PDA

View Full Version : Castle 80HV Explodes in Fire on the Bench!


Pages : 1 2 3 4 [5] 6

Chris_M
05-24-2011, 06:18 AM
Castle needs to go back to the basics with there HV controllers!!! Lets face it, they never had this problem with the 85 or 110HV. Im sure there were a few here and there but nothing like the epidemic with the new 80 and 120. It doesnt take a brain surgeon to realize that and you would think they would already have this worked out after more than a year. I would say get an HV110 instead of a 120 but at this point I think everyone should buy Kontronik and show Castle that as customers we dont trust there product anymore.

vadimr
05-24-2011, 06:42 AM
Folks,

Remember the time when the cases on the CC controllers were separating due to the bad glue? The ESC is composed of two parts, the FETs and all other high voltage electronics that convert DC to 3-phase AC that are attached to the lower part of the case and the control board that is attached to the upper part of the case. They use some surface mounted connector to connect both components and the lower components cannot survive without the control board. I was told my a reliable source that when the case would separate the ESC would burst into a fire.

Let's fast forward this to CC HV ICE controllers, guess what? (My assumption) They don't have an internal BEC that means they rely on external power to be fed to them through servo lead to power on the control board. Now what will happen if you plug in the main pack into the ESC but not connect the ESC servo lead into either a castle link or an RX which is powered by an external BEC/RX pack, well I believe you are essentially creating the same condition as a separating case. I even know for a fact that plugging in castle link programming card into a CC HV ESC does not even power on the programming card, that is because instead of supplying power to the card it is actually looking for power to be supplied to it.

The question I ask, is it possible that the rate of fires from CC HV is due to the fact that they get powered on from a main pack but the control board which needs to be back fed with low voltage (5-6v) is absent?

-Vadim

watdazit
05-24-2011, 07:11 AM
Castle needs to go back to the basics with there HV controllers!!! Lets face it, they never had this problem with the 85 or 110HV. Im sure there were a few here and there but nothing like the epidemic with the new 80 and 120. It doesnt take a brain surgeon to realize that and you would think they would already have this worked out after more than a year. I would say get an HV110 instead of a 120 but at this point I think everyone should buy Kontronik and show Castle that as customers we dont trust there product anymore.

Sorry to point out the HV 110 has been discontinued. I had an HV 110 go up in smoke on my 700 but the HV ICE 80 on my Compass 600 has been OK. Luck of the draw. For peace of mind I would change but what too??? I have Scorpion on my 250 and 450 and they are fine but when I fitted one to my 500 I had problems with it initializing properly. I would have fitted Scorpion to my 700 but they are massive things and I would have had to modify the frame to fit it.

praetorious
05-24-2011, 08:30 AM
How can one simulate real world loads on the bench?

I have another electric motor setup as a brake. The shafts are coupled together so it acts as a dyno. And wow, I never actually thought of using it as an actual dyno to test motors, that is something I could do.

I never vocalized any assumptions about using a lower ESR value, I said a nice ESR value, was thinking about working out the same current limiting effect to some extent. Although I would probably look at a lower ESR as a starting point because I doubt that they would rely on that for current limiting. Current limiting will involve some heat dissipation.
I agree about the ripple, but if the peak of that ripple at the caps exceeds their working voltage then it would start to break down the electrolyte.

Infinions, I had no idea. I will refrain on commenting on the quality of components unless I know what is actually in there. I will have to take an old esc to see what they use. As for the problems on bench plug in, it might be due to the inrush current.

Starcruiser
05-24-2011, 08:37 AM
Folks,

They use some surface mounted connector to connect both components and the lower components cannot survive without the control board. I was told my a reliable source that when the case would separate the ESC would burst into a fire.

They don't have an internal BEC that means they rely on external power to be fed to them through servo lead to power on the control board. Now what will happen if you plug in the main pack into the ESC but not connect the ESC servo lead into either a castle link or an RX which is powered by an external BEC/RX pack, well I believe you are essentially creating the same condition as a separating case.

The question I ask, is it possible that the rate of fires from CC HV is due to the fact that they get powered on from a main pack but the control board which needs to be back fed with low voltage (5-6v) is absent?

-Vadim

This point you made regarding the BEC on the HV ESC's is something I was pondering too. My thought though was the FET's would be in the off state thus not allowing power to the motor buss.

The only thing I could think of would be a static issue triggering the gates on the FET's to allow a connection between the B+ and B- on the same motor leg thus causing the dreaded short and melt down.

The only other potential cause would be the controller software turning both FET's on the same "channel" at the same time, again a short condition as outlined above.

