The following blog discusses some of the common issues encountered when using the very specialized test systems for EMP and Wideband testing, such as those sold by APELC. A brief outline is provide below.
- Producing an accurate waveform from the source
A. Choosing the correct source
B. Tuning the source
C. Managing ground-bounce and reflections
- Measuring the waveform accurately
A. The right sensor/antenna
B. Reducing/rejecting common-mode noise
1. Generating an accurate waveform from the source
More often than not, when a test-operator is looking to produce an EMP or high-power wideband transient waveform for testing purposes, they are following a MIL-STD. As mentioned in prior blogs:
- Increased Threats Lead to Changing Testing Standard: Understanding the Standards
- Increased Threats Lead to Changing Testing Standards: Subsystem Testing Standards and Solutions
- APELCs Custom High Powered Wideband Test Products
MIL-STDs such as 464C, 461G, and MIL-STD-188-125 define specific Environmental and Electromagnetic Effects (E3) environments by detailing waveforms, spectrum, and field-strengths over a specified area and distance.
a. Choosing the correct source
One common mistake made when looking for a high-powered RF or EMP source is relying on poorly defined nomenclature instead of ensuring adherence to the correct MIL-STD. For example, APELC is often asked if our RFSC-400 High powered RF suitcase can be used for EMP testing. The first question we ask is- what MIL-STD are you trying to meet? If the customer is uncertain, we may need to drill down further. This is because the term Electromagnetic Pulse (EMP) is incredibly general. Simply stated, an EMP can be any pulse/transient event that lies within the electromagnetic spectrum. Unfortunately, pseudo-scientific articles on the internet and movies have done a good job of affixing an EMP device in the public’s mind as a portable device used to disrupt or disable electronics. While the RFSC-400 does in some ways meet that description, it certainly does not disable or destroy electronics over a large metropolitan area, as the movies would have us believe. The only thing that is capable of doing this is a High Altitude Nuclear Electromagnetic Pulse (HEMP), which is discussed in more detail here: Increased EMP Threats and Testing Standards.
As a result, we do have MIL-STD’s that define the HEMP threat environment, allowing us to design pulsed power systems that recreate these environments with ultra-fast (nanoseconds) capacitive discharge circuits coupled to radiating structures. (MIL STD 461) Because these devices are designed to efficiently use the energy produced by focusing it on a nearby or co-located device-under-test (DUT, a.k.a. Equipment-Under-Test or EUT), they can recreate the HEMP threat on a small, localized scale without the need for an actual “bomb”.
While there are MIL-STD’s that address non-nuclear directed energy threats (e.g. MIL-STD-464C, wideband and narrowband HPM), we more often than not find that when a customer is referring to EMP, that they are actually wanting a system that recreates the very specific HEMP E1 waveform.
Some of these MIL-STDs provide tolerance ranges with the understanding that the pulsed power behind these sources utilize physical electronics, such as spark gaps, that yield uncertainty such as amplitude and temporal drift. APELC attempts to minimize this uncertainty by providing sources with wide operating ranges and highly repeatable performance characteristics (see this white paper for an example).
b. Tuning the source
By starting with a fast Marx generator, APELC is able to keep the pulse-conditioning (a.k.a. “peaking”) on our HEMP systems fairly simple. A single, air-insulated spark gap is most often used to produce the fast rise-time double exponential pulse, and as a result, most of the adjustment over the desired amplitude range (e.g. 10% to 100%) can be done by simply varying pressure. APELC has attempted to simplify this process for our customer by using custom software that allows the user to enter a desired field-strength, and have the software automatically select the correct pressure and voltage required to generate the specified waveform. An example of this is shown below:
Figure 1 APELC Control and Acquisition software for RS-105 testing
Similarly, APELC has developed software for control of our high-powered wideband systems that controls the pressure and voltage. For these systems, the waveform characteristics are set by the physical dimensions of the antenna and resonator. Therefore, a specific antenna/resonator combination is selected to meet the desired bandwidth as defined in the MIL-STD (e.g. MIL-STD-464C Table A.5).
Once the correct source is selected, the test range must be clearly defined.
c. Managing ground bounce and reflections
While MIL-STD-461G RS-105 attempts to mitigate reflections by requiring the operator to, “Keep the top plate of the radiation system at least 2 times h from the closest metallic ground, including ceiling, building structural beams, metallic air ducts, shielded room walls, and so forth”, other standards are less specific and the test operator must use their own judgement. A good rule-of-thumb is to know that a reflection cannot interfere with the pulse in the test volume, and to base the distance of any reflecting boundaries off of the velocity of propagation for the pulse in air- i.e. 1 ns/foot. Therefore, when operating a MIL-STD-464C wideband system with a pulse-width of ~20ns, we would want to ensure that any reflecting boundaries are much greater than 10 feet (two-way transit time of 20ns) away from the test volume. Similarly, we would want to set the source high enough off the ground, that any reflections from the ground plane are either angled away from the test-volume, or occur much later in time. This can be estimated by using ray-tracing principles similar to those used in optics in combination with the velocity of propagation estimate mentioned previously.
