Battery wire length and inductance

[swbluto]

Hello. Does anybody know what kind of inductance issues there are with long runs of wire? Is the inductance mostly from the wire itself or the loop area of the wire?

For those who don't know, the loop area of a piece of wire partially determines the inductance. More area = more inductance. By loop area, I mean the area enclosed by the wire. If you take a wire and then loop it back right alongside the wire, the loop area will be much smaller than if you were to form a circle using the wire. The circle version would thus have more inductance. More inductance = more voltage spike = more risk of harm to electronics.

Anyways, I'm planning on running a 5-6 foot length of battery wire and I want to get the "skinny" on what to address / avoid. Obviously, the wire's run should be as close together, so I'm thinking about twisting it... or maybe just zip-tie it.

I've also heard that if you parallel two inductors, you get less inductance, so parallel wires should help decrease inductance or not? I also think larger wires would help lower inductance.

If the inductance is inherent in the wire itself, though, then maybe there's not much I can do about that. If so, what kind of capacitances do I need to counteract that inductance?

 

[tostino]

I've heard that twisting the wires will help combat inductance, as well as larger gauge wire lowering inductance. Larger gauge wire is the same as paralleling wires though.

I don't have any real numbers to go from, because I don't know any of this for a fact . It's all just stuff i've heard at one point or another, so wait for someone who knows what they're talking about to comment! lol

 

[Toorbough ULL-Zeveigh]

what prompted ur concern over inductance?
i could see it being a problem for long phase wires but for <2m to the battery, i think straight up resistance would tend to dominate.

enjoy!

 

[swbluto]

what prompted ur concern over inductance?
i could see it being a problem for long phase wires but for <2m to the battery, i think straight up resistance would tend to dominate.

Mainly people who have apparently popped their ESCs from having too long of battery wires. I was wondering if there were "guidelines" that these hapless victims didn't follow that I could.

 

[Miles]

http://www.rcgroups.com/forums/showthread.php?t=952523

 

[Toorbough ULL-Zeveigh]

http://www.rcgroups.com/forums/showthread.php?t=952523
Conclusion from the links below, all controller manufacturers say the same:

what else are they going to say?
of course they're not going to admit they're cheeping out by not installing ample capacitors & point the finger back at the end user.
(same deflect tactic sturmey-archer used to cover up their design flaw in their 3-speed hub for decades).
at a 32 kHz switch speed, even including the harmonics is just too slow for such a small inductance to have much effect.
unless the phd types on here r able to prove different color me dubious, smells like urban-legend to me.

 

[AussieJester]

swbluto I am having issues with ripple on my hv160




Thats using 1/2 8awg and 1/2 10awg with less than 2 foot of wire from battery to esc ...
I have not been able to get it to stay under 5v no matter how many caps i add
to the esc how much shorter i make the wires, only thing that lowered it was taming
down the motor timing obviously in doing so lost some power albeit
only noticeable when viewing the data on the graphs...If you can come up with a fix
i'm all ears...


KiM

 

[swbluto]

swbluto I am having issues with ripple on my hv160

View attachment 1


Thats using 1/2 8awg and 1/2 10awg with less than 2 foot of wire from battery to esc ...
I have not been able to get it to stay under 5v no matter how many caps i add
to the esc how much shorter i make the wires, only thing that lowered it was taming
down the motor timing obviously in doing so lost some power albeit
only noticeable when viewing the data on the graphs...If you can come up with a fix
i'm all ears...


KiM

Have you tried twisting the battery wires together along its entire length?

 

[AussieJester]

swbluto I am having issues with ripple on my hv160

View attachment 1


Thats using 1/2 8awg and 1/2 10awg with less than 2 foot of wire from battery to esc ...
I have not been able to get it to stay under 5v no matter how many caps i add
to the esc how much shorter i make the wires, only thing that lowered it was taming
down the motor timing obviously in doing so lost some power albeit
only noticeable when viewing the data on the graphs...If you can come up with a fix
i'm all ears...


KiM

Have you tried twisting the battery wires together along its entire length?

