Planning an Emissions Sweep That Starts in Kilohertz and Ends in Gigahertz
Few measurements span as many decades as an electromagnetic compatibility scan. One product qualification can begin at 150 kHz on a line impedance stabilisation network and finish at 6 GHz in a chamber — a range of more than four decades, split across different receivers, different antennas and different detector settings. Writing the whole plan in one unit is the only way to see it as a single line, and gigahertz is the unit the upper half already uses.
The three numbers that shape every plan
A Floor at 150 Kilohertz
A Handover at 0.03 Gigahertz
A Ceiling Set by the Fastest Clock Inside
Laying Out Each Segment Before the Chamber Booking
Put the standard's start point in first
Enter 150 in the kilohertz field and the gigahertz side reads 0.00015 — small, but written out in full rather than rounded away. Anything below a millionth of a gigahertz switches to scientific notation, so a 9 kHz start point still shows a usable figure instead of a row of zeros.
Mark the segment boundaries you will change settings at
Work along the plan in the order the receiver will: 150, 30 000, 1 000 000 kilohertz. Each answer lands on the axis your limit lines are drawn against, and the thousands are grouped with a space so 1 000 000 cannot be mistaken for 100 000.
Lift the figure into the test plan document
The copy button returns the digits only, without a unit or the display spacing, so a boundary pasted into a sweep-configuration table or a report template arrives as a clean number your spreadsheet will accept as numeric.
Reverse it when the report comes back in gigahertz
A lab report that flags an exceedance at 0.868 GHz is easier to trace against a switching harmonic once it reads 868 000 kHz. Press the swap arrows, or just type into the gigahertz box — either field accepts input, and a comma decimal is understood as readily as a dot.
Segment Boundaries, Detectors and Sweep Settings
Every row below is a place where something changes: the transducer, the receiver bandwidth, or the limit line itself. Converting the kilohertz edge into gigahertz makes the whole qualification fit on one horizontal axis.
| Sweep segment | Start edge in kHz | Stop edge in GHz | Resolution bandwidth |
|---|---|---|---|
| Lowest receiver band | 9 kHz | 0.00015 GHz | 200 Hz |
| Conducted, mains and signal ports | 150 kHz | 0.03 GHz | 9 kHz |
| Automotive conducted band | 150 kHz | 0.108 GHz | 9 kHz then 120 kHz |
| Radiated, chamber or open site | 30 000 kHz | 1 GHz | 120 kHz |
| Radiated above the crossover | 1 000 000 kHz | 6 GHz | 1 MHz |
| Extended microwave segment | 6 000 000 kHz | 18 GHz | 1 MHz |
Small Edges Stay Readable
Eight decimal places are enough to write 0.00015 in full, and anything smaller flips to scientific notation rather than rounding a start frequency down to zero.
Boundaries Land on the Plan Axis
Grouped digits keep 6 000 000 kilohertz distinguishable from 600 000 at a glance, which matters when a mistyped ceiling costs a chamber slot.
Two-Way Between Plan and Report
The swap arrows turn a lab report's gigahertz marker back into the kilohertz figure your harmonic arithmetic uses, without retyping either value.
Pre-Compliance Questions From the Test Bench
Why do conducted emissions stop around 30 MHz and radiated emissions start there?
It is a practical split based on wavelength, not a law of physics. At 0.03 GHz a free-space wavelength is ten metres, so a mains cord of one or two metres is still electrically short and behaves as a conductor carrying noise out of the product rather than as an efficient antenna. Above that the same cord and the seams of the enclosure become a respectable fraction of a wavelength and start radiating properly. The standards therefore measure current and voltage on the cable below the crossover and field strength above it. Nothing dramatic happens at the boundary itself; a real product usually shows the same offending harmonic on both sides of it.
Which resolution bandwidth belongs to which part of the range?
Three values cover almost everything. Nine kilohertz applies from 150 kHz up to the 0.03 GHz crossover, 120 kHz applies from there up to 1 GHz, and 1 MHz applies above 1 GHz; a fourth setting of 200 Hz covers the band below 150 kHz where it is used. The bandwidth is not a preference — the limit lines were derived with it, so measuring a broadband noise floor with the wrong setting shifts your reading by several decibels and invalidates the comparison. Step sizes follow from it as well, since a sweep normally advances by no more than half the bandwidth in use.
How does a switching converter running in kilohertz cause trouble in the gigahertz segment?
Through edges, not through the switching rate itself. A converter running at 500 kHz produces a comb of harmonics every 500 kHz, and reaching 1 GHz simply means the two-thousandth harmonic. Their amplitude is governed by the rise and fall time of the switch node: a transition of a few nanoseconds still holds meaningful energy into the hundreds of megahertz, and ringing from parasitic inductance resonating with device capacitance places a broad bump somewhere between 0.05 and 0.3 GHz. That resonance, coupled onto a cable that acts as the antenna, is the usual culprit behind a failure hundreds of times above the switching frequency.
What decides how high the radiated scan has to go?
The highest frequency generated or used inside the equipment, not the frequency it transmits on. The rule scales in tiers: a product whose fastest internal activity stays modest may finish at 1 GHz, while faster designs are carried to 2 or 5 GHz, and anything with an internal clock above 1 GHz is normally swept to five times that clock with a practical ceiling around 6 GHz. Two consequences follow. Adding a faster memory bus or a higher crystal multiplier late in a project can extend the test range and the chamber booking, and the fastest clock is often buried in a module somebody else specified.
How closely does a bench pre-scan predict the accredited result?
Well enough to find problems, not well enough to declare compliance. A near-field probe and a low-cost analyser on the bench will reliably show you which harmonic families exist and whether a filter change moved them, and that is where most of the value sits. What it cannot reproduce is the calibrated site attenuation, the antenna factors, the turntable and height scan, and the quasi-peak detector the limits assume. Expect several decibels of disagreement in either direction, treat a margin under about 6 dB as unproven, and keep the pre-scan for iteration rather than for the declaration itself.
No comments yet. Be the first to comment!