This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

INA333: Unstable voltage at differential inputs with a load cell connected

Part Number: INA333
Other Parts Discussed in Thread: TINA-TI

Hi,

I am using INA333 to amplify a differential signal from a FSS1500NSR load cell, which is based on the Wheatstone Bridge topology (I measured ~5kOhms between sensor's differential inputs). The maximum voltage from the load cell is 210mV, therefore I am using gain of ~20, i.e. RG = 5.6kOhms. The output is connected to an 8bit ADC. INA333 is connected through a single supply of 5V.

However, I have trouble achieving stability at the inputs of INA333. I have measured resistance between pins 1 and 6 and 5 and 8, which is supposed to result in ~350kOhms (looking at the schematic in the datasheet), however I measured ~182kOhms and ~307kOhms respectfully - though the values were not constant. 

Consequently, even though the load cell provides Vcm of 2.5V (V+ = 5V), connecting them to the inputs of the amplifier results in varying voltages at the inputs (Vin- = ~1.2 - 2.0V and Vin+ = ~0.7 - 1.2V). When I disconnect the load cell, I measure around 0.2V at both inputs which decreases and eventually goes down to almost 0. Thinking it might have been caused by the input bias current having no return path, I have connected 47kOhms between both of the inputs and ground but it didn't help. 

Since I have my prototype on the breadboard, I used an IC socket adapter to connect the chip to the breadboard, see picture below. 

 

The setup is shown below:

 

The circuit diagram:

Note: I have tried 4 separate INA333 chips and all of them suffered from a similar problem. One of them even managed to start smoking, which was quite a surprise since the 5V power supply is stable and was applied to the correct pin. 

Also note that I have measured the internal resistances of one of them just having taken it out of the box - it resulted in 347kOhms between both pins 1 and 6, and 5 and 8, as expected. Having soldered it on the adapter and measured it again, resistance 5-8 changed to ~250kOhms and became unstable, therefore I suspect that the chip is sensitive to high soldering temperatures? I used 300C. 

I searched the entire forum for a similar problem but couldn't find anything related, therefore I'd be very happy if someone could help me out with this. Thank you!

Regards,

Sarunas 

  • Hello Sarunas,

    I will look into this one for you.

    Best regards,

    Errol Leon
    Texas Instruments
    Precision Op Amps
  • Hi Errol,

    Just one thing I forgot to mention - I have rebuilt the entire circuit that was shown in the datasheet (ignoring the RFI filters) using MCP6004 op-amps and resistors of the same values - it works perfectly.

    Thanks.

    Regards,
    Sarunas
  • Hello Sarunas,

    Looking at the circuit diagram and simulations in TINATI,  you will want to have VREF to mid-supply (2.5V).  Running single supply (+5V) and having VREF at ground will cause the output to saturate.  I have attached a TINATI model to this post.  If you have any questions on the simulations file let me know.

    Best,

    Errol

    INA333 single supply.TSC

  • Hello Sarunas,

    One more thing to note is that the INA333 may not have the bandwidth needed to drive the ADC directly. Depending on your acquisition time you may not be able to settle in time to 1/2 LSB error resulting in inaccurate measurements.

    Best,

    Errol
  • Hi Errol,

    Can I infer that the amplifier in this case should never have VREF grounded? In my design the input voltage values will only be positive, therefore I thought that grounding VREF lets me to utilise the entire dynamic range as I won't need to amplify any negative values.

    Also, just to clarify - the output is connected to one of pololu baby orangutan's digital i/os. 


    Regards,
    Sarunas

  • Hi Sarunas,

    Looking over this it seems like if you have a 2.5V common-mode input from the bridge, you should be fine tying the reference to ground. As you mentioned, if your input is only of a single polarity, this will maximize your range. Just be sure to hook the bridge up to the correct inputs. To be clear, the measurements you took after soldering to your adapter board were with or without the gain resistor populated? I assume you also took these measurements without the board powered. The 47kOhm resistors shouldn't be necessary as the bias current should have a path to ground through the bridge. Did you make sure to thoroughly clean your board after soldering? Excess flux can create leakage paths between pins and cause abnormal behavior, and it sounds like this might be what you are seeing.
  • Hi there,
    I am trying to build a very similar configuration and I was having the exact same problem. Can I ask a few questions here?

