Compensate battery readings for voltage drop

[Nick Metcalfe]

Hi guys.

Should APM battery voltage readings compensate for voltage drop in shunt resistors and PDB wiring? I am having trouble setting a reliable battery failsafe as my power module sees a lot of voltage drop under heavy load.

It occurs to me that even 0.05 ohms from the battery to the sense point and back, when drawing 50 amps, is a loss of 2.5 volts. This is close to what I'm seeing in real life, using cheap Zippy 4s 5000mAh batteries, fairly heavy wiring and a point-source power distribution system. My last build using a non-3DR sensor showed much the same.

Is this your usual behavior? Say my battery is still at 14 volts, under load the PM reads it as 11.5 volts, triggering an early failsafe. Perhaps I could set my failsafe voltage lower, but then If I don't load the vehicle as heavily the battery could be in danger of discharging too low. I see some serious flyers turn battery failsafe off entirely.

Would a simple Ohm's Law compensation calculation given a PDB resistance parameter be useful? I would like to try and add this to APM. I imagine one could reasonably guess a typical resistance value as default and, even if it's a bit off for any particular build it should still be better than none. A simple calibration procedure might also work, but may not be necessary for many and might be fooled by bad batteries.

The goal is to have the reported battery voltage less coupled to current draw and make it more faithful to the actual battery voltage. This should make battery failsafes more reliable.

Or perhaps APM does all this already, I missed it completely and my copter is wonky :-)

Thanks,
Nick

[Jonathan Challinger]

The voltage drop across the shunt is very small. It isn't .05ohms, it is less than that. If you were dropping 2.5v at 50A across that shunt,it'd be red hot and its solder would melt and it would smell bad and a fire might start.

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Jonathan Challinger]

I recently made this same mistake on the skype, by the way. There is an amplifier after the shunt that feeds the current signal wire.

[Nick Metcalfe]

Yes I see what you're saying, 125 watts. Am I seeing the internal resistance of my cheap battery?

[Jonathan Challinger]

Most likely yes.

[Josh Welsh]

Agreed.  Taking a cheap battery from 5-15A at idle/hover to 50A+ on a burst can easily produce a real voltage drop of 2.5V, especially if it has any significant internal resistance and if its discharge rating is low.  The longer you hold that burst, the more the voltage can continue to fall on weaker batteries.

[jolyboy]

I wonder if its possible/feasible to compensate for lipo internal resistance so that your voltage readings appear to drop quite so much. It would just be a function of current; i.e v battery = v in * (current * internal resistance)

[jolyboy]

Correction, v battery = v in + (current * internal resistance)

[john...@gmail.com]

Theoretically yes, but it's complicated by having the battery change behavior depending on temprature.

[jolyboy]

Yeah, not much point really. If users are having trouble seeing voltage drops with big throttle I'd probably just tell them to get a higher C battery. My current machine barely sees much drop, 65C 6s 1550mah, but the old one did quite badly, 15C 4s 4000mah. No big deal to honest though

[Nick Metcalfe]

I have tried a new, freshly charged battery, same type, brand and size, and the voltage drop is nowhere near as bad, a little less than a volt.

So my old battery is about a year now and has done about a dozen cycles. It does seem slightly squashy compared to the new one but I wouldn't have called it puffed. Interestingly, my ESR meters show both batteries as same condition, about 50-60 mOhm each cell.

So it is the battery. This would then appear to make any compensation effort battery dependent. That's no good, who would realistically keep track of all that without smarter batteries? My battery is not smart but it may have some grunt left.

Never mind, I will find another use for my big tired old bag of chemicals. Perhaps an in-field coffee machine?

[john...@gmail.com]

The biggest problem is people just pushing LiPo batteries to hard. Even if the battery has some fantastically high C rating on the label, you just can't pull that much current and expect a long battery life.

In comparison I build some relatively huge battery packs for my eBikes. Typically in the range of 18-24S 10-20Ah. Becouse of the size I can draw several kW and still only have a 1-2C load on each battery cell. And doing this the battery last not hundreds, but thousands of cycles with no problem.

- JAB

[jolyboy]

I agree with what you are saying JAB, but sometimes drawing small loads from giant battery banks isn't quite as fun as massive power on a nimble light machine

[Robert Lefebvre]

Yeah, I thought that resistor wsa 0.005 ohm, not 0.05.  Or was it 0.0025?

[David Pawlak]

I use a 1cm piece of 1,7mm (2.5mm sq) copper wire for 80 amps :)

[Brian DeBusk]

Your best (but most expensive) solution is to use a higher C-rated battery.  And don't believe the C-ratings on batteries -- I have found them to be woefully over-rated.

As a next line of defense, consider running two smaller battery packs in parallel.  In theory it only doubles your C-rating, but because the batteries are over-rated to begin with it will typically help much more than that.  And if you have a dual charger with a sync mode, it makes the charging process equally simple.

And finally, there probably needs to be a couple of APM parameters for a smart battery failsafe. 

1) A flag to turn the feature on

2) A parameter for voltage compensation as a function of throttle in.  For example, a value of 0.02 would allow for a voltage drop differential of 2.0V at 100% throttle, but only allow for a voltage drop 0.08 V at 40% throttle.

And no, the drop-versus-throttle curve isn't linear.  I've tested it.  But in practice, the parameter I stated above is fairly accurate.  For a single battery, I might set it closer to 0.024 and for a dual battery setup I'd probably set it around 0.016 (assuming high-C batteries to begin with).

[john...@gmail.com]

The problem again is that LiPo battery characteristics can change, sometimes radically.

Not everybody is using high quality ultra C rated batteries and sensible currents. And so when the battery heats up during flight, there can be large changes to the voltage drop. Another factor is age. Putting cycles on the battery degrades the quality, leading to more drop as the battery ages.

The most reliable and simplest solution I have found, is still just a simple old fashion timer. Knowing how long you flew and how much you put back in the battery when charging, is all the information you really need to know the health of the battery and be able to track any changes over time. So a timer and the the old 80% rule, and telemetry voltage readings as a bonus to catch fringe problems and you really don't need any more to keep it safe.

[David Pawlak]

I'd prefer not to have the battery reading compensated. I like to see what's going on. I think if you compensate for throttle you may be cheated into  thinking your battery is good when it's not. You have a better feel for time left with the uncompensated reading.

The original problem here shows this. If the battery reading was compensated he would never have realized that his batteries sagged. They'd run along fine, and then boom, cut off.