Some installations defeat a default radar config: dusty grain silos, narrow threaded pipes, small closed metal tanks, and very short ranges. This guide collects the field-proven fixes for these difficult tanks and special applications on our 80GHz FMCW radar level transmitters (wired and 4G models, accuracy down to ±1 mm on wired units, IP68 housing).
For general echo/ signal faults start with the echo & signal diagnosis guide; for wiring, 4–20 mA and RS485 setup see the installation & wiring guide.
Bulk solids such as soybeans scatter the radar signal, so with default parameters the surface echo is misidentified and the height reads inaccurate. On a grain silo with the JJY-YLD128HS-1.5, proven tuning is to raise the threshold/ sensitivity for dusty surfaces and, where needed, enable surface-tracking that tolerates the angle of repose. Case study: inaccurate grain-silo level and the parameters that fix it.
In a narrow pipe the beam must stay centred on the surface. If the M38 radar is mounted off-centre or tilted, part of the beam hits the pipe wall and the sensor locks onto the wall echo (about 1105–1106 mm) instead of the true level (~510 mm). Centre and plumb the antenna, and use the smallest feasible antenna/ window for the pipe bore. Worked example: M38 false reading inside a narrow pipe.
In a small closed metal tank the 80GHz wave reflects repeatedly between the liquid surface, the metal walls and the lid, burying the real surface echo among multipath ghost peaks. Mitigate with a shorter blind zone where possible, narrow the window so the beam cannot see the side wall, and enable spurious-echo learning for the fixed internal reflections. Diagnosis: messy echoes in a small closed metal tank.
Very short ranges (tens of centimetres) stress the near-field behaviour. A short-range 80G unit that reads the wrong distance and throws error codes usually needs the correct near-field setup and a clear window in front of the antenna; lifting the sensor away from the target changes the geometry and hides the real fault. Model-specific case: short-range wrong distance and error codes.
Usually no. Bulk solids scatter the signal; tune the surface-tracking and sensitivity for dusty solids first.
The beam is hitting the pipe wall because the antenna is off-centre or tilted. Centre and plumb it.
Partly. Enable spurious-echo learning for the fixed reflections and narrow the window so the side wall is out of view; physical geometry still matters.
Field-tested fixes for each difficult-tank scenario:
Bulk solids scatter the signal — the threshold/ sensitivity and surface-tracking parameters that fix a dusty grain silo on the JJY-YLD128HS-1.5.
Off-centre or tilted M38 mounting locks the sensor onto the pipe-wall echo (~1105 mm instead of ~510 mm) — centring fixes it.
Multipath between surface, metal walls and lid buries the true echo — spurious-echo learning plus a narrower window.
Tens-of-centimetres ranges stress near-field behaviour — the correct near-field setup and a clear antenna window.
Model: JJY-YGD38M-18L — 80 GHz short-range radar level transmitter
The customer needed the radar to measure correctly at an actual distance of about 0.12 m. With the radar lifted away from the target the reading was about 0.25 m, and in some cases an abnormal error code was output. During testing, even after lifting the sensor the reported distance was still around 0.11 m — far from the real value.
In this very short measuring range the true surface echo competes with antenna and tank-wall spurious echoes. Without proper false-echo learning the unit locked onto a wrong echo.
Upgrade the firmware to the 0901 build and perform false-echo learning. After that the unit met the customer’s requirement; the remaining units are being upgraded and re-tested.
False-echo learning commands:
AT+FALSEECHO=2,0.12,0.18 — learn false echoes in the range 0.12–0.18 m (effective suppression).AT+FALSEECHO=1 — enable false-echo learning.AT+FALSEECHO=0 — disable false-echo learning.AT+FALSEECHO=3 — clear the learned false echoes.AT+FALSEECHO=2,a,b — learn false echoes in the distance range [a, b].Also check and adjust the installation position.


