High Frequency Measurements
Web Page
Douglas C. Smith
Address: P. O. Box 1457, Los Gatos,
CA 95031
TEL:
800-323-3956/408-356-4186
FAX:
408-358-3799
Mobile: 408-858-4528
URL:
www.dsmith.org
Email: doug@dsmith.org
Technical Tidbit - July 2003
Measuring E-Field Coupled IC Chip Noise
Figure 1. Chip E Field Noise Measurement
Abstract: Measuring noise coupled onto a copper foil patch applied
to a chip package can yield useful data. A method of measurement is suggested
and data presented. The data from such a measurement can be useful for design troubleshooting as well as EMC work.
Uses related to reliable system operation and as well as conclusions for EMC compliance are presented.
Discussion: Figure 1 above shows a copper tape patch applied to a
chip package, A 0.8 pF low capacitance active probe (Agilent 1158a)
is shown measuring the signal voltage on the patch. Such an arrangement can yield useful
information about the chip operation and its effect on nearby structures such as heat sinks.
Notice that there is no ground lead attached to the probe. The input
capacitance of the probe is low enough that the capacitance of the probe
body and the hand holding it forms the return path back to the board at high
frequencies. At lower frequencies, tens to hundreds of kHz, the scope
probe shield through the scope chassis and earth ground forms the return
(if the circuit has an earth ground connection). Thus, the ground reference
of the probe for frequencies in the megahertz range and above is an average
of the circuit board ground. The measurement is one of an average potential
from the chip surface and leads to an average ground potential on the board.
The measurement, to a first order approximation, is made through the three
part capacitive divider of the capacitance of the chip and its bonding wires
to the copper tape, the probe input capacitance, and the capacitance from
the probe back to the board. If a probe ground lead is used, its impedance becomes the third leg of the divider.
Ideally, the size of the copper patch should match the size of the chip,
possibly covering bonding wires also. Too large a patch will act to slow
the observed risetimes or lower the signal amplitude for fast risetime signals.
This is due to the added capacitance between the patch and the board as well
as the free space capacitance of the patch. This effect is described in the
November 1999 Technical Tidbit, Transient Suppression Plane.
Figure 2 shows a measurement made in just such a way. The signal is
almost one volt peak-to-peak! This does not in itself mean there is a problem.
Each step on the waveform means an area of the chip was raised in potential
at that time, such as an address bus going from zeros to ones. The greater
the area of the chip changing levels, the greater the step observed on the
waveform.
One possible use for this display is to determine if chip operation
has changed in a logical way as compared to another state. Another use could
be to trigger a closer look at the chip and its surrounding circuits on the
board. A change in the waveform appearance could indicate a significant change
in noise or EMC margins. For example, if an older chip is replaced by one
implemented in a newer technology, a close look at the internal device risetimes
may be in order. A significantly faster risetime on internal signals, as
indicated by this measurement, should be an indication that a closer inspection
of EMC and logic margins in the circuit is in order.
Consider the case of a heat sink attached to the chip package instead of copper tape. Something like the open circuit
voltage from this measurement is coupled onto the heat sink resulting in a current trying to return
to the circuit board. Such a current is at the root of many EMC compliance
problems.
Figure 2. Chip E Field Noise Waveform
(50 ns/div)
A handy use of this measurement technique is to determine internal
risetimes on the chip as mentioned earlier.
Figure 3 shows a similar measurement, but at 10 ns/div instead of 50
ns/div. One can see that the internal voltage risetimes on the chip are on the order
2 to 5 ns. These times are relatively slow for this example because
of the vintage of the chip being measured. Most modern chips will have risetimes much faster than this example.
Figure 3. Chip E Field Noise Waveform
(10 ns/div)
Figure 4 shows the use of a ground reference for the probe. The probe ground
lead is touched to the board ground near the chip. Using such a probe ground
reduces the loop area of the measurement considerably and can significantly
reduce noise pickup into the measurement from other sources, on and off the
board.
Figure 4. Chip E Field Noise Measurement Using Ground Lead
The use of a differential probe makes such a measurement even more accurate.
Figure 5 shows a passive differential probe making the same measurement. Configurations
such as Figure 4 and 5 can give an accurate representation of the open circuit
voltage induced on a heat sink. It is not difficult to come up with an estimated
source impedance for this voltage. The capacitance between the chip and copper
patch or heat sink can result in an impedance at hundreds of megahertz and
higher on the order of a few hundred Ohms or less. This can result in currents
on the order of milliamperes flowing through the heat sink and back into
the board whereas many EMC problems can be caused by tens of microamperes
of current. It is easy to see why heat sink grounding is important.
Figure 5. Chip E Field Noise Measurement Using Passive Differential Probe
Summary and Conclusion: A measurement technique was presented that
can yield useful information about chip operation, including logical operation,
well as potential EMC problems. In particular, the importance of heat
sink grounding is shown.
Equipment used in this article included:
Top of page
Home
Questions or suggestions? Contact me at doug@dsmith.org
Copyright © 2003 Douglas C. Smith
