Read other posts in the “Elephant in the Test Room” series here.
As promised in the last post, here is the start of the next elephant!
Elephant #2 – Disharmony in Harmonic Limits
The Room: RF immunity testing
The Elephant: The curbing of the contribution made by harmonics to a calibrated test-field varies wildly from standard to standard, and within standards
The Culprit: Harmonic limits that are seemingly not well thought through, and / or are open to interpretation, and / or pay no regard to how other standards bodies determine harmonic limits
The Consequence: A customer’s product sent to two different test houses for the same RF immunity test can be subjected to test fields of very different harmonic content
Introduction
As a test house, the amount of notice you take of harmonic noise depends on the type of RF immunity test being conducted. Commercial testing says the portion of the test field created by harmonics should not exceed 25%. Automotive says the harmonic levels from the amplifier must be minus 20dBc or better. Military (last time I looked) is not specific on either of these type of limits.
Let’s start with the automotive 20dBc limit, as I can see no RF engineering rationale behind this seemingly arbitrary and overcautious limit (20dBc means only 1% or less can be harmonic power), and even if you could justify the limit, and I don‘t see how you can, a coach and horses can be driven through the way the minus 20dBc is specified.
Example: It is a given that an amplifier tends to have to work hardest at the lowest frequency point in its band. This is because the antenna it is driving generally has lower gain with lower frequency. Yet, almost without exception, amplifier harmonic levels are worse at the start of the amplifier band.
So let’s say a test house puts a significant dollar investment into an amplifier rated at 1kW, and (not too carefully looked at at the time of purchase) the data sheet states the harmonics are minus 20dBc at a power level of 700W. Should it turn out that the amplifier needs to produce 800W to get the required field strength, does the stipulated 20dBc limit still apply?
To be continued…….
The Magic Number 30
During a reply to a blog reader, the question was posed “Where does the magic number ‘30’ in field strength E = sqrt(30P.G)/d come from?”
Well done those who worked it out for themselves. In order of clicking the ‘publish‘ button, they are 1. Bruce Curry 2. Jim Pollock and 3. Hans Joubert
Here is how, long ago, I worked it out: –
E/H = Zo so H = E/Zo
Power density S = ExH
But H = E/Zo
So S = ExE/Zo
That is power density S = E^2/Zo
(by the way, intuition tells us that since this is watts per square meter, this has to be ERMS, but again, let’s just stick with finding where the magic number 30 comes from)
Power density is also S = P.G/ 4Pi.r^2 (shown in the picture where a point source of power emanates a wavefront spherically, followed by one from an antenna emanating in a particular direction with linear gain over isotropic G. Note: The arrows are for illustration only and simply show the direction of the emanating wavefront through the square meter drawn for each case)
Given that S = E^2/Zo and S = P.G/4Pi.r^2
Then E^2/Zo = P.G/4Pi.r^2
But Zo = 120Pi
Therefore E^2/120Pi = P.G/4Pi.r^2
E^2 = 120Pi.P.G/4Pi.r^2
E^2 = 30P.G/r^2 (ping! – out pops the magic number)
E = sqrt (30P.G)/r
Incidentally. Somewhere around 2008, while investigating testing to 6GHz for the new 61000-4-3, in some instances the calculation was amazingly close to actual measurement for bore-sight field strength (the strength right in the middle of the measurement plane). OK, it needed cladding on the floor around the antenna and between the antenna and the measurement plane, and the antenna was pointing to one corner of the room, but impressive all the same.
The Linearization of EMC Amplifiers
Linearizing TWT Amplifiers
The first approach we could try is simple cancellation of the high power harmonic by adding a phase inverted harmonic frequency at the input to the amplifier. We inject the anti-phase signal at the input and monitor the level of the output harmonic while adjusting the level/phase of the injected signal (ultimately we will get the PC running the software to do this). The concept diagram below shows the key parts of this idea. We will be forced to add many more components later such as circulators, line stretchers etc., and we will replace the second signal generator with a frequency doubler, but let’s stick at the concept level for now (so please let’s ignore easily fixed foibles.
I recommend we press a long-suffering intern* into adjusting the level and phase of the injected signal for now, and then once he/she has got the hang of it, write what they actually do in pseudo code in readiness for writing instructions for the dumbest intern of all, the PC running the EMC software.
