New waveshape definitions for 1.2/50-8/20 and 10/700

14 posts / 0 new
Last post
admin
Offline
Joined: 6 Jan 2011 - 19:02
New waveshape definitions for 1.2/50-8/20 and 10/700

I have been meaning to raise concerns about the forthcoming IEC 61000-4-5 Ed.3. The mismatch differences between IEC 61000-4-5 Ed.2 and other standards have been covered in Chapter 8 of Impulse Generator tutorial. With Edition 3 those differences become more major as the impulse waveshape measurement has changed to pulse measurement waveshape with 10% to 90 % rise and 50 % to 50 % duration. The 1.2/50-8/20 becomes 0.72 (10 %-90 %)/50 (50 %-50 %)-6.4 (10 %-90 %)/17 (50 %-50 %). As the IEEE PES SPDC were the originators of the 1.2/50-8/20 generator one feels they are corrupting our baby.

Read all about it in Elementary and Ideal Equivalent Circuit Model of the 1,2/50 – 8/20 μs Combination Wave Generator by Carlo F. M. Carobbi, Member, IEEE, and Alessio Bonci, IEEE Electromagnetic Compatibility Magazine – Volume 2 – Quarter 4, 2013

 

Edited by: admin on 17 Apr 2014 - 11:13
admin
Offline
Joined: 6 Jan 2011 - 19:02
The dominance of linear EMC thinking in SC 77B

Ed. 3 of IEC 61000-4-5 is pervaded by linear suppression EMC thinking, with scant appreciation for non-linear limiting surge thinking. There are two ways of mitigating surges; linear suppression and non-linear limiting. The Figure below, from Introduction to Surge Mitigation Techniquesshows the two approaches:

Ed. 3 has an informative Annex on the mathematical modelling of surge waveforms. The Annex purpose apparently is to help surge generator design and aid surge simulations on digital equipment. The four wave shapes analysed are; 1.2/50, 8/20, 10/700 and 5/320. The IEEE Std C62.45 is quoted as the source for the 1.2/50-8/20. As pointed out in the Impulse Generator tutorial this isn't a good idea as the C62.45 "8/20" is defined not to have an undershoot so should be inherently of short in duration. C62.45 also tries to comprehend the effects of the AC backfilter on the waveform and quotes 20 µs +8 µs -4 µs for the duration, which could be written as 22 µs ±6 µs. The IEC offers the 8/20 current in two duration values; 20 µs ±2 µs or 20 µs ±4 µs depending on undershoot, but these values are for a current generator, not a 1.2/50-8/20 generator. IEC 61000-4-5 has a 1.2/50-8/20 generator 8/20 duration of 20 µs ±4 µs - but defines it in a variety of ways depending on where you look.

Having got the outcomes of two open-circuit voltage and two short-circuit current frequency spectrums, what do you do with them? You can define the scope frequency/sampling requirements, but this could have been predicted from the waveform front times. Connecting the generator to a circuit will impose a load that will change the frequency spectrums. If that load varies with frequency then this further modifies the frequency spectrum. This is without even considering any non-linear element that might be present. The Annex doesn't contain any information of how the frequency spectra might be used for the purposes given. One feels this Annex is similar to the results of an academic exercise imposed on a pupil provided with Excel, a Fast Fourier Transform (FFT) and the specified waveshapes.

 

   

 

admin
Offline
Joined: 6 Jan 2011 - 19:02
The goal posts move for the duration

This entry looks at how the definition of impulse duration or time to half value measurement changes between Ed 2 and Ed 3 of IEC 61000-4-5. Those of you who read the post #1 Elementary and Ideal Equivalent Circuit Model of the 1,2/50 – 8/20 μs Combination Wave Generator reference will likely be up to speed already.

1.2/50 voltage waveshape

Front time

On this parameter Ed. 2 and Ed 3 agree on the term "Front time", where to measure (at 30 % and 90 %) and how to quote 1.67 x 30 % to 90 % time. Thus nothing changes with front time being 1.2 µs ±0.36 µs.

Time to half value (Ed. 2) or Duration (Ed. 3)

Here's the disconnect. Between Ed. 2  and Ed. 3 the term is different, Time to half value (Ed. 2) becomes Duration (Ed. 3), the former is measured from the the virtual origin (resulting from the extrapolated 30 % and 90 % line) to the 50 % point of the decaying waveshape and the latter is the time the waveshape exceeds 50 %. Was it really necessary to change both the established term and measurement?