Nope i haven't and can't not the entire length anywayz mate. remebering the
throttle interface and turnigy meter are also on that 'battery wire length'

I can do about half way though shall try that today...also spoke with Matt will be limiting the esc to 160amp
i'm pretty confident the ripple will stay under 3 then considering its taken 270amp
to pop it up to 6.5

Cheers for the suggestion shall implement when the sun comes up this morning

KiM

 

[Toorbough ULL-Zeveigh]

http://www.audioholics.com/education/cables/calculating-cable-inductance-of-zip-cord

 

[rhitee05]

The reason inductance matters is because the controller is not drawing a steady, constant current from the battery - it's actually drawing a series of pulses due to the PWM. It will hopefully be obvious that current is only flowing from the DC bus when one of the high-side switches is on and no current is flowing during the freewheeling period. The source inductance (wires, battery, etc) prevents the battery current from jumping instantly, so the capacitors have to supply current in the interim otherwise the voltage will sag.

More inductance -> more caps required

Twisting the wires tightly should reduce the inductance significantly. As a side benefit, it will also reduce any EMI problems from all the pulsed current.

 

[Toorbough ULL-Zeveigh]

Twisting the wires tightly should reduce the inductance significantly

define significantly.

sur inductance matters, but i'm pretty sur that the small amount of difference in cable length (up to 2m) don't matter.
even if reduced 'significantly' down from a microhenry (rough calc) that's not a whole lot of inductance in there to begin with is it?
i haven't seen any numbers to back up how much the slight change in inductance affects the characteristic impedance or propagation delay or slew rate or whatever that the twisting of the wires is gonna provide along a 2m length.
if my own figures aren't way off it could be that i don't have a feel (yet) for the proper context in which to place the appropriate significance onto the numbers?

however it wasn't until miles posted the link to rcgroups that i might have hedged my bets (not knowing or even interested in knowing all there is about the various controller designs out there), but safe to say a controller that can't handle a few extra inches of battery cable in a model airplane has some sort of design flaw.
monkeying with the length of the power feed might band-aid the problem somewhat, but it's not the root cause.
which in AJ's case sounds more like an impedance mismatch in the controller itself causing the ringing, as a wag.


unless these controllers are switching at over a MHz or there's something else going on that i would not be aware of or have overlooked, i just don't see it being that big a factor.
i would very much like to hear what bob mcree would have had to say about this, as he has both the technical expertise & familiarity with rc gear & wouldn't mince words if this is much ado over nothing.

 

[Toorbough ULL-Zeveigh]

http://www.rcgroups.com/forums/showthread.php?t=171111&page=3



Since I was asked, a little background on myself. I am an aerospace engineer, with more than 20 years experience, working at the NASA Goddard Space Flight Center designing and building the ultimate radio control vehicle, Earth orbiting spacecraft. My discipline is electrical and data systems. I received a BS from U MD and am a former electronics instructor. I design, build, and fly my own ESCs.

First let me say Pattrick is correct in his comments on wire size and connectors. A good rule of thumb for all EP systems is that they should use the largest wire gauge practical, reduce the number of or eliminate connectors, and keep wire length as short as possible. Now for the inaccuracies in Patrick’s comments.

The statement “NEVER run long leads to the battery—you will kill the controller” is not correct. My earlier posting shows ripple voltage in both a short and long wire setup (16 cells running a Jeti controller and Nippy 1608/160). There is no appreciable difference between the 2 pictures. I have included pictures, in this post, of power on in-rush current and voltage. Since capacitors oppose a change in voltage and voltage lags current, initial power on results in a high current (2A in the example shown) dropping quickly to a low value (motors are not energized at this point). The voltage at power on ramps up from 0V to applied voltage in 5 times R times C. R in this setup is the total battery-to-ESC wire resistance and C is the capacitance of the filter capacitor. Input filter capacitors serve several purposes. They improve voltage regulation, reduce ripple, and reduce the effects of connecting wires. Ripple voltage is a function of input impedance and motor current. Impedance is represented by the letter Z and is equal to the square root of R squared + the sum of Xc (capacitive reactance) squared – Xl (inductive reactance) squared. Can anyone say Pythagorean Theorem? Voltage is equal to I (current) times Z (Ohm’s Law). In a typical battery to ESC connection, Xc is huge compared to R and Xl. Increasing the length of large gauge wire has almost no effect on impedance. This is why the scope pictures show almost no difference – there isn’t any! If there were a difference, the oscilloscope would show it. If the high voltage application is within the ESC’s specification the input capacitor will not fail prematurely. If it does, then the ESC is not properly designed. As shown earlier, Xc is the only real factor in ESC input connections and heat (Watts) is a function of it not wire length. Some additional points:
- Stray inductances are so small they have no bearing on circuit operation.
- Higher current ESCs may have larger filter capacitors, there is a practical size and weight limit involved, but why not just add another low ESR capacitor in parallel with the one on the ESC if ripple is a concern?
- The actual terminology is in-rush current not avalanche current.
- In-rush (avalanche) current is not caused by spikes it occurs only at power on.
- Spikes are caused by the fast turn on and turn off of the MOSFETs and the motor winding inductance.
- Most ESC MOSFETs are rated for 30V operation. If battery voltage is ~20V then the ESC can survive an additional 20V P-P ripple. In this case the ESC would see 30V maximum and 10V minimum.

