|
RF Emissions Certification for Large Systems:
By Darren J Carpenter BSc (Hons.), MSc EMC Consultancy, BTexact Technologies
Introduction
The European Electromagnetic Compatibility (EMC) Directive (89/336/EEC – EMC) affects BT because of the inclusion of Systems within its scope. This is because BT’s business is fundamentally dependant upon the operation of a large number of diverse switching, transmission and network management Systems. In the provisioning of these Systems, BT operates as a Systems Integrator: bringing together a wide variety of Commercial Off-The-Shelf (COTS) equipment (Apparatus sourced from the marketplace that individually comply with the Directive) to deliver the required functionality. Under the Directive, the Systems Integrator is legally responsible for the compliance of it’s Systems.
In the absence of Harmonised Standards containing technical requirements specific to Systems, a Systems Integrator can only compare its Systems performance with the technical requirements of the Harmonised Standards applicable to the Apparatus from which they are built.
This situation is particularly problematic when applied to the radiated emissions requirements, because these are the primary means of a System causing interference to adjacent radio reception. There is also the more fundamental issue: when a number of Apparatus, each of which is independently compliant with an emissions requirement, are brought together to form a System, what will the cumulative, System-level emissions performance be? Will a composite System meet the same emissions requirement as it’s constituent Apparatus? Does CE + CE = CE?
The System-level radiated emissions therefore introduce a risk of the System undergoing an enforced shutdown and remedial engineering to address an interference complaint. BTexact Technologies has developed a compliance strategy to manage this risk that developed from consideration of the implications of adopting a strategy based upon System-level emissions testing and/or mathematical modelling.
System-Level Emissions Testing
The economics and practicalities of performing System-level radiated emissions testing were investigated. While it was clear that such an activity would demand a physically large test facility (to accommodate the various Systems of interest) with commensurately high capital cost, the investigation indicated that the actual costs to the business of doing System-level emission testing would be far higher. This arose from the combination of the following five factors.
Firstly, there are the logistics costs associated with the transportation to and from the test facility of both the System’s constituent Apparatus and the considerable amount of exercising equipment required to drive it during the test. For testing to be practical, the delivery schedules for all Apparatus and exercising equipment would have to be synchronised: something that is often practically difficult, particularly when working within a multi-vendor supply environment. Hence the ability to perform System-level emissions testing is often limited to very narrow time-windows.
Secondly, there are the configuration costs. These have three components: the initial installation and commissioning of the System and it’s exercising equipment at the test site, prior to testing; the installation and commissioning of additional Apparatus during execution of the test plan (discussed hereafter); and the final decommissioning of the System after testing. Given the complexity of the Systems of interest, each part of the configuration process is a non-trivial task in itself that can each take days to complete and requires the availability of a skilled workforce for many mandays of effort.
Thirdly, there are the costs of performing the actual testing. For Systems, there exists the need to define and execute a test plan that covers all of the various configurations in which the System will be deployed within the network. Typically, the minimum deployed configuration is tested first, then the System is reconfigured to the next deployed configuration and so on. For each configuration, it is necessary to test at several points around its perimeter (since the System’s physical size is likely to rule out the use of a turntable). Hence the need to consider many configurations and typically many positions around each configuration leads to the development of a test plan that can take many days/weeks to systematically execute.
Fourthly, there are the support costs arising from the need to have expert personnel available throughout execution of the test plan to correctly configure the System and confirm correct operation during the test.
Finally, there are the losses associated with tying up both the System’s constituent Apparatus and the associated exercising equipment for the duration of the test. For vendors, the Apparatus is not in the supply chain and generating revenue from sales. Also, if the System contains prototype Apparatus the often considerable venture capital invested in the prototype is tied up in testing.
When estimates of the practicalities, timescales and monies associated with these ‘costs to the business’ were collated, it was clear that System-level emissions testing would be prohibitively expensive. This conclusion occurred at the same time that concerns were expressed regarding the fundamental value of such an activity: what was the confidence that the results obtained would be representative of the emission levels seen among the population of deployed Systems?