Answer: The radar was mounted off-center (or tilted) in the 108 mm pipe, so part of the beam hit the pipe wall and the sensor locked onto the wall echo — about 1105–1106 mm — instead of the true 500–510 mm level. Adjusting the installation position so the antenna sits centered in the pipe and perpendicular to the liquid surface fixes the reading.
The M38 thread radar level transmitter (38 mm body, model JJY-YLD128D-W-32) was mounted on a pipe with a diameter of 108 mm. When the actual level was around 500 mm, the output jumped around and read about 1106 mm. With the true distance at about 510 mm, the sensor reported 1105 mm. Aimed at open ground the same unit read normally, which confirmed that the beam was hitting the pipe wall.
Field installation photos — the M38 transmitter mounted on the pipe:


Echo spectrum at the moment of the fault — the sensor reported a false distance of 1.1056 m:

In a narrow 108 mm pipe, an off-center or tilted antenna lets part of the beam cone reach the pipe wall. The wall returns a strong echo from a longer path, so the radar locks onto the wall instead of the liquid surface. The output therefore shows a much longer and unstable distance, and it jumps whenever the beam footprint touches the wall.
After the installation position was corrected, the readings became stable and matched the actual level.
If you meet a similar application, contact our support team with your installation photo and the echo spectrum from the upper-computer software — we will help you select and tune the right radar level transmitter.
Answer: In a small closed metal tank the 80 GHz wave reflects repeatedly between the liquid surface, the metal walls and the lid. The real surface echo gets buried among multipath ghost peaks — in this case a strong interference peak appeared at about 17 cm — so the sensor cannot lock onto the correct echo. Upgrading the sensor to the latest ranging algorithm restores a correct and stable reading.
The 80 GHz compact radar distance sensor (model JJY-YGD38M-18L) was installed on a stainless steel tank with a diameter of only 30 cm, and the tank was closed with a lid. The returned echo was very messy with multiple reflections, and a strong interference wave appeared at about 17 cm. The sensor could not read the correct level echo.
Field installation — the radar mounted on the closed 30 cm stainless steel tank:

Echo spectrum returned by the sensor before the fix — several overlapping peaks instead of one clean surface echo:

A small, closed, all-metal vessel is a highly reflective environment: the 80 GHz signal bounces between the liquid surface, the cylindrical wall and the lid, producing strong multipath echoes. The true level echo is no longer the strongest peak in the spectrum, and the ranging algorithm locks onto a ghost peak (here at about 17 cm) instead of the real surface.
Upgrade the sensor to the latest firmware (ranging algorithm). The updated algorithm identifies the true surface echo among the multiple reflections and suppresses the ghost peaks caused by the metal walls and the lid. After the upgrade the sensor returned the correct level reading.
If you meet a similar application, contact our support team with your installation photo and the echo spectrum from the upper-computer software — we will help you select and tune the right radar level transmitter.
Answer: Bulk solids such as soybeans scatter the radar signal, so with default parameters the surface echo is misidentified and the height reads inaccurate. Field-proven tuning for the JJY-YLD128HS-1.5 on a grain silo: raise the antenna peak threshold (100 → 150), enable suspended installation and enlarge the blind zone to 0.3 m.
The 1.5-inch 80G radar level transmitter (model JJY-YLD128HS-1.5) was installed on a grain silo to measure the level of soybeans. The measured height was inaccurate — the reported level did not match the real fill height.
Installation on the grain silo:

Echo spectrum during tuning — the antenna threshold command AT+ANTPEAKTHRE is being sent from the upper-computer software:

Granular products return weaker and more scattered echoes than liquids: the surface is loose and uneven, the fill forms a cone, and dust attenuates the signal. With the default antenna peak threshold (100), noise and secondary peaks compete with the true surface echo, and the default installation type and blind zone do not match a suspended installation above bulk solids.
Raise the antenna peak threshold from 100 to 150, switch the installation type to suspended installation, and enlarge the blind zone to 0.3 m.
AT+ANTPEAKTHRE=0 — disable the antenna thresholdAT+INSTALLTYPE=1 — suspended installationAT+BLINDZONE=0.3 — blind zone 30 cmAfter these changes the measured soybean level matched the actual fill height.
If you meet a similar application, contact our support team with your installation photo and the echo spectrum from the upper-computer software — we will help you select and tune the right radar level transmitter.
Send us the vessel drawing, medium and mounting constraint — our engineers will specify the right antenna, blind zone and parameter set. Module JJY-YLD138M-32: MOQ 1 with free prototype.
Copyright @ 2026 JiaJia Eye

Scan to add WeChat