Leakage current measurement with non-frequency-weighted device

[maxruben]

Hi,

I hava a question about the measuring device, MD according to figure 12, when measuring with a non-frequency-weighted device.

Clause 8.7.3 Allowable values in IEC 60601-1 ed. 3.1 paragraph e) "Additionally, regardless of waveform and frequency, no leakage current shall exceed 10 mA r.m.s. in normal condition or in single fault condition when measured with a non-frequency-weighted device."

I am using a Fluke ESA612, electrical safety analyzer, to measure leakage currents wich includes the MD according to figure 12. However this has no setting for measuring non-frequency-weighted currents so I have to do this with another instrument. I assume that the MD that I need to use together with a true r.m.s. measuring DVM then only includes the resistor R2, 1000 ohm, without R1 and C1.

Am I assuming correct?

 

[MedMartin]

Hi,

Correct, you can measure the voltage on R2 only.
We measure during development using the resistor R2 and monitor the waveform with an oscilloscope and check the measurement using the current measurement function of a calibrated true-RMS multimeter.

Best regards,
Martin

 

[3xute]

On the same subject and paragraph 8.7.3 e) do the words "no LEAKAGE CURRENT" apply to all leakage tests under 8.7.3? LEAKAGE CURRENT includes EARTH LEAKAGE CURRENT, TOUCH CURRENT, and PATIENT LEAKAGE CURRENT by definition. Therefore in addition to applicable measurements made with the MD from Figure 12, should all of the same measurements be made a second time with the non-frequency weighted device? If I understand correctly EARTH LEAKAGE CURRENT (8.7.4.5), TOUCH CURRENT (8.7.4.6), and PATIENT LEAKAGE CURRENTS (8.7.4.7 a, 8.7.4.7 b, 8.7.4.7 c, 8.7.4.7 d, 8.7.4.7 h) should be performed twice, once with the MD from figure 12 and a second time with a non-frequency-weighted device. Am I interpreting that correctly?

 

[Benjamin Weber]

Yes, you are correct. All leakage currents should be done twice - with and without frequency weighting. Only if you can show, that there are no higher frequency components >1kHz (e.g. by use of an oscilloscope), you can skip the unweighted measurement.

 

[Benjamin Weber]

Hi,

Correct, you can measure the voltage on R2 only.
We measure during development using the resistor R2 and monitor the waveform with an oscilloscope and check the measurement using the current measurement function of a calibrated true-RMS multimeter.

Best regards,
Martin

The standard says you should measure the voltage drop over the MD and converting them to the corresponding currents by dividing the voltage with 1kOhm. This is acutally current measurement with a quite high shunt resisitance.

I am not sure if measuring the currents directly with the current measurement function will give equivalent results.

 

[Benjamin Weber]

Hi,

I hava a question about the measuring device, MD according to figure 12, when measuring with a non-frequency-weighted device.

Clause 8.7.3 Allowable values in IEC 60601-1 ed. 3.1 paragraph e) "Additionally, regardless of waveform and frequency, no leakage current shall exceed 10 mA r.m.s. in normal condition or in single fault condition when measured with a non-frequency-weighted device."

I am using a Fluke ESA612, electrical safety analyzer, to measure leakage currents wich includes the MD according to figure 12. However this has no setting for measuring non-frequency-weighted currents so I have to do this with another instrument. I assume that the MD that I need to use together with a true r.m.s. measuring DVM then only includes the resistor R2, 1000 ohm, without R1 and C1.

Am I assuming correct?

That is why test labs usually use custom-made equipment for this measurement.

 

[3xute]

Thanks for your confirmation on this Benjamin, that was how I interpreted the standard as well. My test lab will be using custom-made equipment for the measurement.

 

[Peter Selvey]

There are a few technical side points here: measurements above say 10kHz are unlikely to be practically accurate nor legally traceable, even if using an oscilloscope, even if using an accredited test lab. There are no off the shelf systems, including DMMs, scopes that are OK up to 1MHz, and even the parts used to create the filter are unlikely to be stable up to 1MHz, e.g. stray capacitance, capacitor ESR, inductance and other high frequency effects such as leakage paths through the equipment to ground, common mode effects (which are a nightmare at higher frequencies). With some care measurements up to 100kHz could be traceable, but above 100k is not realistic. Fortunately, it's rare that higher frequency measurements come anywhere near the limits, so it's not a critical issue.

