Oscilloscope input safety ratings

[Salas]

Hi, oscilloscopes have input marks like 300VRMS CAT I. Probes have some max voltage spec too. Are they safe enough to those claims after possible high voltage practical experiences and tear down observations? Are some brands better than others to work with vacuum state circuits for instance, or to watch mains noise?

[Lightages]

I am no expert on oscilloscope front ends, but you should never connect an oscilloscope directly to mains. It should be done with a differential probe or other isolating device.

[alm]

CAT I 300 V would be enough for low energy circuits up to 300 V. This includes measuring on the secondary side of any reasonably sized power transformer. Any mains circuit will be at least CAT II.

Many scopes are also rated for something like CAT II 100-150 V. A 10x probe rated for CAT II 300 V will attenuate signals up to 300 V by 10x, so the scope only gets to see 30 V. Pay attention to the probes limits, if you exceed them the attenuation resistor/cap might short and the scope suddenly gets to see the full voltage. If the probes are of the switchable type, make sure you don't accidentally switch them to 1x. CAT II 300 V should be fine for connection to mains circuits as long as there are no high power motors or substations nearby. See Fluke's appnote 'The ABCs of multimeter safety' for a description of the overvoltage categories. Note that these ratings are for DC or low frequency AC; they are derated with frequency. The probe won't be able to handle 300 V of RF.

Since 99% of the scopes out there are grounded, you can't directly clip a probe across a mains circuit like you would do with a DMM. Never defeat the scope's ground connection, it's critical for safety. You can get a decent approximation of the mains signal by just measuring between ground and live. In most circuits there is only a few volts between ground and neutral. Don't clip the ground lead to the neutral, though, large currents can flow. If you want to directly measure between live and neutral, you can either use an isolation transformer or a differential probe. A poor man's differential probe (common-mode rejection ratio will suck) can be made by using two channels, connecting the ground clip together, and put them in subtract mode. You can then connect the tip of the two probes to live and neutral.

Of course common precautions for working with dangerous circuits should be followed, like avoiding working live and following the one hand rule.

[kg4arn]

The scope input (at the BNC has a CAT rating) e.g. CAT 1 300 Vrms.
Adding the probe may increase the CAT rating of the scope plus probe system.

For example:  My Agilent DSO-X-3024A has a CAT 1 300Vrms input rating at the BNC  on the front of the scope.
The supplied N2863B probe is rated 300Vrms CAT I, II.
So the scope plus the probe may be used CAT II.  This is my understanding.

Also beware the probe's voltage derating curve with frequency.
As frequency increases the probe may be damaged by fairly low voltages (certainly lower than the 300Vrms figure). 

I attached a typical curve.

The curves are not exactly the same for all the different passive probes/manufacturers, but they seem to have the same general shape, in my experience.
Also, some of the less expensive probes seem to be fairly well made but the manufacturer wont provide any derating curves.

[kg4arn]

... to watch mains noise?

Take alm's advice.  Get a step down transformer, say mains voltage to 12 or 6 Vrms and monitor the mains noise from the isolated low voltage side.  Fuse the primary on the HOT wire not the neutral.

[Salas]

Alright guys, thanks. So its better to avoid high energy even if the probe's spec is CAT II. 

[ejeffrey]

... to watch mains noise?

Take alm's advice.  Get a step down transformer, say mains voltage to 12 or 6 Vrms and monitor the mains noise from the isolated low voltage side.  Fuse the primary on the HOT wire not the neutral.

His flag is greek.  I believe they use unpolarized schuko and europlug outlets.  You have to fuse and switch both leads or risk leaving the circuit live.

[kg4arn]

His flag is greek.  I believe they use unpolarized schuko and europlug outlets.  You have to fuse and switch both leads or risk leaving the circuit live.

Thanks for correcting me 

[Salas]

Will a transformer inevitably filter out some of the original mains higher frequency garbage to observe? Is it better be a toroidal type that its higher primary to secondary capacitance due to windings on top of another may pass more to see?

[SeanB]

For the HF noise use a 1n class Y cap in series with a 10n class X cap as a capacitive divider.

[tlu]

I am no expert on oscilloscope front ends, but you should never connect an oscilloscope directly to mains. It should be done with a differential probe or other isolating device.

Can this isolation be done by just cutting off the earth ground prong on the power cord itself? I always thought that if you did this, you would have only the floating ground. Please correct me if I'm wrong.

tlu

[Salas]

For the HF noise use a 1n class Y cap in series with a 10n class X cap as a capacitive divider.

Connected as a series system across where? Live to earth?

