Floating a function generator

[salbando]

I thought I was getting a handle on floating, but I was reading the instructions to a function generator and it says the following:

"The shield of this output may be floated up to +/-40V relative to earth ground."

What does that mean?

Thanks

[Bored@Work]

Sounds like some Chinglish where the meaning got mangled in translation. My guess: Iit should mean you can add up to +/- 40V DC between earth and the shield of the output.

[salbando]

Later the manual goes on to say (in reference to the FUNCTION and SYNC BNC's):

"Both of these outputs are fully floating, and their shields may be floated relative to earth ground by up to +/-40V."

Does that sound like what you are saying about adding up to +/-40V DC? Because to me, it still doesn't make sense. If a device is fully floating then I don't understand why there would be any limit between it and a potential to ground. I mean, if I had (which I don't) a 1000v supply, and the 1000v was referenced to earth ground, why couldn't I connect that to the floating ground of any particular device? That device shouldn't have any "knowledge" about what is connecting to its ground rail and any voltage within that device will just be referenced to its internal ground... right?

Now I understand that if I (an earth grounded person) were to touch this device then I would see a potential of 1000v plus or minus the relative voltage of the device itself.

[Bored@Work]

Because there is an isolation between the shield and the ground that is not rated for more than 40V?

[salbando]

Oh... Because the entire function generator is not floating, just these two ports, then you're saying that internally the BNC's have been isolated through something like a 1:1 transformer, and that transformer or whatever has a 40V rating...

Thanks, that's one less thing to wake me up a two in the morning now.

[alm]

Which model function gen are we talking about?

The individual outputs are likely fed by separate transformer windings. The shield of the BNC connectors is not connected to ground, as you might expect. This prevents ground loops and allow you to give it an offset relative to ground by connecting a DC source between ground and BNC shield.

The 40 V limit is most likely because the BNC connectors are not insulated, so they should be safe to touch at all times.

[salbando]

I have a Wavetek 190 and on the front, next to the BNC's it says float 42v peak and these words appear between the bnc and a ground symbol, but the manual says nothing about this. So I found other manuals for other function generators to  see if I could learn what it meant.

Now since my posting, I tried some experiments with my Wavetek to see if the bnc's are truly not earth referenced, but now I am more confused because when I used my multimeter to check to see if there was a voltage present between the center of the bnc and some other earth ground (I used the shell of the bnc on my oscilloscope) it said there was full voltage. I used the LoZ setting on my Fluke 117 to see if it was a ghost voltage and it wasn't.... Hmmm... Then I tried using my scope probe without the ground clip and hooked it to the center output of the generator and lo and behold a fine looking, undistorted sine wave. To me, this means the bnc shell of the generator is earth referenced.  Hmmm indeed.

[alm]

What's the resistance between the BNC shell and ground? The symbols strongly suggest floating outputs, which is actually fairly common. Does it have terminals somewhere that allow you to connect the shield to ground for grounded operation? Maybe a modification by a previous owner? Any hints in the service manual (if available)?

[salbando]

The resistance from bnc shell to ground is about 1.5 mega ohms (so that sounds like floating to me) and no, I don't see any ability to connect the shell to ground.

But if they are floating, then why did my experiments seem to indicate otherwise?

[alm]

I would say capacitive coupling between ground and shell, although the loZ result is somewhat puzzling. I would just connect a low, current-limited DC voltage between ground and shell and see what happens to the voltage between them.

[salbando]

I limited my power supply to 50mA, but saw no voltage between ground and bnc shell. Then I tried 250mA briefly, still nothing. I was afraid to do more so I guess my mystery continues.

[Mechatrommer]

"The shield of this output may be floated up to +/-40V relative to earth ground."

Iit should mean you can add up to +/- 40V DC between earth and the shield of the output.

sound semantically equivalent to me. though maybe i have greater translation ability (or disability) since i frequently mixed up with english, chinglish and manglish except the DC part, i believe the manufacturer means any instantaneous absolute voltage magnitude compared to earth, DC or AC (did i make things more confusing?)

I limited my power supply to 50mA, but saw no voltage between ground and bnc shell. Then I tried 250mA briefly, still nothing. I was afraid to do more so I guess my mystery continues.

my guess is the unit is fully floating, no fancy path stuff to earth whatsoever except the "natural" capacitive coupling, which is what you saw scoping the shield. the ±40V limitation probably because the manufacturer figured out thats the limit before you start to feel the tinggling sensation by touching the shield (AC).

[salbando]

Ok, I found this page from Agilent
http://dsnra.jpl.nasa.gov/technical/cp_LibraryFAQDetail.jsp.htm
and it shows how to add a DC offset through the modulation bnc to "float" the output up to 42V

It all make sense and seems to answer my questions except for the use of the word "float" because as you can see in the figure, both the generator and the DC supply are grounded, although the generator is grounded via 1 Meg ohm.

Also step 3 talks about connecting the power supply to ground and the figure shows two grounds at that point... so I guess I am just getting more confused. 

[alm]

You can pretty much ignore the 1 Mohm resistor when driving it from a low impedance source of the power supply. What the schematic is showing is that your signal (with the extra offset) is between ground and BNC center pin, not BNC shield and center pin. The modulation input is just a convenient point to connect to the shield of the output connector.

The connection between shield and ground in your function gen appears to act like a dead short. This suggests to me that either the silkscreen is wrong, or some previous owner (accindentally) shorted it. For example, someone might have skipped some insulating washers. You could try to trace the short using similar techniques as suggested for the Lecroy scope Dave was recently troubleshooting: shove some current through there and measure millivolts with your DMM. The lower the voltage, the closer you are to the short.