FAQ for Passive IM Products

Through interaction with engineers and technicians in many countries throughout the world, Summitek Instruments has had the opportunity to discuss and evaluate measurement techniques as they apply to a wide variety of items. We have discussed the problem of IM with component manufacturers, infrastructure providers, site managers and service providers. This list of Frequently Asked Questions and their answers is an attempt to consolidate the information we've gathered into one convenient location.

1. What is Passive Intermodulation (PIM)?
2. What are the Typical Causes of PIM?
3. What is Meant by "IM3" and "IM5"?
4. What is a "Good" PIM Level?
5. Does PIM Vary with Power Level?
6. Does PIM Vary with Frequency?
7. Can I use a DCS-1800 Band PIM Test System to Measure Cables for Use in the 900 MHz Cellular Band?
8. How is PIM Specified?
9. My PIM Test System has a Noise Floor of -140 dBm. Can I measure a -140 dBm PIM Response?
10. Does PIM Change with Time?
11. Is There a Difference in Forward and Reverse (Reflected) IM Levels?
12. Is There a Difference in IM with Carrier Frequency Separation?
13. Should by RF Sources be "Phase Locked" to "Align" the Carriers?
14. What is the Difference Between a 16-Tone and a 2-Tone PIM Test?
15. Why Would I Care About PIM?
16. Are There Measurement Standards for Passive IM?
17. Other Questions?

1. What is Passive Intermodulation (PIM)?

Passive IM, similar to Active IM but occurring in passive devices, is present whenever RF signals at two or more frequencies are simultaneously present in a conductor of RF energy. Every passive RF device generates passive IM products when more than one frequency is present in the device. The signals are mixed  by the non-linear properties of junctions between dissimilar materials. Typically, it is the odd-ordered products (e.g. IM3=2*F1-F2) that can be very problematic should they fall within an uplink, or receive band of the base station because they appear to the receiver as interference. The result can be a receiver desensitization which is independent of the receiver's random noise floor.

Back to Top

2. What are the Typical Causes of PIM?

In RF components (antennas, cables, filters, etc.), there are typically three causes:

1. Poor mechanical junctions in the RF path
2. RF components fabricated with materials which exhibit some level of hysteresis (e.g., stainless steel)
3. Contaminated surfaces or contacts within the RF path. Examples might include flux (which can attract other contaminants) and metalic particles from the machining process.

 

In integrated base stations, significant levels of passive IM can be generated within any of the passive components between the high power amplifiers and the receiver filter. Passive IM can also be generated on the tower ("rusty bolt noise") or by nearby metallic objects in the direct beam of the transmit antenna.

Back to Top

3. What is Meant by "IM3" and "IM5"?

This annotation is commonly used to specify the order of the IM product being discussed. The IM stands for "intermodulation." The numeric value that follows is the sum of the integer multipliers used for each of the two parent tones to realize the given IM product. This is best understood by reviewing the following table:

Most commonly, the lower order tones are of the largest magnitude. However, in frequency-selective systems, it is possible that an IM5 product might actually appear larger to the receiver than an IM3 product.

Back to Top

4. What is a "Good" PIM Level?

The required PIM performance for a given RF device is a strong function of where that device is located in the final system. For example, an antenna must have excellent PIM performance as the PIM generated in the antenna is both received and radiated by the base station. Further, the transmit antenna is subjected to nearly the full carrier power of the base station. On the other hand, the PIM performance of a receive "clean-up" filter need not be so stringent. This filter might be located on the other side of a diplexer thus preventing the full carrier power level from reaching its input connector.

Ultimately, it is up to the buyer to specify the maximum acceptable PIM level and carrier power levels. Commonly seen specifications for antennas are -100 to -110 dBm IM3 levels with two, +43 dBm (20 Watt) per carrier tones.

Back to Top

5. Does PIM Vary with Power Level?

Yes. However, the relationship between the generated PIM power level and the parent carrier power levels is not always straightforward.

