We have had the same experience with the variety of our analogue and
digital meters.
One solution can be found within the IEC 60990:99 (which is referred to
for touch-current measurements e.g. by the newest versions of the IEC 60950
and IEC 60065) and which defines measurement methods for DC and AC
sinusoidal or non-sinusoidal waveforms from 15Hz upwards.
The interesting part is Annex L, which is about performance and calibration
of your measurement network.
As the required frequency range of the leakage meter is usually 15Hz to
1MHz (e.g. IEC 60950, Annex D), this Annex L of the IEC 60990 lists tables
with input and transfer impedances for a selection of frequencies that make
it rather easy to verify the adequacy of your equipment.
When using a True-RMS meter for the measurement of switching type power
supplies that are operating at frequencies in the range of 100kHz it must
also be ensured that the meter is capable to cover that.
Many so-called True RMS meters on the market reach up to only 5kHz, crest
factors of 2 or 3.
/Randolf
The IEC 60990
Rich Nute
<[email protected] To: [email protected]
.com> cc: [email protected]
Sent by: Subject: Re: Leakage Current
Measurements
owner-emc-pst
[email protected]
04.10.00
06:41
Please
respond to
Rich Nute
Hi Ken:
> I found something interesting today. While measuring leakage current
with a
> digital multimeter, I noticed a high amount of leakage. The unit
> incorporates a switching power supply, with some EMI/EMC circuitry.
Anyhow,
> I read somewhere that DMMs are some times inaccurate with high
frequency
> line conducted emissions. Took a old Simpson analog meter (took
forever to
> figure it out, since I've never touched a analog meter hehehe) and
found
> measurements that were rather low. The questions is, I'm wondering if
UL or
> for that matter, any other testing agency actually know about this
fact, or
> merely fail their clients with incorrect data.
*
The following is Copyrighted 2000 Hewlett-Packard
Company.
*
When measuring leakage current, and, if the current
waveform is non-sinusoidal, then the measurement
MUST be made with a true rms meter (whether analog
or digital). (The Simpson leakage current meters
are true rms.)
If a voltmeter is not true rms, it measures the peak
of the waveform, and then reads 0.707 of the number.
So, for any non-sinusoidal waveform, the voltmeter
reading will generally be higher than the rms value,
and will not represent the rms value of the current.
UL and most other testing houses do understand this
requirement, and ALWAYS use true rms meters.
For switching-mode power supplies, the leakage
current waveform can be sinusoidal or it can be non-
sinusoidal. The waveshape is dependent on how the
EMC filtering is done. The non-sinusoidal portion
of the waveform is the leakage from the primary
circuits ON THE LOAD SIDE of the rectifier. And,
conversely, the sinusoidal portion of the waveform
is leakage from the SUPPLY SIDE of the rectifier.
I believe that the non-sinusoidal leakage current
waveform does not exceed the bandwidth of most
voltmeters. The frequencies contained in the non-
linear leakage current waveform do not have
significant current magnitudes above the 40th
harmonic. If we consider all frequencies up to the
40th harmonic of 60 Hz, we have 2400 Hz. Most ac
voltmeters have a bandwidth of at least 10 kHz,
well above 2400 Hz.
The body network is likewise not particularly
significant to the leakage current measurement.
This is because leakage current is derived from a
source that approaches a current source. For the
sake of this discussion, a current source is a
source that provides a constant current regardless
of load.
Consider that the source resistance for 0.5 mA
leakage current is 120 volts divided by 0.5 mA, or
240 kohms. Adding 1.5 kohms (the body impedance
model) to 240 kohms gives 0.4969 mA. So, any
reading without the body impedance network is a
very, very slightly pessimistic reading.
Likewise, the 1.5 kohm resistance and 0.15 uF
capacitance have a pole at about 1 kHz. So, the
capacitor has no effect for frequencies below 1 kHz.
So, if the non-linear current waveform has lots of
harmonics, then the network will give a lower
reading for those frequencies above 1 kHz. Again,
any reading without the body impedance network is a
very, very slightly pessimistic reading.
My advice: re-equip your lab with only true rms
meters.
Best regards,
Rich
-------------------------------------------
This message is from the IEEE EMC Society Product Safety
Technical Committee emc-pstc discussion list.
To cancel your subscription, send mail to:
[email protected]
with the single line:
unsubscribe emc-pstc
For help, send mail to the list administrators:
Jim Bacher: [email protected]
Michael Garretson: [email protected]
For policy questions, send mail to:
Richard Nute: [email protected]
-------------------------------------------
This message is from the IEEE EMC Society Product Safety
Technical Committee emc-pstc discussion list.
To cancel your subscription, send mail to:
[email protected]
with the single line:
unsubscribe emc-pstc
For help, send mail to the list administrators:
Jim Bacher: [email protected]
Michael Garretson: [email protected]
For policy questions, send mail to:
Richard Nute: [email protected]
