A test report for radiated emission carries an expanded uncertainty, and CISPR 16-4-2 decides what the laboratory does with it. If the laboratory's value is no larger than the value CISPR publishes, the result is compared with the limit as measured. If it is larger, the excess is added to the result before the comparison, and a product that was 1 dB under the limit can fail. Laboratories tend to treat that number as fixed. It is not. The chamber's own validation result, the NSA deviation below 1 GHz and the site VSWR above it, is an input to the calculation, and the standard allows the measured value to be used in place of the tolerance.
- NSA validates a site from 30 MHz to 1 GHz: the measured site attenuation between two antennas, normalised by their antenna factors, shall be within ±4 dB of the theoretical value at every position, height and polarisation. Site VSWR validates it from 1 GHz to 18 GHz: the variation of received signal as a transmit antenna is moved along the measurement axis shall not exceed 6 dB.
- CISPR 16-4-2 carries each into the budget as a "site imperfections" term: δAN with a ±4 dB half-width and a triangular distribution below 1 GHz, giving a standard uncertainty of 1.63 dB; δSVSWR with half the SVSWR as the half-width above 1 GHz, giving 1.22 dB for a 6 dB site.
- In the 3 m OATS and SAC budget that sets UCISPR at 6.3 dB, the site term is second only to the antenna's directivity correction; at 10 m it is the largest term by a wide margin. The standard lets a laboratory substitute its own measured deviation, and a chamber validated at 2 dB instead of 4 dB takes about 0.7 dB off the expanded uncertainty at 3 m and 0.8 dB at 10 m.
The two validations every chamber carries
NSA, 30 MHz to 1 GHz
Normalised site attenuation, in CISPR 16-1-4 and ANSI C63.4, is the transmission loss between a transmit and a receive antenna at defined positions, with the two antenna factors subtracted so that what remains is a property of the site alone. The receive antenna scans from 1 m to 4 m in height and the minimum loss is recorded. For a chamber the transmit antenna is placed at the centre, front, back, left and right of the test volume, at two heights and in both polarisations, so that reflections from the walls, ceiling and ground plane edges are all exercised. The measured value is compared with the theoretical NSA the standard tabulates for an ideal ground plane, and the difference shall be within ±4 dB at every frequency, position, height and polarisation. Nothing in the standard relaxes that figure: a site that misses it is not a valid site.
Site VSWR, 1 GHz to 18 GHz
Above 1 GHz the absorbers, not the ground plane, decide the site. Site VSWR measures how much the received signal varies as a transmit antenna is moved along the line towards the receive antenna in six steps over 40 cm, at positions across the test volume and in both polarisations. In a perfect free-space site the received level would follow the distance change and nothing else. Reflections make it ripple, and the peak-to-peak ripple after distance correction is the site VSWR. CISPR 16-1-4 allows 6 dB. It is called a VSWR because that ripple is exactly what a standing wave between the direct and reflected paths produces.
Where the site enters the budget
CISPR 16-4-2 builds the measurement instrumentation uncertainty from a list of input quantities: the receiver reading, the cable attenuation, the antenna factor and its corrections, the receiver's pulse and sine-wave corrections, mismatch, and a group of site corrections. Each is given an estimate of zero, a half-width and a probability distribution, and the standard uncertainty is the half-width divided by the divisor for that distribution: 2 for a normal distribution quoted at k = 2, √3 for rectangular, √6 for triangular and √2 for U-shaped. The combined standard uncertainty is the root sum of squares of all of them, and the expanded uncertainty is twice that.
The site enters as one line in the group of site corrections.
| Range | Input quantity | Half-width | Distribution | Standard uncertainty | Why that distribution |
|---|---|---|---|---|---|
| 30 MHz to 1 GHz | δAN, site imperfections from the NSA result | ±4.0 dB, the CISPR 16-1-4 tolerance | Triangular | 1.63 dB | Comment D7: a site that meets the 4 dB tolerance is unlikely to cause errors of 4 dB in a disturbance measurement, so the extremes of the range are treated as unlikely. |
| 1 GHz to 18 GHz, method 2 | δSVSWR, site imperfections from the SVSWR result | ±3.0 dB, half the 6 dB SVSWR tolerance | Triangular | 1.22 dB | Comment E6: the SVSWR is the maximum of 15 or 20 comparison measurements, so the error in any one measurement is most likely near zero. |
| 1 GHz to 18 GHz, method 1 | δSVSWR | ±4.0 dB, the deviation from ideal transmission loss that a 6 dB SVSWR corresponds to | Normal, k = 3 | 1.33 dB | Comment E6: the 4 dB is not exceeded anywhere in the band, so it is treated as a very-high-confidence bound. |
The published UCISPR values in Table 1 come from these budgets: 6.3 dB for an OATS or SAC from 30 MHz to 1 GHz, 5.3 dB for a fully anechoic room over the same range, 5.2 dB from 1 GHz to 6 GHz and 5.5 dB from 6 GHz to 18 GHz. Each is the worst case among the tables that feed it, and for the OATS and SAC value that worst case is vertical polarisation from 200 MHz to 1 GHz at 3 m, Table D.4. Above 1 GHz there is no separate horn budget: Tables E.1 and E.2 are written for either a double-ridged guide horn or an LPDA, the only antenna-specific row being cross-polarisation, which the standard treats as negligible for a horn, and the site term is the same for both.
