There’s an ongoing debate about how deeply you can discharge your battery and what the long-term effects are. A common method to maintain a safe discharge amount is to use a voltage alarm. In this article, I set out to explain how to set up a reliable alarm and the effects of choosing different set points for that alarm. Fair warning: it isn’t really going to accomplish either of those things. If you use a voltage alarm when you fly, you should read it anyway.
Stressing a battery to the point of failure is useful for container testing, but generally something we all want to avoid—we really want to know the best way to keep our batteries healthy. So, we set out to learn about voltage alarm set points.
The test laid out like this: procure three identical batteries and use them under specific test conditions. Fly one down to 3.5V per cell each time, one to 3.4V, and one all the way to 3.2V. Measure voltage during flight through the flight controller, transmitted by telemetry to my Taranis. Program the Taranis with an audible alarm that selects between these set points based on a switch position. Record flight time, time × throttle %, total voltage, and battery temperature for each flight. Record individual cell voltages, internal resistance, and battery thickness after a rest period. Record this data again after each storage and charge cycle.
Based on what we’ve read, draining to 3.5V in flight should leave a comfortable margin in the battery. Draining all the way to 3.2V in flight causes a significant strain on a battery and uncomfortably nears the “irreparable damage” zone. These end points should provide very different results over time. As the batteries degrade, we should see a rise in internal resistance or measure puffing through battery thickness.
On the very first outing, I learned something about these batteries: 3.2V is just too low a set point for the alarm. The instant the 3.2V alarm sounded, the battery dramatically stopped delivering power. There wasn’t enough left in the tank to bring the quad back in for a soft landing; it went down right where it was. Afterward, sitting disarmed in the field, the telemetry reported the battery voltage plummeting: 3.0V. 2.9V. 2.7V. This didn’t sound good; I’d used up the entire capacity—the camera and radio equipment were draining out the dregs. 2.5V. 2.2V. I was rushing to the quad at this point, expecting to find a stressed battery puffing out. When I got there, everything looked perfectly normal, but I still disconnected it right away, set it aside from the other equipment, and treated it with extreme care. The last readout before telemetry cut off was 1.8V per cell: just 7.2V across the entire 4-cell pack. Everything I’ve ever read about LiPos was perfectly clear: I’d killed it. Once disconnected from the load, the voltage jumped up a little when tested with a voltage checker: I recorded 10.8V into my notes. That’s 2.7V/cell: 10% below the 3.0V figure often quoted as an irreversible self-destruct point.
I got it home and let it rest and stabilize for a few hours. When I went back to read its voltage, it had recovered to 12.8V (3.2V/cell). Usually a battery recovers a few tenths of a volt at rest, but this was pretty significant. I set the battery to storage charge and everything went normally. After bringing it to storage charge, I took some measurements. Both thickness and battery resistance were exactly the same as when I’d received it. I charged up a few days later without issue. Did it really survive unscathed?
I persisted with the experiment, leaving the alarm set point at 3.2V for that battery, and cringing every time I took a long walk out to wherever in the field the battery gave out. Each time there was a similar story: dropping below 3.0V before I could disconnect the battery, recovering several volts at rest, then measuring near normal after all is said and done. I hammered this battery into the ground at least five separate times. It just did not want to show even the slightest form of wear.
Meanwhile, the battery with a set point of 3.5V had a cell go out of balance during a storage operation, reading about .4V higher than the others. It took over 2 hours to get the battery balanced back to storage, and then again took longer than usual to balance charge the next time. This battery was treated very well from day one and almost never went below 3.4V, even under load—yet it was the first to behave badly and show signs of wear.
At this point in the experiment, it became clear that we weren’t going to get useful, reliable data from the test. We called it off before pounding our 3.2V battery any further outright caused a fire. You’d be absolutely justified in calling out the small sample size here. Getting reliable information for this kind of test would require a significant investment of time and materials, and probably needs more tightly controlled test parameters and environment.