[ejeffrey]

Can this isolation be done by just cutting off the earth ground prong on the power cord itself? I always thought that if you did this, you would have only the floating ground. Please correct me if I'm wrong.

Sure, if you want to end up dead.  Seriously, don't do this.  You shouldn't do it with any device, but especially never and T&M gear.  It will cause the entire oscilloscope to float at whatever voltage you clip the ground lead to.  If that is a line voltage, the whole oscilloscope becomes an electrocution hazard.  The most obvious problem here is that the BNC shields on unused channels, trigger, the calibration terminal, any communication interfaces, or any exposed metal parts become live.  Less obvious is that even apparently insulated components like buttons, switches, knobs, and the screen have not been designed for safety isolation, and do not maintain necessary clearance margins.  Finally, if you connect the scope to anything else (USB, serial, LAN ports) the device will become grounded again, and un-defeat your ground defeat, only now instead of a correctly designed ground terminal, the fault current will flow through your USB cable shield or something equally horrible.

There are purpose designed scopes that run isolated, and they have insulating shells on their probe jacks and so forth to reduce the risk.  Use one of them, use a differential probe, or just measure relative to ground.  Don't ever connect your probe ground clip to anything but a ground connection.

[tlu]

Can this isolation be done by just cutting off the earth ground prong on the power cord itself? I always thought that if you did this, you would have only the floating ground. Please correct me if I'm wrong.

Sure, if you want to end up dead.  Seriously, don't do this.  You shouldn't do it with any device, but especially never and T&M gear.  It will cause the entire oscilloscope to float at whatever voltage you clip the ground lead to.  If that is a line voltage, the whole oscilloscope becomes an electrocution hazard.  The most obvious problem here is that the BNC shields on unused channels, trigger, the calibration terminal, any communication interfaces, or any exposed metal parts become live.  Less obvious is that even apparently insulated components like buttons, switches, knobs, and the screen have not been designed for safety isolation, and do not maintain necessary clearance margins.  Finally, if you connect the scope to anything else (USB, serial, LAN ports) the device will become grounded again, and un-defeat your ground defeat, only now instead of a correctly designed ground terminal, the fault current will flow through your USB cable shield or something equally horrible.

There are purpose designed scopes that run isolated, and they have insulating shells on their probe jacks and so forth to reduce the risk.  Use one of them, use a differential probe, or just measure relative to ground.  Don't ever connect your probe ground clip to anything but a ground connection.

Thanks for the information on this matter. So as I as I do not clip the ground of the probe to a mains line, this set up with the cut-off prong will be ok or like you said just go with the differential probe? I do not want to spend any money on this if at all possible.

tlu

[alm]

Thanks for the information on this matter. So as I as I do not clip the ground of the probe to a mains line, this set up with the cut-off prong will be ok or like you said just go with the differential probe? I do not want to spend any money on this if at all possible.

Both ejeffrey and I have stated quite explicitly not to defeat the ground connection. A ground connection is essential for safe operation. Don't ever cut of the ground pin, ask your family how much a funeral will cost them if you think it will save you money. Experienced technicians have died doing this. Both of us have described several alternatives that don't require an expensive differential probe, although you should get one if you want the best signal integrity. The simplest of them is just not connecting the probe's ground lead, probing the neutral or live wire (taking the usual safety precautions), and relying on the ground pin on the scope's power cord to provide ground. This will be a very high inductance ground, so don't expect great signal integrity, but it should show you the basic sine. The next step would be the poor man's differential probe that I described.

[tlu]

Both ejeffrey and I have stated quite explicitly not to defeat the ground connection. A ground connection is essential for safe operation. Don't ever cut of the ground pin, ask your family how much a funeral will cost them if you think it will save you money. Experienced technicians have died doing this. Both of us have described several alternatives that don't require an expensive differential probe, although you should get one if you want the best signal integrity. The simplest of them is just not connecting the probe's ground lead, probing the neutral or live wire (taking the usual safety precautions), and relying on the ground pin on the scope's power cord to provide ground. This will be a very high inductance ground, so don't expect great signal integrity, but it should show you the basic sine. The next step would be the poor man's differential probe that I described.

Thanks alm. After hearing both of you experts on this matter, I'll get another powercord with the ground prong. I guess paying for a differential probe is cheaper than risking anything dangerous. I'm looking forward to your poor man's differential probe diy write up.

tlu

[alm]

A poor man's differential probe is very simple:

A poor man's differential probe (common-mode rejection ratio will suck, especially at higher frequencies) can be made by using two channels, connecting the ground clip together, and putting the channels in subtract mode. You can then connect the tips of the two probes to live and neutral.