WiFi signal - 65dBm, finally distinguishing dB, dBm, and dBi clearly

Let's leave the conclusion here: dBm is the reading, dB is the difference
dBm is the absolute reading of power, with a reference of 1 milliwatt. 0dBm is 1 milliwatt, 20dBm is 100 milliwatts, -70dBm is 0.0000001 milliwatts—one ten-millionth of a milliwatt. Why not just write 'milliwatt' directly? Because the wireless signal range is absurdly large: routers transmit 100 milliwatts, while phones in the bedroom receive only one-ten-millionth of a milliwatt, a trillionfold difference from one end to the other. Writing in milliwatts, with a string of zeros after the decimal point, makes anyone dizzy. The logarithmic scale reduces this trillionfold to double digits: transmitter +20, receiver -70—just one glance and you know it.
dB is a gap without a benchmark. Saying '90dB attenuated' means the power has become one billionth of its original power; Saying 'difference of 3dB' means a difference of double. It's not a reading, but the distance between two readings. Remember this difference in one sentence: dBm measures height, dB measures height difference.
I've listed the common conversion ranges here, but I'll repeat them later
| dBm | convertedto milliwatts | : |
|---|---|---|
| 20 | to 100 milliwatts, | 2.4G routers to national standard transmission limits |
| , 10 | to 10 milliwatts | ,commonly used settings for mobile WiFi backhaul |
| 0 | 1 milliwatt | baseline |
| - 30 | milliwatts of a signal close to the router | |
| - 60 | milliwatts | can still be usedby a wall |
| 0.70 | trillion milliwatts | , the rate starts to drop gear |
For every 3dB difference, the difference doubles: the gap between -60 and -70 is tenfold, not
tenThis is the most counterintuitive one. On the logarithmic scale, a 3dB difference equals a 10dB difference, and a 10dB difference equals a tenfold difference. So -60 and -70 are not just '10 differences', but a tenfold difference; My living room is 44 to bathroom -79 ( the blind spot self-check log), a difference of 35dB, which adds up to over 3,000 times.
This conversion rule explains two things I've always been vague about. One is the misleading signal bars: a phone drops from four bars to three, possibly only 3dB, with power cut in half but the feel almost unchanged; Dropping from three bars to one often means a difference of more than ten dB, and the rate has long since dropped from hundreds of megabytes to tens of megabytes. Another issue is the router and phone secretly shifting gears ( QAM covered the entire gear chart): each step of modulation consumption consumes about 6dB of signal-to-noise ratio, so the signal drops sharply, and the speed is cut in half, losing just one bar. Watching the frame count is always a beat slow; watching dBm gives you a chance to get a headshot.
By the way, let me share a pitfall I once made: a couple of years ago, when choosing antennas, one marked 5dBi, the other 8dBi. I thought, 'Isn't the difference just 3? I'll just buy the bigger one with my eyes closed.' Later, I realized that 3dB is the account of doubling power—but that's not the calculation at all; the dBi section is a separate topic.
One chain can handle three scenarios: 20dBm when going out, dropping 90dB on the road, and -70dBm in hands
Link the gap with the reading, and the whole chain connects. The national standard for 2.4G router transmission power is 20dBm, which is 100 milliwatts ( I read the national limit in the transmission power section). After the signal goes out, every loss is a dB account
For 5 meters in the same room, just the "radio spreading in all directions" means the 5G band consumes about 62dB. There is a ready-made formula: free space loss equals 32.44 plus 20 times the logarithm of distance kilometers, plus 20 times the logarithm of frequency megahertz. Substituting 5 meters at 5800MHz gives 62dB. 20 minus 62 equals -42. In my living room, it was -44, with a margin of error of 2dB. The formula is surprisingly reliable. Eat another 10 to 15 after a brick wall, restaurant test -55; Adding distance to the load-bearing wall in the master bedroom, actual measurement -71. All five points of reading can be reconciled, and this unit is instantly revitalized.
Using the same formula for 2.4G: the same 5 meters only consume 54dB, which is 8dB less than 5G, and the wall penetration also absorbs less than 5G. This is the physical ledger of "2.4G wall penetration is good" ( frequency band real-world test), not mystical but dB-recorded accounts.
Looking at the fiber optic chain, the style is completely different. The optical modem transmits power around +3dBm (visible in the background, usually between 0.5 and 5), and at my home it receives -22.5dBm, with only a 26dB drop throughout the process. Why does wireless lose 90 while fiber only drops 26? The laser is locked inside a fiber core as thick as a hair, running by total reflection and preventing leakage; Radio waves really spread throughout the entire house. Moreover, of the 26dB fiber optic fiber, the main end isn't even a wire—the 1:64 splitter in the hallway naturally consumes about 20dB (dozens of households share one beam of light per block at a building, theoretically 18dB, but the engineering leaves a 20dB margin). The remaining dB is the fiber itself (0.35dB per kilometer) separated from the fusion joint.
