AD converter input protection

9 posts / 0 new
Last post
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi!

I'm facing now a problem, namely that I have to measure analog signals in very rough industrial environment. The measured analog signal comes from a sensor over a long (5-10 meters) cable. The inputs must be protected agains EMI and ESD, and even a the case must be considered, that the 440V mains voltage is suddenly present on the inputs. So the inputs must be very high impedance inputs with lot's of protection.
The signal is a slow signal so the capacitance of the protection devices in not really important. But the leakage current should be as low as possible.

Does anybody have experience in that field? Any suggestion, links, books, tutorials are welcome.

Thanks,
Mik

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm not an expert, but I'll share a thought or two to get things started.

The input impedance for AVR ADCs is (usually?) spec'd at <= 10K. This means you can't just put a large series resistor on the input.

You may well end up with an op-amp stage at the input, with input protection for the op-amp's input, and the op-amp feeding the ADC.

The op-amp can easily have a large input resistor, and then diode voltage clamps to V+ and gnd.

I would use a shielded cable, also.

Consider whether or not you can put the micro next to the sensor, and send a digital signal over the 5-10 m distance to whatever. Transmitting a digital signal in a noisy environment may be easier.

Obviously in noisy environments you may want to consider taking multiple samples and averaging them. The software equivalent of puting an analog filter on the front end. You probably want to do both, but you did not mention how fast you need to be measuring the signal, or your accuracy requirements.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for the advice!

I found an interesting application note from Analog Devices:
http://www.analog.com/static/imported-files/tutorials/MT-069.pdf

Do I remember correctly, that the input resistance for the instrumentation amplifiers really high is? Does this mean, that looking at the schematic on the 2nd page (of the appnote I attached) I can add a few 100kOhm resistor in series with the input pin instead of that 500 Ohm resistor (and of course using protection diodes)?

DocJC wrote:

I would use a shielded cable, also.

And to what, and how would you connect the shield? I have seen many solutions so far: For example connectin the shield through a capacitor and a varistor(paralell with the capacitor) to the protective ground of the AC line.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Wiki Instrumentaion Amplifier is a nice summary of instrumentation amplifiers.

I do not believe you mentined what your sensor is. If it has a single ended output you probably do not need an instrumentation amp configuration.

Instrumentation amps are very useful for measuring Wheatstone bridge circuits and bioelectric signals, for example, but not needed for many sensors with analog outputs.

There are many op-amp configurations. Wiki Op Amps is somewhat helpful, but more importantly has links to other good sites. National Semiconductor also has several good Application Notes on using Op Amps.

The question about how to ground a shield is actually a very interesting and complex topic, and way over my head. I was looking for my favorite article on modeling such systems and can't seem to find it at the moment. There have been several threads discussing this on this forum.

With the little information known about your system one might make a general purpose recommendation: Use a twisted wire cable with a shield, and ground the shield at the uC end only, (directly, not through an RC filter). This will no doubt generate some further discussion on this topic.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Devices like LittleFuse SP720 may help. They divert transients to power rails, very fast 2ns, with low leakage 5nA. Power rails must be able to asorb the energy dump. Great protection in a small multi-device packages.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You might do better by looking for an existing solution in industrial electronics. This is not a new problem. You might start with by googling signal conditioners.

Yes, a lot of this stuff is expensive. But it's generally designed for reliable performance in unfriendly environments -- both electrically and physically.

Just a thought.

If you really want to learn a lot about this area, you might try and pick up a copy of "Noise Reduction Techniques in Electronic Systems" by Henry Ott. I was fortunate enough to take a short class based on his book years ago. He worked or consulted for the then Bell Telephone -- and they had a vested interest in low cost methods of minimizing electrical noise. This guy knows his stuff.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

DocJC wrote:

The input impedance for AVR ADCs is (usually?) spec'd at <= 10K. This means you can't just put a large series resistor on the input.

The source impedance must be less than 10Kohms so the ADC can charge the ~20pF sampling capacitor without too much error.

The source resistance can be higher, but to keep impedance requirement less than 10k, you can also put a capacitor there. 100nF capacitor should be enough. But then the sampling rate must be slow enough that the 100nF capacitor can charge to the nominal voltage before new conversion. Of course 100nF capacitor will limit the signal bandwith too with the extra series resistor.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

4...20 mA current loops are often preferred in industrial environments.

Ross McKenzie ValuSoft Melbourne Australia

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

As Ross suggests, 4-20mA is the 'usual' method. For strain gauges and thermocouples, you can get a 'puck' that converts the signals to 4-20mA. This gets located close to the sensor.

The device is called a 'puck' as it looks like a ice hockey puck - usually round and 2" dia abouts.

http://www.omega.com/ppt/pptsc.asp?ref=TX91_92