Actions
Forum Thread Details
  • Replies 118 replies
  • Subscribers 313 subscribers
  • Views 14019 views
  • Users 0 members are here
Related
See a helpful answer?

Be sure to click 'more' and select 'suggest as answer'!

If you're the thread creator, be sure to click 'more' then 'Verify as Answer'!

Have a question about ADCs or DACs? Ask our Expert, Nick Gray

ChristyZ

This thread has been closed to new questions.

However, we welcome you to Post Your Question about Communications in the element14 Community Wireless Communications Technology group. You'll find many fellow members and experts who have just the answer you're looking to find! 

 

Thank You, Your Friends at element14 Community


Nick Gray

 

Nicholas Gray

Nicholas has worked in the Semiconductor industry for over 30 years and has authored a number of published articles about data converters (ADCs and DACs) and signal integrity issues.

 

Parents

  • I have a problem with temperature drift using a AD7714 (http://www.analog.com/en/analog-to-digital-converters/ad-converters/ad7714/products/product.html). I was hoping for someone to put me in the right track. I'm not an expert in electronics, just starting.

    What I'm trying to do is use it for a weight scale. I have a typical load cell that gives a signal of 2mV/V. I'm connecting the ADC using SPI to an arduino.

    I have it working and I can set filter, gain, and so on. What I do not understand is the behaviour I'm seeing.

    To check what was going on I have tried different set ups. In the simplest case I have a 0.1% resistor divider generating the signal. The positive reference of the ADC is ratiometric with the excitation of the divider. When I leave it connected through the day the reading changes way above the resolution that the load-cell can give. It is a 2000 kg load cell that should be good to measure 3000 divisions. The oscillations that I see in the reading of the ADC are equivalent to more than 2 kg, which is way to much.

    I have also tried with a fluke 715 calibrator that a friend lend me. It can generate a very precisse signal; however, results were pretty much the same.

    I have been reading about voltage offset and so on and I thought that the ADC would take care of that I used in continuous calibration. It does not seem to be like that. I have tried to configure it in continuous calibration as well as doing a full calibration before each sample but I got the same result.

    Am I doing something wrong? If I should measure temperature and compensate for it I will, but then I see no point in using such a complex (at least for me) ADC.

    I habe been going crazzy with this for weeks. Any help would be greatlly appreciated.

  • in reply to Former Member

    Hello Alberto,

     

    Thank you for your question. Let's see if I can help you.

     

    The type of problem you are seeing is not at all unusual and there are a FEW possible causes. There are so many things that could affect your system performance. Let’s look at some of these. Here are the things I suggest you try, in order of ease of doing them. All of these could affect the outcome.
     
    50/60 HZ INTERFERENCE
    It sounds as though you may have power line noise in your output as you mention an “oscillation”. If the frequency of this “oscillation” is the same as your power line frequency (50 Hz or 60 Hz) then it is almost certainly not really an oscillation but is power line noise getting into the conversion process.
     
    Power line frequency interference is quite common, especially in high resolution system. It is very difficult to reduce frequencies this low to levels that will not affect the system. Consider that, at 24 bits and a 5V reference, one LSB corresponds to just 298 nanovolts! The best thing here is to operate from batteries, but that is often impractical. It may also be necessary to shield the entire circuit, but shielding at low frequency is not trivial. The AD7714 has an internal notch filter that you should set equal to your power line frequency and set the ADC clock frequency to 2.4576 MHz. Set the output data rate to be the same as your power line frequency.
     
    PLAIN OLD NOISE
    Noise can come from a number of sources. Noise in the reference can play havoc with your signal. To minimize this, I suggest a three-pronged approach. First, use a low noise reference source that is bypassed it with three capacitors: one of 47uF to 100uF, one of 0.1uF and one of 0.0047uF to 0.01uF. Second, Place the capacitors and reference source as close to the ADC reference pin as you can, with the smallest capacitor closest to the ADC pin. All capacitors should be on the same side of the board as the ADC and have two vias connecting the grounded end of the capacitors to the ground plane. Third, the ground plane must be in its own board layer with nothing else in that plane. No signals should be in the vicinity of the reference area, not on any layer. Noise can also be affected by PCB (printed circuit board) layout.
     
