06:30:22 +0000

VRLA’s are fairly well characterized – the open circuit voltage (Voc) correlates with the state-of-charge (SoC) and with temperature. There is a memory effect whereby Voc will be mis-read if the battery was previously under load – need to wait some time with no load. Keep in mind that the internal impedance increases with discharge (and Voc drops with discharge). If your load is well characterized, other than temp comp, you just need to know the terminal voltage. Lead Acid tempco is roughly -30 mV/°C. See: http://www.panasonic.com/industrial/includes/pdf/Panasonic_VRLA_LC-R127R2P.pdf and http://www.altestore.com/howto/Batteries-Measuring-State-of-Charge/a81/

Most gadgets these days either don’t monitor SoC or use some type of coulomb counter (aka current integrator). This would require a low-value resistor (limit measurement loss) in the current return path and a jellybean opamp (with rail-to-rail capability… not so jellybean I guess) with some gain/lowpass for your ADC.

By: Merlin Skinner

Merlin Skinner — Tue, 22 Feb 2011 07:54:03 +0000

Another thing to be careful of (as if you didn’t have enough to consider) is that ADCs “like” a relatively low input impedance. If you are using a potential divider with high values to minimise current drain, you won’t have that. This is particularly a problem if you are using a multiplexer so you can read several inputs. In this case, you have the capacitive load to consider as well. As mentioned above, a op-amp buffer will solve that for you but at the cost of more power and complexity.

I like Luke S’ idea of using a FET to turn off the divider when you are not using it. That way, you can have a “stiff” divider that doubles as a battery test load and still keep minimal (average) battery consumption.

By: Christioher Biggs

Christioher Biggs — Tue, 15 Feb 2011 03:00:50 +0000

Some years back, I wrote the software for a solar/SLA powered installation desgined by a “real EE” that did exactly that.

Worked for us, and I’ve always done it the same way since.

By: Ian

Ian — Sun, 13 Feb 2011 20:15:43 +0000

Sounds like for what you are doing its a suitable solution (and obviously already been debated). The only consideration I will suggest is what kind of lead-acid you are planning on using? Most are only suggested to be discharged up to 50% of their capacity which with this topology would be quite difficult to predict. If you want to get really fancy you might be able to look into a coulomb counter (or add a differential across a small shunt). Assuming you aren’t too worried about the depth of discharge I think you can reasonably approximate monitoring your ~C/360 point through trial and error (this could be affected by C1 depending on the magnitude). If you want to maximize your ADC range you can just buffer the output with a unity op-amp.

By: Christian Vogel

Christian Vogel — Sun, 13 Feb 2011 12:25:32 +0000

Ed Davies: A Zener in itself is not really “stiff”, so the zener-voltage will depend quite heavily on the current going through it. As you want to run as little current as possible through the zener, it will not regulate well. Also keep in mind the variations with temperature that also will be different for z-diodes with different nominal voltages.

(an exaggerated U-I-graph is on http://hyperphysics.phy-astr.gsu.edu/hbase/solids/zener.html )

Just use a voltage divider made up from precision resistors (and a trimpot to adjust away the remaining variations) and go into a differential ADC. If the remaining leakage current is too much cut the line with a simple mosfet when not in use.

Chris