Vibration is one thing. These things are subjected to some serious abuse. Bottoming out from a 20ft jump is nothing to sneeze at.

Neither is 8 G's

Jeff

Edit: which is what we test at for 2 minutes.

yes it is if there is a guy that your in a battle with and he can do the same thing repeatedly with the exact same components and his truck.


Nothing against castle- i know that things happend and not blaming them- but when two cars fail for the the same component- even further the same chip on a component when there are nearly identical cars in each of the respective classes doing the same thing something is wrong.

someone else can have the soap box.

castle's replacing them. I'm not complaining...

Not buying it. No two cars will be exactly alike and the difference in suspension can change a gentle bottom-out to a not-so-gentle bottom out.

To be clear, I'm not saying this IS the case, just something to think about. Depending on the nature of the failure, the failure rate differences could be something as simple as a slight variation in component tolerance, temperature, ESC mounting, etc etc.

8 G's might be strong for vibration but it's not very much when you start talking about SHOCK.

True that.
Yeah its a really low chance of it happening as it did to Mike and myself though..:neutral:

Mount the escs upside down, so the chips arent subjected to the weight of the esc pressing down upon landing a big jump- same principle as having rear facing seats on a bus or train etc...

Granted it will look stupid though, but it might just work... :P

A vibratory table will sustain forces for a much longer time. A single shock load may be higher but a shaker table will tell you. Since implementing this test we have yet to have a single failure for “lost” components in the field. A ferrite inductor isn't really that heavy (I’ll weigh some in the morning), it may be heavier than the Cap that's sitting next to it but because of the way they broke you can clearly see that the solder joint was not a good one. It happens; they are fixing it under warranty. Excellent customer service as I have seen in the past. I am not saying anything bad about them or their process, hell we are a small company and we screw up all the time. When we find it we fix it and put in place measures so that it doesn't happen again. I am not even saying they made a mistake, a solder joint can get past the best of us, even with a 10X Loupe. And with some of the new solder and components out there it is a problem until you get used to it. I have soldered HPM solder for tools used in the oil field industry. I can freehand any component you put in front of me surface mount a PLCC chip not a problem, SC70 transistor, you betcha, leadless packages all day long. I have done solder inspection till I am blue in the face, it’s a bad solder joint. No big deal. You guys are always blaming the user I don’t understand it.

Jeff

J57:

Wave solder? Is this the 80s?

We use 100% reflow soldering -- there is no wave soldering done. You couldn't wave solder these parts anyway, they are much too small and placed too densely for wave.

The joints look a little grainy because there is a silicone conformal coating over the whole board. Those are beautiful solder joints -- not a cold joint in the bunch.

(Remember, we are 100% lead free, so the solder joints are not as shiny nor as strong as leaded solder joints... but we have no voiding, and 100% of the solder joints are inspected by AOI prior to conformal coating. -- We are lead free because we sell worldwide, and we cannot sell items containing lead in Europe...)

The joints in the picture failed, not because of poor joints, but because of shock. I imagine (even though we didn't see the parts) that the end caps on the inductor failed, the part rotated, and tore the solder joints cleanly. It's pretty obviously a torn joint. A cold joint would have shown voids, not a cleanly torn surface.

Shock can be enormous in these cars -- I've seen 10ga wires ripped in half just from their own weight (think about how many Gs that is!!), 96 mil thick FR-4 circuit boards bent 90 degrees, circuit boards where EVERY pad on one side of the board is torn off the FR-4... you name it, we've seen it. (And 8Gs is nothing... We routinely see forces over 70Gs in RC cars.)

So soldering is very important to us. We use a Heller 7 zone full convection oven (accurate to +- 1 degree in every zone, and specifically designed for lead-free operation), and we profile every batch. We use a very sophisticated AOI system that literally inspects every single solder joint on every single board we produce.

We use a VERY expensive tin/copper/silver solder formulation for our solder paste (printed,) bar solder and wire solder. This is a much stronger (physically) formulation than SN100 (100% tin) which is used in most lead-free industries today. It's still not as strong as 60/40 -- but it's close.

We take quality very seriously -- we aren't perfect (see MMM V1...), but we utilize tools that make the quality easy to maintain, and we check our quality continuously.

Castle isn't a basement operation - - we have some of the most sophisticated equipment in the world in our surface mount line (including the fastest and most accurate chipshooter available today), and we produce around 30,000 controllers a month.

Thank you for clarifying Patrick, now is there anyway you can scoot our esc's to the front of the line :lol:

But there are parts on both sides of the boards.... :whistle:

You know, since we are talking theoreticals here, how about this: Design the ESC so the FETs are pointing down. Then, attach an aluminum heatspreader that is exposed to the outside world. Then put four 3mm threaded holes at each corner. Attach an L heatsink or something as the default option (or not, whatever). This would allow people to alternatively attach it to the chassis or their own heatsink. Wouldn't involve opening the case that way. Just thinking out loud...

I know it's reflow I misspoke, we have ours wave soldered, because we still use a lot of thru hole components. I am not saying your product is bad, that you don't use checks on your equipment I am sure you do (especially after seeing the load tester for one of your ESC’s. in another thread:wink:). I think you make a great product, I never said anything to the contrary. And I am not saying you don’t use top of the line equipment, I have worked with many AOI systems (nothing is perfect) as well integrating them in systems we build.

I know first hand about conformal coating and lead free, we are just not there yet. (We don’t sell overseas). To me if the joints at the inductor failed as you described it would have left one of the leads behind. Maybe it did I was just using my experience to make an informed theory on what might have happened.

I understand the amount of shock that these things are subjected to, and you probably have more solder needed for that purpose (since it is known that that is what is going to be subjected to). I understand that you take quality seriously I have read about the exemplary customer service that you have. That’s why I bought my Mamba Max, and in the near future a MMM or a Mamba Max pro for RC and an industrial application in the works.

In my line of work I have seen just about everything (because as soon as you say it, something else comes along) as well I have been with the same company for nearly 15 years and do a lot of repairs, I have seen my fair share of defective components, absolutely wrong installations, people that think the A/D card needs 110 plugged into it, lightning strikes, forklift damage, cards sent in without a thing wrong with them. People who have no business trying to troubleshoot a piece of equipment using a DMM, or any real clue what the problem is in the first place (The old it’s broke, it don’t work deal). Again I am not bashing your product, but to have the same failure mode to me would raise a flag.

Jeff

I agree -- after I saw this thread, I went and asked for statistics from the repair folks here at Castle. They hadn't seen another failure like the one shown here. At that point I had to assume it was a fluke.

But one of my engineering techs (Mike) who is also a racer told me he had seen a failure like that when we were first wringing out betas of the Mamba Monster. It was only a single failure, and it was caused by the inductor cracking (it didn't fall off the board, it just cracked and failed.)

With only three know failures (from about 40K units sold), I can't justify changing the design at this point to address this issue. Maybe in V4. :)