Instrument Amp Problem

Discussion in 'The Projects Forum' started by bnfdvn, Dec 8, 2008.

  1. bnfdvn

    Thread Starter Member

    Dec 8, 2008
    21
    0
    I inherited a design when someone left the company and I am trying to figure out why it is performing the way it does. We are using an Instrument amp to provide some ground isolation on an analog input (4-20 mA). From the input terminal +, the signal goes through a zener diode (1N5221, 2.1 V), through a 46.4 Ohm resistor and another 1N5221. We are using a INA118 Instrument Amp with a 12.1K Ohm gain resistor, giving a gain of 5.13. Since all signals are positive voltage, we are using a single 24VDC power supply to power the Instrument Amp.

    The problem is that the gain decreases with an increase in signal. At 4 mA, the gain is 5.14 and at 20 mZ, the gain is only 4.79. The output of the INA118 goes through a 100 Ohm resistor to a MicroChip 16F819 analog input.

    I had a INA128 available and installed it. The gain at 20 mA decreased to 3.81.

    Any help or suggestions will be appreciated. Thanks for looking!
     
  2. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    Can you post the circuit? I imagine you are trying to convert from the current loop to a voltage equivalent.
     
  3. bnfdvn

    Thread Starter Member

    Dec 8, 2008
    21
    0
    See attached. The circuit in question is on the lower left corner of the diagram. The jumpers around the zeners are not installed.
     
  4. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    I kind of question the function of the zeners. Voltage across the 46.4 ohm resistor will be 185 mv with 4 ma in, and 928 mv with 20 ma in. Having the zeners in series with the current loop is going to do strange things. Certainly, you won't have a linear response to the change in loop current.
     
  5. bnfdvn

    Thread Starter Member

    Dec 8, 2008
    21
    0
    I thought that this might be a problem also. I measured the voltage drop across the 46.4 Ohm resistor (actual measured resistance = 46.1 Ohms) and it tracks over the range of input current. At 4 mA, it is 0.184 V and at 20 mA, it is 0.918 V. With 46.1 Ohms at 20 mA, the voltage drop would be 0.921. My volt meter only reads XX.XX so this is probably within the tolerance of my equipment.

    According to the notes of my predecessor, the zeners are there due to Common Mode concerns. Other devices can be installed in the current loop and the diodes keep the input at least 2.1 V above ground and 2.1 V below the 24 V loop driving voltage.

    For my testing, I am using a 2.5 KOhm Pot, taking the center tap through a current meter to the input and taking the return to ground.
     
  6. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    I don't think you need be concerned with the driving voltage. The only thing is the loop current, which will be from 4 to 20 ma. Only enough voltage to produce the current will be passed through the resistor.

    You might place only a similar resistance across the current loop and see what actually develops across it.

    The INA118 is protected to +/- 40 volts on the inputs.
     
  7. bnfdvn

    Thread Starter Member

    Dec 8, 2008
    21
    0
    I installed the jumpers on the input, shorting out the zeners and removing them from the circuit and leaving only the 46.4 Ohm resistor. When I did this, the output of the INA118 went to zero and stayed there over the entire range (4-20 mA). The culprit was the jumper across D3. With this installed with my test setup, the loop's return is tied to the system ground. Removing the jumper brought the 118 back to life.
     
  8. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    Sure, you need D3 to satisfy the input common mode range spec. I can't see how the zeners can affect the gain, unless the source is somehow getting limited. Is it working now?
     
  9. bnfdvn

    Thread Starter Member

    Dec 8, 2008
    21
    0
    No - there is no change. Thanks
     
  10. bnfdvn

    Thread Starter Member

    Dec 8, 2008
    21
    0
    I replaced the bottom zener on the input (D3) with a 470 Ohm resistor and the performance of the Instrument Amp is much better. The gain is still not linear over the range of inputs but it is much better (5.13 at 4mA, 5.15 at 20 mA).

    Ron H. mentioned the 'input common mode range spec.' above. I looked at the data sheet but there is no spec for using a single power supply. Could someone please explain this principle to me?? Why does it matter what the common mode voltage is as long as it is slightly above 0 V potential?? What is the minimum acceptable common mode voltage for the inputs and how is this determined?? Thanks for any help you can offer!!
     
  11. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    Read "Single Supply Operation" on p.10.
    At 20mA, you will have 9.4V dropped across the 470 ohm resistor. Will that eat into the margin for other devices on the same loop?
     
  12. bnfdvn

    Thread Starter Member

    Dec 8, 2008
    21
    0
    Thanks, Ron! I had an incomplete datasheet and downloaded the one on the TI website and read the referenced section. It makes more sense to me now. I still don't get why it didn't work with the 2.1 V Zener but now I know how to design the input so it will work. Thanks again for pointing me in the right direction!
     
  13. bnfdvn

    Thread Starter Member

    Dec 8, 2008
    21
    0
    That's true, there is a much larger voltage drop at 20 mA. Typically, the other devices in the loop have much lower input resistances (50 Ohms) so the loop should support the larger voltage drop. Now that I have read the spec, I can see that I can get by with a 270 Ohm resistor which will have a 5.4 V drop at 20 mA.

