Castle HV Controllers - Use Extra Caps!

[johnrobholmes]

12s lipo, 10ah turnigy.

It keeps happening around half DOD too. Never near the start of the run, always after about 20 minutes of hard riding.

 

[swbluto]

12s lipo, 10ah turnigy.

It keeps happening around half DOD too. Never near the start of the run, always after about 20 minutes of hard riding.

Ok. I'm not as intimate with the phoenix ESC design as its engineers, but I'd think that ripple voltage failure would be more likely at the beginning of its discharge when voltage is at its highest (The ripple voltage oscillates around the average voltage, and the higher the average voltage, the higher the peak ripple voltage). Their might be the exception that the inductance may increase as the battery discharges, but I've never heard of that but it seems understandable no one would investigate a battery's ESL. It's not exactly easy to investigate, the cause doesn't seem obvious, and it hasn't caused problems with regular ebike controllers in the past.

I'd then suspect heating problems in which case increasing the capacity with low esr caps as close to the ESC's input capacitors should decrease the likelihood. At some point, it should eliminate it and if you're at some absurd capacity and it's still present, then I think heating problems can be eliminated as a suspect. The thing is, the ripple voltage reduces with increased capacity, so increasing capacity would help both problems at once (And if it solves it, it masks the real root of the problem but who cares if it's solved?).

Hmmmm... I wonder if there's a relationship between temperature and maximum voltage rating? It seems increasing the temperature of a capacitor increases its capacity and lowers its ESR (It's a chemical based energy storer), but I'm not sure about the max voltage rating. I wouldn't think it'd change but maybe it does? If it does change, I'd imagine it'd be lower since the electron band gap would be lower but how large is the electron band gap compared to the width of the dialectric? Like 10000x thinner? If some huge magnitude like that, which seems likely, then it doesn't seem like it should have more than a 1% effect on the max voltage.

 

[johnrobholmes]

From what I was told, the lower the ESR, the more likely the Cap will pop. Maybe shenzen controllers don't pop caps because they are not high quality caps? Also, the throttle advance on the shenzen controllers is much slower. The caps are not getting warm at all either, I checked them all right after the blow. Even the blown one was cool.


Standard recommendation- shorten wires, better batteries, more capacitance.

 

[swbluto]

From what I was told, the lower the ESR, the more likely the Cap will pop. Maybe shenzen controllers don't pop caps because they are not high quality caps?

That would certainly be worth testing! That would reverse pretty much all the general wisdom spouted on the forum and on the DIY brushless controller page at the rcgroups forum which has a lot of freakin' smart electrical engineers. (Let alone matt's confirmation)

But, I wonder, were they referring to the stock caps (The lower the stock cap's ESR, the more likely the stock caps will pop) or the supplemental caps (The lower the extra cap's ESR, the more likely the stock caps will pop)? It seems the "general wisdom"'s reasoning kind of assumes all capacitors are identical for modeling simplicity, but that isn't the case when we're adding different caps. However, I don't understand how that would change anything but that shouldn't be alarming - I'll admit I don't completely understand your phenomena which is what I'm trying to figure out.

I did notice that they heavily suggested to not run batteries more than 75% of their rating for fear of destroying the controller. Given they likely have a lot experience with that, I wonder why. Is it as simple as warmer batteries = warmer environment = hotter ESC? Or is there some electrical reason? Something like "Batteries show funny behavior when they're stressed" (As in, what, their output voltage bounces all over the place?).

 

[johnrobholmes]

in regards to the ESR,

The lower the ESR, the faster that it can charge and discharge. The controller chopping causes the ripples in current and voltage, which the cap tries to smooth. The lower ESR caps will work harder to smooth out the ripples, in comparison to a high ESR cap. The peak voltages and currents will be higher.


When a battery gets voltage depressed, the ripple voltage gets larger. I don't have means to test this myself, but i have seen graphs depicting larger ripples as the current is increased. It has been a while since I studied the occurrence, so I am not fresh on why it happens. Probably related to the current rushes as the fets chop.

 

[swbluto]

in regards to the ESR,

The lower the ESR, the faster that it can charge and discharge. The controller chopping causes the ripples in current and voltage, which the cap tries to smooth. The lower ESR caps will work harder to smooth out the ripples, in comparison to a high ESR cap. The peak voltages and currents will be higher.

That's hard to accept. Maybe my mental model is far too simplistic, but I was told that a capacitor tries to smooth the voltage and it does so by storing charge and releasing current when need be (An inductor tries to "smooth current change" by "releasing voltage"; the energy for that voltage is stored in the electromagnetic field). The lower the ESR, the "easier" it is to provide the current it needs to smooth the voltage change due to the lower resistance. But, not to be convinced simply by my understanding, I did do some simple simulations. It seems that lowering the ESR does lower the output voltage. But at 470 uF and testing between 10 mOhm and 100 mOhm and a 10khz sine signal, it made a very negligible difference. It only became obvious when the capacitance was something like 100 F or 100,000,000 uF.