I personally prefer a while box approach where the source of the leakage is understood, including higher frequency components. For example if the ECG leads also have a 60kHz ac signal for measuring respiration, the test lab (guided by the designers) should be familiar with the circuit that creates this signal, the expected currents in normal and single fault condition. Blindly measuring leakage with a custom filter and Fluke 87 DMM isn't going to cut it, and potentially overlooks the single fault analysis.

Leakage associated with switching power supplies often have frequency components above mains frequency, but in my experience they are never anywhere near the limits. For example, you might measure 0.12mA earth leakage with the filter, and 0.14mA without the filter. Even if that second measurement is wildly inaccurate, it's nowhere near 10mA.

 

[3xute]

Thanks for your response and time on this Peter. It is fortunate that higher frequency measurements rarely come near the limits. I appreciate your comments on tracing the source of the leakage to make sure it is understood and make any necessary modifications. It seems like the non-frequency-weighted measurements are rarely vastly different from the regular measurement, but still always good to check.

 

[Benjamin Weber]

There are a few technical side points here: measurements above say 10kHz are unlikely to be practically accurate nor legally traceable, even if using an oscilloscope, even if using an accredited test lab. There are no off the shelf systems, including DMMs, scopes that are OK up to 1MHz, and even the parts used to create the filter are unlikely to be stable up to 1MHz, e.g. stray capacitance, capacitor ESR, inductance and other high frequency effects such as leakage paths through the equipment to ground, common mode effects (which are a nightmare at higher frequencies). With some care measurements up to 100kHz could be traceable, but above 100k is not realistic. Fortunately, it's rare that higher frequency measurements come anywhere near the limits, so it's not a critical issue.

I personally prefer a while box approach where the source of the leakage is understood, including higher frequency components. For example if the ECG leads also have a 60kHz ac signal for measuring respiration, the test lab (guided by the designers) should be familiar with the circuit that creates this signal, the expected currents in normal and single fault condition. Blindly measuring leakage with a custom filter and Fluke 87 DMM isn't going to cut it, and potentially overlooks the single fault analysis.

Leakage associated with switching power supplies often have frequency components above mains frequency, but in my experience they are never anywhere near the limits. For example, you might measure 0.12mA earth leakage with the filter, and 0.14mA without the filter. Even if that second measurement is wildly inaccurate, it's nowhere near 10mA.

Peter I have to disagree at some point with you ;-)

Frequency response of the measuring network at >10kHz can be an issue. But if you take care during the development of the network, you can achieve quite good results. So we did not just take "any" resistors or capacitors. First we chose parts with very good nominal tolerances (10 times better than recommended in fig. 12, IEC 60601-1). Then we purchased a certain number (10 to 20) for each part and selected the best fitting manually. Additionally we did not take just one single part per component. We actually used multiple single resistors/capacitors to achieve the required values of R1, R2 and C1 - let's say 4 x 2.5 kOhms in series for R1. By clever selection of each single part, you can actually level out the remaning deviations from the nominal values. This leads to very good results also at higher frequencies during calibration at an accredited callibration lab (<5% overall deviation at 1 MHz).

Regarding the actual voltage measuring device: You are correct, standard DMMs are usually limited to 100 kHz. But oscilloscopes have much higher bandwiths. We use a fully isolated/floating scope with 60 MHz bandwith. We can assume, that the frequency response up to 1 MHz is linear and not an issue. But AC measurement uncertainty is bit of a problem. All scopes are specifed by the bandwith (-3 dB deviation) and mostly by the DC gain accuracy. But you will not find a value for AC accuracy. So the conservative approach is to use the -3 dB (appr. -30%) also for low frequencies. But you can try to perform AC callibration for the whole measurement chain (network + scope) at discrete freqnuencies, e.g. 50Hz, 60Hz, 100Hz, 500Hz, 1kHz, 500kHz and 1MHz.

Regarding those aspects it is possible to have a good measurement setup also for higher frequencies.