Unity gain buffer on AI lines

[johnsold]

What makes you think your source has a high impedance? The only mention in the data sheets you posted shows a minimum load resistance of 2000 ohms. That is NOT the output impedance. It more likely is related to the current limit of the output in voltage mode.

Is the ground of your 24 V supply connected to earth ground through its chassis or mains power cord? If not, try the RSE connection with the single ground point at the DAQ device AI GND terminal.

Can you post a schematic diagram showing how your sensors, sensor power supply, and DAQ device are wired? Include more than one channel but not all 80. Also make sure that all ground, sense, and reference connections are shown as actually wired. Just dropping ground symbols on a diagram may not reveal any ground loop or improper connections.

In the images of your signals you posted a few days ago it is hard to see what is happening on a millisecond time scale. If you could expand the data around the transient which occurs at about 1.7 on the sample count scale so that 0.1-0.2 seconds of data is shown (and specify the time scale), it might give us some clue about the nature and source of the interference.  My guess is power line frequency interference.

Lynn

[Vikash_Kumar]

1) The coupling symptoms I was observing was explained in NI-Knowledgebase as Ghostings. Since then I am trying out the solutions proposed in these knowledgebases.

2) Yes, my 24V supply is connected to earth ground through both its chassis and its main power cord.

3) Schematics are attached

4) The original matlab figures are also attached in the same response. Assuming you donot have Matlab to open those figures. The data used for generating those figures is attached below, incase you want to have closer look.

- each column is one AI channel 

- Sampling rate of the acquisition is 10 kHz

[Henrik_Volkers]

Some words to add:

If you look carefully at the spec page 2 you find these diagrams:

The top one tells you about your problem of ghosting AND why adding additional GND channels won't help you. (Log scale for the error!!)

The only difference is, that your ghost has only one value(=0), but the error is still there. So using buffers and the lowest possible samplerate (= long settle time)  seems the only way to me that migth help.

As you already found out, different OPs have different output impedances 🙂

If you going to build buffers anyway and you have many channels in a bigger setup, think about differential input design and keep the trace(s) to the AI-Sense short and clean, this is the OUTPUT REF for your buffer!  TI has a forum too, maybe ask an OP expert there??

If all your sensors are of the types above: A look into the spec told me something of a MINIMUM LOAD of 2k. So check that (ask an app. ing. from SICK if you are in doubt)

The 2k input impedance help you against noise, but add significant current. So look for your current pathes (ground loops).

[johnsold]

The Omron S8VS-24024A is NOT grounded on the DC side. The specifications allow connecting them in series and the block diagram clearly shows capacitive coupling between the -V output and ground.

Try connecting the -V power supply connection and all the sensor negative (M) connections to the AI GND.  This should be a single point ground in a star configuration. The use RSE.

With all your sensors connected to one power supply, they all have a common reference and it is the -V output of the supply. Differential does not gain you anything except a lot of extra wiring in this situation.

Lynn

[Vikash_Kumar]

Are all AI-GND channels internally connected?

Is connecting one AI-GND channel to the power supply -ve terminal enough OR I need to connect all of them?

[johnsold]

The manual is your friend:

Lynn

[Vikash_Kumar]

Not much of a difference. Here are the plots. Data and Matlab figures are also attached. 

(forum is acting weird. Its reporting some error. I'll try to upload things one at a time) 

[Vikash_Kumar]

[Vikash_Kumar]

[Vikash_Kumar]

Matlab figures of plots above