1

Our robot is supplied trough our own "supply block", which is converting 3 phase AC (without neutral) into DC current and powering a 2 smaller devices directly between 2 pairs of phases.

schematic

simulate this circuit – Schematic created using CircuitLab

Phase to phase voltage is 380V nominal

I was tasked to measure power (active, reactive and apparent) for this whole system.

The issue is that we don't have an appropriate "watt-meter" to compute those powers, so I will have to do with what I have :

  • a fully isolated 2-channel scope + probes (max 600V, cat III 600V between channels, cat III 600V to earth, cat III 600V to any exposed connector)
  • one current probe (clamp) with 1000V Cat III isolation, that I can connect to the scope
  • 2 multi meters with at least 600V Cat III ratings (not sure they will be of any use)

The rest of our measurement equipment is not rated for working on mains.

So I can measure with the scope simultaneously either 2 voltages, or one voltage and one current. Consumption should be rather reproducible, so I can switch off, unplug, displace the probes and start again.

To complicate things further, as you can see on the schematics, the load is really balanced (even if the unbalanced part are less than 20% of real power, but might be a significant part of reactive power, as the AC/DC claims power factor >0.99).

So how to proceed to get the best possible approximation of power (active, reactive and apparent), given the tools we have?

Sandro
  • 5,519
  • 6
  • 24
  • It sounds to me like you are going to have a hard time getting an accurate result. – Andy aka Apr 05 '23 at 10:44
  • I agree that what I have is far from ideal. On the other hand, I don't really need very accurate results (the main goal is to know what supply we will need for field tests). So the goal is mainly to have an idea of the power (and an idea of the max error) to know what cable section and what generator we need for field testing. – Sandro Apr 05 '23 at 10:58
  • If we get something a bit more accurate, then we can do some additional tests, like measuring the impact of different motors/movements/configurations/... but this is just a nice to have for now (and we can also do it indirectly, by measuring power on the DC side, which is easier to do accurately) – Sandro Apr 05 '23 at 11:00
  • So far, the best idea I've found is to suppose the load is balanced, and measure just one current and one voltage, and compute the power based on balanced load formulas. Then I start again measuring the 2 other currents, and computing slightly different powers. And finally averaging the 3 powers. Is there any better approach? Any idea how to estimate the error I'm making? (can I safely assume that the power is somewhere between the smallest of the 3 and the biggest of the 3?) – Sandro Apr 05 '23 at 20:22

0 Answers0