🚀 go-pugleaf

Joerg <invalid@invalid.invalid> wrote:

>Nico Coesel wrote:
>> John Larkin <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:
>> 
>>> On Tue, 30 Jun 2009 11:38:10 -0700, Joerg <invalid@invalid.invalid>
>>> wrote:
>>>
>>>> John Larkin wrote:
>>>>> On Tue, 30 Jun 2009 17:27:21 GMT, Rich Grise <richgrise@example.net>
>>>>> wrote:
>>>>>
>>>>>> On Thu, 25 Jun 2009 12:39:08 -0700, John Larkin wrote:
>>>>>>
>>>>>>> On Thu, 25 Jun 2009 13:48:35 -0400, Phil Hobbs
>>>>>>> <pcdhSpamMeSenseless@electrooptical.net> wrote:
>>>>>>>> John Larkin wrote:
>>>>>>>>> On Wed, 24 Jun 2009 21:42:07 -0700, <castlebravo242@att.net> wrote:
>>>>>>>>>
>>>>>>>>>> ... was reading it would get data errors. The problem was that the
>>>>>>>>>> shield on the cable from the heads to the preamp was grounded on both
>>>>>>>>>> ends.
>>>>>>>>> That's usually the best way to ground shields.
>>>>>>>>>
>>>>>>>> Depends.  With floating transducers and shielded twisted pair, it's best
>>>>>>>> to ground the shield only at the transducer end.  That way any
>>>>>>>> capacitive imbalance doesn't produce any currents, because the shield is
>>>>>>>> at the transducer's idea of ground.
>>>>>>>>
>>>>>>>>
>>>>>>> At the receive (instrument) end, the ungrounded end of the coax shield
>>>>>>> becomes the hot side of a resonant "vertical" antenna. Any voltage there
>>>>>>> becomes 100% common-mode signal on the inner twisted pair.
>>>>>>>
>>>>>>> What's wrong with a little current in the shield? The signal pair isn't
>>>>>>> affected. But volts, or tens of volts, of common-mode RF can be nasty.
>>>>>>>
>>>>>> I think half of us are talking about low-noise audio frequencies, and
>>>>>> half of us are talking about RF. The way I heard it, you leave one end
>>>>>> unshielded for audio preamps and stuff, but ground both ends for RF;
>>>>>> I think it has to do with wavelength and, well, propagation
>>>>>> characteristics. Of course, the thing that was drummed into me while
>>>>>> they were teaching me this was "GROUND LOOPS!". :-)
>>>>>>
>>>>>> Cheers!
>>>>>> Rich
>>>>> Audio: if one end of a shield is disconnected, the delta-voltage (60
>>>>> Hz, RF, esd spikes) between chassis is greater. If the signal is
>>>>> single-ended, that increases noise directly. If differential, it
>>>>> increases common-mode noise.
>>>>>
>>>>> Why would you want to do that?
>>>>>
>>>>> A 60-Hz (and RF, and ESD) "ground loop" can result from different
>>>>> boxes being plugged into different outlets which have some 60 Hz
>>>>> difference voltage (or form a relative antenna for RF.) Connecting the
>>>>> shield on both ends shorts out some or most of this difference voltage
>>>>> and REDUCES the effect of the "ground loop" on the signal.
>>>>>
>>>> Unless the shield slowly commences a faint red glow, followed by some 
>>>> "amperage scent" wafting through the room ...
>>>>
>>>> Seriously, I've seen that happen.
>>> Yup. I've seen some ghastly wiring, some done by licensed
>>> electricians.
>>>
>>> The most 60 Hz I've experienced (between outlet ground pins, in an
>>> office building) was a few volts RMS, and plenty stiff. Messed up some
>>> printer signals pretty good.
>>>
>>> The real way to pipe signals around is fiberoptics. A kilovolts of
>>> common-mode is no problem.
>> 
>> I wish... I'm working on interfacing a current sensing coil from a
>> welding transformer. During welding there is a couple of hundred volts
>> everywhere. This project starts to give me a headache :-)
>> 
>
>Try differential transfer, shielded twisted pair or Twinax soldered to 
>the burden resistor, from there to your measurement box, place a 2nd 
>signal transformer there. Make sure the twisted pair is source- and/or 

I already went for a differential input. Solves most of the problems,
but not all but I'm pretty sure the inputs where overdriven as well. I
fixed that, but didn't field-test it. A 2nd signal transformer won't
fly because the pulses may be quite long (up to 100ms). 

The problem is that nobody seems to know what the pulses look like
exactly. The transformer is pulsed by a 1000Hz PWM signal. This makes
me assume the current sensor sees pulses as well. But the shape
probably depends on the induction in the welding circuit while
carrying many kA...

>end-terminated properly, usually 100-120ohms. And never, never use this 
>connection as the burden resistor even if there is a fire extinguisher 
>at hand ...

That might be worth something to explore. The burden resistor (1k is
specified) is now on the measurement board itself. I'd like to keep it
there though to keep the installation procedure easy. The wiring in
the test setup is pretty short (2 meters) but bad (not twisted,
unshielded). I'll order some shielded twisted pair tomorow. I already
figured this must be part of the problem.

-- 
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
                     "If it doesn't fit, use a bigger hammer!"
--------------------------------------------------------------