TEM method, issued for USA SAE J1752/3
For the measurement of parasitic emission radiated by the integrated circuit, the most popular method is based on a TEM cell (i.e Transverse Electromagnetic assumption), consisting of a shield box and a window for inserting the board 10x10cm.
The chip under test is fixed in the inner side of the board, and radiates inside the chamber, not outside.
A metal plate, inside the box, captures the chip emission, which is converted and plotted in the frequency domain by the spectrum analyzer.
The frequency limit of this method is around 1GHz, due to the physical dimensions of the chamber.
An example of test board designed for TEM cell radiation measurement is shown in this slide. The inner side of the board only contains the device under test surrounded by a ground plane. The outer side gathers all switches, supply and connectors for setting up the chip under test.
Once placed in the Tem cell, the DUT is completely isolated inside the chamber, while all connections are provide out the cell.
The TEM cell is "Adapted 50 Ω"
The TEM cell is adapted 50 ohm. This means that at any location, the geometry of the septum relative to the shielding gives an inductance and capacitance value per unit length such as the square root of the ratio (L/C) is 50ohm. The cable is also adapted 50 ohm. A 50 ohm resistance is placed on the far end of the cell. The analyzer has a 50ohm input resistance by construction.
Measure in TEM cell
One problem with the TEM cell is the influence of the emitted spectrum with the chip orientation within the chamber. In these measurements, orientation A gives a spectrum 20dB under the customer limit, which sounds like a good news. Unfortunately, when rotating the board 90 degrees (position B), the spectrum is very close to the customer limit, and even goes above near 100MHz. This is the reason why the TEM standard specifies that the 4 orientations should be considered, and the maximum spectrum should be kept for qualification.
The TEM cell frequency limitation
The TEM cell has a frequency limitation due to its physical characteristics. The measurement of the attenuation factor between a sinusoidal input Vin and the output Vout shows that a 0dB gain is observed below 1 GHz, but that losses appear in some particular frequencies above 1 GHz. This due to stationary waves inside the TEM chamber, corresponding to its physical dimensions (30 cm corresponds to λ at 1 GHz). This means that the TEM cell would significantly attenuate some energy rays generated by the IC, that would not appear accurately on the spectrum analyzer.
The TEM cell at work
The video shows a 1 GHz emission of an IC and the emergence of stationary waves (permanent color) on some part of the chamber.
GTEM cell: up to 18 GHz
One very important measurement method for parasitic emission above 1 GHz in the GTEM cell (Giga Hertz TEM). Several sizes of GTEM cells are proposed, from IC size (10 x 10 cm aperture) up to complete system characterization. The GTEM cell enables measurements up to 18 GHz. Again the physical dimension of the shielding and the inner septum are 50 ohm adapted.
GTEM cell
The GTEM cell is 50 ohm adapted. The inner plate, also called septum, is situated inside the shielded room. At the far end of the tem cell, a 50 ohm termination is distributed near the absorbent material. The absorbent material prevents from wave reflection.
German Std VDE UK 767.14 -> IEC 61967-4 International Standard
Concerning conducted mode emission, the most common method is based on a 1 ohm resistance added in serial to the ground supply. The voltage sent to the spectrum analyzer is the image of the current flowing on the ground pins of the integrated circuit.
The method is issued from the German standardization group VDE. The method is limited to 1GHz, although measurements may be performed at higher frequencies. The problem is the inductance behavior of the 1 ohm resistance above 1GHz.
Use of the 1ohm conducted emission method
An example of conducted emission in the range 1-1000 MHz shows a huge amount of harmonics, with a peak at 72 dBµV around 32 MHz. The noise floor is around 22dB.
Magnetic Probe SAE J1752/2(Japan) -> IEC 61967-3 International standard
An other method radically different from the 1ohm conducted probe is the magnetic loop. This loop is connected directly to the spectrum analyzer. It enables rapid investigations of high peaks of radiated emission over the device under test.
Magnetic Probe (IEC 61967-3)
An exploitation example of peak emission is given in this slide. We define Amax as the maximum peak appearing in the instant spectrum. At each X, Y location, the Amax [X, Y] is extracted, and plotted according to a palette of colors. This method give valuable information about the location of high magnetic fields, specifically over the IC die area and the package leads. Some scanning systems have spatial resolution as small as 1µm.
Summary
This slide summarizes the most common emission measurement methods. The internal standardization has been completed in 2001 for TEM, 1ohm. The process for standardization is undergoing for the GTEM cell.
| EMC for ICs > Measurement Methods > Measurement of Emission |  |