In simple, broadband devices terminated into a broadband termination, the IM3 response typically increases approximately 3 dB for every one dB in carrier power level (assuming equal carrier powers). However, there are many factors which tend to work against this nice, simple relationship. These include:

• High return loss values at n*F₁ and/or m*F₂
• Extreme slope variations on the hysteresis curves associated with ferrite devices
• Non-Linear behavior of electromechanical junctions as they approach a breakdown potential
• The interaction of multiple IM sources as the impedance of each IM source changes with incident power level




In general, as the transmitter power increases, the importance of PIM on the overall system performance becomes of increasing concern. As a TDMA system fills available frequency and time channel slots, or as a CDMA system increases forward power levels to increase capacity, PIM levels typically increase.

Back to Top

6. Does PIM Vary with Frequency?

It depends. One postulate is that a single IM source which is located at a single point (not spatially distributed) and is matched in impedance to the incident transmission line (or source of stimulus RF energy) generates frequency-independent IM isotropically. This is the analog of the classical "Point Source" of RF in antenna theory.

Given that this point source of PIM exists (at least theoretically), real-world RF devices can be modeled as being comprised of multiple PIM sources. These sources generate IM which has a phase relationship with the parent RF carriers. Once the PIM is generated at each point source within the device, the PIM signals themselves can vectorially combine (either constructively or destructively) to produce a composite PIM response. The phase relationship between the PIM sources will depend upon their physical separation, the dielectric through which the RF must travel between the sources, and the frequency of the parent carriers.

Given that all real world devices have more than one source of PIM, it is quite probable that the device will have a frequency-dependent PIM response. However,

• if the device is electrically small, or
• if the bandwidth of interest is relatively small compared to the device under test, or
• if the device is dominated by a single, large IM source the measured frequency response may appear frequency independent.

Back to Top

7. Can I use a DCS-1800 Band PIM Test System to Measure Cables for Use in the 900 MHz Cellular Band?

Even though electrically long (more than one-half wavelength) cables can have a frequency-dependent PIM response in the reverse direction, the periodicity of the ripple can quite often be related directly to the electrical length of the cable assembly. Yes, the cable PIM response is frequency dependent. However, if the cable assembly PIM is measured across a swept-frequency bandwidth that includes both peaks and nulls, the worst-case combination of PIM sources can be captured across the test band.

So long as the individual sources of PIM do not change in magnitude dramatically with frequency, and so long as the loss of the cable assembly does not change appreciably with frequency, there is a good change that the PIM results measured at 1800 MHz will be representative of the performance expected at 900 MHz.

This topic does bring up an interesting note, however. As the carrier frequencies increase, the RF skin depth on the conductors of the device-under-test tends to decrease. For equal carrier powers, the current density at 1800 MHz will be higher than at 900 MHz. For this reason, testing a cable at 900 MHz may produce a PIM result which is better than the results which might be obtained if the same cable were tested at 1800 MHz.

The bottom line is as follows: To be absolutely certain of the PIM level for a cable assembly in a given band, you should test in that band. To characterize the approximate performance of a cable assembly (or the integrity of the mechanical connector-cable interfaces), testing in one band will most likely yield results which are representative of the overall cable performance.

Back to Top

8. How is PIM Specified?

Passive IM is typically specified in absolute power (units of dBm) or power relative to only one of the test tones (units of dBc). For example, a -110 dBm IM signal caused by two +43 dBm tones is also specified as a -153 dBc IM level. In the case of unequal carrier power levels, Summitek Instruments has established the convention that units of dBc are relative to the largest of the incident carriers.

It is important to note that a carrier power level must always be specified with the given PIM performance level. This applies equally to PIM performance specified in units of dBm and dBc.