The 80-80 Rule
The Harmonised Standard for radiated emissions from Apparatus contains a method by which manufacturersmay acquire statistical confidence in the compliance of their entire production run. The method requires that a number (3 or more) of Apparatus selected at random from a production run are tested. For those frequencies at which the highest emission levels are recorded, compliance of the production run is achieved through satisfaction of the following condition:
X + Sk £ L
Where X is the arithmetic mean of the measured emission levels at a specific frequency; S, is the standard deviation; and k is a defined factor corresponding to the condition that 80% of the population (of manufactured Apparatus) can be expected to fall below the limit level, L, with 80% statistical confidence. The method is therefore referred to as the 80-80 rule.
Thus, for a Systems Integrator to obtain statistical confidence that it’s population of deployed Systems will meet the emissions requirements, using the 80-80 approach several (3 or more) System-level emissions tests would be required, significantly increasing the costs of compliance.
Mathematical Modelling
A review of the available mathematical modelling techniques (such as Method of Moments (MoM), Geometrical Theory of Diffraction (GTD) and Finite Difference Time Domain (FDTD)) and codes (such as the Numerical Electromagnetics Code (NEC)) applied to electromagnetics problems highlighted the fact that most were designed to calculate the electromagnetic field values around a defined conducting structure with defined excitations.
To apply these techniques to the analysis of a System’s radiated emissions, it would be necessary to develop a model that captured the exact geometry of the System and all radiating elements contained therein. To do this, a huge amount of information would need to be amassed and the computer resources required to perform the analysis would be well beyond the capabilities of a commodity PC.
Again, concerns were expressed regarding the fundamental value of such an activity: the results obtained would be specific to the model considered, but how representative would this model be of the population of deployed Systems? Given the tremendous variation that exists between the exact physical details of just one deployment configuration of a System, this question could only be addressed through the development and analysis of many models, something likely to be ultimately as expensive and as time consuming as practical testing.
BTexact Technologies’ Strategy
Neither conventional testing nor conventional modelling therefore offers a means of compliance for Systems that is both cost-effective and generates statistical confidence that can be applied to the entire deployed population.
To address this situation, BTexact Technologies has developed a System-level emissions compliance strategy based upon risk management. This is based upon the exploitation of an innovative, high-level mathematical modelling approach that is supported by on-site measurement. This allows exploitation of the Technical Construction File (TCF) as the route to compliance. The approach developed naturally from consideration of a conventional high-level approach to System-level emissions prediction.
System-Level Emissions Prediction: A Conventional Approach
If it is assumed that the System-level emissions, ESystem, are the absolute worse-case combined field generated by some number, n, of identical Apparatus, each independently radiating some level, EApparatus, at some common frequency, then it is possible to write:
ESystem = EApparatus + 20 log10{n}
To prevent the System-level emissions exceeding the limit, Elimit, it is therefore necessary to restrict the Apparatus emissions to a level below the limit by the 20log10{n} margin, i.e.:
EApparatus = Elimit – 20log10{n}
Adoption of this approach would lead the Systems Integrator to define a new, tighter emissions limit as part of its procurement process. As the number of Apparatus within the System increases, the restricted emissions are unlikely to be met by COTS Apparatus. As a result, the Integrator is likely to have to organise the customised reengineering of an existing product. This will generally cause both significant delay to System deployment and significant increase to the unit costs (a result of reduced economies of scale in manufacturing and the need to recoup redevelopment cost). Fundamentally, this approach may lead to the specification of an emissions requirement that is just not physically possible.
System-Level Emissions Prediction: BTexact Technologies’ Approach
Basing System-level certification upon the absolute worse-case is a fundamentally flawed, since a second question must be considered: how likely is this to occur? Consideration of this question led to the development of an alternative approach to the System-level certification issue that is now described.
The worse-case approach assumes that the individually radiated Apparatus emissions are in phase at the measurement point and hence interfere constructively. So, how likely is this to take place? Consideration of this led to the conclusion that the phase relationships between the individual Apparatus emissions are both unknown and uncontrolled: in principal, any value is possible. Given the a-priori lack of specific knowledge, the individual phase terms are assumed to be random: capable and equally likely to adopt any of the possible values (i.e. between 0 and 2p radians).
