Reply to thread

Message: <blockquote data-quote="John Roberts" data-source="post: 26283" data-attributes="member: 126">Re: Ebtech HumX It seems I've told this story before so stop me if you've heard it... Back in the '80s I used the diode trick, to reduce some ground noise between rack mount products due to a path between chassis and the rack rails. For safety ground bonding you need to accommodate enough ground current to trip a fuse or breaker. One thing that isn't obvious to casual observers is how diodes fail... When diodes get too much current, resulting in too much heat, they melt and fail as a dead short... but wait, a dead short between the two grounds is exactly what we want in a fault condition, and once the diode fails as a dead short, the power dissipation drops dramatically. So the magic to making this work, is to use a diode robust enough that it can handle the initial heat pulse without mechanical failure since the heat after it shorts out is very much reduced. The bottom line is you can't just use some tiny signal diodes since they will vaporize when hit with 50-100a (I know, I did that on the bench), but using something more substantial like a power supply diode bridge, with some more thermal mass to handle the initial transient, and even a heat sink on the larger power bridges. is suitably robust to take out every breaker I ever tested it on.  Of course after it shorts you no longer have a floating ground, but safety bonding is preserved.Later when I was working at Peavey and able to communicate with UL using other people's money, I had our in-house guy ask UL what they thought about the diode scheme (using real diodes).  They did some preliminary bench tests and confirmed that it worked like I said, but to formally approve it, they would want something like $10k or more to set up a case file with a bunch of proof of concept tests with full documentation, specifying approved parts for this application, etc..By then I'd figured out that you don't really need to float a ground if your differentials are set up properly so we never pursued it further, but UL was not discouraging us. If Ebtech is indeed using small signal diodes, YOU DO NOT HAVE A SAFTEY GROUND, that will pass ground bonding (50a), and protect you,  so walk away.  You could roll your own using a relatively inexpensive power supply diode bridge, but I am not suggesting that, since you won't have UL (or me) sitting with you in court. Using properly designed gear, you don't need to game the safety grounds. JRPS: I think I contacted EBtech the last time this came up and told them how they might pursue UL approval. I suspect they are a small company and playing the odds that they don't kill anybody. They may not have room in their current unit to fit a large enough diode bridge.</blockquote>

Verification

Top Bottom