Back to Top

9. My PIM Test System has a Noise Floor of -140 dBm. Can I measure a -140 dBm PIM Response?

There is a clear distinction between random noise floor (kTBF) and the "PIM Noise Floor." The latter is more accurately restated as "Residual IM Level". Each of these two parameters are discussed below:

• The noise floor of the PIM test system is typically defined as the mean value of the measured signal when the receiver is terminated into 50 Ohms and the RF is turned off. If there is a coupling mechanism for noise from the high power amplifiers to appear in the receiver, this source of noise must also be included in the noise floor test. Noise is random and is typically due to a combination of phase noise in the local oscillator, kTBF noise from the receiver's pre-amplifier(s), and noise from the transmitter. The noise floor of a PIM test receiver (or spectrum analyzer/LNA combination) typically varies from approximately -120 dBm to -140 dBm depending upon the selected averaging level (or resolution bandwidth). You cannot make a meaningful IM measurement at a level below the noise floor of the receiver.
• The residual IM level of the analyzer is caused by internally generated IM within the analyzer's cabling, internal connectors, filters, and duplexers. This level is typically larger than the noise floor of the receiver for the third-order IM product (IM3). When an IM measurement is performed on a DUT whose true IM level is near that of the analyzer, significant measurement errors can occur. This is because the residual IM of the analyzer vectorially combines with the true IM of the device-under-test thus producing a measurement with a high uncertainty level. The residual IM level of a test system can be reduced through the use of high-quality diplexers, filters, and carefully constructed, Low-PIM interconnecting cables.

Note: Averaging (or reducing the resolution bandwidth of a spectrum analyzer) cannot reduce the residual IM level. Averaging is only useful in reducing the level of the receiver noise floor. For the most efficient measurement time, and to maximize the test system's responsiveness to transient PIM, use only enough averaging (or use the widest possible resolution bandwidth) to maintain the receiver noise floor at least 10 dB below the expect minimum PIM level.

Back to Top

10. Does PIM Change with Time?

Possibly.

Two types of PIM generation are typically found. The first type is of a "burst" nature and is commonly associated with the periodic breakdown of poor mechanical junctions exposed to high RF power levels. With this type of PIM, the IM will appear as a short (less than 1 second) burst of broadband, noise-like energy. On some devices and systems, these bursts have been measured at random intervals from 2 or 3 seconds to several hours.

The second type of PIM generation is more steady state, and coherent in nature. RF heating within RF conductors and around RF interfaces can causes minute changes in the contact integrity. The result is a PIM level which changes with time. A classic example of this can be found by measuring the PIM from a cable assembly which is poorly constructed or has been subjected to mechanical stress. The PIM performance of the cable assembly may appear quite good at first, only to degrade as the assembly heats up. Interesting enough, the opposite has been found to happen. The cable assembly is poor at first, but as the RF heating causes the mechanical interfaces to expand (and compress), the PIM performance improves with time.

Consider an operational and fielded base station. Wind, Rain, and Sun-induced thermal cycling are all at work to continuously stress the mechanical interfaces within the antenna, the cable assemblies, and the connections to the shelter. As the sun rises and heats the RF connections, the PIM levels can rise (or fall) if the cables, connectors, and antennas are not functioning properly. The result can be increased levels of IM only at certain times of the day.

Back to Top

11. Is There a Difference in Forward and Reverse (Reflected) IM Levels?

Because the intermodulation signals created at various PIM sources within an assembly are vectorial in nature, their relative phase relationships will determine the overall magnitude of (scalar) PIM measured at a particular location in a Device Under Test. Using the model developed in the application note Measuring the Passive Intermodulation Performance of RF Cable Assemblies, we find that all the IM responses arrive in-phase at Port 2 of a through IM measurement, independent of the IM frequency, while the reflected IM response present at Port 1 is a combination of the Port 1 response plus a phase-shifted response from the IM sources at Port 2. Because there is a vector combination of IM sources with differing phases, it is expected that the reflected IM response is a function of both frequency and the electrical length of the assembly.

In the real world, however, the forward IM response may not measure as being frequency independent and may not be the worst-case IM response. This is due to differences between the real world and the very simple model used in the above reference. Using more complex models which account for complex impedances and losses at not only the IM frequencies, but also at the harmonics of the carrier frequencies is a step forward towards more accurately predicting the results of a passive IM measurement.

