All error values are given in ppm of Full Scale (1ppm = 1 part per million = 1E-6).

A LHC600A-10V converter operating at 10 ppm of precision (10 ppm x 600 A = 6 mA) can then produce theoritically an error of up to 6mA whatever the current level of operation, even at 1 A where the relative error would be of 6 %.

topConverter accuracy classes

In the context of HL-LHC, a new revision of performance metrics describes better converters accuracy classes. This short page explains and details the terminology and concepts which are used in this Converter Accuracy Table widely used at CERN. (This page is entirely based on the referred document below)

Design Requirements
Accuracy class Class 0 Class 1 Class 2 Class 3 Class 4 Class 5 Class 6 Class 7
0.5 0.5 1.0 1.0 1.0 1.0 1.0 1.0
Initial uncertainty after cal.
[2x RMS ppm]
2.0 2.0 3.0 7.0 10.0 50.0 100.0 200.0
[|.| max ppm]
2.0 2.0 5.0 8.0 9.0 20.0 50.0 100.0
Stability during a fill (12 h)
[2x RMS ppm]
1.0 2.6 15.5 33.0 39.0 50.0 100.0 200.0
Short term stability (20 min)
[2x RMS ppm]
0.2 0.4 1.2 2.0 5.0 10.0 20.0 50.0
Noise [0.1; 500] Hz
[2x RMS ppm]
3.0 5.0 7.0 15.0 19.0 50.0 100.0 200.0
Fill-to-fill repeatability
[2x RMS ppm]
0.7 1.6 14.5 32.0 38.0 60.0 100.0 200.0
Long term fill-to-fill stability
[2x RMS ppm]
9.5 9.5 26.5 56.0 64.0 200.0 500.0 1000.0

Measurements    -    Requirements

*   The table above already includes temperature coefficient correction. For more details place cursor above class header

Newly defined performance metrics

Figure 1: Newly defined performance metrics

  • References
    • Accuracy class definition (Miguel Cerqueira Bastos, CERN) .edms
    • HL-LHC Power Converter Requirements (Miguel Cerqueira Bastos, CERN) .edms
    • HL-LHC High Precision Current Measurement - a roadmap 2012 (Miguel Cerqueira Bastos, CERN) .edms

top[Old] - LHC Classes

This short page explains and details the terminology and concepts which are used in this LHC (or HL-LHC) Converter Accuracy Table widely used at CERN. (This page is entirely based on the referred document below) There are several categories of converters in LHC and corresponding accuracy classes were agreed in the early design stages. A summary is given below.

LHC Design Requirements
Converter category Accuracy class 1/2 hour stability
24 hour reproducibility
1 year accuracy
LHC Inner Triplets
1 3 5 50
LHC4-6-8kA-08V* 2 5 10 70
3 10 50 200
4 50 100 1000

Measurements    -    Requirements

*   These converters, by design will show a Temperature coefficient of up to +/-1ppm/°C.
** These converters, by design will show a Temperature coefficient of up to +/-3ppm/°C.

  • References
    • Current measurement transducer for power converters, 2012 (Miguel Cerqueira Bastos, CERN) .pptx
    • High Precision Aspects of DC Power Converters, 2010 (Miguel Cerqueira Bastos, CERN) .pdf
    • Presentation of the LHC power converters calibration results, 2012 (Miguel Cerqueira Bastos, CERN) .edms
    • Current Measurement: Transducers, Analogue Signal Conditioning and ADCs, 2012 (Miguel Cerqueira Bastos, CERN) .indico



    • Smallest increment in current that can be induced or discerned


    • The closeness of agreement between a test result and the accepted reference value. (ISO)


    • Non-negative parameter characterizing the dispersion of the quantity values being attributed to a measurand. (VIM)

    The term accuracy, when applied to a set of test results, involves a combination of Random Components (which affect precision) and Systematic Error or bias component (which affect trueness).
    Sometimes we refer to Absolute Accuracy if the reference value is given by an accepted Standard Value. If the reference value is not a Standard, we often talk about Relative Accuracy.

    Accuracy vs Uncertainty:

    • Accuracy is a qualitative concept ! It cannot be quantified. Although common usage of the word Accuracy for quantitatively describing the characteristics of measuring instruments, is incompatible with its official meaning, in many situations the difference really doesn't matter at all and it remains much more attractive to say 'This instrument is accurate to…' rather than 'This instrument is uncertain by…'

topQuantitative Notions

    Figure 2

    Figure 2

    All measurement results must be accompanied by quantitative statements of uncertainty. The most common approachs to expressing measurement uncertainty are given below.

    Standard Uncertainty:

    • Represent each component of uncertainty that contributes to the uncertainty of the measurement result by an estimated standard deviation.

    Combined Standard Uncertainty:

    • Determine the combined standard uncertainty of the measurement result, by combining the individual standard uncertainties using the usual “root-sum-of-squares” method.

    Expanded Uncertainty:

    • Determine an expanded uncertainty by multiplying the combined uncertainty by a coverage factor, k. The purpose is to provide an interval within which the specific quantity subject to measurement can be asserted to lie with a high level of confidence. For example, for a level of confidence of 95%, k=2.

topAccuracy Characterization

    The term Accuracy is a qualitative concept, used to describe the quality of a measurement. At CERN (and elsewhere) a measurement’s systems capability is often characterized in terms of Gain and Offset errors, Linearity, Repeatibility, Reproducibility and Stability.

    Figure 3

    Figure 3

    Gain and Offset errors:

    • They are systematic errors that relate to the trueness of a measurement.
    • The offset error refers to the systematic error at zero and the gain error to the systematic error at full scale.


    • Difference in the systematic error of a measuring device, throughout its range.


    • Closeness of the agreement between the results of successive measurements of the same measure and carried out under the same conditions. (VIM)


    • Closeness of the agreement between the results of measurements of the same measure and carried out under changed conditions. (VIM)


    • Measurement of the change in a measurement system’s Systematic errors with time. We can more specifically refer to Gain Stability or Offset Stability.
    • Noise can also be seen as a measurement of a device’s stability, although normally the term stability is used only for the low frequency range (≤Hz).
    • When specifying stability, it is therefore important to clearly state the time / frequency ranges we are referring to.