There’s some formula I’ve seen on rcgroups telling what’s needed per foot of 12awg but don’t know. I wonder what would be needed to see what that split second voltage and current was when the magnetic field colapses, instead of assuming based on an inductance measurement. @Battery_Mooch can you? I’m assuming not with your battery thing. Or can you run the battery while testing?
I read on the vedder forum the vesc does better than expected related to that formula.
And forget what type of caps specifically needed. Low sl or something.
It’s easy to see what the voltage spikes are and what effect caps have on that using an oscilloscope. I regularly use my scope to check on inductive spiking in my designs.
DC current levels can be checked with an ammeter or a current shunt and a voltmeter. We don’t necessarily need to know the current level for the spikes themselves.
My battery analyzer can be used while the battery is running, great for tracking internal resistance over time as the pack warms up. But the analyzer won’t help us here as the inductance and capacitance probably won’t change as the pack is used.
Low impedance capacitors are needed, low ESR (equivalent series resistance) they are called. Basically, the lower this resistance value is the better the cap can “swallow” voltage spikes and the less it heats up from the battery voltage “ripple” caused by the ESC pulsing the battery current thousands of times a second.
But the lower the ESR, the larger the cap is and (typically) the more it costs. At a certain point the little bit of lower resistance you can get isn’t worth the $$$ and room it all takes.
It’s better to parallel several smaller capacitors rather than try to do it all with one or two. Paralleling capacitors lower the ESR and spreads out the heat, helping the caps to shed the heat faster. But spreading them out too much increases theIr total inductance and reduces their effectiveness. It’s a balance between a bunch of different factors but for most applications it doesn’t matter much.
This is a pretty interesting conversation guys. I know shit all about batteries, but its nice to follow along.
I do know quite a bit about pulsed loads and getting FETs to survive them. I routinely check dc power supply’s for how well they handle pulsing demand currents. I would do the same as Mooch and put a scope on the output to monitor the voltage, checking to see if improves with caps. However I am sure the voltage spikes would be smaller with no load on the motors?
Mooch, what’s the typical battery self inductance you see at the cell level? Maybe one could build a pack model?
Besides binding the battery leads together, decoupling capacitors are actually used to reduce inductance (they are placed on the ESC side). I don’t know the formula or calculation on top of my head as I would need to dig up my EE notes.
Ya, I assume once the current to the motor shuts off that inductance becomes unclamped bringing the voltage up on the esc. The cap should provide a place for that current to go. The key is know the magnitude and the frequency content of that spike to select that cap and size. What are the motors pulsed at? 30kHz? Maybe a film cap or a hybrid polymer would do. I usually just test this empirically with a scope.
A quick check shows that a Samsung 30Q has about 2-3nH of inductance. The inductance of a two foot length of 12AWG cable ranged from about 250-650nH though, depending on how the wire was positioned. A huge difference. I have no idea what the inductance of strip is.
A model of the cells themselves, connected as a pack with estimates for the inductance and resistance of wire connections, would be easy. The problem is how do we accurately model the strip and the different arrangements of the main wire leads…so many ways to do it.
IMO, since we would have to check several packs to validate the model anyway it’s probably just easier to measure a bunch of packs and/or just go by the rough numbers given in the R/C groups and the esk8 ESC manufacturers for the filtering cap values and their recommendations for any additional capacitance needed.
Since too much capacitance isn’t a big issue (if anti-spark is used), other than size and cost, it doesn’t hurt much to overestimate what’s needed to handle longer battery wires.
Not a huge cap…the right cap(s).
A bunch of smaller “low ESR” caps is better (more effective and runs cooler) but you still need to roughly determine the capacitance value. The ESC manufacturer will probably have some recommendations and/or you can check the R/C forums.
All of my packs are drastically different except for the evolve packs. Those are 10S4P 30Q with the pgroups welded of 10mm x 0.15mm nickel which is jumped into series with 1" 1/4" (ish) lengths of 1/4" copper braid. I’m sure the field would be fun looking visuals.
this seems like as good place as any to ponder a 4x4 battery wiring arrangement…
instead of a traditional battery arrangement where the leads come off one end, and a second set of leads run lengthwise along the pack to the other motor controller, I’ve been pondering making a battery where the leads come off the center of the pack and then leads run only 1/2 length of the pack:
schematic:
my actual intention is running the battery leads down the center of the pack.
Rather than having all the ESCs at an end, you can pack them all in the middle
But the splitter in the middle like you drew is probably the best option, unless you are super aggressive on the throttle it’s probably fine, that’s not an unusually large board
I’m curious what kind of effect going to say, 8AWG and twisting the power cables over those 2 feet. Would you think that twisted cables would produce less inductance based on your testing? As significantly different as your #s show? Does half the distance (1ft) result in less than half the inductance?
Also, there is enough variation in pack building, it will be hard to test them all. I’m more interested in a known good battery and its inductance interactions with the system as it relates to wire interconnect length.
I thought I had a more in progress picture than this… but this ended up coming together really nicely. It’s basically how you see it here, only imagine the batteries both flipped so the terminals and balance wires are facing in. There is a single antispark and two 10S BMSs under the ESCs. It has two 10S2P charge loops separated by loopkey and could be charged all together, but will almost always likely be split and charged simultaneously.
Key takeaways, ultra short battery wires, sorta long phase wires, but everything is the same length across sides.
Twisting the cables would reduce the inductance. I don’t know by how much. You lose a lot of flexibility though.
Halving the length of a cable reduces its inductance by a bit more than half IIRC, assuming the same spacing and physical arrangement with its return wire to the pack.
I’d love to measure the inductance of some packs and come up with some general recommendations for building but it would require testing several different packs.
Just stumbled upon this thread. Do you guys think coaxial cables could be a solution to running long battery wires? Not sure if coax cable thick enough to handle our amps are even obtainable for less than a few grand though…