Back to Top

12. Is There a Difference in IM with Carrier Frequency Separation?

Because the phase of the individual sources of IM within a device-under-test is related to the phase of the parent carriers, changing the carrier frequencies will change the phase relationship of the PIM signals. Depending upon where the composite PIM response is measured, the resulting composite PIM level may change as the carrier frequencies are changed.

Back to Top

13. Should by RF Sources be "Phase Locked" to "Align" the Carriers?

Not if you are testing with two tones. When two carrier tones are used, the relative rate of phaser rotation between the carriers is determined by the frequency separation. The carriers will periodically combine in and then out of phase at a fixed rate for the given frequency separation. Phase locking the carriers together will force the carriers to cross at a known instant in time, relative to the phase of one of the carriers. However, this won't impact the magnitude of the generated PIM levels.

If three or more carriers are utilized for testing, the phase of the third carrier now becomes important. By phase locking the three carriers together, and adjusting their relative phases, a specific phase point on the third carrier can be made to align with a known phase crossing point of the first two carriers. This could be used, for example, to establish a worst-case current density at a set of fixed frequencies at a specific point within the device-under-test.

Whether you are using 2, 3, or 100 carriers to perform PIM testing, it is good practice to connect the clocks together to minimize the impact of RF frequency drift on the measurement. This is especially important if you are using a very narrow receiver to perform the PIM testing.

Back to Top

14. What is the Difference Between a 16-Tone and a 2-Tone PIM Test?

When measuring PIM with a 2-tone test, there is only one IM response of each order which is of interest. This is in contrast to a 16-tone PIM measurement. In this case, there is a "picket-fence" of, say, IM3 responses to choose from displayed on the spectrum analyzer. Depending upon the characteristics of the device- under-test, the picket-fence may be flat, or have a more complex shape. Typically, the largest of the observed responses which is reported as the device's PIM level.

The 16-tone test has the advantage of allowing the device's PIM frequency response to be observed in a single measurement. This same frequency response display is obtained with the Summitek Instruments Passive IM Analyzer (or other computer-controlled PIM systems) by sweeping the carriers across a pre-defined band, much the same as a conventional network analysis measurement.

Comparing the 16-tone and 2-tone measurement results is a difficult task. Some users have reported the swept 2-tone test is much more difficult to pass, while others say the 16-tone test is more rigorous. As results from comparison testing become available, Summitek Instruments will continue to update this FAQ.

Back to Top

15. Why Would I Care About PIM?

Ultimately, it is the performance of the integrated base station that is important. Although most wireless transmit and receive frequency bands are carefully selected to avoid landing the largest IM products within the receive band, self-generated higher order products (IM5, 7, 9) do land within some communication bands. More frequently, IM products from a nearby (or co-located) competitor's site can become troublesome sources of interference.

To the receiver, PIM products appear as interference. Once the PIM power level rises above the random (kTBF) noise floor of the receiver, the system C/I becomes adversely impacted. Because PIM products typically increase significantly as the average transmit power level increases, the impact of PIM on a base station may only become significant when the base station becomes fully loaded. Just when the most capacity is needed, passive IM level can rise up and interfere with normal base station operation.

Back to Top

16. Are There Measurement Standards for Passive IM?

The International Electrotechnical Commission (IEC) has formed a Technical Committee (TC46/WG6). The assigned task of this committee is as follows:

"To prepare test methods and to investigate relevant limits, for Passive Intermodulation in the RF and microwave frequency range for passive components (i.e. connectors, cables, cable assemblies, waveguide assemblies and components...). To closely liaise with TC 102 for matters relevant to antennas and with SC 48B for connectors with respect to PIM. To liaise with other relevant committees, subcommittees, working groups, organizations and individuals, in order to ensure the widest appropriate awareness of, and the greatest relevant participation in and contribution to the work being carried out."

This group has been meeting for several years, and the first release of a standards document is imminent. Contact the IEC http://www.iec.ch for additional information or to obtain a copy of this document when it becomes available.

Back to Top

17. Other Questions?

Please use our contact information to reach us via telephone, facsimile, or email if you need further information or if you have a question you don't see answered here.

Back to Top