You have three carriers to put through one BUC. The datasheet gives you a single linear-power number, and the temptation is to subtract a backoff, add a margin, and call the link closed. That number is only meaningful if you know which measurement produced it. For three carriers, the figure of merit that matters is not the one most datasheets lead with.
This is not a story about finding a magic backoff. It is about refusing to compare incomparable numbers, and about knowing which measurement method actually describes your operating case.
The carrier count is the first branch point
Amplifier linear power is defined by at least three different measurement methods, and the method is chosen by carrier count. Communications and Power Industries (CPI) states the taxonomy directly in its amplifier-selection guidance: spectral regrowth (SR) for single-carrier operation, third-order intermodulation products (IM3) for two to five carriers, and noise power ratio (NPR) for multi-carrier and high-order modulation operation, typically more than five carriers. The same source notes that these definitions need further refinement depending on modulation type and where and how the intermodulation products are measured.
Three carriers sits in the IM3 domain by that taxonomy. That does not mean NPR is irrelevant to your link, but it does mean a datasheet IM3 number is the one you can compare against your case, provided you resolve what the number actually references. An NPR number quoted for a many-carrier case is not a drop-in substitute for a three-carrier IM3 figure, and vice versa.
The IM3 reference ambiguity is a 3 dB trap
CPI is blunt about this: customers should exercise extreme care to determine whether the IM3 level in a datasheet is for the sum of two equal carriers or for the individual single carrier level of two equal carriers. The same source states the consequence: the difference either improves the IM3 level by a full 3 dB, or allows for a 1.5 dB increase in linear power.
Work the arithmetic on those quoted figures, clearly as arithmetic and not as a measured result. Suppose Vendor A quotes −25 dBc IM3 referenced to the sum of two equal carriers, and Vendor B quotes −25 dBc referenced to the single carrier level. If you compare them at face value, you will conclude they are equal. They are not. Resolving both to the same reference moves one of them by 3 dB in IM3, which corresponds to a 1.5 dB difference in the linear power you can extract at that IM3 level. On a three-carrier uplink where you are trying to hold Es/N0 through a rain event, 1.5 dB of linear power is not a rounding error. It is the difference between closing and not closing.
CPI also lists examples of vague or misleading IM3 wording it has seen in datasheets, including phrases like −25 dBc two signal 5 MHz apart at P(linear) relative to total power, and −17 dBc at 4 dB total output power backoff from rated power with two equal carriers. The pattern to reject is any IM3 number that does not state the reference, the carrier count, the spacing, and the backoff definition. CPI’s own guidance is that in such cases it is necessary to contact the manufacturer to determine how the IM3 is specified before a like-for-like comparison can be made.
Bandwidth dependence hides inside family datasheets
IM3 performance is not constant across an amplifier’s rated band. CPI notes that IM3 products improve when the amplifier bandwidth narrows and when RF output power is lowered, and gives a concrete example: an amplifier operating from 5.850 to 6.425 GHz will tend to have better IM3 performance than one operating from 5.850 to 7.075 GHz. The same source states that this difference is often not mentioned when a datasheet specifies a family of amplifiers.
For a three-carrier BUC, this matters because your operating band may be a subset of the family’s headline range. A family IM3 number quoted across the full band is not the number that applies at your frequencies. Ask for the IM3 at the specific operating band, at the specific carrier spacing, and at the specific backoff you intend to use. If the vendor cannot supply that, you are comparing a family average against your specific case, and the comparison is not like-for-like.
What NPR does and does not tell you
NPR is the figure of merit CPI identifies for many-carrier operation. It defines the ratio as power density (signal plus intermodulation distortion) divided by intermodulation distortion power density, or (C+I)/I. In practical terms, it is the quietness of an unused channel when nearby channels are transmitting.
Two cautions follow. First, CPI states that the intricate setup and equipment to measure NPR is quite expensive, and that some manufacturers may not have the ability in-house to perform this measurement. That is a procurement question: if a vendor cannot measure NPR, they cannot give you a many-carrier NPR curve, and you should not accept an IM3 number as a substitute for a many-carrier case. Second, the retrieved CPI text does not provide a three-carrier NPR procedure or a mapping from NPR to Es/N0. Any specific backoff number or Es/N0 target presented as derived from NPR for a three-carrier BUC would be invented. The honest move is to request the vendor’s NPR test data for your carrier count and bandwidth, and to close the link in your own budget with your own modem thresholds and propagation assumptions.
