I am working on a 5 axis rotational table with 6 sensors. These sensors allow me to know the position based on an analog signal from each of them. Each sensor provide me 4 signal from 0 to 5 Volts with a current range from 0-20 mA. The four signals are A+ A- B+ and B-. Using differential mode I obtain signal A and B. To obtain the position I find the angle using arc-tan(B/A) and the length of the vector using sqrt((A^2)+(B^2))

I have been doing some test and my readings have some noise. The main noise source is some leaking power from the sensor that is affecting the readings. I think that using an optocoupler will make a galvanic isolation that will allow me to separate the grounds from the sensors and the ground from the DAC. I think I am going through the right direction but this is my first noise isolation project.

I also though about electromagnetic noise but all the cables are shielded and separated from high voltage cables. Also, the DAC system is placed outside of the machine with a Faraday shield and all the cables from and to the DAC are completely shield.

I have been trying different setups and designs to implement the optocouplers but my first and second prototype had a lot of problem with the output voltage and I had to use a graduable resistance and tweek it to be able to obtain a similar result for the output signal but I cannot do this for all the signals. My current design is "Attachment 1" .

For prototyping I used an Avago HCNR200 optocoupler and as my amplifier I used an TL074CN fro Texas Instruments. The resistances R1 and R3 were 10.000 ohm and R3 was 220ohm. I did not use capacitors because they will slow down my reading and I need a recording frequency of at least 15000 Hertz.

 

hi Blas,
I understand your noise problems, but I am not sure what question you are asking.?
E

EDIT:
you say:
The resistances R1 and R3 were 10.000 ohm and R3 was 220ohm.
Is this your circuit.? ref image.

 

Each sensor provide me 4 signal from 0 to 5 Volts with a current range from 0-20 mA.

Is that sensor output current range a limit rating or does the sensor have separate voltage and current outputs?

You will need some small feedback capacitance to stabilize the circuit, otherwise it likely will oscillate (which I observed in my simulation of a similar circuit).
That may be what was causing the erratic behavior you noticed.
You should be able to find a capacitor small enough to give you the desired frequency response while still stabilizing the circuit.
Start out with 1 nF or so.

 

hi Blas,
I understand your noise problems, but I am not sure what question you are asking.?
E

EDIT:
you say:
The resistances R1 and R3 were 10.000 ohm and R3 was 220ohm.
Is this your circuit.? ref image.

Sorry about the misunderstanding. I used a 10.000 ohm resistance for R1 and R2 and a 220ohm resistance for R3. The design is the practical design provided by Avago in the spreadsheet of the HCRN200 optocoupler.

I would like to know, why am I having a different voltage in the output and how can I improve the design to obtain an accurate result without decreasing the recording frequency. I was thinking also to use this design (attached) but it requires transistors instead of capacitors and amplifiers. In the spreadsheet it is referred as High-speed Low Cost Analog Isolator. Do you recommend this design over the other based on my type of signal and my recording speed?

 

Is that sensor output current range a limit rating or does the sensor have separate voltage and current outputs?

You will need some small feedback capacitance to stabilize the circuit, otherwise it likely will oscillate (which I observed in my simulation of a similar circuit).
That may be what was causing the erratic behavior you noticed.
You should be able to find a capacitor small enough to give you the desired frequency response while still stabilizing the circuit.
Start out with 1 nF or so.

Hello Crutschow,

The sensor output range its just a limit rating. I am interested in the frequency and the voltage of the signal.

Thanks for the recommendation I will make some test, the past capacitor I used was 50 microFaraday and it decrease a lot the recording frequency.

I will reply you back with my results.

 

hi Blas,
I ran a LTSpice simulation for the HCNR circuit, refer to image for result.
It looks OK for the required 15kHz BW you require, the Gain is low but this could be corrected by a following OPA,
E

 

See Fh=157kHz

 

hi Blas,
These two images show the LTSpice circuit and plots for 0v thru 5v input to 0v thru 5v output.

E

 

hi Blas,
I ran a LTSpice simulation for the HCNR circuit, refer to image for result.
It looks OK for the required 15kHz BW you require, the Gain is low but this could be corrected by a following OPA,
E

Hello Eric,

Thanks for all your help I will implement your design and I will let you know about the results.

 

hi Blas,
These two images show the LTSpice circuit and plots for 0v thru 5v input to 0v thru 5v output.

Why not just increase the gain of U9 instead of adding U1?

 

hi Carl,
It is just a personal preference, sometimes it is more convenient to use the final output OPA in a buffer/filter mode, also I do not know the requirements of the circuitry he is driving.
The TS did say he had some 'interference' problems, he is planning to post back his findings.

If you have a better idea please post it, I am always open to alternative solutions.

Eric