Safely probing flyback primary with scope

[es1]

I am working on a flyback AC:DC switching converter design which converts wall power to DC. The design is working but I wanted to probe it for some characterization data.

I do have Earth ground in my design and it is wired up exactly like this TI EVA schematic.
http://focus.ti.com/lit/ug/sluu263c/sluu263c.pdf

So I plugged the design and the scope into the same power strip, both use the third prong. Told the scope to use Earth ground as its reference and as soon as I connect the probe ground to the design the input fuse popped.

This is exactly the behavior described in the app note below, page 2 warning 2. But the National design has no Earth ground (but maybe this is irrelevant because my design also has no Earth ground per se during the rectifier dead time).

http://www.national.com/an/AN/AN-2097.pdf
"The ground connection on the evaluation board is NOT referenced to Earth ground. If an oscilloscope ground lead is connected to the evaluation board ground test point for analysis and the mains AC power is applied (without any isolation), the fuse (F1) will fail open. For bench evaluation, either the input AC power source or the bench measurement equipment should be isolated from the Earth ground connection. Isolating the evaliation board (using 1:1 line isolation transformer) rather than the oscilloscope is highly recommended."

I do have access to a line isolation transformer, and following the warning I isolated my design which does work. But I am trying to figure out what happened in the first setup and why using the scope with the Earth ground setting failed.

Maybe the scope's probe ground is not actually physically connected to Earth ground but only some copy? I guess it must be because while typing this I happened to think that this signal has to move from the primary to secondary side inside the scope somehow, no?

Also, why do you think isolating the design rather than the scope is recommended? Shouldn't it be equivalent from a measurement point of view? I figured maybe it was a safety thing as the scope presumably has isolation but if one is probing the primary (w/o the iso transformer) then there is no isolation.

Does anyone know of a general reference on how to safely take accurate measurements on systems connected to the line with a scope?

 

[berkeman]

Without the isolation transformer, it looks like the reference for the "ground" of the switcher is Neutral. So when you connect your 'scope ground (which is Earth grounded) to Neutral, you can get big enough fault currents to pop the fuse.

It's much more preferable to isolate the switcher from the AC Mains than to "cheat" your oscilloscope (using a 2-prong power plug adapter thing), because you are still at risk of ground faulting the circuit yourself if you happen to touch some part of the circuit. With an iso transformer, you can touch one thing at a time (but usually not 2!), either accidentally or on purpose.

 

[es1]

That makes sense.
It looks like using an isolation transformer on the scope is the equivalent to just cutting the third prong.
http://www.tek.com/Measurement/App_Notes/Technical_Briefs/tds3000-float/eng/safety.html

From searching online it looks like the preferred way to safely do these measurements is with a special probe that can convert a floating measurement to a low voltage single ended signal.

http://www2.tek.com/cmswpt/psdetails.lotr?ct=PS&ci=13415&cs=psu&lc=EN

 

[berkeman]

The preferred way is to make the circuit under test isolated from the AC Mains, so that a ground fault (like you touching part of the circuit) does not generate a ground current.

It can be as simple as touching a heatsink to figure out how hot something is running... Do that with a non-isolated switcher, and you may get a surprise...

 

[mdjensen22]

If you've got deep pockets, there are some nice differential scope probes that take the worry out of trying to do this.

For example, I use the Techtronix P5205 probe for this type of measurement.

 

[mheslep]

Battery powered (thus isolated from ground) scopes are inexpensive these days. Handy for field work too.

 

[jsgruszynski]

Back when there were more CRTs around, every scope maker sold optional HV probes exactly for this application.

 

[likephysics]

You could also use isolated scopes like the tek TPS2021