If you need to lengthen wires in a specific installation, doing it between battery and ESC is the safest approach.



if it helps any, my bike has the power leads running from the battery pack in the front wheel up the fork leg along the length of the top tube & down the seat stay to the controller embedded in the rear hub with the motor.
it measures at least 6 feet long, closer to 6 and a half, probably the longest cable run between battery & motor you're likely to find on any full size MTB.
the cable is nothing more than a pair of 10 gauge wires zap-strapped to each other along the frame with 2 sets of anderson pp connections in the way, one at the battery & one at the motor.
never had a single burp from the controller or the motor in nearly 10 thousand miles & i'm not the only one with the same bike with similar mileage that's never had a circuit or motor failure.

so a 6 foot battery cable shouldn't be a concern, if rc controllers are having problems stray inductance is a red herring & need to look elsewhere to beef them up.

 

[swbluto]

So, basically, Toorbough ULL-Zeveigh (What kind of name is that?), you're saying that high current controllers just don't have enough capacitance to minimize the ripple voltage occurring as a result of that high current? is adding bank capacitors supposed to help?

I'm only running my controller at 15 amps battery current so maybe i have nothing to worry about... I've already installed 1600 uF of extra capacitance or so.

 

[BatteryMooch]

- Spikes are caused by the fast turn on and turn off of the MOSFETs and the motor winding inductance.

IMHO, this is the most critical point. The frequency of the waveform isn't a problem, it's the speed of the voltage rise and fall in those waves that can cause problems.

You can have a 100Hz waveform with 100mS rise/fall times and you're not going to have any problems. If that 100Hz waveform has 1pS rise/fall times, every little thing you can think of will cause problems.

 

[liveforphysics]

As far as real world testing goes, people that have had reliability issues with controllers in large scale RC stuff, then moved the ESC to be butted right up against the batteries rather than right up against the motor, and have cured the controller reliability problems they were having.

Adding loads of caps is kinda a band-aid solution IMO. Setting something up with a design that inherently minimizes the need of the caps seems like a logical winner IMHO.

 

[Tiberius]

So many words, so few calculations.
Can I sell anyone some linear crystal, oxygen free, low inductance battery cables?

Reactance = 2 * pi * freq * Inductance

The inductance of battery cables will be of the order of nanoHenries. Even if we assume 100 nH, the reactance X at 10 kHz is 0.006 Ohms. Its not significant.

What is more likely to be going on is not effects at the PWM frequency or even in the DC-100 kHz range, but ringing and resonances in the 1 to 10 MHz range. There will be excursions in voltages due to ringing all over the circuit; they are very difficult to capture and measure, they depend on fine details of the circuit and layout, but they can take the FETs and other components outside their voltage ratings.

Its possible in some controllers that the battery leads interact with these. In fact its likely in most controllers since the electrolytic caps won't be effective at these frequencies so there is little isolation. So there could well be cases of controllers sensitive to the battery leads. The battery leads are lengthened, the pattern of spikes and ringing inside the controller changes, something blows up, and the myth of battery lead inductance is born. The fact that the effects are not repeatable even adds to the mystique.

A better solution would be to isolate the leads at high frequencies. Ie., remove the extra electrolytic caps and add some series inductance. Even better would be to have the isolation designed into the controller. Do any of these controllers or ESCs have EMC approvals?

Nick

 

[Toorbough ULL-Zeveigh]

As far as real world testing goes, people that have had reliability issues with controllers in large scale RC stuff, then moved the ESC to be butted right up against the batteries rather than right up against the motor, and have cured the controller reliability problems they were having.