So my cousin's question is answered: different media means different family backgrounds. Fiber -22.5 is only 26dB away from the transmitter, so the link budget is generous and healthy; the -71 fiber is 90dB from the 20dBm transmitter, so the phone's reception sensitivity is already struggling. Light decay -27 and WiFi -71 are both called "still usable," but you can't directly compare the values to size.
Classification tables for four occasions: For the same unit, check your own table
dBm is everywhere, but each scenario's health interval is independently calibrated and cannot be viewed in a single sense. I have compiled a table listing the four frequently checked items at home:
| Where to check | thehealth zone | when it deteriorates? What to do first | |
|---|---|---|---|
| when receiving WiFi | on your phone? WiFi Analyzer | can be used anywhere within -50, -60~-70 works, beyond -70 don't get your hopes | up Adjust the router to adjust the position; a 10dB difference is a tenfold difference |
| optical modem receives optical power | 192.168.2.1 Status page | -8~-27 healthy, stick -28 to smooth the cord and wipe the head, outside -30 and ask the technician | to bend first and then wipe the head. Three tricks are here the light decay article |
| Transmitter power | router backend, national standard | 2.4G with a maximum limit of 20dBm, 5.8G up to 33dBm | but factory soldering and no adjustment, The test article mentioned |
| signal-to-noise ratio SNR | WiFi Analyzer, | background-70 signal paired with -97 noise base, leaving 27dB, enough for mid-range 64QAM | to distinguish whether the signal is poor or the noise is high |
There's a detail hidden in the table that needs to be singled out: SNR is measured in dB, not in dBm. Because it's calculated by subtracting two dBm — signal -70dBm minus noise at -97dBm, subtracting two absolute readings and canceling out the baseline, leaving a pure difference of 27dB. dBm minus dBm gives dB; dBm plus or subtract dB is still dBm. This logarithmic arithmetic follows just these two rules: think of the last m of the unit as 'meters', subtract the absolute value from the difference, and it works smoothly in one minute.
dBi is a different matter: the gain doesn't produce power, only the spread method is changed
The routerantenna is labeled dBi, and the 'i' at the end means 'relatively ideal omnidirectional antenna.' 0dBi is essentially a theoretical perfect spherical ejection, spreading evenly in all directions; 5dBi isn't about increasing power; it's about flattening the sphere into a pie shape, thicker horizontally, and thinner vertically. I swapped out three antenna gains and tested it. After switching to 9dBi, the bedroom signal shifted from -72 to -70, because all the extra gain was filled horizontally, making vertical upstairs and downstairs even thinner in the vertical direction.
There is also dBd, which is based on a half-wave oscillator antenna; 0 dBd equals 2.15 dBi. Sellers love to label dBi because the same antenna is labeled one size larger than dBi, which looks intimidating. When choosing an antenna, first identify the last letter of the wire, then compare the numbers.
The real power ceiling is the national standard EIRP—Equivalent Omnidirectional Radiation Power, which equals the dBm number of transmitted power plus the antenna's dBi number. 2.4G is limited to 20dBm, 5.8G to 33dBm, and both are limited to this general ledger. So when the antenna gain is increased, the machine has to lower the transmission power to make up the total, and the ceiling is welded shut from the factory. This is also the root of the 878 article's long test where "wall-penetration mode only yields two or three dB."
Order of counting: set the date first, then get started
By using this unit system, the order of household inspections can be half the hassle.
WiFi is slow, so I first took out my phone to check the dBm, which is much more reliable than the signal grid. Don't blame the signal for within -50, check channel interference ( packet loss method); Between -60 and -70, change routes and locations, and follow the blind spot self-inspection method; After -70, just consider overlaying solutions, Mesh or relay, and stop competing with a single router. At home, whether to choose between 2.4G and 5G, it's all about using dBm.
On the optical modem side, if the base is unstable during network disconnection, it reads light degradation. According to the table above, the dilution is a hard wire. Fluctuations inside the line should be smoothed out first, then repaired offline.
Finally, leave three sentences—it's enough: dBm is absolute, dB is about difference—3dB doubles, 10dB is ten times more. Next time you see a reading of negative tens, don't panic—it's not 'how many signals are owed,' it's the abbreviation for one millionth of a milliwatt—humans invented this awkward but truly useful scale to avoid writing a string of zeros.