The change is likely to prompt the question "what is the difference between the Time to half value and Duration times?" From the diagram below it is obvious that Time to half value is greater than Duration. If one assumed a linear voltage rise, then the time to reach 50 % would be 0.6 µs making nominal values; Time to half value is 50.6 µs for a Duration of 50 µs or Time to half value is 50 µs for a Duration of 49.4 µs — a 1.2 % difference.

8/20 current waveshape

Front time

On this parameter Ed. 2 and Ed 3 agree on the term "Front time", where to measure (at 10 % and 90 %) and how to quote 1.25 x 10 % to 90 % time. Thus nothing changes with front time being 8.0 µs ±1.6 µs.

Time to half value (Ed. 2) or Duration (Ed. 3)

Here's the big disconnect. Between Ed. 2  and Ed. 3 the term is different, Time to half value (Ed. 2) becomes Duration (Ed. 3), the former is measured from the the virtual origin (resulting from the extrapolated 10 % and 90 % line) to the 50 % point of the decaying waveshape and the latter is the time the waveshape exceeds 50 % multiplied by a bodge factor of 1.18. Was it really necessary to change both the established term and measurement?

The change is likely to prompt the question "what is the difference between the Time to half value and Duration times?" From the diagram below it is obvious that Time to half value is much greater than time the waveshape exceeds 50 %. Hence the appearance of a scaling factor claimed to be derived from empirical data. Empirical means "derived from experiment and observation rather than theory" yet the document doesn't provide any further information as to where the data was gathered from. Engineering suspicious are raised when people don't provide informative supporting materials for assertions. One can't even apply a simplistic evaluation, as in the case of the voltage wave, because the ratio of the waveshape designation values (8/20) is only 2.5.

 

 

 

admin
Offline
Joined: 6 Jan 2011 - 19:02
Grandfathering

Grandfathering is a North American expression meaning to exempt pre-existing conditions from new requirements, legislation, or restrictions. One wonders why this technique wasn't used in Ed.3.

The voltage and current impulse front times Ed. 2 and Ed. 3 are the same, either 30 % and 90 % or 10 % and 90 % measurements are taken. Inherently these measurements allow the determination of the virtual origin. 

Ed. 2 time to half value required the measurement of the 50 % point in the decay time period. Ed. 3 requires measurement of the 50 % points on the rising and decaying impulse edges. 

All the required values are there or can be derived to give both time to half value and duration results. If the Ed. 3 was truly equivalent to the Ed. 2 value, why not give the option to use either the old (Ed. 2) way or the new (Ed. 3) way? Another reason to not to like the Duration approach is that there are a bunch of other surge waveforms used in surge resistibility testing and Ed. 3 duration only specifically deals with only two surge generator types; 1.2/50-8/20 and single output 10/700. Put another way, time to half value is almost universal and Duration isn't.

Ed. 3 stipulates that the generator must include a series, either internal or external, capacitor of 18 µF for voltage and current calibration measurements. Guys, who are purely dealing with communication lines, now have to include the 18 µF for calibration even though the equipment testing does use an 18 µF capacitor! This seems an unnecessary imposition which could have been overcome by text like "If the generator is to be used for testing a.c./d.c services then open-circuit voltage and short-circuit current waveshapes shall be measured with a series capacitor of 18 µF. However, I note that any guidance on the 18 µF capacitor tolerance, voltage rating and parasitic values seems to be missing. 

admin
Offline
Joined: 6 Jan 2011 - 19:02
Empirical duration factor

To have a "one size fits all" 1.18 multiplication factor for the 20 µs duration seems fortuitous. The referenced Elementary and Ideal Equivalent Circuit Model of the 1,2/50 – 8/20 μs Combination Wave Generator article came up with a factor 1.19, but the analysed circuit produced too much undershoot.

Intrigued, I dug out some old and rather crude calculations on a simple CLR 1.2/50-8/20 generator circuit that produced the following plot of short-circuit current waveshapes for various values of damping factor k.