The regulatory boundary you cannot back off past
Backoff is not only an amplifier-linearity decision. ITU-R S.524-5 states that off-axis e.i.r.p. density levels are determined by the side-lobe gain, the transmitter output power level, and the spectral distribution of that power. The recommendation sets maximum permissible off-axis e.i.r.p. density values for earth stations transmitting in the 6 GHz and 14 GHz bands, with the limits expressed as functions of off-axis angle and referenced to specified bandwidths.
The same recommendation includes a provision relevant to rain-fade operation: when up-link power control is used and rain fades make it necessary, the limiting values stated in section 3 may be exceeded for the duration of that period. That is a regulatory boundary condition on any backoff or power-control strategy, not a backoff recipe. The current in-force version of this recommendation is S.524-9 (2006); the retrieved full text was the 1994 S.524-5 version, so confirm the applicable limits for your band and administration against the current text before finalizing a power-control design.
A datasheet-interrogation checklist
Before you compare two BUCs for a three-carrier uplink, get answers to these questions in writing:
- Which measurement method produced the linear-power number: SR, IM3, or NPR?
- How many carriers were used in the test?
- What was the carrier spacing?
- Is the IM3 level referenced to the sum of the carriers or to the single carrier level?
- What bandwidth was the amplifier tested over, and does that match your operating band?
- How is backoff defined in the datasheet: from rated power, from P1dB, or from some other reference?
- Can the vendor supply NPR data at your carrier count and bandwidth?
If any of these cannot be answered, treat the linear-power number as unverified for your configuration. That is not a reason to reject the amplifier; it is a reason not to pretend the number means what you need it to mean.
What to do when you have three carriers and a datasheet
Match the specification method to the carrier count. For three carriers, that means IM3 is the relevant figure of merit, and the IM3 reference must be resolved before any vendor comparison. If the vendor cannot state the reference, contact them; CPI’s guidance is explicit that this is necessary for a like-for-like comparison.
Do not substitute an NPR number for a three-carrier IM3 case, and do not substitute an IM3 number for a many-carrier NPR case. The methods answer different questions. If your carrier count will grow, ask for NPR data now rather than discovering later that the vendor cannot produce it.
Finally, remember that the amplifier is one element in a chain. The off-axis e.i.r.p. density limits in ITU-R S.524-5 depend on side-lobe gain, transmit power, and spectral distribution, and the rain-fade exception for up-link power control is a boundary condition on your operating strategy. Your link budget, your modem thresholds, and your propagation assumptions determine whether the chosen backoff holds Es/N0. The datasheet, once you have resolved what it actually says, only tells you what the amplifier can do before you add everything else.
FAQ
My BUC datasheet quotes NPR. Is that better than an IM3 number for three carriers?
Not automatically. NPR is the figure of merit CPI identifies for many-carrier operation, typically more than five carriers. For three carriers, IM3 is the method that matches your case. An NPR number may still be useful context, but it is not a direct substitute for a three-carrier IM3 figure.
The datasheet says −25 dBc IM3. Can I compare that directly to another vendor’s −25 dBc?
Only if both numbers use the same reference. CPI states that the difference between referencing IM3 to the sum of two equal carriers versus the single carrier level is 3 dB in IM3, or 1.5 dB in linear power. Resolve the reference before comparing.
Why does the datasheet quote one IM3 number for a whole family of amplifiers?
CPI notes that IM3 products improve when the amplifier bandwidth narrows and when RF output power is lowered, and that this difference is often not mentioned when a datasheet specifies a family. Ask for the IM3 at your specific operating band and backoff.
Can I exceed the ITU off-axis e.i.r.p. density limits during rain fade?
ITU-R S.524-5 includes a provision that when up-link power control is used and rain fades make it necessary, the section 3 limiting values may be exceeded for the duration of that period. Confirm the applicable limits and conditions in the current in-force version (S.524-9, 2006) for your band and administration.
What if the vendor cannot provide NPR data?
CPI states that the setup and equipment to measure NPR is quite expensive and that some manufacturers may not have the ability in-house. If you need many-carrier performance data and the vendor cannot supply it, that is a procurement constraint to weigh, not a reason to accept an IM3 number as a substitute.