Below 1 GHz at 3 m the site term is second only to the LPDA antenna's directivity correction, and well ahead of the antenna factor itself. The directivity term is a 3 m artefact: at 10 m it falls from 1.80 dB to 0.29 dB, the same budget gives 5.2 dB, and the site term becomes the largest contributor with close to 40 percent of the combined variance. Above 1 GHz the site term is second only to the mismatch between antenna and preamplifier. In every case it is worth between a fifth and two fifths of the variance, which is a large share for something most laboratories enter once and never revisit.
What the standard lets you do with a better result
Both comments carry the same permission. Comment D7 states that if the measured Dmax, the largest difference between the theoretical and the measured site attenuation, is less than 4 dB, the laboratory may take the correction δAN as zero with a triangular distribution of half-width Dmax when it calculates Ulab. Comment E6 says the same for site VSWR: a measured SVSWR below 6 dB may be halved and used as the half-width under method 2, or scaled as 4 × (SVSWR / 6) with k = 3 under method 1. The chamber's actual validation result, not the tolerance, becomes the input.
| Site result entered | Standard uncertainty of the site term | Ulab, Table D.4 at 3 m, all else as tabulated | Against UCISPR 6.3 dB |
|---|---|---|---|
| Dmax = 4.0 dB, the tolerance | 1.63 dB | 6.3 dB | Equal |
| Dmax = 3.0 dB | 1.22 dB | 5.9 dB | 0.4 dB below |
| Dmax = 2.0 dB | 0.82 dB | 5.6 dB | 0.7 dB below |
| Dmax = 1.0 dB | 0.41 dB | 5.5 dB | 0.8 dB below |
| Perfect site, 0 dB | 0.00 dB | 5.4 dB | 0.9 dB below |
| Site result entered | Standard uncertainty of the site term | Ulab, Table E.1 at 3 m, all else as tabulated | Against UCISPR 5.2 dB |
|---|---|---|---|
| SVSWR = 6.0 dB, method 2 | 1.22 dB | 5.2 dB | Equal |
| SVSWR = 6.0 dB, method 1 | 1.33 dB | 5.3 dB | 0.1 dB above |
| SVSWR = 4.0 dB, method 2 | 0.82 dB | 4.9 dB | 0.3 dB below |
| SVSWR = 2.0 dB, method 2 | 0.41 dB | 4.6 dB | 0.6 dB below |
| Perfect site, 0 dB | 0.00 dB | 4.6 dB | 0.6 dB below |
Two readings of these tables. First, the gain is real but bounded: even a perfect chamber removes less than 1 dB from the expanded uncertainty below 1 GHz, because the antenna factor, the receiver's pulse response and the LPDA's directivity at 3 m stay where they are. A chamber is not a way to buy a small uncertainty on its own. Second, the gain is available now, from a certificate the laboratory already holds, provided that certificate reports the deviation in a form that gives Dmax, and provided the laboratory's quality system records the substitution. Method 1 at the full 6 dB tolerance lands above UCISPR, which shows that the published value assumes method 2, and a laboratory using method 1 should check which side of 5.2 dB it ends up on.
Why the number matters at the limit line
Clause 4.2 of CISPR 16-4-2 is the reason any of this has consequences. If Ulab is no greater than UCISPR, compliance is decided on the measured value. If Ulab is greater, the difference Ulab − UCISPR is added to the measured value before it is compared with the limit. A laboratory whose antenna factor uncertainty, cable losses or mismatch push it to 7.0 dB below 1 GHz gives every product a 0.7 dB handicap. Entering a measured Dmax of 2 dB instead of the 4 dB tolerance returns roughly that 0.7 dB at 3 m, and 0.8 dB at 10 m where the same budget falls from 5.2 dB to 4.4 dB, and can move the laboratory back under UCISPR without touching a single instrument. That is the practical value of a good site validation result: not a pass mark, but a number that can be carried into the budget.
What to take from the validation certificate
NSA from 30 MHz to 1 GHz and site VSWR by the CISPR 16-1-4 method from 1 GHz to 18 GHz are both on the accredited scope we validate under, and the certificate reports the deviation at every frequency, position, height and polarisation, so that Dmax and the largest SVSWR can be taken straight from it into a CISPR 16-4-2 budget. The site validation guide lists the parameters. If your budget still carries the 4 dB and 6 dB tolerances and your chamber does better, send us the certificate and the budget, and we will show what the substitution is worth for your site.