ESCs and extended power wires: Should we add capacitors?

Motopreserve

Motopreserve

Drone Enthusiast
fltundra said:
I recap vintage tube gear on average every 10 to 15 years max. 30 hours is unrealistic.
Click to expand...

Ironically, recapping vintage gear will actually lessen the value of vintage gear. I've worked on a few Neve's that were ruined (ruined is pretty harsh - but they were never the same) by recapping. Same with 1176. There's something magical in the leaking of the caps that we all love...

ok, ok, sorry to be off-topic :)
 
Last edited by a moderator:



J

jes1111

Active Member
fltundra said:
I recap vintage tube gear on average every 10 to 15 years max. 30 hours is unrealistic.
Click to expand...
It may be... but: without objective data to suggest a more realistic interval, and since I rebuild the machine after 30 flight hours, it's convenient (and cheap) to change the caps during that exercise (as well as the bearings, the power connectors and most of the power wires).

I don't think that analogue amplifiers are a comparable application - an ESC is chopping high current at a very high frequency - the worst possible environment for a piddly little 50 cent component. Certainly I believe the quoted durability rating is to be ignored.

However, I'm also not wanting to argue on this peripheral point. I'm as interested as anyone else in econfly's willingness and ability to get scientific about what's really what with one of the "mission critical" components of our MRs.
 

R

RotoTwit

Member
econfly said:
The above just isn't helpful. Feel free to start your own thread on capacitors and ESCs (or on Amazon, audiophiles, general and obvious electronics observations, etc.). I very much want to keep this thread to a tight and orderly objective discussion. I will address the one quantifiable (and unsupported) statement made:



In the OP you can find a link to a review I did here that included three very popular ESCs (by Jeti, T-Motor, and Castle), all rated to 6s, and all using capacitors rated at 35v. So while you can believe anything you want, what you can't do (and did not do) is offer objective evidence that all three ESC makers don't understand their product and used the wrong components. In addition, I can say that my observations of voltage fluctuations on the oscilloscope and with no capacitors at all did not reveal voltage fluctuations at 6s power (25.2v at full charge) anywhere near 35v.

Please, let's keep this objective and on point.
Click to expand...

So many makers use poorly engineered designs and manufacturing - SURPRISE!, called 'just get by' engineering - cheap, you seemed concerned for safety and reliability - then why listen to an engineer?

I guess then it is 'pearls before swine', as a electrical engineer who once actually designed digital circuits for a living and designed and built many of my own analog audiophile designs, using very expensive capacitors and resistors and such, to good review by others, I 'have no knowledge' on this subject !

You say that there were NO voltage variations above 25.2v (then why the concern-you said otherwise?), was this with an oscilloscope, assuming you know what that is or even a signal analyzer (which actually is not as good - it pre-conditions signals displayed on screen) ? A lowly multi-meter is NOT a good measure of what you are doing - TBH, but then I think you do not want good analysis/debugging?

An oscilloscope lets you see the 'real' signals and voltages occurring, ac & dc, spikes and all - but of course you are more knowledgeable than I !

So, do NOT use my proven 'rule of thumb' and use the practical/simple engineering practice of 2x or greater voltage rating on capacitors, certainly an untrained person is more knowledgeable than an electrical engineer !

As well as cheap Chinese or any nation capacitors are just that - cheaply made thus cheap in price, high failure rates - 'garbage in - garbage out', so Please ignore any of my suggestions, I do not know what I speak.

'Pearls before Swine' !! Knowledge is scary!

Happy Crashing !! (Liability suite awaiting)

LOL ;)

Now on with the blind leading the blind!
 

R

RotoTwit

Member
jes1111 said:
That's what I'm hoping you can tell us :)

Just common sense :)

A few pages back it was suggested that temperature should be the focus and I'd agree with that: temperature is what destroys the caps, thereby exposing everything else to danger. Observing voltages may lead to an understanding of what's happening (with different lead lengths, cell counts, ESC ratings, etc.) but the final goal must surely be to provide a mechanism for estimating the appropriate replacement schedule for any given setup.
Click to expand...

Temperature is the 'indication' of internal leakages/elec shorting/material breakdown - caps do ok with external temps normally encountered - sub 200F or so, but not with internal shorting/self generated heat - which will continue to degrade the cap until failure - which could cause battery and ESC failure.

If you can get 'higher quality' caps (not cheap/poor Chinese made) with higher voltage ratings (>= 2x) and similar size/weight, then a single replacement should suffice !
 


R

RotoTwit

Member
Motopreserve said:
Thanks for the clarifications. Just one more.... :)

when you describe how to solder the caps on (if using multiple caps), are you saying to put the cap pins almost touching - so that they act almost like a single unit - or the opposite (staggered by a hair)? I understand parallel vs serial, but I have also seen premised units (like the ones castle sells) that are physically staggered because they're on a PCB.

I dont know that I need this after reading your assessment - but it's helpful actually knowing which caps would be best, and how to do it - since I wondered down the path of this research inadvertently :)
Click to expand...

