Noise on the Baseline

Licel Logo


If you follow good laboratory praxis and record dark measurement files (see the excellent High Power Lidar: Standard Quality Assurance Procedures for NF operation Chapter 9 (page 18) from ACTRIS /CARS) you might notice that the baseline is not flat.

If you don't record those files, there is a serious question if your results are scientific valuable, usually they are not.

What you expect then is a flat line with white noise on it. Sometimes it is not flat and then comes the question how to improve it. And the first step is analysis.

Why Range corrected lidar signals are not so useful for searching noise

Typically errors in the range squared signals or backscatter time series are the starting point to search for instrument problems. However the instrument electronics have nothing to do with scattering in the atmosphere and even less with the molecular density. In the following it is assumed that you can trace back the problems to raw data files without any manipulation.

Analog versus Photon counting

The first question is in which channels do you see the problem, is it in the analog signal or in the photon counting signal or in both. This gives a good hint what to search for. High frequency sources like mobile phone or even worse cordless phones can spoil photon counting measurements, but are usually completely invisible in the analog signal. The analog signal preamplifier has an anti-alias filter which suppress high frequency signals.


The next good question is: if this is something new or has always been there. If it is a regression, then there should be a dark measurement file where it was OK.

Is this problem stable

Is the structure/feature stable in time, it appears in each data file at the same time/height? Or does it appear at different places?

Can you see the problem in the Live Display

Can you also see the problem in Live-Display with 100 shots or in Single shot mode. If it is only sometimes visible in Live-Display take your mobile phone out and make a video of it. Asynchronous spikes, as from cheap AC/DC converters, which are not trigger related, are very easy to spot in Live-Display

Is this problem laser related

Usually the laser is one of the key noise sources.

Laser is not the trigger source

If laser is the controlled by an external trigger source remove the trigger source from the laser and repeat the dark measurement.

Does the problem go away?

Laser generates the trigger

If the laser generates the trigger either as QSW out or via photodiode pickup turn off the laser and supply via a function generator a trigger with the same trigger rate to the system and repeat the dark measurement.

Does the problem go away?

Problems reproducible without the laser

Synchron to the trigger generator

The problem ist stable and does not move relative to the trigger. Try to remove as much cables from the system, where you still see the problem.

This is then something, that can be reproduced at Licel and we can assist.

Ansynchronous Problems

The problem is either not stable in time relative to the trigger or comes and goes. As above try to remove as many cables as you can while the problem persists.

Once you have reduced the number of participants connect the signal cable to an oscilloscope and terminate with 50 Ohm. Search there for noise.

Violations of the trigger isolation

The transient recorder rack has an isolated trigger input which helps to isolate the transient recorder and detectors from the laser noise. In an ideal system the between the BNC trigger input housing and the BNC signal input housing with all cables attached you measure high impedance with a multimeter.

The intentional PE violation

The signal GND is connected to the protective earth of the transient recorder rack. If the laser has a trigger output which is also connected to protective earth then they meet at the wall plug. This is OK, as the noise has to pass two filters to enter the transient recorder rack. The recommendation is of course to use different power plugs for the laser and the rack to maximize the distance that the noise needs to travel.

If the trigger is not connected to PE (like after an opto-coupler) there might still be isolation violations as shown below under External violation of the trigger isolation - Housing

However if you unplug the transient recorder and detector rack from power this connection should go away.

Internal violation of the trigger isolation

If you then remove the trigger cable from the transient recorder rack and there is still a measurable resistance starting from 0 Ohm, you take photos of your setup and contact Licel.

External violation of the trigger isolation - Network cable

Look at the rear side of the rack for the Ethernet cable. The network receptacle creates a connection to the PE and signal GND. Using there an unshielded twisted pair cable ensures the trigger isolation. On a shielded twisted pair cable you see usually a metallic sheet on the cable plug. This breaks the isolation. So replace it. All cables that you get from Licel are UTP cables.

External violation of the trigger isolation - Housing

This is a more difficult question, which one can diagnose with a usual handheld multimeter.

Unplug in addition to the power and network cables the laser from the power outlet. Find on the laser power supply a point which is still connected to the PE Pin of the laser power cable. This is reference point 1.

From there find a point at the laser head housing which is connected to reference point 1. This would be reference point 2.

Try to identify, where you see a electrical connection between the shield of the BNC input of the transient recorder and the reference points 1 and 2.

If you identified a point stay calm and follow the recommendations below for changing the grounding

Change the grounding

Before you do anything you grab your lab book to write down what you will do and plan where you will record dark measurement files. The worst (usual) thing that will happen when you don't do this, is that after 2 or 3 iterations you are lost, your system is worse than at the start and you do not remember which actions you did take and which actions did you reverse.

Back to index