Measuring 240v 3-phase circuits with an oscilloscope

Question

I have an injection molding machine that has some odd behavior* going on in it which I need to diagnose. I've probed around with a multimeter, but I think I need to understand deeper what is happening. So, I want to connect up an oscilloscope. However, I'd also like to avoid frying my o-scope or electrocuting myself.

The house main circuit breaker runs to a 3 phase converter, which then serves out 240v, 3-phase power to the machine. The machine's protective earth (PE) is tied to the phase converter, which is tied to the house grounds.

I want to probe the different individual legs being used to power heating elements and probe between them to compare what is going on. Here are my questions:

1. Can I use a passive probe for this with either 1x or 10x attenuation? The datasheet[1] for the o-scope has max input voltage of 400 Vpk. If I tie the ground of the probe to machine PE/ground, this should be ok, right?

2. Would a differential probe be more appropriate? I ordered 2x of these, thinking they would be the way to go, but I'm concerned that if I have an inappropriate voltage (ie: short) on two of the legs at the same time, the common mode will reject this. It would be more appropriate to use the probes on separate legs, measuring leg to PE, and then use the o-scope to subtract the two waveforms, right?

* More detailed description of what is going on: when the heating bands of the machine are turned on, power is applied from a 240v leg through a solid state relay, and then goes back to another leg. All of the solid state relays are brand new (I replaced the old ones). You can see the schematics in reference [2]. Pages 6 and 7 are the heating elements. My machine has the K85, K84, and K81 zones (no K82). I have verified that points 85, 89, and 95 are not tied together (no neutral wire for these elements). I'm having an issue where if I turn on K84, K81 also has energy flowing through it. I've tried to use a multimeter to find a short, but I haven't had any luck so far. What's worse is when I pull out the plugs for E81, E84, and E841 that would go to the heating bands and measure point 94 to PE with the SSRs energized, I see 115v RMS (expected) and when I measure point 95 to PE with power on, I see ~80v RMS (not good).

[1] https://res.cloudinary.com/iwh/image/upload/q_auto,g_center/assets/1/26/SDS1000X-E_DataSheet.pdf

[2] https://drive.google.com/file/d/1_mbgeQGryCEPEZo5bxtJjDMLyaaw9okz/view?usp=drive_link

Answer 1

Phase to ground of a balanced three-phase 220 V line will be \$\frac{220\sqrt2}{\sqrt3}\ \mathrm V ≈ 180\ \mathrm V\$, which is well inside the voltage rating of your scope, even with a 1× probe¹.

However, phase to phase is a different matter. The vast majority of oscilloscopes² have the ground of all their inputs connected directly to earth ground, and if you connect the ground lead of your scope probe to anything other than earth ground, you will short out the equipment and probably damage your oscilloscope.

If it's possible to isolate your equipment under test³ from earth ground, you can measure the phase-to-phase voltage with a standard oscilloscope probe; however, note that it will have a peak voltage of \$220\sqrt2\ \mathrm V ≈ 311\ \mathrm V\$, which may be more than your oscilloscope probes are rated for (the ones I use regularly are rated for 300 V peak, for instance). There's a solution to this, of course: high-voltage passive probes, which are usually 100× probes rated to a few kV. These are cheaper and usually higher-bandwidth than high-voltage differential probes, so a good choice when you don't need the differential functionality. I use a LeCroy PPE4kV high-voltage probe, which it seems has been obsoleted and replaced with the PPE6kV-A and the HVP120, both 100× probes rated to 6 kV peak.

For measuring phase to phase voltages without a fancy isolated oscilloscope or isolating the equipment, you need a differential probe. Not just any differential probe will do; most things sold as simply "differential probes" will be designed for measuring differential signalling protocols, and will have a peak voltage limit of less than 20 V. You want specifically a high-voltage differential probe. High-voltage differential probes usually have pretty poor bandwidth, typically maxing out at around 100 MHz, but that's probably enough for what you're working on. I use a Cal Test CT4072, but if you don't need that 100 MHz bandwidth there are cheaper options.

There are a couple caveats with differential probes, first among them being that they are not isolated. The electronics inside the probe body convert the differential signal to a single-ended signal, and attenuate it by a large amount, but there is no galvanic isolation inside there. That means the output of the probe should be treated as if it were live.

Another caveat to bear in mind is that diff probes have complex voltage ratings. The figure usually labelled as "maximum input voltage" is the maximum voltage between the positive and negative input terminals, and depends on the attenuation setting--at lower attenuations, the maximum input is lower. The figure usually labelled as "maximum common-mode voltage" or "maximum input voltage (input to ground)" or similar is the maximum voltage that can be applied to either input with respect to earth ground; again, high-voltage diff probes are not isolated. This figure usually does not (but sometimes it does; check your probe's manual) depend on attenuation setting. Both of these limits must be separately observed. Using the aforementioned CT4072 as an example,

^{CT4072 datasheet, page 2.}

On the 1000× attenuation setting, the limit for differential voltage is 3500 V peak, but the limit for either input to ground is 2500 V RMS. So you would be within both limits if you measured between a +2000 V DC line and a -1000 V DC line, but if you only read the 3500 V number and measured between a +3000 V DC line and ground, you'd damage the probe.

At the voltages you're working with, that's not anything you need to worry about, though.

---

Footnotes:
_{^{¹ Assuming you can even find a 1× probe that's rated for such a high voltage. I recommend using a 10× probe; there's almost never any reason to use a 1× probe.
² Those that don't are specially designed to have isolated inputs, and you pay a premium for that. I know Fluke makes (made?) a handheld oscilloscope/multimeter ("ScopeMeter") with two isolated inputs, and Tektronix used to make a four-channel scope with isolated inputs. There are probably others.
³ Always isolate the equipment under test, never your oscilloscope! There are a number of reasons isolating the scope is a bad idea, which other people have already expressed more eloquently than I could elsewhere on this site; give it a search if you're curious.}}