When it comes to batteries, there’s a lot of variability. One manufacturer’s batteries don’t perform the same as another despite having the same specs on the label. Even with the same manufacturer, some perform better than others. The battery that won’t die surprises many people I talk to and might be a good review for that brand, but it’s the same brand and spec of battery that is showing the unbalanced cells already. We couldn’t say if this is normal variance or a pair of unlikely outliers.
Here’s What We Did Learn
Mostly, we learned that there will never be a simple answer to where your voltage set point should be, and that you need to do some work on your own. But let’s break it down.
It’s Really Hard to Hit a Voltage Set Point Consistently
Flying a quad for racing or acro involves a lot of quick throttle changes. Your voltage is constantly bouncing up and down depending on how much load your battery has on it. In this kind of environment, it’s often difficult to decide whether your alarm going off is a temporary symptom of a quick maneuver and will stabilize, or whether it’s time to land. This might be fine if you’ve given yourself a wide range that’s acceptable to finish at, but it’s nearly impossible to hit a very specific endpoint. Take a look at this Blackbox graph. Could you have decided when to land for a specific result based on this kind of voltage?
Because of the extreme variability of voltage during flight, alarms often work differently. Some will alarm immediately when a voltage threshold is reached, others wait for the voltage to stay consistent below that level for a period of time before activating. Most alarms shut off if the voltage rises again, but some don’t. These features are designed to make the alarm more reliable, but this ultimately only works to your advantage if your flying style and your understanding of the alarm matches what the developer intended. Alarms designed differently require different set points to get the same result.
Other People Aren’t You
If you do an image search for LiPo curve, you’ll bring up discharge test graphs for batteries that look somewhat similar. Take a look at the voltage when 80% of the capacity is used. You can find examples that show 80% capacity is used from 3.2V all the way up to 3.7V. This depends on discharge rate, but also the specific batteries in use. That’s a huge range.
Let’s assume that you’re reading a recommendation for an alarm set point online. The voltage given has a certain set of expectations behind it: the performance of the battery being tested, the expected discharge rate, and even what happens between the alarm sounding and getting the battery disconnected. Do these apply to you? You’ll get very different results from battery performance, the weight of your craft, passive or aggressive flight styles, the age of your batteries, and setting down right away vs. flying back in from half a mile away. The alarm set points for these conditions need to be very different if the power draw is to end up the same.
Get Your Own Data
This doesn’t have to be hard. Check your battery voltage after each flight. You almost certainly already have the equipment for this: your charger, flight controllers, or multimeter may read it for you. If not, a basic voltage checker is only about $2. Also, pay attention to the amount of energy that’s fed back into the battery at charge time. If you notice that these values are off from what you expect, adjust your alarm monitor up or down a little. Do this each time you fly and charge. It only takes a moment, but allows you to constantly update your monitoring to match all of the different variables in play.
Use More Than One Method
If you can, use a secondary method to estimate your battery capacity. I like to use the “Throttle %” timer available on my Taranis. It’s like getting a second opinion: sometimes it feels like the voltage just isn’t reading the way you expect, so the second method makes sure you don’t overdo it. This can also pick up the slack if your primary method fails—like when your OSD finally gives out.
Deciding When to Land
Our conclusion is basically that everyone has to solve this problem on their own. Your equipment, how you fly, your risk tolerance, and even your values, (such as how you feel about performance vs. longevity,) affect your approach. The best approach can’t be prescribed by any other pilot on a forum—regardless of being any sort of expert in the field—because they don’t understand your situation. While the voltage alarm might be a terribly unreliable method for race quads, it’s only one of many tools available. If you’re looking for voltage set point advice, we don’t have it right now. As a result of our experiment, though, we put together an article on deciding when to land which outlines how to use voltage as well as other methods. Give it a read! We hope that it will give you the tools you need to make the best decision.