The only thing I could think of is to power the HV ESC lower power B+ first and then the main B+ to allow the controller to initialize first as you pointed out.

John Cook
05-24-2011, 08:45 AM
lol people saying the 85 and 110hv's didn't break or catch fire, either can't remember or weren't around 2-3 years ago when they had similar issues. People had airframes catch fire in the air and total loses.

Castle speed controls work well when they WORK. But, they seem to have WAY WAY higher failure rates, at least in heli use. I don't trust them myself.

WBFAir
05-24-2011, 08:54 AM
The only thing I wonder about this thinking about no power via the control side of things when it is plugged in is, and this is just pure speculation as I don't really know how these work on these levels per say.

But for ESC's there seems to be an initialization procedure right?


Just to state the known, on my regular ICE unit, if I plug it in, it beeps the packs cells, and then so long as the throttle is at zero, then it initializes and can send power to the motor.

Then if it isn't set to zero, then it will not allow this to happen no matter how much I play with the throttle. Unless of course I bring it to zero.

So again, I don't know how all this works really but I would have to assume that there must be some kind of power interruption to keep parts that until this full initialization happens, not powered.

So I think if you don't have a controller of some kind hooked up to the ESC's controller plug, that this is basically the same thing.

It takes power first to power this controller circuit, which by default has the main part of the ESC un-powered. Then keep sit that way until it sees a zero throttle setting.

So I think it basically is of no difference whether there is a controller hooked up and there is no zero, or there just isn't one connected at all.

The bottom line is, until it sees a zero throttle reference, the rest of the ESC is dead?

I would almost think this is the way it would have to be or if not, then you would be sending something to the motor, no?


Then even if that is all wrong, then how is it if not having power to the controller side of this is the case, do you explain how many of these smoke when 30second to several min into a flight, and do so with no reports of control loss till is on fire?


In other words too, isn't there a fail safe default it goes to if it were to loose the controller power?

vadimr
05-24-2011, 09:05 AM
So again, I don't know how all this works really but I would have to assume that there must be some kind of power interruption to keep parts that until this full initialization happens, not powered.


The bottom line is, until it sees a zero throttle reference, the rest of the ESC is dead?

In other words too, isn't there a fail safe default it goes to if it were to loose the controller power?

Pretty much No across those three questions as far as I understand. There is no power interruption, the controller board just does not turn it on but no extra circuit to protect it as far as I understand. It's all a logical circuit about 0 throttle and all that.

ESC is not dead until it sees 0 throttle it is just not powering the motor, the controller is taking care of this, if the controller is not powered well I guess anything is possible.

I doubt there are any fail safe defaults anything goes into if the power is lost.

I cannot explain why these things happen 30 seconds in.

-Vadim

John Cook
05-24-2011, 09:06 AM
I think the problem with ESC's and castle in particular. Is that the mosfet switches that control the current the motor "gets"; are in series. If one goes bad, or even gets "stuck" as in the switches aren't open or closed in correct timing, the the rest will get overloaded and you basically get a meltdown of the mosfets that is catastrophic to the ESC. Now sometimes you could get fire, sometimes just pop the esc. I'm not sure why it would go one way or another. Anyway, from what I've seen at the field, Castle ESC's are more widely used, but also seem to suffer a higher failure rate than their peers that I see used, cheaper alternatives included.

what I don't understand is...... For the $ castle charges for their units, and the KNOWN problems that there are, that are very APPARENT from reading forums and more importantly from seeing these meltdowns at the field. WHY hasn't it been corrected???? A small failure rate is expected and it's just the nature of manufacturing. I think we are beyond a small failure rate though, and the nature of the failure being so often damaging to other components is disturbing.

My personal suspician, and I'm NO WAY an electrical engineer. Is that the mosfets they use are barely spec'd to grade. If they used better, more expensive mosfets, which in computer motherboard use really weren't much more $, maybe 10-20 cents ea, so maybe $5-10 more at the MOST I'm guessing for the ESC, about 80-90% of teh problem would be fixed there. I just think Castle thinks things are acceptable and they are not willing to spend more on components as long as sales are "good". Why fix it??? People are still buying "junk" so why bother? If they turned around and put better stuff in their current designs that would be the same as admitting a mistake and legacy users would be upset they have "old" junk. So they are in a bit of a pickle IMO. I woulnd't be surprised to see a new revision or ICE 2 series come out with "fixes" and a larger price tag not only to cover the tiny increase in component costs, but to try to rape users for the fixes that should have been made in teh first series, and to cover any costs of warranty of old units they still are servicing.