The maximum allowed undershoot is 0.3, making the k range of interest 0.35 and above. The old calculations already had front times and times to half values, mean only the time to 50 % was the only additional parameter to be extracted. Having done that the ratio of time to half value to the time above 50 % could be plotted against k. This gave the following plot:

 

The ratio plot shows a variation from about 1.19 to 1.11. In practise, the ratio range is smaller as k values below 0.35 are not allowed and timing limits of the front time (8 µs ±1.6 µs) and time to half value (20 µs ±4 µs). Imposing these limits gives the allowable k ranges shown below.

The vertical heights of the front time boxes have no significance and are arbitrarily chosen to allow range differentiation of the front time limit and nominal values.

These crude calculations show there is reason to examine the "one size fits all" 1.18 multiplication factor in greater detail.

 

 

admin
Offline
Joined: 6 Jan 2011 - 19:02
US 1.2/50-8/20 generator duration multiplier

The following is the analysis of the Excel calibration file of a well known and widely used US produced combination wave generator. The "8/20" waveform was normalised to unity amplitude and input to a curve fitting program. In the past I have used programs like curvefit (freeware) and Tablecurve 2D. The neat thing about Tablecurve 2D is the evaluation after curve fitting. Setting amplitude evaluation points of 0.1, 0.5 and 0.9 gives all the information needed to calculate the generator current waveshape parameters. The analysis results were:

Front time: 7.99 µs
Time to half value: 22.8 µs
0.5 to 0.5 time: 19.7 µs
Time to half value/0.5 to 0.5 time: 1.16

 

For this generator the multiplication factor is 1.16 not 1.18. From the previous materials, a front time of about 8 µs and a time to half value of 22.8 µs indicates a k value of 0.55 and an adjustment factor of 1.159. However, both 1.16 and 1.18 are not necessarily good values to use generally, because post #5 shows that, depending on the front time, the possible multiplication factor can vary from 1.19 through to 1.11. The wrong value multiplier could either underestimate or overestimate the effective time to half value calculated from the 0.5 to 0.5 time.  

admin
Offline
Joined: 6 Jan 2011 - 19:02
10/700 - 3 wrongs don’t make a right—Part 1

The 10/700 generator is circuit defined. When the ITU-T published the 10/700 circuit, only nominal component values were given. This is a pity, because had the component tolerances had been given, it would have defined the waveshape tolerances rather than leaving it to other organisations to set what they thought the waveshape tolerances should be.

IEC 61000-4-5 Ed.2 came up with 10 µs ±3 µs/700 µs ±140 µs voltage and 5 µs ±1 µs/320 µs ±64 µs current.

IEC 61000-4-5 Ed.3 has 10 µs ±3 µs/700 µs ±140 µs voltage and 5 µs ±1 µs/320 µs ±64 µs current, using the new decay time currency.

The people putting together TIA-968-B seem to have studied things more intently and came up with 9 µs ±2.7 µs/720 µs ±144 µs voltage and 5 µs ±1.5 µs/320 µs ±64 µs current.

Logically the TIA-968 has the same ±30 % tolerance on the voltage and current front time. IEC 61000-4-5 has ±30 % tolerance on the voltage front time and ±20 % tolerance on the current front time. The current front time circuit uses one extra component (R3) yet the tolerance tightens up from 30 % to 20 %, implying R3 is trimmed to bring the tolerance in from ±30 % to ±20 %. Hardly the way in which one would naturally design such a circuit, indicating that IEC 61000-4-5 applied standard waveshape practice, rather than tolerancing the given generator circuit.

admin
Offline
Joined: 6 Jan 2011 - 19:02
10/700 - 3 wrongs don’t make a right—Part 2

Many of the ITU-T test circuits use the dual output 10/700 generator version for testing twisted pair lines.

For the 10/700 common-mode surge test the two conductors were driven by the generator R3 and R4 outputs. Differential mode surge testing is interesting, because it is often assumed the differential surges are caused by asymmetrical surge protector operation or joint flashover. To simulate this condition one of the two outputs could be shorted either by a link or a dynamic short provided by a suitable switching element such as a GDT. The unshorted generator output would be applied the tested conductor and the untested conductor returned to the generator common connection.

In practice this doesn’t happen. One generator output is taken to the tested conductor; the other generator output is left open and the untested conductor is returned to the generator common connection.

This arrangement makes the differential-mode surge is more severe than the common-mode surge on the tested conductor, which doesn’t make sense. A debate in the ITU-T ended by the group leader admitting shorting one output was the correct way of testing differential-mode resistibility, but to change from doing it the wrong way would confuse people.

