Read other posts in the “Elephant in the Test Room” series here.
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
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
Continuing…
Reminder – commercial RF immunity testing says no more than 25% of the test field can be created by harmonics, automotive stipulates minus 20dBc harmonics from the amplifier, and military has nothing set in stone.
So on what grounds did the automotive industry determine the 1% harmonic limit? For instance why is 1% harmonics acceptable yet 2% is not?
I see the need for limiting the proportion of the test field created by harmonics, but 1% seems absurdly tiny. Maybe the limit setting got caught up in the ‘unintended acceleration’ maelstrom and the industry wanted to be seen as taking strong action regarding car safety, and as a result overcompensated for other weaknesses in the suite of automotive tests.
To my mind, all the 20dBc limit did was cost test houses a small fortune and create a mini-bonanza for amplifier sellers.
So what harmonic limit would be required if automotive had gone for the commercial limit?
Easy enough to calculate. Let’s go for a 1GHz test with the harmonic at 2GHz. So the 200v/m total test field would be made from 150v/m at 1GHz and 50v/m at 2GHz (25% of total or minus 6dB)).
We assume we are using an antenna with 10dBi (reasonable) at 1GHz and 16dBi (excessively cruel but let’s go with it) at 2GHz.
Let the power required to generate the 150v/m with antenna linear gain G1 = 10 be P1, and let the power required to generate 50v/m with antenna linear gain G2 = 40 be P2. The powers are the power required at the antenna connector.
That is a difference of 4dB compared to the automotive limit. This may not look much on paper but it means the automotive harmonics have to be way less than half those of an amplifier used in commercial testing. Put alternatively, you can use an amplifier with 16dBc harmonics for commercial testing, but need one with 20dBc for automotive.
To be continued…….
The Linearization of EMC Amplifiers
Training the Intern
If we are going to use an intern for mundane tasks, it is only fair that in exchange we provide them with knowledge, combined with hands-on practice in the use of that knowledge.
We start by providing knowledge on the characteristics of RF power amplifiers likely to be met in the course of the intern’s compliance testing career.
Here we go–the input-output characteristic of power amplifiers is monotonic. Monotonic is a flash way of saying that as the input power is increased, the output increases too, and the situation never arises where an increase in input power results in a decrease in output power. All amplifiers exhibit gain compression as the maximum output power is approached, and ultimately the amplifier reaches saturation, where an increase in input power produces no discernible increase in output power. However the output power does not drop, it just no longer increases with increased input power.
Sounds like we are stating the obvious, but TWT amplifiers are non-monotonic. As the input power is increased the gain compresses as maximum power is approached, then the amplifier saturates, then as the input power is increased further, the output power drops.
So what is the big deal? Well, here is the big deal. We want to use cancellation to reduce the harmonic power of a TWT amplifier (see picture). We are using an intern to apply active pre-distortion via the second sig gen level and the phase changer, and if unaware of this characteristic, the intern could find themselves in a declining spiral.
To be fully cognizant, the intern needs to know one more thing about power amplifiers. Again it seems obvious, but we will spell it out for them all the same, and in this case the snippet of knowledge applies to solid-state and TWT amplifiers.
It refers to the most you can expect from an amplifier. Let’s say you have an amplifier that at its best can output 450 watts. So you can reasonably expect the amplifier to produce 450W at one spot frequency, 225W (each) at two frequencies, 150W (each) for three frequencies, etc. What you cannot expect the amplifier to do is produce two frequencies at 450W. That would be crazy, the total power cannot exceed 450W.
So when the intern is adjusting the phase and amplitude of the cancelling signal, and during the circumstances where to get 200v/m the amplifier is running at or close to full power, the intern should anticipate weird and wonderful things happening to the fundamental signal power as the cancelling signal is aligned with the harmonic.
It all sounds very frightening, but we will get the intern to fully characterize the amplifier by recording its behavior under different power situations. They can then refer to these ‘look-up’ tables when conducting actual testing. That of course, is what the PC will to do when it takes over this mundane task.
To be explained further when we continue ……
The Cellphone Threat
In this post, we dig a little deeper into the 3G and 4G air interfaces.
More on 3G Orthogonal Codes
Reminder – Each user is allocated a special code that ‘chips’ the user’s data, and this self-same code is used at the receive end to recover the data. Cyclical, and pseudo random in nature (so noise like when viewed on a spectrum analyzer), any two of the special codes have low cross-correlation. That is they have little similarity. Happily for us this is digital electronics, and there is a logic gate that provides the level of similarity between any two codes – the EXNOR gate. The output of an EXNOR gate is only true when the bits of the two codes presented to it are the same.
To make progress, we just need to do one more thing, and then we will be all set to use the EXNOR gate to determine level of similarity. For logic 1 we will allocate +1 volts (could be +5V or +3.3V or whatever), and for logic 0 we will allocate minus the logic 1, that is -1 volts (again, could be whatever).
The new truth table:
Now we are all set to use the EXNOR gate to determine the level of similarity between any two of the special chipping codes.
As a preliminary test, we will compare the similarity of a code with itself (should be all TRUEs, right?), and then we will compare two special codes. I will stop here to allow you to take a shot at this yourself. Here are two sample codes:
01101001
00110101
Can you establish the average voltage when auto (=self) cross-correlated? And then when the two codes are cross-correlated?
More On the 4G Air Interface
A key feature of 4G OFDMA is the splitting of very fast serial data into many slow parallel channels (see picture), and then reversing the process at the receive end. Seems like a lot of messing about, but each slow parallel channel will have a superior bit error rate (from the simple fact that the data rate is slower) and moreover there is scope to add guard periods and so forth, making the bit error rate better still. The parallel channels are separated by frequency and spread across a wide band, with the result that 4G is less prone to multi-path fading (a serious issue with 3G).
To be continued…
-Tom Mullineaux
