ESC signal wire is important - wire vs twisted pair vs coax

Support my work / pawelspychalski
We usually do not think too much about a cable that connects a flight controller to an ESC. It's just a cable and if it works, then it works. However, it's not that simple. Poor quality and too long wire can be a cause of serious problems. Signal degradation lost frames, drones, and airplanes falling from the sky. Let's connect a flight controller and an ESC to the oscilloscope ( Rigol DS1054Z bit.ly/2RJ0gHB ) and see how the signal of PWM, OneShot, Multishot, and DSHOT looks on the plain wire, twisted pair, and a proper coaxial, shielded, wire. Oh yeah, there is a difference!
The coaxial cable used in this experiment is bit.ly/3uP6fsZ

[@thirtythreeeyes8624]

I've always twisted all wires on all my builds I learned how big of a difference it made on the early days of fixed wing FPV and have never even thought about it even my micro builds haha.

[@marcknab7456]

You know the kind of stuff that works right out da the box. That's my favorite.

[@yokonunz]

Interesting but you are forgetting about the ground loop you are creating when connecting two grounds to the same device. Can you try to disconnect the ground from the signal pad on the ESC and redo the test leaving the rest untouched?

[@Tonynux]

The twisted gnd wire must be soldered on the ESC side only, to avoid Ground Loop. Ground Loop is something very bad!

[@MCsCreations]

Fantastic tips, Pawel! Thanks a lot! 😃 Stay safe there with your family! 🖖😊

[@rougeneon1997]

Cool. Ive always twisted appropriate wires when possible. I remember learning to use both sig and gnd wires for the esc signal somewhere years ago. Sadly the only fixed wing I have with a fancy esc protocol is my Nano Goblin but it has a rather short sig wire....Good Video

[@GivenRhythm]

Haha. Thanks for putting the machines to this and getting conclusive data, Pawel. This has been on my mind for a while. Awesome up! I feel that was that an inadvertent / vague J Bardwell reference at the end....

[@power-max]

You forgot a CRITICAL aspect of testing these. That would be interference from the high currents of the motor controllers and other electronics. Also you can get some voltage drop across the ground cable carrying power to the ESC. This results in ground rise at the ESC. So while the flight controller ground is say 0V, the ESC ground is more like ±100mV which could be enough to upset the single-ended 3.3V logic from the flight controller.

[@Tonynux]

But the twisted gnd wire has to be soldered on the ESC side only. Absolutely not in the FC, otherwise you will create a Ground Loop. Even if you use a coax cable, it must be soldered on the ESC side only.

[@power-max]

@@Tonynux I'm not even sure the shield even does much if anything in that case. You still have an issue of what is essentially a series noise voltage with the signal. If it exceeds 1.5 volts or whatever the logic level threshold is, then it will cause problems. And if the ground wire could be severed with one or more of the high side FETs conducting, the current will find a path to ground via the GPIO pin on the FC. That's very unlikely though if it is soldered correctly. In terms of magnetic immunity, only the center conductor of the coax conducts the AC current while the ground wire of the ESC carries the return current instead of the shield. The loop area will be the area between the center conductor of the coax and power ground wire looping back to the PDB and to the uC. That large loop introduces series inductance which tend to low pass filter the signals, generate radiating magnetic field and conversely be equally suspectable to external magnetic fields coupling into the loop. The shield does nothing. The shield only terminated at one end might still help with electrostatic EMI field shielding, as the surrounding shield will block the electric field from reaching the center conductor. But such electric fields only really affect very high impedance nodes on a circuit, like floating analog input pins on a micro or high gain amplifier inputs. That isn't the case with this anyway since the microcontroller dshot output is a low impedance drive, maybe around 10 ohms or less depending on the output resistance of the GPIO pin and series protection resistance. Really the best solution is to have a galvanically isolated differential signalling method. Ethernet for example uses transformers and low voltage differential signaling. Canbus also similarly relies on differential or pseudo differential input stage to help reject common mode noise. Optocoupler can be used but these can introduce bandwidth and propagation delay issues, and would also require 2 wires for TX and RX. Actually the bestest best solution would be fiber optics but that's getting into absurdity haha 😜 Actually, another practical possibility is to design ESCs that use entirely isolated gate driver ICs so that the microcontroller on the can be powered by the flight controller (using the flight FC ground reference) while the motor inverter MOSFETS are independently powered by the completely galvanically isolated battery bus. This is what higher power motor controllers have to do, guaranteeing isolation usually up to between 300V to 1000V

[@nuhbiwan]

Hi Pawel, what is the exact type of the coaxial wire you used? I want to use a wire like that for camera video signal and ground, and same for connecting the VTX. Would you recommend a setup like that at all?

[@MrLewta]

Very clear , never thought about it. Today testing a tiltrotor rig,.. had a rpm reduction on the left side motor... And can not find why !!!