NORMALLY you add caps in a parallel configuration to increase the effect of energy storage, more voltage stability (does NOT increase the voltage rating of the cap!), and exactly how 'closely' they are physically connected is a non issue, saving .00001 ohm of impedance for ultra close connection is irrelevant !

But connect close just to stay tidy/packaging, and remember you may want to remove them later so leave some room for ease of removal and no impact to other devices.

Spacing on a PCB is for assembly purposes, if you look at other devices you will see surface mounted caps, resistors, etc which are VERY small, nearly impossible for human fingers to position well, always automated machined placed !

These are simple caps (capacitors), don't get carried away, but multiple caps are more reliable, simply due to redundancy, if 1 fails you still have another - KISS - keep it simply simple ! ;)

PS: Same batch run caps will generally have the same failure rates and possibly flaws - duplicate caps can help here.

Remember: Higher voltage rated caps are better !!
 

Motopreserve

Motopreserve

Drone Enthusiast
Whoahhhhh.....

lets take it down a notch here Roto. You're making some pretty broad (and I might add inaccurate) assumptions here. If you read the thread, you'll see that Econ has been (graciously) testing with various devices , including an occilliscope. If you had read far enough back, I think that you'd see, shockingly, that you are not the only one who understands what and occilliscope does.

While its impressive that you invented electrical engineering, and apparently "hi-fi" sound systems, I think we are talking apples to oranges here. The main objective of this thread was to expose a need (or not) for additional capacitors in the use of longer power wires from the ESC to the battery. I see it as an attempt to clarify, and prove or disprove a common perception that without these additional caps - our ESCs may fail.

The ripple current (and its effects) is what comes up most often in this subject, so I have to admit, when your original post did not mention ripple current, as well as your misquote of the max 6S voltage, I sorta discounted your expertise in this matter. But that said, I don't think myself, or anyone else, was looking to get negative on you (which is why I didn't lambast you for your omissions or misquote), because that's the civil way to handle it, unlike your last posts.

I have seen nothing in your posts that is A) offering to conduct tests the way YOU see for, or B) even a civility that would encourage people to listen or accept your advice.

Please take the negativity elsewhere.
 


econfly

econfly

Member
R_Lefebvre said:
Wouldn't an inductor before the caps help reduce the ripple current, voltage fluctuation, etc?
Click to expand...

It would if our current demands were stable. But in the presence of high frequency changes in the load at the ESC an inductor would make things worse. Ideally, we want the current to fluctuate (that is what the ESC wants in order to perform output pulse-modulation). It's the inherent inductance in the system that is getting in the way of this desired current fluctuation and, as a result, causing the voltage spikes that threaten the capacitors and, ultimately, other ESC components.

More broadly, the root of so much trouble here is the practice of controlling motor RPM by flipping the power on/off thousands of times per second. On the input side of the ESC that creates an AC component in the DC load (though certainly not the usual sine wave form). That AC component is the problem, and it's reasonable to think that an inductor (or choke) could strip out that AC component and solve the problem. But it won't work because the source of the problem is downstream -- at the load -- and the ultimate result would be ever greater voltage spikes there.
 

T

tregtronics

New Member
Econfly
Thank you for this thread. This has been a very useful thought experiment.
1) Have you tried a larger power (on the order of a Farad, or a large fraction of a farad) cap at the output of the battery or the input to the power distribution plate? The table you are showing is adding muffs and puffs (uf and pf) to the equation, I am just wondering if anyone has tried adding a whole Farad to the equation, right at the source of the power distribution? I would think this would really dampen out any power fluctuations for all the downstream ESC's by giving a giant well upon which each ESC could draw power from. The local caps would not have to work so hard.

2) Temperature, this is how I test, and I really wanted to thumbs up the suggestion. On a safety check sheet, we run laser temp readers (<$20.00 at Radio Shack) and log our motor temps, ESC FET's and ESC Caps temperatures. When one starts trending hot, the whole ESC gets replaced. Some people we work with replace anything that has not been replaced in the last 24 months just for safety, independent of the temperature measurements. At some point you just have to draw a line in the sand.

3) Can you label a little bit of scale on your table at the beginning? I may have missed it, but it would be useful to know the induced voltage ripple, and the frequency at which this is measured. (18 kHz, 4.2 uVpp) or some reference so we can have an idea of scale of the measurement?

This is really good, important work. Thank you for taking the time to write this up and share.
 

econfly

econfly

Member
I can't edit the first post, but in the table the baseline value (stock capacitors, stock power wires with no extension) is 170mV AC RMS, 1.6V peak-to-peak. The peaks arrive at a frequency of 8kHz -- the same frequency as the tested ESC's output modulation.

Adding a capacitor upstream of the load won't address the problem -- at least not directly. The path between the source and the load is causing the issue, so whatever we do upstream or at the source to stabilize voltage will ultimately be undone as current is demanded through the path to the load. For example, I can see similar ripple/spikes at the ESC even if I use a bench power supply (with lots of internal capacitance and voltage stabilization) instead of a battery.