When a battery gets voltage depressed, the ripple voltage gets larger. I don't have means to test this myself, but i have seen graphs depicting larger ripples as the current is increased. It has been a while since I studied the occurrence, so I am not fresh on why it happens. Probably related to the current rushes as the fets chop.

That's interesting behavior. The only simple model that'd explain that would be "ESL" (effective series inductance) on the battery's part and wires, since a larger current would be capable of a greater current change and so it'd be more prone to higher voltage shocks and maybe lower C batteries tend to have larger ESL. But there could possibly be some odd impedance interactions with the ESC, maybe.

 

[recumpence]

This has go to have something to do with your hub motor. I have never popped a cap with RC motors. But, I have fried FETs.

Nic Case (top speed world record hold for RC cars) pops caps all the time, but only when he plugs in his 54 volt pack.

You should shorten your pack wires, thicken them, and increase capacity of your pack and see if the problem persists.

It is easy for me to recommend that because I am not the one spending the money or the time on it myself.

Matt

 

[swbluto]

This has go to have something to do with your hub motor. I have never popped a cap with RC motors. But, I have fried FETs.

Oh, yes. A hub motor might have something to do with it. I wonder if the inductance might be too great for it? I can't really say if a hub motor has inherently higher inductance that a typical RC motor, but given that the torque constant is inversely proportional to KV and the KV of a hub motor is typically far lower than an RC motor, then the hub motor has an inherently higher torque constant. Now... what would the relationship between inductance and the torque constant be? More coils means a higher torque constant and more coils = higher inductance, but I'm not sure if it's that simple for comparing RC motors and hub motor... I think hub motor's typically have a greater copper pour factor.

If it was me, I would just measure the inductance of the hub motor directly and compare it to an RC motor. But... that does require some specialized knowledge to measure. For barebones inductance measurement, find a suitable resistor (Get the right value), hook up an ammeter and find how long it takes to get to 67% of the equilibrium current or the current it'll eventually settle to. That amount of time is equal to the time constant which is equal L / R, and since you know R because you chose it, you can calculate L (inductance) as L = time-to-67% * R. If your ammeter is a digital one that refreshes once every second (Or otherwise has poor time resolution), then it seems you'll need a sufficiently low value of R (That doesn't seem intuitive, though). But... hmmmm... I'd have to think this over and research more to confirm the above thoughts and I don't want to do that now. Just look up RL circuits to get an idea of what's going on.

Looking at http://en.wikipedia.org/wiki/RL_circuit, it seems like there might be better methods by measuring the voltage. You're less likely to damage something by measuring voltage. :wink:

 

[johnrobholmes]

I would assume that hub motors have higher inductance just from the crappy laminations. Non optimized design, cheap steel, and thick laminations. Otherwise I do know that outrunners typically have higher inductance than a similarly powered inrunner.

Wish I knew more info on the subject. All I know is that I am ditching the hub motors for my high power apps. Too much trouble for me.

 

[12p3phPMDC]

What is the voltage rating on the caps that come with the Castle HV controllers...?
60V? 75V?

 

[swbluto]

What is the voltage rating on the caps that come with the Castle HV controllers...?
60V? 75V?

Nominally, 50 volts. Castle creations puts a rating of 50.4 volts on it, though and it appears Ggoodrum has had good luck so far with 53 volts. Somewhere around 54 seems to be near the Guaranteed Capacitor Killing region.

 

[12p3phPMDC]

IMO, there is not enough margin voltage wise. I would replace them with a cap that is double the rating voltage like a 100V cap.

how can you run stuff at the edge and expect reliability? :?

 

[johnrobholmes]

For the intended purpose of the HV110, the caps are sufficient.

The caps I am using as replacements are 100v units. I haven't had one get warm or pop yet.

 

[swbluto]

I started measuring the input current into the ESC at partial throttle to see what the waveform looked like. I used a .333 Ohm shunt resistor and the voltage was ~18 volts and I slowly increased the throttle to 100%. This was done under no loading and I suspect the peaks during the PWM switching would be amplified under load since they seemed amplified at startup and I suspect it's these peak currents that's stressing the input caps at partial throttle - Also, the current is variable at 100% throttle so the caps would also be vulnerable there too, but it seems like dI/dt isn't as great as it's wavey instead of jagged so it'd seem less hard on the caps and, also, its "switching" less often (Instead of switching+commutating, it's just commutating and the commutating period is lower than the PWM.)