Adding loads of caps is kinda a band-aid solution IMO. Setting something up with a design that inherently minimizes the need of the caps seems like a logical winner IMHO.

as u know since u work with power supplies of all stripes & sizes, u can get the same oscillations/ringing problems running long lead lengths in a poor layout/design of even a lowly 7805 (or any, really) linear reg.
setting it up properly & adding a few pennies worth of extra appropriately placed components can eliminate the problem.

 

[rhitee05]

I hate to burst some bubbles, but battery lead inductance is not a myth, nor are the effects it can have on a controller. The values seem tiny and insignificant, but they have real effects. I set up a simulation to demonstrate.

Assumptions:
- 100V system running at 100A
- Battery leads have 2 mohm resistance and 600 nH inductance (should be equivalent to roughly 1 meter of 8 AWG wires)
- Dual 1000 uF input caps with 100 mohm ESR each (a good quality electrolytic)
- 20 nH trace inductance b/w caps and FETs (about 2" of PCB)
- 2.2 uF ceramic cap located right at the FET (4 nH ESL, 4 mohm ESR)
- Rise/Fall times of 500 ns for the FET current, which should be appropriate for a Xie Chang-type controller

Here's the model:


Here are the results (blue is FET voltage, red is main cap voltage):
View attachment 1

Note that despite all that capacitance, the voltage at the FETs is spiking ~8V above and below the battery voltage. If you're running 100V FETs at say 95V, poof. That's the risk. Note that this model still ignores some other factors, like inductance in the battery itself.

 

[swbluto]

Is 500 nS really the rise/fall time of the xie-chang controllers? I thought it was closer to 1000 nS at one time...

 

[rhitee05]

The total switching time has been reported as about 1 us = 1000 ns, but the voltage and current do not change simultaneously. The current rises first and then the voltage drops. It's a reasonable assumption that the rise/fall time for the current is about half of the total switching time.

 

[Tiberius]

Hi Eric,

I'm surprised at the 600 nH figure; it seems too high to me.

Apart from that, we're singing from the same hymn sheet, except that you are working in the time domain and I tend to work in the frequency domain.

Any battery cable effects are not happening at the PWM or commutation rate; they are at the nanosecond rate. They are not, as the myth goes, a problem of the cable inductance stopping the battery supplying current in PWM pulses.

I can't really tell from your plot, but it looks as if the ringing on the FETs is somewhere above 1 MHz.

Its not the PWM pulses that are the problem; its the switching transients and these should not be on the battery cables in the first place. The switching transients, etc, should be dealt with close to the FETs and not allowed to get near the cables exiting the controller. What is needed is more sophisticated capacitor arrays and filtering, and I hope the better controllers have them.

Nick

 

[rhitee05]

Actually, I think 600 nH/m is fairly conservative. 1 uH/m might be more realistic, and that's assuming the two wires are very closely paired for the entire length. I wasn't able to quickly find the dielectric constant for standard wire insulation, but if you assume a permittivity of 2, use 0.16 cm as the radius of 8AWG wire and a separation of 0.6 cm, the result is almost exactly 1 uH/m.

http://www.cvel.clemson.edu/emc/calculators/Inductance_Calculator/wire2.html

The root cause of the problem is that current demand from the bridge FETs is not continuous. That causes the source inductance to behave somewhat like a boost converter. Capacitors are the solution, but at the nanosecond time frame all the little parasitic contributions become important. For my example, the bypass caps have a net ESR of only 50 mohms, but at 100A that works out to a voltage ripple of 5V. You can look at this in the frequency domain if you prefer, but I think it's more straightforward in the time domain.

I wasn't trying to imply it's an unsolvable problem, obviously it's not. Just trying to point out that it is a real effect and that there are a lot of small factors which influence how bad the problem is.

 

[swbluto]

The total switching time has been reported as about 1 us = 1000 ns, but the voltage and current do not change simultaneously. The current rises first and then the voltage drops. It's a reasonable assumption that the rise/fall time for the current is about half of the total switching time.

Oh, doi. Switching time = rise + fall. That makes sense.

 

[rhitee05]

Sorry, I think I might have confused more than I helped.

It's been reported that the switching times are ~1 us, that is the turn-on and turn-off times are each about 1 us.

However, the turn-on time can be divided into two discrete portions. During the first, the current is rising from zero to the full value, and during the second the drain-source voltage is falling from approx. battery voltage to approx. zero. Turn-off is the same in reversed order. So, I'm making the assumption that the time over which current is changing is roughly half the switching time.