    It seems to me like your TINATI model is still saturating. I got the same thing on the models I built for my issue.

    My configuration is similar but not identical, so let me lay it out:

    Sensor: load cell (wheatstone bridge), 2mV/V, on 0-5V supply. This means I should get a 0-10mV signal out of it.
    ADC: It has 12bit resolution and 500kHz sample rate. I can either input a 0-5V or -10+10V analog signal.
    Power source: 0-5VDC, 0-3.3VDC, -15+15VDC.

    My idea with the INA333 was:
    - power with 0-5VDC;
    - use a 200ohm Rg for 500V/V gain.

    My problems with the INA333 are:
    - My TINA models don't do what I was expecting. Either saturate at 2,5V or marginal gains on the 10mV even with the 200ohm Rg;
    - The INA333 is really small. I ordered online and was expecting something I can use on a proto-board;
    - I never considered capacitors or large resistors for the supply. But the suggested configurations have them and I am not sure I should/have to.

    My expectation:
    Nothing too fancy. I can live with some error margin if less than 10%. I would be happy with any samplying rate >100Hz and I am willing to have to recalibrate the sensor because of influence from the amplifier.
    I just want to measure something. After that I can start optimizing, reducing noise, searching for smaller and cheaper components.

    I hope someone can shed a little light into this. I am a little frustraded. Thanks!
  • Hi Bruno,

    Please include your TINA circuit so we can get a better idea of how you are setting up this circuit and where the issue might be.
  • Hello Bruno,

    We have some INA333 parts coming in and would like to recreate the design and compare it to simulation. Could you send your schematic as well as the TinaTI files that Zak requested?

    Best,

    Errol Leon
    Texas Instruments
    Precision Op Amp Applications
  • HI guys, thanks for your interest.
    I got my MSOP8 adapter from TI his week and now I am able to do test on hardware.



    170826_test2_sboc264d.TSC

    I am trying to upload my simulation, just not sure how to upload a file here. Ok, now it worked.

    I was able to recreat my load cell with the resistances I measured. I also compensated one resistance to get me the theoretical max. output from it.

    Some components are intentionally not connected, since I did some testing trying to figure out what the problem was.

  • I have made some further experimenting and I still can't get it to work right.

    Now I get the component saturated at 5V.

    170827_test2_sboc264d.TSC

  • I am still fighting with my setup so I will update again with hope that someone will help me.

    I gave up on the 0-5VDC supply for the INA333 and I am using a -2,75/+2,75VDC supply.

    The load cell is using a 0-5VDC supply as before. However a different source and therefore different fround. However I attached the shield from the wheatstone bridge to the ground on the supply for the INA333.

    Now the thing that I can't understand.Tina now works, but the experiment on real parts is not working and I can't figure why. This should be an easy thing to do, just amplifiend a signal and measuring.

    With the +/- supply on the INA I was finally able to get the Tina schematic working (see attachment). It works just like it should.

    I assembled exactly the same schematic as in Tina. I even build two complete different protoboards to make sure it wasn't any component or problems with my assembly, but the results are the same. So I have two INA333 on different protoboards, with different components working the same and different from the Tina schematic.

    This is what I can't understand (see picture with voltage measument in RED from Osci). I get a saturated OUT voltage of 2,5V. And if I measure input singal + against input signal - I get the expected 0,01mV. However when measuring any input signal against GRND I get 2,5V just like the OUT.

     170831_test3_sboc264d.TSC

  • Hello Bruno,

    I tidied up your TINA circuit so that there weren't so many unconnected components and did away with the hard wires that could be replaced with a common ground. When I did that it was evident that the output of the INA333 was being driven into the positive rail. 

    The gain of the INA333 is set to 501 V/V, when RG is 200-Ohms. Having a 10 mV differential input voltage and a gain of 501 V/V, would result in an output level of +5.01 V, plus the VREF level applied to REF pin. Since that was +2.5 V in the schematic, the output would attempt to swing up to +7.51 V, which is above the positive +5 V supply rail and therefore, not attainable. 