[youtube]OzenQU_lcBY[/youtube]

I kind of wonder why my motor sounds like it's "stuttering" in the 20-60% throttle range but not at 100% throttle. I think that might be the source of so much racket on my scooter at lower speeds.

 

[spinningmagnets]

I notice Matt is using 50V caps on a 12S system (44-48V?), and AJ is using 65V caps on a 12S system. Then others are using 100V caps on 12S and 14S systems. If the cap rating is closer to the system voltage does that neans that the clipped parts of the voltage spikes (having been passed through the cap becoming current, and causing cap heat) results in a smoother voltage to the ESC, but hotter caps?

How useful is it specifying a cap that has a voltage rating indexed to the system voltage? or, should I just use caps with a much higher voltage than my system?

I am planning on using a 6S LiPo pack (22V -24V) on a friction-drive (typically lower amps than an axle-driven bike). I have found low-ESR caps that are 35V, and Luke has a big bag of 40V caps, would they be better on a 6S system than using 50V caps? (clearly I don't need 100V caps, even at the same low price)

I understand the uF rating should be as big as possible (caps are cheap, small, and light), however several 1,000uF caps are better than one very big 3,000uF cap due to heat shedding being better.

Any advice on cap voltage ratings appreciated, thanks in advance...

 

[liveforphysics]

Here is the simple low-down on caps.

ESR

The voltage rating SHOULD be at least 20% above the peak voltage you ever expect to see. You can run them closer, but it's definately not something any professional product is going to do, but OK for a tinkerer who doesn't mind failures now and then.


Capacity doesn't make a lick of difference for these caps. The entire function is to clamp spikes and soften dips. The lower the ESR, the tighter they hold the voltage.

As a rule of thumb, as caps decrease in capacity for a given physical volume, they are being made with thicker foils and higher current current conductors etc.

So, look for big bulky caps with low capacity, or if you can find it, seek out caps specifically designed around being as low of ESR as possible. Don't even look at capacity, it's just meaningless for this.

 

[neptronix]

Hmm...

What about using two ESC's.. everything paralleled?
Then you could spread the load out between the two.

Because it seems like they don't make an ESC big enough for eBikes! or one that is big enough is too freaking expensive.

 

[liveforphysics]

Hmm...

What about using two ESC's.. everything paralleled?
Then you could spread the load out between the two.

Because it seems like they don't make an ESC big enough for eBikes! or one that is big enough is too freaking expensive.

Yes we arrived here a couple years back.
Join the club.
Some of us are working hard to solve this problem.

 

[recumpence]

Hmm...

What about using two ESC's.. everything paralleled?
Then you could spread the load out between the two.

Because it seems like they don't make an ESC big enough for eBikes! or one that is big enough is too freaking expensive.

You cannot parallel two ESCs. They will fight each other. It just doesn't work. The only easy (easy?) way around this is two motors with two controllers to share the load.

Of course, the truely best option is just a huge freakin controller. This is being dealt with on a number of fronts right now.

Matt

 

[Byte]

Castle is working on a 16S controller and in the feature maybe a 20S controller. But first they will release an 8S RC-Car controller and a 12S...

 

[spinningmagnets]

I hate to be a bother, but I'm 'almost' done asking about caps...I have been scanning the digikey web-catalogue, and the first page says they have over 150,000 choices for capacitors! I have looked at the capacitor specs of the model most often listed here at ES, then I backtracked, then tried several minor variations in the catalogue filter.

http://www.digikey.com/

search: capacitor, then aluminium, check lead-free and ROHS-compliant, filter: general-purpose, radial-can, through-hole, 1000uF

...and then, when I enter in (as an example) 35V, 63V, or 100V, I am still presented with a page of choices where, the only difference is the ripple rating. This column can range from 0.9A to 3.3A, and they are all about the same price. For some reason, the ESR of most of them is not listed...but these are the same ones as the 50V unit that is already enjoying success on many builds.

Is a lower ripple number better? by that I mean is a low ripple rating number the maximum the cap can process without blowing up (high being good), or is it the maximum ripple the cap will allow to occur (low being good)?

 

[liveforphysics]

I started a thread on this very subject of how to pick the right caps for your needs:

http://endless-sphere.com/forums/viewtopic.php?f=30&t=22194

It's a WIP though until I can finish it at work tonight.

 

[spinningmagnets]

Thanks, LFP. Thats just what I needed. I appreciate your time and effort.

 

[neptronix]

Hmm...

What about using two ESC's.. everything paralleled?
Then you could spread the load out between the two.

Because it seems like they don't make an ESC big enough for eBikes! or one that is big enough is too freaking expensive.

Yes we arrived here a couple years back.
Join the club.
Some of us are working hard to solve this problem.

Facepalm.
Sorry. I have only been on here for 3 months or so and didn't read every single post. ( these threads get pretty daunting )