Maybe I'm being chicken little and the sky really isn't falling. Just depressing that people still have their "pride and joy" burned up by faulty equiptment, and it's been an ongoing issue for YEARS, and across various product lines with Castle.

I sincerely hope they prove me wrong and become a benchmark for the industy. They offer some really great options in their units and their pricing is reasonable, and service from what I hear is exceptional. If they couldn't only become more reliable they would really earn my respect and trust.

WBFAir
05-24-2011, 09:19 AM
Pretty much No across those three questions as far as I understand. There is no power interruption, the controller board just does not turn it on but no extra circuit to protect it as far as I understand. It's all a logical circuit about 0 throttle and all that.

ESC is not dead until it sees 0 throttle it is just not powering the motor, the controller is taking care of this, if the controller is not powered well I guess anything is possible.

I doubt there are any fail safe defaults anything goes into if the power is lost.

I cannot explain why these things happen 30 seconds in.

-VadimYeah I again, I really don't know how these are internally constructed and I don't mean to say its "all completely dead", but I'd have to assume the main power circuity has to be off and or at least not under any real loads as simply put, if these 3 phase motors are seeing any power or any real current, they will spin.

As well, three phase motors are very sensitive to any major levels of unbalanced current, so its not like there is anything like, two legs of it has current and one does not.

Basically I gotta think that all the power and current side of things are either un-powered, and or have no real current to speak of.

But again that just from pure speculation on my part as I admit I really don't know for sure.

DonDuc
05-24-2011, 09:35 AM
ICE 120HV Failures
The Castle team has noted a number of reports of failure of our Phoenix 120HV in the field. We do our best to design and build brushless motor controllers that outperform all others in this industry. While no failures are acceptable, our records indicate that this controller’s actual rate of return – for any reason – are quite low.

We have not been able to determine any particular cause for the failures that we have studied. The common thread amongst them appears to be increased stress on the controllers’ capacitors. Unfortunately, this can result from any number of reasons including:
• battery condition & performance
• length and type of wire between the ESC and the batteries
• condition of the solder joints in the wire harness
• quality of connectors used
• flying style

We are always working to improve our controllers, and we are looking at a number of measures that may assist the 120HV’s ability to operate even with the worst of conditions described above.

In the interim we suggest that users:
• use high performance batteries that are stabilized at room temperature or above before flight
• use connectors rated for 120 amps continuous operation, such as 5mm bullets
• keep the wiring harness as short as possible
• mount the controller on the rear deck of the heli whenever possible to improve cooling.



Patrick del Castillo, president

_________________
Castle Creations, Inc.



One thing I don't understand:
if these solder joints, wire length, battery quality etc. putting stress on the caps is really so important, shouldn't the "ripple" voltage (that can be logged) reflect this somehow?

I haven't seen a clear explanation from CC what the ripple actually says and have never seen a clear answer to "how much is too much"

I regularly checke the ripple in the logs but upt ot now I could not see any useful pattern or change when switching to new packs or whatever.
If this is so critical, they should make it clear what to look for.

Anyone know more about this?

Chris_M
05-24-2011, 09:59 AM
Ripple values should be about 5% or less of supplied voltage. Old lipos with higher ohms or longer wire leads will raise ripple values obviously. They should definitely make that more clear as the capacitors they supply the 120HV with are only 50V which are really under rated in my opinion.

Here is some good reading material..http://www.rcgroups.com/forums/showthread.php?t=1203522

DonDuc
05-24-2011, 10:17 AM
Ripple values should be about 5% or less of supplied voltage. Old lipos with higher ohms or longer wire leads will raise ripple values obviously. They should definitely make that more clear as the capacitors they supply the 120HV with are only 50V which are really under rated in my opinion.

Here is some good reading material..http://www.rcgroups.com/forums/showthread.php?t=1203522

Thanks, interesting stuff!
I usually see max ripple voltages of 0.7V and average of 0.4V on both 3S and 6S setups.

WBFAir
05-24-2011, 11:40 AM
Ripple values should be about 5% or less of supplied voltage. Old lipos with higher ohms or longer wire leads will raise ripple values obviously. They should definitely make that more clear as the capacitors they supply the 120HV with are only 50V which are really under rated in my opinion.

Here is some good reading material..http://www.rcgroups.com/forums/showthread.php?t=1203522Yeah not to be hijacky here but its stuff like this that is why I made a post asking if people could start maybe doing some uploading of these reports with what ever relevant info that might be useful about what they are from and any particulars to the flight to here: http://www.helifreak.com/showthread.php?t=308587 so that others might have at least something to compare to theirs to so that we might be able to see if there is something that can be view that might not be right about the ESC you have before it goes up in flames.