    PCB LAYOUT
    Unfortunately, this is a difficult one to discuss with less than a white paper or technical article. However, there are some key things to remember. Use at least a 4 layer board. The ADC and its reference should be on the same side of the board with each other. This layer is a signal layer, as is the other side of the board. The next layer from the ADC should be the ground layer, which should have NO TRACES, just a solid ground. The third layer should be the power plane with each voltage having its own, unbroken area. No traces should be in this layer. Signals should, ideally, not cross the boundary from one power plane to another, but it is not always practical to avoid this. You must consider the current flow through traces and allow the return current to stay very close to the outgoing current path. Realize that return currents sometimes flow in a power plane or two. Sometimes you must put capacitors between ground and a power plane to allow short paths for AC return currents.
     
    Please let me know whether or not anything here helped you.
     
     
    Nick Gray
Reply
  • in reply to Former Member

    Hello Alberto,

     

    Thank you for your question. Let's see if I can help you.

     

    The type of problem you are seeing is not at all unusual and there are a FEW possible causes. There are so many things that could affect your system performance. Let’s look at some of these. Here are the things I suggest you try, in order of ease of doing them. All of these could affect the outcome.
     
    50/60 HZ INTERFERENCE
    It sounds as though you may have power line noise in your output as you mention an “oscillation”. If the frequency of this “oscillation” is the same as your power line frequency (50 Hz or 60 Hz) then it is almost certainly not really an oscillation but is power line noise getting into the conversion process.
     
    Power line frequency interference is quite common, especially in high resolution system. It is very difficult to reduce frequencies this low to levels that will not affect the system. Consider that, at 24 bits and a 5V reference, one LSB corresponds to just 298 nanovolts! The best thing here is to operate from batteries, but that is often impractical. It may also be necessary to shield the entire circuit, but shielding at low frequency is not trivial. The AD7714 has an internal notch filter that you should set equal to your power line frequency and set the ADC clock frequency to 2.4576 MHz. Set the output data rate to be the same as your power line frequency.
     
    PLAIN OLD NOISE
    Noise can come from a number of sources. Noise in the reference can play havoc with your signal. To minimize this, I suggest a three-pronged approach. First, use a low noise reference source that is bypassed it with three capacitors: one of 47uF to 100uF, one of 0.1uF and one of 0.0047uF to 0.01uF. Second, Place the capacitors and reference source as close to the ADC reference pin as you can, with the smallest capacitor closest to the ADC pin. All capacitors should be on the same side of the board as the ADC and have two vias connecting the grounded end of the capacitors to the ground plane. Third, the ground plane must be in its own board layer with nothing else in that plane. No signals should be in the vicinity of the reference area, not on any layer. Noise can also be affected by PCB (printed circuit board) layout.
     
    PCB LAYOUT
    Unfortunately, this is a difficult one to discuss with less than a white paper or technical article. However, there are some key things to remember. Use at least a 4 layer board. The ADC and its reference should be on the same side of the board with each other. This layer is a signal layer, as is the other side of the board. The next layer from the ADC should be the ground layer, which should have NO TRACES, just a solid ground. The third layer should be the power plane with each voltage having its own, unbroken area. No traces should be in this layer. Signals should, ideally, not cross the boundary from one power plane to another, but it is not always practical to avoid this. You must consider the current flow through traces and allow the return current to stay very close to the outgoing current path. Realize that return currents sometimes flow in a power plane or two. Sometimes you must put capacitors between ground and a power plane to allow short paths for AC return currents.
     
    Please let me know whether or not anything here helped you.
     
     
    Nick Gray
Children
  • in reply to nickgray

    Hi Nick,

    I'm working on a project that need to split the RTD signal into two  Data  Acquisition System (or one goes into DAS and the other goes to a   controller of another system). I don't know any if you have any  suggestion for me.

    Thank you.

  • in reply to Former Member

    Hi Tran,

    I am not sure just what you mean by "split" the RTD signal, but I think you probably mean that you want to send the voltage across an RTD to two different places. If this is what you mean, it is really pretty simple if the two destinations have a high input impedance. If they both have a high input impedance and you are using a two-wire RTD, I would simply run lines from the RTD output to the input of the two destinations. I really like the idea of including a precision resistor in series with the RTD and monitoring the voltage across that resistor so that you know exactly what is the current through it and the RTD. If you are using a three-wire of a four-RTD, please let me know which you are using and any details about your need that you can. I will do my best to answer your specific question or questions. Be sure to tell me the specifics of your DAS and controller inputs. It might also help me understand your needs if you explplain how you would drive just one of them.

     

    Nick