Like most long articles, this started off as a short one. It all stemmed from a customer question:
“We had some issues using a LED driver that could not cope with load dump and volt spikes. Do you have any provisional tests that could determine the circuit reliability? It doesn’t have to be to [ISO 7637-2]”
The ISO 7637-2 standard defines automotive conducted transient test pulses on vehicle power lines (12V or 24V). It is called up by standards including:
- UNECE Regulation 10.06 for E-marking
- EN 50498 (aftermarket automotive equipment)
- ISO 13766-1 (earth-moving and building construction machinery)
I don’t have an ISO 7637-2 pulse generator. Automotive surge generators are less commonly found in many EMC test labs due to their more specialised nature.
Systems are available to hire; budget for €/£1000/week for a generator that will cover Pulses 1, 2 and 3. They are also available to buy new; expect to pay around €/£15k. If you need to cover pulse 4 then this will increase the costs yet again, mostly for the bipolar amplifier.
But, like most EMC test labs, I do have an IEC 61000-4-4 (EFT) and IEC 61000-4-5 (Surge) generator capable of 1.2/50us and 10/700us pulses.
Question: Could I use the IEC generator to simulate the surge pulses from the ISO generator?
This question comes with caveats:
- The aim here is pre-compliance / confidence testing with the tools available. Not to replace the ISO 7637-2 tests entirely.
- We are only looking at the potentially destructive Pulses 1, 2a, 3a and 3b.
Unit 3 Compliance can perform pre-compliance and full CE Marking testing to EN 50498. We can also perform pre-compliance testing for many of the R10 tests for E marked products.
Please get in touch for a chat if this is of interest.
ISO Pulse 1
- IEC 10/700 pulse generator can be used as a close substitution for a 12V system
- For a 24V system the 10/700 pulse is not as good a match. Follow the flowchart to select the test compromise and set the surge voltage based on the values in the tables.
ISO Pulse 2a
- Not a good match, recommend a compromise between current and energy as shown in these tables
ISO Pulse 3a, 3b
- IEC EFT generator is a good match and can be substituted for ISO pulse 3a and 3b
Pulse Parameter Comparison
Comparing the pulse widths and impedances against each other gives a mixed picture.
For Pulse 1, neither waveform is a great match with both of the ISO pulses having a longer pulse width than the 10/700 generator. Whilst the 24V bus pulse has a much higher impedance, this could be corrected with an additional series resistor in the IEC generator output.
For Pulse 2a, the 1.2/50us IEC generator appears to be an excellent match.
For Pulse 3a and 3b, the 5/50ns EFT generator is pretty close but the width of the ISO pulse is three times bigger.
However, as we shall see below, this approach is incorrect as it does not tell the whole story.
Pulse Width Definition
The problem comes from how the pulse widths are defined in the standards. Let’s take the comparison between ISO Pulse 1 to IEC 10/700 comparison as an example.
e can see that the ISO pulse width is defined at the 10% crossing point, whereas the IEC pulse width is defined at the 50% crossing point.
This is not helpful.
How do we compare a ISO 1000us @ 10% with a IEC 700us @ 50% waveform?
Open Circuit Ideal Waveform Comparison
I found some information over on the PSCAD website that showed the equation for the waveshape (from IEC 61000-4-5)…
…along with some Matlab optimised coefficients for alpha, beta and k.
From the PSCAD website “Standard Surge Waveforms” https://www.pscad.com/webhelp/Master_Library_Models/CSMF/Surge_Generators/Wavelet_Transformation_(WT).htm
ISO 7637-2:2011 gives the equation for the falling edge only of the pulse waveform. It also states that “The influence of the rise time is not taken into account (tr << td), which is allowed for all pulses specified in this part of ISO 7637”
After watching a Numberphile video on coronavirus infection curve modelling I decided to give Geogebra a try for modelling these waveforms. It’s quite a useful graphing calculator package, much more powerful than I’ll ever need to use.
I also modified the equation for the IEC waveshape equation to take into account the generator and load impedances by taking the first term of the ISO equation and adding it to the start of the IEC equation.
A required surge voltage of 1V was used for simple direct comparison.
Review of Waveform Comparisons
For Pulse 1 we can see that the 10/700 IEC waveform is actually a really good match for Pulse 1 for a 12V bus.
The same cannot be said for the 24V bus requirement. Some further thinking is required here.
The 55 ohm impedance for the 24V version of the pulse is the 15 ohm 10/700 generator natural impedance with a series 40R resistor in addition.