    What do you think will happen if the driving loop is isolated?? (Grounds not connected) I finally got a hold of the original designer and he included the zeners to get the common mode voltage above what he felt was a reasonable difference between the two grounds. Does that make sense?? For instance, if the two grounds are 1.2 volts different potentials, the zeners would get the bottom input in the linear range of operation. The top zener (D2) would keep the + input suitably below the top rail. The spec sheet doesn't address the top end of the input voltage requirements. Does it matter where the + input is relative to the +24 volt rail??

    The purpose of this input circuit is to receive the analog input and translate it to a usable signal for the PIC (16F819). Is this even the right approach if it is possible that the driving analog signal can be isolated from this circuit's power??
     
  14. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    The datasheet does spec the high common mode limit. The minimum is V(+) - 1. See the attachment.
    The inputs must have a path back to the power supply ground, one way or another, to supply input bias current. For instance, you couldn't connect a floating battery across the inputs and expect a meaningful output (even assuming that you used a low voltage, or reduced the gain). I ran some simulations on your circuit, and strangely enough, the model doesn't work with a battery in series with the low signal line unless it is around 1V. Resistors above about 400Ω seemed to work. I ran the same sim using an LT1789-1, and it was very flexible as to the component used to satisfy the low end common mode limit. It may just be a problem with models. Of course, that doesn't explain your problems.
    See the schematic below. If you connect 1Meg from the negative input to GND, it should not introduce any error, and you can then guarantee that the input bias current will have have a path to GND. A similar circuit should work with your INA118, except you may have to change the 2 diodes to a 400-500 ohm resistor.
     
  15. bnfdvn

    Thread Starter Member

    Dec 8, 2008
    21
    0
    I have six INA118Ps so I checked out what minimum resistance (in place of D3) would make the circuit work. It turns out that 340 Ohms will make all the IAs work. This gives a measured voltage drop of 1.34 Volts across this resistor. Next, I installed two diodes in place of D3 and the circuit didn't work (voltage drop of 1.42 V). Then, I tried three diodes in place of D3 and the circuit didn't work. If I install four diodes (1N4002 is what I have), the circuit works over the entire 4-20mA range. The four diodes give a measured voltage drop of 2.87 Volts. Three diodes give a measured voltage drop of 2.15 Volts.

    It seems wierd to me that three diodes didn't work since the voltage drop across them was higher than the voltage drop across the minimum resistance.

    At 20 mA, the minimum voltage drop to keep the circuit in the linear range (using resistance) was 184 Ohms, giving a voltage drop at 20 mA of 3.65 volts. I checked the IA that gave me this reading with the four diodes and it worked across the range.

    These results don't make a lot of sense to me. I would think that the common mode voltage wouldn't depend of the type of device used to develop this voltage but it seems to.

    I don't have any LT1789s to test with but I will get some and repeat the above test. It sounds like it is better behaved in the analysis that Ron H did above. (Thanks, Ron, for taking the time to look at this!!)
     
  16. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    I agree that the results don't make sense. If I were you, I would contact TI technical support and summarize your problem, or simply ask them to read this thread, and then comment on it.
     
  17. bnfdvn

    Thread Starter Member

    Dec 8, 2008
    21
    0
    I was poking around on the TI website and came across a little app at this link.

    http://www-k.ext.ti.com/SRVS/CGI-BI...00000000013359558,K=5449,Sxi=0,Case=obj(42948)

    This app shows the operating range of the input voltage vs. the output voltage after you configure the parameters. For my circuit, V+ = 23.3, V-=0, Vref=0 and Gain = 5.13. This software tells me that the INA118 was not the best choice for this application. The INA122, 125 and 126 all require much more reasonable Common Mode Voltages. Am I reading this right??
     
  18. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    Test your link. I couldn't find where you went from there.
     
  19. bnfdvn

    Thread Starter Member

    Dec 8, 2008
    21
    0
    Sorry about the link. It changes everytime I get into the website. The steps to get to the App are:
    1. Go to www.ti.com
    2. On the left side, sellect 'Products'
    3. Near the top of this window, place cursor over 'Design Support', then click on 'Knowledgebase'
    4. Select 'Analog & Mixed Signal'
    5. In the Knowledgebase Search box, enter the following:
    Keyword: INA118
    Product Type: select 'Op Amps/Comparitors'
    Document Type: select 'Product Folders'
    Click on 'Search'
    6. Select Item 1, 'INA118 Product Folder...'
    7. Scroll down to 'Tools & Software'. This is over 3/4 of the way down the page.
    8. Click on 'Calculate Input Common-Mode Range of Instrumentation Amplifiers'
    9. On this screen, click on 'Download'
    10. After it downloads, unzip it and run the .exe file.

    When the App is running, click on the the 'Configure' pull-down menu to enter the parameters.
     
  20. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    I ran the app, and if I am interpreting it correctly, you are right.
    Actually, AMP04 seemed to have the widest input and output range, but curiously, it seems to be an Analog Devices part. What's up with that?:confused:
    Maybe they included it because it is inferior to TI parts in some applications? Maybe they quit making it after the software was created?
     
Loading...