2. Measuring the waveform accurately
Once the correct source has been selected and the test range is clearly defined, diagnostics must be carefully selected and arranged to make sure that what is measured is actually what is present in the test volume. One of the biggest mistakes a test-operator can make is attempting to adjust their source before vetting their diagnostics. Many a tail has been chased by starting to turn knobs on the source, when all that is wrong is a bad attenuator or poorly placed cable.
a. Choosing the right sensor or antenna
When defining the signal chain for diagnostics, it is best to start from the source (sensor) and work back to the scope. Because we must avoid reflections on our diagnostic cabling, we select a 50 Ohm scope input. This also ensures a good match with attenuators and filters upstream of the scope input. Because we are limited by the dynamic range of the oscilloscope at 50 Ohms (1V/div), we must carefully select an antenna or sensor that provides the maximum signal without saturating the input to the scope or optical link (more on that later). Moreover, the sensor must operate within the bandwidth of the signal being measured and not saturate itself in the presence of high electric fields. One of the few diagnostics capable of meeting these criteria is the Asymptotic Conical Dipole (ACD), such as the Prodyn D-Dot free-field probe.
b. Reducing/Rejecting common-mode noise
Common-mode noise, or common-mode voltage, is the noise that is present on both inputs of an amplifier. This is typically the result of a ground in the signal chain (shield on a cable) having an impedance that is significant enough to develop a voltage other than the desired signal appearing between the shield and center conductor of the cable. Here are some steps APELC takes to lower noise on the signal chain:
- Get rid of RG-58 in your signal chain! The poor shielding quality on RG-58 results in considerable common-mode noise that can create massive offsets in integration and completely erroneous signals. 50-Ohm coaxial cables with high shielding effectiveness ratings such as RG-214, RG-400, and RG-223 are good choices.
- Ensure all cables are perpendicular to the electric field. Cables that are parallel to the electric field act as great receiving antennas and can become major noise sources.
- Place ferrites/chokes along the cabling. By selecting a ferrite that has a low-impedance for the frequency of the noise you are trying to eliminate, the noise on the shield can be significantly reduced.
- Get rid of long cable runs by using optical links. This point cannot be overstated- long cable runs (especially those electrically longer than the pulse you are measuring) are always going to be a noise source. Using analog optical links, or shielded, remote sampling units with digital optical links will always be the best option for getting rid of noise. You still need to make sure noise is not coupling onto the small length of coax between the antenna and the fiber-optic transmitter, but replacing copper with glass is ALWAYS a good idea for eliminating noise.
- Put you oscilloscope in a screened enclosure! This almost goes without saying, but the oscilloscope and/or optical link receiver can still couple noise directly into their chassis, either through apertures or external power and signal cables. If you don’t shield your scope, there is no point in doing any of the above.
Because antennas and probes by nature differentiate the signal they are measuring, an integration must be performed to resolve the actual signal. Prodyn Technologies does a great job of educating their customers on the math behind this, so I’ll simply link to that here instead of taking up more room on the page.
There are two main types of integrators used for fast transients: passive integrators, and numerical integration. Passive integrators are simple R-C circuits that use analog hardware to perform the integration. While these are simple to build, they can be prone to noise and do require droop correction in post-processing. These do have their place, but most often, APELC prefers the use of numerical integration of the waveform once it has been acquired. This can be done directly on the oscilloscope, or in post processing using Excel, Matlab, MathCAD, Python, or any other programming language capable of importing the data from the oscilloscope and performing a numerical integration. With all numerical integration, care must be taken to remove any offset present in the resultant waveform. Because a digital integration is essentially a summation of points, any noise or offset present on the incoming signal is summed and results in a monotonically increasing, diagonal offset in the integrated waveform. One method of removing this offset on the oscilloscope (specifically Tektronix oscilloscopes) is to subtract the mean of the signal prior to integration. In a math waveform editor, this would look like: INTG(CH1-mean(CH1)).
We hope this blog provides some guidance on sources and testing in this very specialized field, but if you need further assistance, we are always here to help! Contact us any time at firstname.lastname@example.org.