For now we want the intern to familiarize him/herself with the working of the system. We ask them to get the system up and running with the upper signal generator set for say 50W of fundamental (at 1GHz) output power from the amplifier. He/she should be able to see the fundamental and the harmonic at 2GHz on the spectrum analyzer as shown. Now the second signal is introduced at say at 1.8GHz so it can be seen on the screen, and the level is adjusted to something similar to the level of the harmonic. The frequency is then changed to 2GHz so it coincides with the harmonic, and this increases or reduces the harmonic level depending on the relative phase. The intern now plays with the phase and fine tunes the level of the signal until the harmonic level on the screen is down by 10dB.
I can see at least one juggernaut that is going to hit us hard later, but let’s see how far we get first.
* No interns were harmed in the writing of this post
To be continued ……
The Cellphone Threat
As mentioned in the last posting, I am of the view that those involved in EMC testing should be familiar with the most prevalent radio frequency interferer out there – the cellphone. A cellphone in close proximity to a victim is the threat we should be concerned about, not the signals from the base station, even though at their source these are of far higher transmit power. Nonetheless, to get a grasp of the working of the 3G and 4G air interfaces as presently in use, we really need to cover base station transmissions. Once covered, we can then look at what is transmitted by the real threat, the cellphone.
In this particular post we will start with one fundamental principle, key to all. That is orthogonality, and how it applies to both 3G and 4G air interfaces. Once we have this down, we will start to fill in the details (frequency bands, transmitted waveform signature, etc.).
Orthogonality
Simply put, orthogonal signals allow users to share a composite transmitted band, and yet the signals can still be extracted/separated from each other at the other end. Both 3G and 4G rely on orthogonality. With 3G’s WCDMA, the codes used to modulate data are orthogonal to each other. With 4G’s OFDMA, the channel frequencies are orthogonal to each other.
3G’s WCDMA Codes
Each user’s data is modulated by a special code allocated to the user and the user alone. The special code runs at high speed, ‘chipping’ the data as shown in panel [a] below. At the receive end, the self-same allocated code is used to extract this one user’s data [b]. Panel [c] shows how another user’s allocated code does not extract the data. This is because all the special codes are orthogonal to one another.
4G’s OFDMA Pulsed RF Spectrums
In regular frequency division schemes, channels are separated by frequency and guard bands are placed between the channels. OFDMA takes advantage of the fact that a repetitive RF pulse in the time domain produces a sin(x)/x shaped spectrum in the frequency domain as shown below (Panel [d]). Note the frequency domain characteristic central peak with well-defined nulls either side. Unlike regular frequency division, the sin(x)/x spectrums of adjacent channels can be placed right next to each other with the peak of one corresponding with the null of the other (i.e., orthogonal to one another), see [e] and [f]. This allows very tight packing of the channels without cross-channel interference, and without the need for guard bands.
Orthogonality concepts will be continued ………
Unintended Acceleration
No doubt you have noticed that the subject of unintended acceleration has reared its head again as reported recently in the Interference Technology News section.
I have often wondered about possible reasons for unintended acceleration and my mind tends to go back to early days and discrete digital electronic design, where often, to save putting in another multiple JK flip-flop IC, it was common to use spare NAND gates on a hex NAND gate chip to create a flip-flop. In those days medium scale integration was in it‘s infancy, and there was a phenomenon called ’latch-up’, where for no apparent reason a flip-flop (this applied to both NAND gate based and the JK flip-flop ICs) would latch-up in the ‘set’ state, and only removing the dc power from the chip could release it from this state. It has crossed my mind that if the most significant bit of the flip-flop bank that is the memory address holding the accelerator position, was to latch-up, then even when the disturbance that caused the latch-up went away, the engine management system would continue to provide maximum acceleration as seemingly requested by the accelerator memory address.
And my musing got me to wondering about how the current EMC standards check potential RF interferers attached to the wiring harness before the new car design is released. I recently wrote on this in the guise of a mock design review, where the existing test set-up is put through the rigors of a real design review as required of RF systems destined for the commercial marketplace. In the real world of RF product design there are mandatory reviews written into the design cycle, where the first review is often of the internal design specification written by the design team itself. As well as ensuring designed inputs and outputs interface with other parts of the system being designed by others, this also ensures the design team knows the particular performance requirements that the proposed design will be measured against. The article will appear in the 2014 Interference Technology Design Guide.
-Tom Mullineaux
