Two numbers, two reference planes
A 100 W disagreement between an HPA front panel and a link budget is not a mystery to be solved by picking the number you like. It is a measurement problem with a known structure. The front-panel reading and the coupler-referenced reading are produced by different detector chains, at different reference planes, under different assumptions about load match and waveform. A persistent offset between them is expected. The engineering task is to characterize that offset, bound its uncertainty, and decide which number the link budget is allowed to use.
This article is about the procedure, not about diagnosing any particular amplifier. No source establishes why a specific HPA would read 400 W while a link budget says 300 W, and this article will not invent one. What the sources do support is how to build a coupler-referenced measurement, how to calibrate the coupler path, and how to reconcile the result against a built-in meter so the link budget carries a defensible figure.
What the front panel actually is
Many BUCs and HPAs include a built-in power meter. That is a useful operational indicator: it tells you the amplifier is producing output, it trends over time, and it can drive alarms. It is not, by itself, a traceable measurement of output power at the HPA flange. The internal chain is typically a coupled detector, a scaling factor, and firmware correction, and its accuracy depends on the detector’s frequency response, its temperature behavior, and the load match at the output. None of that is a criticism of the design; it is a statement about what the reading is for.
The alternative is a calibrated external coupler and a power meter head. At a teleport hub, output power can be measured using a high-loss calibrated cross-guide coupler and a broadband power meter head, or the HPA’s own built-in meter can be used. The two paths are not interchangeable until you have shown, by measurement, that they agree within a stated tolerance over the range you care about.
Define the reference plane before you measure anything
Most reconciliation arguments are really reference-plane arguments. Three planes matter here:
- HPA output flange. The amplifier’s rated output power is usually specified here. This is the plane the front panel is trying to report.
- Coupler coupled port. The external measurement is made here, and the coupler’s coupling factor and insertion loss translate it back to the main line.
- Antenna feed flange. The link budget’s EIRP calculation needs power at the feed, after waveguide runs, switches, and diplexers.
The distinction is not academic. In at least one inter-operability test specification, antenna gain for calibration purposes is explicitly referenced to the transmit coupler output, not the HPA flange. If your link budget assumes one plane and your measurement assumes another, you have manufactured a discrepancy before you have even turned on the power meter.
Write down which plane each number belongs to. Then convert everything to a single plane — usually the HPA flange for amplifier health, and the feed flange for EIRP — using measured losses, not catalog losses.
Build the coupler calibration table
The coupler is the reference only if it has been calibrated. A nominal 50 dB coupling factor is a starting point, not a measurement. The same inter-operability specification that references antenna gain to the coupler output requires the transmit waveguide coupler assemblies to be calibrated across the operational band at frequency points no more than 5 MHz apart, including points at least 15 MHz below and 15 MHz above the band of interest. That is a demanding but sensible requirement: couplers are not flat, and the band edges are where the surprises live.
For each polarity, record:
- Nominal coupling value and the measured coupling table versus frequency.
- Insertion loss of the coupler in the main line.
- Waveguide type and connector type at the coupled port.
- The calibration date and the method used.
If you do not have a calibration table, you have a nominal number with unknown frequency dependence. Treat it accordingly: your uncertainty budget just got larger, and the link budget should say so.
Measure the output power the coupler way
The standard computation is straightforward. A calibrated directional coupler with attenuation G0 sits between the HPA output and the load. The coupled port feeds a power meter reading P1. With a known input level P0, the gain is computed as G = P1 − P0 + G0. Rearranged for output power, the HPA output is the measured coupled power plus the coupler attenuation, corrected for the coupler’s frequency-dependent coupling factor and any loss between the coupled port and the meter head.
Practical points that decide whether the number is worth anything:
- Use a real load. A dummy load rated for the power, or the antenna pointed safely away from the geostationary arc. Do not calibrate against an unknown mismatch and then apply the result to a well-matched antenna.
- Measure at band edges, not just center. The coupler table tells you the coupling factor varies; the measurement should confirm the correction is applied where it matters.
- Record the cable and connector losses between coupler and sensor. A short coax jumper and an adapter can easily account for a few tenths of a dB. That is not noise; it is bias.
- Note the waveform. A power meter head responds to average power. If the HPA is amplifying a modulated multicarrier signal, the meter reading and the internal detector may weight the waveform differently. Measure under the conditions you will operate under, or state the difference.
