Low (<16A) current monitoring, 22Ω vs 120Ω and SCT013 vs SCT006?

I’ve done a quick search and didn’t find an exact answer in a form I understood so I was hoping someone could enlighten me on a few details.

I have a emonTx 3.4 with 3 SCT013 connected to CT1, CT2 and CT3 with default 22Ω burden resistors. CT2 and CT3 are never going to see above 3KW and much more likely to be in the 0-2.2KW range.

What I’d like to change:
CT2 and CT3 will almost always be in the bottom 10% of the SCT013 range, from what I understand, this isn’t ideal as it’s less accurate.


  1. Ignore the inaccuracy and leave things alone
  2. Use SCT013 on high sensitivity inputs
  3. Use SCT006 with 22Ω burdens
  4. Use the existing SCT013 but loop the wires

I’ve had a quick look through the reports by @Robert.Wall, unless I’m misunderstanding, the SCT013 looks pretty good down to 0.5A. Should I even be trying to improve things?

If the SCT013 is used on a high sensitivity input it has a 18.8A/4.5KW max which fits well with my scenario. How does this effect the accuracy at the lower end of the SCT013 range? Is it improved when compared to a normal input or is it inherent due to the CT? Converting CT3 looks easy enough.

The SCT006 should be perfect for the range I need and again, Robert’s report seems to suggest it should work well with a 22Ω burden. I assume the sketch would need updating to account for the alternative CT, are there any suggested settings?

The last option is probably the easiest, I would need to extend two wires to give enough slack to loop through. I’d need to check how much the cable will be de-rated giving it’s effectively a bundle. Assuming the max current would be 13A, a single 2.5mm² can probably be looped 4/5 times and still be fine. The CT may act as a heatsink given it’s a hunk of metal and the cables will be touching it?

As always, any thoughts even if it’s just to correct me are much appreciated.

Thanks again,

My first problem is I can only infer that you’re on a 240 V supply (from “18.8A/4.5KW”), so I’m assuming that you are.

As the SCT006 report says, using it is roughly equivalent to gaining one more bit in the ADC. In your case, and assuming that you’re not going to be measuring the maximum loads you quote all the time, then there’s probably a small advantage is changing. But I’d caution against expecting too much.

Any c.t. is at its most accurate when it’s working at minimum power, i.e. into a short circuit. Unfortunately, we need some voltage to measure :anguished:
So the lower the burden resistance, the better as with all engineering, it’s a trade-off between the c.t’s and the ADC’s inaccuracies and noise.
Unfortunately again, all c.t’s become inaccurate at low currents. That’s because the magnetising current is no longer negligible. The definition of “low” depends largely on how good your c.t. is and how much you paid for it. A revenue grade measurement c.t. will be the best in that area.

If you’re using the SCT006 below 16 A, it should be fine, but in the 50 Hz world it struggles above that.

If you do put several turns through a SCT-013-000, I wouldn’t worry too much about derating. Derating is all about temperature rise and lack of cooling. The wires will only be bunched for 15-20 mm and round the outside of the c.t. they can be fanned out to leave air circulating between them. In any case, you’ve got to get to 7 or 8 turns before a 2.5 mm² derates to 16 A, and I think you need 6 turns maximum. It’s cheaper to put 2, 3, 4 or 5 turns through a SCT-013-000, and that would give you a 50 A, 33.3 A, 25 A or 20 A c.t. And the last two at least could end up with a better nominal accuracy than the SCT006. (I’d need to do the sums to prove that though.)
That would be the most accurate, but hard to implement, and what you’re gaining in accuracy may not be worth the trouble.

If you use the SCT006 and the 22 Ω burden, it’s certainly easier.

I wouldn’t advocate increasing the burden value if it’s possible to use multiple turns in a SCT-013-000, and an SCT006 in the 'high sensitivity" input definitely doesn’t make sense.

As far as the sketch goes, you need to correct the calibration. The current calibration constant is the current needed to deliver 1.0 V across the burden resistor.

If you’re using the CT006, it’s 36.36 (the c.t. is 20 A : 25 mA). Or if you’re using multiple turns with the SCT-013-000, just divide by the number of turns (e.g. 90.91 ÷ 5).

I think you’ve got to weigh up what is practical and what you’re happy with doing.

Yes, I’m in the UK so it’s a 50hz/240V single phase supply. I’ve updated my profile to add a location :slightly_smiling_face:

Based on what you’ve said, extending the wires and looping them around the existing SCT013’s sounds the most plausible plan. It’s not much work and I can do the division in emonCMS to avoid needing to customise the sketch yet.

The valve in the sketch also makes sense now you’ve mentioned it’s the current needed to give 1V: (1 / burden resistance) / CT output A * CT input A

It’ll certainly be useful if I end up trying other CT’s in the future!

Thanks again,

1 Like