Finally, a 1F capacitor is HUGE. But let's say we had an ideal 1F capacitor -- i.e., with no internal resistance even at our 8kHz+ ripple frequency. The time constant of a basic RC circuit is just R*C, and the capacitor can charge/discharge about 63% of it's rated capacitance in that time period. If we could live with about 2/3 of the capacitance utilized, we would need the effective resistance in the series to be low enough to get R*C to 8kHz (i.e., 1/8000th of a second). Since C is 1 here, that means R = 1/8000 ohms or 125 micro-ohms. That is pretty much impossible (superconductor?). And, even if we could somehow get resistance that low the current flow into the capacitor would be massive. At, say, 15v the current would be 15v / (1/8000 ohms) = 120,000 amps. We would need an 8,000C battery for that too. And a blast shield... All of this (admittedly silly) speculation goes to a major constraint in the system, and that's the time available for capacitors to help at all.

What are typical temps you see on your ESCs capacitors when things are working well?
 

C

cybergibbons

New Member
Excellent work. I'd embarked on something similar last week, though I was comparing extending the motor wires to the ESC wires. There is so much advice and guidance that just has no evidence or thought behind it, and work like this is really good at clearing issues up.

Do you have more detail on how you measured the noise i.e. what scope did you use, was the noise bandwidth limited at all (I'm seeing a lot of power high frequency harmonics above the switching frequency), what power supply did you use (I see very different results from a LiPo compared to a bench PSU), what load was on the motors etc. Sorry if you have already mentioned all of this, I just can't see it.

I didn't get to the point of any formal testing, but I did note that a 30A ESC, when at lower throttles and with smaller motors, the noise was at far lower levels than a full throttle run at high power. Moving the throttle rapidly caused some fairly large spikes as well. I don't have anything that can fully load test a 30A ESC though - really a big motor with a big prop is going to be needed.
 

econfly

econfly

Member
cybergibbons said:
Do you have more detail on how you measured the noise i.e. what scope did you use, was the noise bandwidth limited at all (I'm seeing a lot of power high frequency harmonics above the switching frequency), what power supply did you use (I see very different results from a LiPo compared to a bench PSU), what load was on the motors etc. Sorry if you have already mentioned all of this, I just can't see it.
Click to expand...

I'm using an Agilent MSOX2024a oscilloscope. Voltage spikes for the ESC I tested were at 8kHz (same frequency as the output PWM), and while there is higher frequency ripple and noise visible, the main issue of pulsing power / spiking voltage is best seen at that output switching frequency. I used a LiPo for tests to deal with the issues you mention, where bench power and its internal regulation produce different results than you get with a battery. From what I can see, the problem is the worst at heavier loads, but is significantly decreased at 100% power. This is reasonable given that the switching all but disappears at that level. I was primarily interested in relative measurements, so I arbitrarily chose 60% power as my benchmark measure (see the table in the 1st post for props, motors, etc.). For the relative results I decided to use AC RMS as a measure of voltage fluctuation rather than a peak-to-peak measure. On the scope and with various testing I found both measures to be highly correlated (that is, the PTP values I could see correlated very well with the automated AC RMS values on the meter). It was easier to measure RMS, so I went with that and used an Agilent 34461a DMM for this (its AC RMS measurement is rated to 300kHz). The table values have no practical absolute meaning, but they do give a very good indication of the relative (ordinal) comparisons made.
 
Last edited by a moderator:


C

cybergibbons

New Member
econfly said:
I'm using an Agilent MSOX2024a oscilloscope. ... of the relative (ordinal) comparisons made.
Click to expand...

Thanks, that's really helpful. My gear isn't as high end but does a similar job. There's a lot of stuff around electronics in multi rotors that has no evidence behind it, and it is great to clear it up.
 

econfly

econfly

Member
cybergibbons said:
Thanks, that's really helpful. My gear isn't as high end but does a similar job. There's a lot of stuff around electronics in multi rotors that has no evidence behind it, and it is great to clear it up.
Click to expand...

Thanks a lot. That is what motivated me -- the over abundance of opinion absent much fact. This is pretty easily observable stuff. A $50 used oscilloscope on ebay could see what I'm seeing (my equipment is overkill for this). I've had a bit of an epiphany recently as to how valuable very basic electronics test equipment can be in this area. For the price of a LiPo battery or two in test equipment (e.g., a used scope, and maybe a decent multimeter), there is a ton of fun and learning that can be had.
 

C

cybercrash

New Member
Thanks for doing all that research.

To me it sounds very complicated that's why I ask this question:

If I connect the ESC with it's standard cables to my power distribution board and extend just the three motor wires up to ...cm I will be ok?
 

econfly

econfly

Member
cybercrash said:
Thanks for doing all that research.

To me it sounds very complicated that's why I ask this question:

If I connect the ESC with it's standard cables to my power distribution board and extend just the three motor wires up to ...cm I will be ok?
Click to expand...

Yes, you will be fine. Extending the motor wires is not a problem, generally speaking.
 

You must log in or register to reply here.
Top