Heck, as we have this feature with this ESC, why not use it?

Anyway, I know its only been about 4 hrs on the initial post and perhaps I should have put it in the 700 section or something, and don't mean to be whiny, but so far there's no posters and for sure it will be buried in no time.

As I mention on it, I don't my 700 running yet or would have posted one from me, but hopefully some will post to it as I think if there is something there, it could be very helpful to others.

Starcruiser
05-24-2011, 12:06 PM
The issue with MOSFet's is they are susceptible to static electricity and any static charge can trigger the MOSFet into conduction, so if the airframe is generating static electricity it can cause a catastrophic failure mode.

Check this out http://www.vialab.org/Bioe_1310/Labs/lab_11.pdf

this is a static detector and it used a MOSFet to do it, note the warnings about touching the gate lead. My guess here in bench and in flight would be a static issue. Can the heat-sink or the B- line be attached to the frame (if conductive) or the motor mount to eliminate this possible event?

This issue would indeed point to a mfg or design flaw that exposes the switching devices to static discharge or field causing an unwanted conduction state of a MOSFet that should be off. Now this is just a theory.

Can anyone check and see if the heatsink is connected to the B- line? or does it float? The heat sink is in close proximity to the MOSFet's by design, is it grounded to reduce it being an antenna for static charge?

extrapilot
05-24-2011, 04:22 PM
Prae- please, you implied low ESR. ‘Nice’ is never implied to mean a lower cost, less capable, higher-esr cap… Im sure CC has selected caps based on simulations and field testing. Of all things in the design, that is the easiest to modify. They are not installing ‘marginal’ caps to save $.30. They are installing the smallest/lightest caps that can do the job. But, and assuming the caps have some culpability in the failures, there is always the caveat about operating environment. Ripple is a serious stressor for caps in this type of application.

For the designs I have seen from CC, they use an external half bridge driver chip. The chip has built-in protection to eliminate the possibility of shoot-through on a phase.

If there is no motor connected, at least for the ESCs I have seen, there is no other path between phases. If there is a case of a CC ESC melting down on the bench with no motor connected, that would be very telling. If the bridge controllers prevent a shoot-through, and there is no motor providing a path between 2 phases, there would have to be a physical problem (i.e. a short on a gate, damaged silicon). If a motor is connected, then you have that path in place. With that, seems feasible that software could be causal (due to the bridge chip preventing shoot-through, the only plausible cause would be failure to initialize PWM properly, so they have a very-high current flowing in a non-commuting motor).

The FETs I have seen in use on CC products are by NO means ‘barely spec’d to grade.’ Example, one on device in their mid-level lineup, they are using an Infineon part, which is rated at 100Amp. They have 4 of those for each side of each phase, so that means each phase is rated to 400amp (continuous). The pulse rating on those fets is 400A, so that is 1600amp pulse capacity PER PHASE. This is needed to keep the resistance (heat buildup) low- so generally, ESCs have way more FET than they need for the current rating…

Static with FETs is always a concern, but FAR less so once the FET is mounted, and its gate is connected to the circuit. Plus, most of the support circuits, MCUs, etc on the board have internal protections- which by their nature protect other devices on their bus.

John Cook
05-24-2011, 04:29 PM
well I'm glad to be wrong about the FET's if that indeed is true they use good german/Infineon parts.

Still worried about why these controllres are burning though :(

WBFAir
05-24-2011, 04:48 PM
The issue with MOSFet's is they are susceptible to static electricity and any static charge can trigger the MOSFet into conduction, so if the airframe is generating static electricity it can cause a catastrophic failure mode.

Check this out http://www.vialab.org/Bioe_1310/Labs/lab_11.pdf

this is a static detector and it used a MOSFet to do it, note the warnings about touching the gate lead. My guess here in bench and in flight would be a static issue. Can the heat-sink or the B- line be attached to the frame (if conductive) or the motor mount to eliminate this possible event?

This issue would indeed point to a mfg or design flaw that exposes the switching devices to static discharge or field causing an unwanted conduction state of a MOSFet that should be off. Now this is just a theory.

Can anyone check and see if the heatsink is connected to the B- line? or does it float? The heat sink is in close proximity to the MOSFet's by design, is it grounded to reduce it being an antenna for static charge?The only thing I would mention on this thought is that the 500's are known to be major static creation kings, or at least they were a few years ago when I used to read more of that section.

