First off, I've been following this thread off and on since it started. Paul is hilarious, and I am actually kinda touched that you all pulled it off... even if lots of the dirty work was actually done by Fran Sabolich
However, you all might not realize just how small the market is for DC motor controllers for DIY EV conversions. We estimate somewhere in the 100-500 per year range.
And there were three companies, two of which are uppity upstarts (EVnetics and NetGain Controls) battling it out with the venerable and entrenched Zilla (which appears to be on hiatus, at least temporarily) for this market. So, this is a very competitive business and every advantage counts. When I see open speculation about what IGBTs or capacitors, etc., are used in the Soliton1 it frankly makes my stomach churn. Some people (e.g. - Dan Fredericksen) think they are
entitled to know what components are used in these controllers, even though it took people like me or Otmar Ebenhoech hundreds of hours, and often times thousands of dollars, to arrive at the choices made.
So, I simply ask that you folks not speculate so much on what's inside the Soliton1 because, well, making it is how I make a living, and not even a remotely lavish one at that. If EVnetics is going to be a financial success then it will be from making
other products besides EV motor controllers.
Quote:
Originally Posted by MPaulHolmes
The panasonic ts-ed caps should last quite a few years though, since they don't get too hot.
|
Depends on how much ripple current they are exposed to, at what percentage of their voltage rating, and at what ambient temp. Why don't you collect some operating data from people driving around to get a better idea of how much ripple the capacitors actually see "on average". Plug that into the Arrhenius equation to see how that, and the ambient temp, affects the lifetime. Remember - you don't get to use both temperature and frequency multipliers at the same time!
Quote:
Originally Posted by MPaulHolmes
You only need like around 5% of the max current for ripple rating, so they say. 50% is the worst case scenario.
|
That is an oft-cited rule of thumb but hell if I know how someone came up with it. Design for 50% ripple current and you don't have to worry. Design for 25% if you can guarantee that the capacitors will never be operated long enough above their rating to develop an internal hot spot that exceeds 105C (or 85C, though such has no business inside a motor controller). At only 5% ripple, though? Man, that's like Russian Roulette.
Also, you need to make sure that each cap sees its fair share of the current (and same with each MOSFET/IGBT and FWD). The tool usually employed for testing this is an expensive one, though: a Rogowski coil current probe. Good layout/wiring practices goes a long way towards solving this problem before it even has a chance to crop up.
This, btw, is one of the main reasons Kelly controllers blow up so often.
Quote:
Originally Posted by sawickm
A while back I looked into the Capacitors that Soliton Controllers used.
|
And the survey says?
Quote:
Originally Posted by sawickm
They are using CORNELL DUBILIER TYPE 947C Film Capacitors....
|
BUUZZZZ. Nope... but the 947C series is a pretty good guess. Not the best choice for this particular application - aimed more at the dc link function in mains-fed inverters - but not a terrible one.
Quote:
Originally Posted by edown
I think some of you are being a little over concerned about the capacitors, I just use 3 old 4400uf 300V screw mount 85C caps. They never even get warm. I just got them off ebay for about $3 each. And I can’t believe I will have this set-up long enough to worry about there life span. Anyway, that’s my 2 cents worth.
|
The part of the capacitor that gets hot is the very center, and it's a long way - thermally speaking - from the very center to the case... in fact, the internal temperature limit is reached with as little as a 5C rise of the case.
Perhaps you won't use the controller long enough to matter - that's entirely possible - but sometimes the failure happens within seconds or minutes, especially if your layout results in one capacitor taking the brunt of the ripple current.