[@MrLewta]

May this long wires ( not braided) be causing a change on esc rpm condition...?

[@perldition]

How does a bit of overshoot matter for the protocols we use to speak with ESCs? All the rising edges look perfectly crisp.

[@houdinifpv7423]

^ this

[@perldition]

Or perhaps phrased slightly differently: I'd really enjoy watching a video demonstrating a configuration bad enough to cause practical problems such as bit-flips on the receiving end, and if the checksum that's part of the protocols is enough for the occasional flip to not matter in practice or if perhaps you can make flips likely enough to have an actual impact on performance.

[@BBFPV]

But shouldn't a coax actually just shield the cables from each other and not improve the signal from a single cable. The peak is only gone because the impedance is higher, I think 100µF/m. With a longer cable, the signal edges are then no longer so steep. At least that's what I learned in my studies.

[@Dron1973]

Thanks!

[@boactech3076]

Better also add pictures of you test setup. Now I have to guess if it also apply to my quad setup. Wire is 2 cm wrapped around the only ground wire to the esc.

[@Fleche_FPV]

Too bad we can not see more the evolution of the signal over time to realize the scale compared to the rest of the transmitted pulse train ....

[@Ender_Wiggin]

Where did you get the coax wire?? What was the impedance of the wire? are you sure the ringing has to do with impedance, not the shielding??

[@FPVUniversity]

bit.ly/3uP6fsZ

[@jimbrown563]

Very informative ...... But what about actual external Noise-Rejection ? Is external Noise even a consideration ?

[@litterbug4023]

I use coax for my analog Video on FPV Planes for a much similar reason. Better video clarity......

[@rolisaenz]

Pawel, i was looking for something like that but for video, i think that those coax mini wires that the video heads use on camcorders will do, but i was not able to find one mini video coax. have you have any ideas ? btw, i tought with coax video you will clean up a lot your VTX feed , great job on this video !!

[@WinstonSmith1984]

Thanks. Good info but what do you mean with "ringing"? Also will wire thickness make any noticeable difference? I assume thicker guage wire will have more capacitance with associated side effects.

[@Jasper_Vis92]

the "ringing" he is talking about is the small overshoot, the small bump at the beginning of the signal when the signal goes from low to high. not sure about wire thickness. too thick is just harder to work with if you are working in a small space. too thin would not be a good idea because the resistance goes up causing a lower voltage at the end of the wire. it won't affect a small mini quad but if you are building a large multirotor or airplane it will have an effect.

[@GivenRhythm]

The "ringing" literally refers to the unwanted resonances and / or feedback causing distortions in a sampled waveform. In this case, the surrounding high current paths + motors can generate sufficient electromagnetic radiation to affect the sensitive nature of signal path. Signal wire leading to and from the ESC need not have heavy conductors. In my opinion, the coax suggested by Pawel is sufficient for almost all RC jobs, because of the low current requirement of signal wire and the protective component to coax. I think it's only going to exhibit voltage drop at lengths over 2+ feet.

[@vasiliyklx450r]

The shape of the pulses doesn't matter, what matters is the count of pulses, no?

[@sparrow082]

How long can a single signal wire be before you get signal degradation that's noticeable?

[@FPVUniversity]

That really depends on the wire itself and the conditions. In general, i should never happen on quads < 8inch and not really on foamies as long as you do not keep FC on one wing tip and ESC on the other

[@sas2dj]

Talking about interference I have a Caddx Tarsier that makes me fail safe on 4 different receivers I shielded the Tarsier and still the same Interference can be very dramatic

[@thirtythreeeyes8624]

Did you shield the camera cable with a ground?

[@sas2dj]

@@thirtythreeeyes8624 Yes

[@Blox117]

idk man, i have several feet long wires but it doesnt affect any ebikes performance

[@FPVUniversity]

ebikes do not ecen try to update the motor speed 8000 per second

[@magicalpencil]

I wonder if the skin effect is causing the issue at higher frequencies? We will have to start using copper tubes lol

[@godsun358]

How do you add coax to connect FC to ESC

[@rougeneon1997]

Id assume the center wire, the one being shielded, is the signal and the outer wire "jacket" is ground. Since P.S notoriously pays little attention to comments Id figure Id reply. That's how I would solder it up.

[@godsun358]

@@rougeneon1997 Bravo Bravo brother I will subscribe to your channel 👍

[@rougeneon1997]

@@godsun358 lol I appreciate the gesture however I don't make educational stuff. That being said feel free to msg me questions in the future.

[@godsun358]

@@rougeneon1997 I already subscribed cheers

[@marcknab7456]

Most coax is steel coated copper lmao

[@frost5268]

EWM ;)

[@misfit333fpv6]

Great job frenchbread! Jesus Christ that accent is so annoying!!!