Despite Pulse 2 looking like a good comparison initially, the modelling shows that it is actually a very poor match.
For Pulse 3, the IEC EFT generator is a very good match and should be able to be used without any issue
Dealing With Pulse 1 (24V) and Pulse 2a
How could we go about compensating for the poor match between Pulse 1 (24V) and 10/700 IEC and between Pulse 2a and 1.2/50 IEC?
We need to ask ourselves: are we more interested in the peak voltage & current or pulse energy?
To answer this, first we need to understand the power input design of the Equipment Under Test (EUT)
EUT Design Assessment
It is useful to establish the following EUT design parameters:
- Is there a discrete reverse protection diode? What is the Vrrm and Trr rating (reverse recovery time) of this part?
- What is the maximum clamping voltage of the TVS diode and can the downstream circuitry survive this voltage?
It is important to remember that Pulse 1 is a negative going pulse caused by the disconnection of a large inductive load in parallel on the vehicle power bus. If the EUT has a reverse protection diode fitted then it’s Vrrm and Trr will change the effect of the test on the EUT.
W2AEW has a good video on diode reverse recovery time over on YouTube.
It is also important to test at full current consumption if a reverse recovery diode is present as this will affect recovery time and therefore surge performance.
EUT Surge Suppression
The assumption is that we are testing an EUT that contains some basic low voltage electronics of some kind. The extension of this assumption is that it has some kind of surge suppression component connected across the power inputs.
This could be a Metal Oxide Varistor (MOV) or a Transient Voltage Suppression Device (TVS). These have a non-linear impedance with voltage and will restrict or “clamp” the input voltage to a defined level. Perhaps a component like a SMBJ26CA-TR.
This clamping voltage is dictated by the impedance of the part when conducting. This would be a diode-like VI curve for a TVS or the current-dependant resistor of a MOV.
Peak current is dictated by available peak voltage and generator impedance. So we need to be interested in the peak current to ensure that the correct clamping voltage is met.
Also, because the MOV or TVS absorbs some of the pulse energy internally, these components will have a datasheet rating for pulse energy. Exceeding this could cause significant damage to the part and affect its capability to handle future surges.
Pulse 1 Peak Voltage & Current or Pulse Energy?
Our main tools for adjusting an IEC pulse to suit an ISO pulse are:
- Peak voltage
- Series impedance
The surge generator has an easily adjustable peak voltage through the control panel or software so this is the main method that will be used.
The Peak voltage is a significant consideration if the system has the reverse protection diode but the compromise test will depend on it’s voltage rating.
I’ve produced a flowchart to help selection of the right test level for using IEC 10/700 instead of ISO Pulse 1
Pulse 1 Best Compromise Voltage
I ran some more simulations in Geogebra adjusting the ratio between the IEC and ISO peak voltages and tabulated the results.
The best compromise is to minimise the total difference between current and voltage when expressed as ratios. This works out at a V_iec or around 0.6 * V_iso.
This yields the following test voltages, peak currents and pulse energies for the different severity levels.
It is interesting that the series impedance for the 24V version of ISO Pulse 1 is up at 50 ohms. This higher impedance implies that the surge expected in such a system would be induced from a parallel adjacent cable in a wiring loom rather than something directly connected to the ignition switch / inductive load circuit directly.
Pulse 2a Best Compromise Voltage
Same approach as for Pulse 1
Test Practicalities & Further Compromises
Pulse 1 Power Disconnection
The waveform for Pulse 1 shows a synchronised disconnection from the DC supply and application of the surge voltage. Since this is not easily done without
It is the surge pulse that will cause the damage rather than the momentary disconnection of voltage therefore, for these compromise tests, this is being ignored.
Coupling/Decoupling Network Requirements
The CDN inside the IEC test generator for mains coupling is adequate for the task of decoupling but the options inside my KeyTek ECAT test generator preclude the coupling of the 10/700 waveform. Instead, some creative front panel wiring with banana plugs will be required.
Since this CDN is designed for decoupling of surge and EFT impulses from the mains, I’m sure it will adequately protect the 12V linear power supply being used and also prevent the power connection from unduly affecting the test.
In may case, input is through a 16A IEC mains plug/socket but it is easy to make an adaptor. Output is via a BS1363 socket or, more convieniently, 4mm banana plugs.
This took way longer to research and write that I was hoping. Something in the order of three days of work was spent going backwards and forwards, thinking about it whilst doing DIY at home (nearly painting the cat as a result) and half listening to Tiger King on the TV.
I’m quite pleased with the result and I hope this eventually proves useful to someone.