In the US the TIA-968 formulation group understood the common-mode/differential-mode test problem and came up with an alternative approach. For differential-mode testing they, like the ITU-T, had one generator output is taken to the tested conductor; the other generator output was left open and the untested conductor was returned to the generator common connection. The difference was that the generator voltage was reduced to 2/3 of the common-mode test voltage. A TIA-968 1.5 kV common-mode test was accompanied by a 1 kV differential-mode test. This reduction in differential-mode test voltage resulted in a similar stress to the situation where the generator unused output was shorted.

admin
Offline
Joined: 6 Jan 2011 - 19:02
10/700 - 3 wrongs don’t make a right—Part 3

Many people use ITU-T Recommendation K.44:Resistibility tests for telecommunication equipment exposed to overvoltages and overcurrents - Basic Recommendation and IEC 61000-4-5:Electromagnetic compatibility (EMC) - Part 4-5: Testing and measurement techniques - Surge immunity test interchangeably as references for the 10/700 generator.

However, the two “10/700” generator circuits for testing twisted pair lines are different!

The difference is the ITU-T uses 25 Ω resistors for R3 and R4 and the IEC uses 50 Ω resistors for R3 and R4. Obviously the ITU-T generator is capable of nearly 50 % more current than the IEC generator. How did this come about?

It seems that the IEC decided to use a single wire, hence single output 10/700 generator, as the reference for the twisted pair case and to maintain the same output current waveshape as the single wire case doubled the output series resistance to 50 Ω. In talks with people who work in this group it was stated magnetic induction is considered to be happening and hence the sum of the AT (ampere-turns) in the wires must be constant. This justifies increasing the series resistance values by a factor set by the number of surged wires – two wires surged double the resistance.

In the linear world you can get away with this arrangement, but in the non-linear world the resistive feed arrangement doesn’t mimic reality. Consider 100 AT of induced current in a twisted pair, if the system switching overvoltage protector is asymmetric and the element on one wire turns on that element will conduct 100 A of current. Had a balanced protector been use, such as a single chamber 3-electrode GDT, then both elements would conduct (one initiating the other), each taking 50 A.

For increasing numbers of twisted pairs the ITU-T adds further output 25 Ω resistors to feed additional wires. As mentioned earlier, the IEC series resistance values are related to the number of surged wires and so progressively stifles off the available wire current as the number of wires increase (up to a certain limit). This introduces the possibility of dangerously under testing the protection.

The ideal solution here would be to magnetically couple the surge such as is done by an injection transformer for RTCA DO-160G aircraft lightning testing. A rough and ready test would be to common-mode surge all the wires connected in parallel, verify the protection elements survive and the equipment is still functional.

admin
Offline
Joined: 6 Jan 2011 - 19:02
Opportunity for improvement

The "Know your Standards" section of the April 2014 The EMC Journal comments on the third edition of IEC 61000-4-5. One comment, related to this thread, was there are three different definitions for "duration". I guess this refers to the normal way from the virtual origin, the open-circuit voltage 50 % to 50 % measurement and the "questionably adjusted" 8/20 50 % to 50 % current measurement. The entry ends with the phrase used to title this post "There seems to be an opportunity for improvement". Those familiar with British understatement humour will recognise the sentence as restrained in ironic contrast as to what might have been written.

admin
Offline
Joined: 6 Jan 2011 - 19:02
What 1.2/50-8/20 Ed 2 generators will fail to meet Ed.3?

A little bit more work on post #5 established the critical areas where a generator (just) meeting Ed.2 fails Ed. 3. The 8/20 waveshape requirement is for the time to half value/Duration to be 20 µs ±4 µs (16 µs to 24 µs range).

The critical correlation area is when the front time is 8.4 µs or less and the Edition 2 time to half value is over 22.5 µs. Under these conditions the Ed. 3 measurement technique can give a value over 24 µs.

The table below shows:

Column 1 - front time, column 2 which Edition values are for.

Row 3, column 3 onwards circuit damping factor value. Rows 4 to end, columns 3 to end Time to half value (Ed.2) or Duration (Ed.3)

Orange highlighlighting is Ed.3 values that exceed 24 µs, whereas the Ed.2 value does not.

 

 

Pages