It's time to restore the 11/45.

js at cimmeri.com js at cimmeri.com
Thu Feb 5 14:12:33 CST 2015



On 2/5/2015 1:04 PM, tony duell wrote:
>>     >  From: tony duell
>>
>>     >  A plain Variac does not provide isolation from the mains so you are
>>     >  actually now working on mains-connected circuitry (read : dangerous and
>>     >  you can't clip a 'scope on it).
>>
>> I'm clearly a little slow this morning: why can't you use a 'scope on it?
> With most 'scopes one side of the input is grounded -- you have a coaxial cable
> with the shield connected to the 'scope case and true ground. In most, if not all,
> countries, one side of the mains is connected to ground too. So if you have
> something directly connected to mains (the common example on this list
> would be the the chopper circuit of an SMPSU) then you can't just clip the
> 'scope ground lead on it. If the input is a bridge rectifier (as it is in most
> SMPSUs and in the DEC regulator bricks that started this), then clipping the
> 'scope ground to the obvious point -- the -ve output of that rectifier -- will
> effectively put a diode across the mains. This is a good way to let lots of magic
> smoke out.
>
> The correct way round this is to run the item under test off an isolating
> transformer. Then no point on the secondary side of that transformer is
> necessarily grounded (that is the whole point of it being an _isolating_
> transformer), so you can arbitrarily connect any single point in the circuit
> to ground. So you can clip your 'scope ground lead where you like.
>
> The dangerous way is to 'float the scope' -- that is to disconnect the mains
> earth from the 'scope (in the mains plug, for example. You can now connect
> the 'scope ground lead where you want, but since that point is not necessarily
> the same voltage as true ground, you can end up with the metal 'scope case
> at a high voltage wrt true ground. Very nasty!

I'm glad you said this as, at times, 
I've put my scope on the isolation 
transformer instead of the item itself.

Thank you-
- J.



More information about the cctech mailing list