Back then that I remember reading a lot of people were doing all kinds of things to keep that from developing, and granted while those are smaller in ratio per what ever size ESC those guys are using, they don't seem to be having these flaming issue. (at least not yet).

Then you might say, well but these with the 80/12HV's are bigger helos so they could amass more static.

But then if they did have this much, wouldn't we be seeing all the other deficiencies the guys with the 500's were seeing when they had static issues?

For sure this is all just pure speculation on my part as I'm not all that formally educated on this as much as some of you guys are, but I just thought I would throw that point into the pot.

Starcruiser
05-24-2011, 08:32 PM
if the gate is protected by the anti-shoot thru chip as you put it then it has to be a silicon failure of some sort that is turning on the fet's to cause the meltdown.

I am not trying to defend CC but to understand the cause and determine if it is an external one or is it just plain bad QC at CC that causes this problem.

extrapilot
05-24-2011, 09:19 PM
What would be very helpful in identifying the cause would be to identify the ‘mode’ that is affected. Example, if the fires are always at init, that says a lot about where the failure is happening. If it is always in flight at high power, that says something else. Is there any data on this? And, does this happen with no motor connections?

praetorious
05-24-2011, 10:18 PM
extrapilot,

didn't say what I wanted at first, but yes, I will go with low. Anyways, Thanks for the information about the fets, I didn't know what was there, never saw the designs or never picked the things apart.
Also, thanks for the info on the H-bridge driver(is that Infinion as well). You are correct about the protection, there wouldn't be a chance of turning on the same side. But these controllers have a brake function, so they would have to turn on either the high or low pair of the bridge to do that(would they?). Could it then be a software issue where both sides switch on happens and it lets the smoke out?

Santini
05-24-2011, 10:42 PM
Ok so I recieved today an email from Steve Rogers at CC indicating they will send me out a new one right away, so I'm very pleased about the response. Still mulling this over tho, what to do w my Logo 600.


Hi Joseph
We apologize for the mishap - we will get a new one out to you.
I am going to set up an AWR where we send out the replacement & when it arrives you then send the one you have back.

extrapilot
05-24-2011, 11:36 PM
P

For braking, at least for the designs I have seen, the approach is to run all 3 phases to either high or low. So in that model, you just set each of the 3 half-bridge controllers to the same config (i.e. high=0ff, low=on), and that’s it… There remains the interlock (cross conduction) prevention logic regardless.

As regards bridge drivers and FETs etc, most engineers will adjust parts as supply changes, and tech evolves. I think you will find several different OEMs participating over the past couple years on the CC ESCs (i.e. IRF, Intersil, On Semi, Infineon, etc). I don’t study their work- I just recall Patrick posting some tidbits recently on some of his designs.

Cheers
R

gerardobrandao
05-25-2011, 05:18 AM
Hello guys...


Going to give mine input...
I have a long journey in Rc Hobby ,specially in Rc Boats, wich is by far the hardest concerning speed controller´s and motors.
I have seen a lot of boaters having huge motors=huge amp draws, normally, with chinese speed controller...They were almost condemned to insucess!
I have always had less boats then my other friends, but the best equipment i could afford, i got it inside!
NEU motors,Lehner, SCHULZE and YGE speed controllers!
I never had a problem with them...and i´m talking about 200, 250 AMP continuous amp draw....
Now ,that i´m in the heli scene, at 4 months ago, i have a scorpion 5025 motor, and a YGE 160HV speed controller......None issue, and everything works right..
YGE is the same quality as Kontronik, the employers and the engineers are most of them ex-kontronik, and 2/3rd of the price....I just say go with it....
I never had a problem of connecting the bullets and seing flames out of it....
I won´t imagine driving mine heli, with the lack of trust of mine speed controller catching fire...this is of course mine humble opinion...
The difference between good and bad speed controllers is ,IMHO, Pro Active safety...if we can call it that...
Schulze speed controllers, have a safety starting routine wich analyzes all working factors..inductance, amp draw, even the capacitors..if the motor isn´t good..it won´t work..if one of the 10 caps are broken ..it won´t work...
the same applies to YGE...
When talking about chinese crap...what you put is what you have...safety margin= 0 !




:hug:

WBFAir
05-25-2011, 06:45 AM
Hey watdazit,
Might consider these for a solution:thumbup:.... no need to add any wire :clap

http://www.helifreak.com/attachment.php?attachmentid=229010&stc=1&d=1306007886
Hello Nightflyer, I was wondering if you could provide a link for these long packs.

I was looking for them on the HK site for 6S 5000 and all I could find were 25C ones.

Are these them?
http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=10480

Thanks