Noise performance on sensors

[iamdink]

Hello all,

I have a question which I was thinking of. Most sensors have some sort of gain stage to bring a small signal out of nosie. If a sensor is good deal from processor then a twisted pair is used to bring signal into the processor.

Has anyone built a circuit with sensor, gain stage, and discrete ADC on a circuit far from processor. Then the data and clock lines and power supply and ground would go out. I wonder if the component has a high rated PSRR if this would be a worthwhile approach.

 

[keith1200rs]

It is not unusual to put the sensor electronics and ADC remotely.

Keith

 

[iamdink]

keith1200rs, thank you for response.

I just wonder if it's worth the extra effort. I'd be supplying 5 lines (SCLK, SDA, VCC, GND, and CS) as opposed to 2 (VCC and GND). I think if you were in an noisy RF setting it might make sense where the long transmission line would be an antenna.

 

[keith1200rs]

It all depends on the environment you are in, the sensitivity of the signals and size of them. I have used remote ADCs on occasions.

Keith.

 

[kender]

I just wonder if it's worth the extra effort. I'd be supplying 5 lines (SCLK, SDA, VCC, GND, and CS) as opposed to 2 (VCC and GND). I think if you were in an noisy RF setting it might make sense where the long transmission line would be an antenna.

It all depends on the environment you are in, the sensitivity of the signals and size of them. I have used remote ADCs on occasions.

Keith.

I'll second Keith.

Another advantage is that you can have a multichannel A/D, which allows you to make a node with multiple (2, 4, 8, etc) sensors, while still using only 5 wires.

Also, have you considered an A/D converter that has an I2C interface? I2C doesn't need CS# lines; it uses address bytes. I2C tends to be more extensible, because you don't need to run additional CS# lines. A word of warning, though: I2C is not too immune to EMI, but that can be dealt with. Both I2C and SPI were designed for short-haul communication within 1m or so. Nevertheless, there are "legitimate" methods for boosting I2C and SPI. (Here's a sample menu showing what can be done with I2C: https://www.ccsinfo.com/forum/viewtopic.php?t=38463.)

 

[pauloynski]

I agree with the posts above but you have also to consider some extra advantages. Your ADC temperature coefficient wil never match the temperature coefficient of your sensor. Placing both at the same place will equalize the temperature of them and you will get a combined temp coef you can manage to cancel out. Also, you can in factory calibrate it and replace a defective one without recalibration. And, not less important (if noise or interference is a problem), you can more easily opto-couple digital signals than analog ones. There are lots of " digital load cells" in the market using this principle.

 

[iamdink]

With regards to the I2C. Yes I like this approach better as single bus can be distributed. When you quote 1m, for what voltage is device operating at? I'm using 3.3v logic so I really wonder if +5 is quoted for 1m.

I guess for longer network you might want to use CAN. High bit rate for larger distance (500m). https://en.wikipedia.org/wiki/Controller_area_network

pauloynski: Yeah something I was considering. Much easier to transmit optical signal if sensor is far away. I should say though that my sensor is maybe half meter away from processor.

I think I will experiment with 2 implementations. I will mount sensor, gain stage and I2C ADC and transmit over 3/4m. Then I'll do same thing but have ADC 3/4m away from gain stage.

 

[keith1200rs]

I have remotely digitized data a few times. 5km down fibre at 8Mbps and 2m at 280Mbps. Both were unusual cases.

Keith