The test instrumentation for this kind of work is ordinary: signal generator, spectrum analyzer, waveguide couplers of roughly 50 dB attenuation, and microwave power meters. The discipline is in the calibration and the bookkeeping, not in exotic hardware.
Reconcile against the front panel
Now compare. Sweep the HPA output over a range of power levels — not just the one operating point — and at several frequencies across the band. At each point, record the coupler-referenced output and the front-panel reading. The difference is the offset you are trying to characterize.
Do not collapse that into a single fudge factor. A single offset is valid only over the power and frequency range where it was measured. If the offset is 1.0 dB at center frequency and 1.4 dB at the band edge, a single 1.0 dB correction is wrong at the edge by 0.4 dB. Store the correction as a table or curve with explicit validity limits: power range, frequency range, load condition, and waveform. Re-derive it after any waveguide work, coupler change, or HPA swap.
A hypothetical example makes the uncertainty arithmetic concrete. Suppose the coupler calibration table shows 0.4 dB variation across the band, and the power meter head contributes 0.2 dB uncertainty. The combined correction uncertainty is not 0.6 dB by simple addition — root-sum-square gives about 0.45 dB — but either way it bounds how finely you can trust the front-panel offset. If the disagreement you are trying to explain is 1.2 dB, a 0.45 dB measurement uncertainty means you have explained part of it and the rest needs another cause. If the disagreement is 0.3 dB, you are arguing about noise.
What goes into the link budget
The link budget should carry the coupler-referenced value, corrected to the reference plane the budget uses, with an explicit uncertainty term. That converts an unexplained 100 W gap into a bounded error. A budget that says “300 W ± 0.5 dB” is honest and usable. A budget that says “300 W” and ignores a front panel reading 400 W is neither.
Two rules keep this honest:
- Never round in the flattering direction. If the corrected output is 297 W and the uncertainty is ± 0.5 dB, do not write 300 W as if it were exact. Write the value and the tolerance.
- Re-check after configuration changes. An HPA swap, a waveguide re-run, a redundancy switchover, or a coupler replacement invalidates the offset table. The link budget’s number is only as good as the last calibration that supports it.
Keep the calibrated number honest over time
An HPA can degrade slowly. An operator running a 10 W carrier on a 150 W HPA may be completely unaware that the maximum power capability has dropped over time to, say, 16 W, because the front panel still looks healthy at the operating point. That is the argument for periodic offline testing: use redundancy to put the offline HPA into a dummy load, measure it properly, and switch over. Annual tests of this kind catch slow degradation that a single operating-point reading will not.
Multicarrier operation adds a second reason to keep the calibration current. In hub multicarrier operation, the BUC should be rated for roughly twice the actual composite power to avoid unacceptable intermodulation interference. During rain, uplink power control can raise each carrier substantially; the composite power the amplifier actually sees is what the calibration and the headroom calculation must reflect, not the clear-sky number.
Frequently asked questions
Which reading should the link budget use?
The coupler-referenced value, corrected to the budget’s reference plane, with a stated uncertainty. The front-panel reading is an indicator until you have characterized its offset against a calibrated coupler over the operating range.
Can I just apply a single correction factor to the front panel?
Only if you have shown the offset is constant over the power and frequency range you operate in. Otherwise store a table with validity limits and re-derive it after any hardware change.
How often should the coupler path be calibrated?
The sources cited here do not specify an interval. What they do establish is that accurate uplink power measurement requires precise coupler calibration across the band, and that periodic offline HPA testing catches slow degradation. Set the interval from your own uncertainty budget and maintenance history rather than from a number invented here.
Does the internal meter have a specified accuracy?
That depends on the specific HPA or BUC and its documentation. This article does not assert an accuracy figure for any internal meter, because the manufacturer documentation needed to support such a claim was not available. Check the vendor’s data sheet for your unit before treating the front panel as anything more than an indicator.
The short version
Define the reference plane. Calibrate the coupler across the band with margin either side. Measure with a real load and a power meter head whose losses you have recorded. Compare against the front panel across power and frequency, and store the offset as a bounded correction, not a fudge factor. Put the corrected value and its uncertainty into the link budget. Re-check after every hardware change, and test offline into a dummy load on a schedule that catches slow degradation. The 400 W and the 300 W can both be real; the job is to know which one the link budget is allowed to believe, and by how much.