An algorithm was developed to capture the radiated fields generated by the individual Apparatus at some point and then generate both the Probability and Cumulative Probability distributions that describe the System-level emissions.
Compliance of the System is demonstrated through use of the Cumulative Probability distribution and the 80-80 rule.
Example
Consider the System-level emissions that result at a distance of 10 m from ten identical integrated Apparatus items that each emits their highest emission level of 25 dBmV/m at 100 MHz. Such Apparatus would be considered as extremely well engineered, complying with the Class A emissions limit of 40 dBmV/m by a margin of 15 dB.
The worse-case emission level generated would be (25 + 20 log10{10}) 45 dBmV/m, i.e. 5 dB in excess of the limit. Hence consideration of the worse-case would lead the Integrator to reengineer or seek alternate Apparatus that emits at or below a level of (40 - 20 log10{10}) 20 dBmV/m.
Figure 1: Example Probability Distribution
Figure 1 however indicates that the System-level emissions are most likely to be within the 30 – 35 dBmV/m range, i.e. at least 5 dB below the limit. Also, Figure 2 indicates that the System-level emissions will meet the limit with 96% confidence.
BTexact Technologies’ approach indicates that while the System-level emissions can exceed the limit (by 5 dB in the worse-case), this is only 4% likely to occur (whereas the 80-80 rule allows at least 20% likelihood of failure), the System-level emissions being instead most likely to comply by a margin of at least 5 dB. BTexact Technologies would therefore claim that this System meets the emissions requirement and would recommend deployment without any remedial engineering whatsoever. Hence the System-level emissions can be shown to comply with the Apparatus emissions limit even when COTS Apparatus is used.
Figure 2: Example Cumulative Probability Distribution
Note that while this example considers a System containing only identical Apparatus, this is simply for the sake of simplicity: the approach is equally applicable to Systems containing non-identical Apparatus.
BTexact Technologies have filed a PCT Patent Application in respect of this strategy, publication number WO00/11483.
Confirmation of Approach
The statistical descriptions of the System-level emissions written into the TCFs are always supported by the results of a limited number of on-site emissions measurements performed with a mobile test facility. To date, the measurements serve to validate the statistical approach: measured levels tend towards the maxima within the predicted probability distributions, not towards the absolute worst-case values.
System Upgrades
This approach can easily be applied to System upgrades: the installation of additional Apparatus into an existing System to expand the existing capacity, the functionality, or both. Where the System has already been analysed using this approach, the effect of upgrade is obtained simply through adding the emissions performance of the new Apparatus to that already installed and repeating the analysis.
Where the System has not been analysed using this approach (where for instance the System was originally deployed prior to adoption of the EMC Directive), a series of emissions measurements would be performed on the constituent common Apparatus types and the results used to analyse the pre-upgrade performance. By adding the emissions performance of the upgrade Apparatus, the post-upgrade performance can be determined and used to determine the overall effect upon the System.
Conclusions
The inclusion of Systems within the scope of the European EMC Directive places a series of mandatory requirements on the very fundamentals of BT’s business. Radiated emissions are of particular concern, since these are the most likely means of a System generating an interference complaint under the Directive.
When BTexact Technologies investigated the options for demonstrating compliance of it’s Systems, applying conventional testing and mathematical modelling approaches at the System-level were noted to be inadequate. Hence BTexact Technologies developed an innovative strategy that exploits the TCF route to compliance and adopts a risk-management approach. The System-level radiated emissions performance is predicted on a statistical basis against which risk of failure is quantified and compared with precedents presented in other Harmonised Standards. The strategy has delivered significant reductions in both the costs and timescales of declaring compliance.
The strategy is of topical interest given the likely inclusion of Fixed Installations within the scope of the revised EMC Directive currently being prepared under the Simple Legislation in the Marketplace (SLIM) initiative: no other approach is perceived as delivering compliance with the combination of low costs and quantified statistical confidence.
Darren J Carpenter BSc (Hons.), MSc EMC Consultancy, BTexact Technologies BT Advanced Communications Technology Centre, Freephone: 0800 169 1689 (UK only) Tel: +44 (0)1473 607080 Email: btexact@bt.com Website: www.btexact.com
|