    I dropped the gain to 51 V/V by setting RG to 2000-ohms, for illustrative purposes. The VREF was moved down to 2.0 V. Now using the 10 mV differential input voltage, the gain of 51 V/V, and a VREF of 2.0V, the output is at +2.510 V, which is correct. If VREF is moved down to 1.0V, the output is 1.510 V. You can see how the amplifier differential voltage sums with VREF. See the TINA Spice circuit below for details.

    If you set the gain, and VREF to values that keep the INA333 output within its linear range the simulation model should perform well for you.

    Regards, Thomas 

    Precision Amplifiers Applications Engineering

    INA333_bridge_amp_01.TSC

  • Hi Thomas,

    Thank you for answering.

    Unfortunately you worked on an earlier model I posted, not the latest one.

    Anyhow I get an error with your model. Using it "as is" I get the error message: Operating point not found. See the picture below.

  • Hello Bruno,

    Okay, I was able to reproduce the convergence issue with TINA-TI V7. I was using a later, the more powerful TINA Industrial V9 and didn't have the problem.

    The issue is realted to the default shunt conductance setting that TINA-TI V7 uses, "0." That can be an issue for some nodes inside a model becuase they are loaded by infinite resistance, which can lead to convergence issues. Setting the shunt conductance to a very small number resolves the issue. I used 1n, for 1 nano-Siemen.

    Select the "Analysis" tab across the top line of the TINA screen. Then, select "Set analysis parameters" and scroll down to "Shunt conductance [S]." Overtype the 0 with 1n. Close the Analysis Parematers box.

    When I did this, the circuit easily converged. You will likely need to do this with your more compex INA333 circuit too.

    Regards, Thomas

    Precision Amplifiers Applicaitons Engineering

  • Hi Thomas,

    Thanks for that. The TINA model is working now. Check below my screenshots.

    Despite that I still have problems with my amplifier.

    My application is a load cell and I am trying to follow a reference design from Texas Instruments called: "Single-Supply Strain Gauge in a Bridge Configuration
    Reference Design".

    I have an input signal that ranges from 0 to 10mV. I am aiming at a gain of 500x using a 5V single suply.

    When I build the system like described above (see the pictures below) I get an output signal of 0,6~0,7V independent of the input signal. Varying the input signal gets no change at all in the output.

    I have made some tests using Rg=100kohm=2x gain and the system responds as it should. I used an input signal varying from 1 to 2V and the signal is showing the desired gain on the output. For this reason I am confident that the INA333 component is working properly.

  • Hello Bruno,

    "When I build the system like described above (see the pictures below) I get an output signal of 0,6~0,7V independent of the input signal. Varying the input signal gets no change at all in the output."

    When I review your INA333 circuit I find that it is behaving as it should. It may be the difference in the common-mode and differential behaviors that is causing you some misunderstanding about the results you are obtaining from the circuit.

    VM2 indicates that the bridge is producing a differential voltage of +1.27 mV across the INA333 inputs when R3 is 293.7 Ohms. That will be multiplied by the INA333 gain of 501 V/V which will move the output off the negative rail to a level of +637.7 mV. Remember you have the REF voltage set to 0 V so the output will try to reside as close to 0 V as possible when the input differential voltage is 0 V. It can't get quite all the way to 0 V, but will move down to a level a few tens-of-millivolts of 0 V.

    When R3 is further decreased to 291.66 Ohms the input differential voltage is incread to +9.96 mV. When multiplied by the gain of 501 V/V, the output will move up to the +4.98 V. If the input differential voltage is increased any more, the INA333 output will hit the positive swing rail limit; a few tens-of-millivolts below the positive supply voltage of +5 V.

    Now if VS2 which powers the bridge is increased or decreased from +5 V, that would result in the common-mode voltage changing at the INA333 inputs. Both inputs will move up, or down, from the nominal +2.5 V level. Because of the INA333's very high input common-mode rejection its output voltage will barely move from the level attained by the differential gain conditions.

    If you set up everything correctly the INA333 should perform very well in your bridge circuit.

    Regards, Thomas
    Precision Amplifiers Applications Engineering
  • Hi Thomas,

    Thanks again for your support.
    I believe however, you didn't understand my problem completely.

    You are right when pointing out that the input voltage of 1,27mV will give me a 0,6V output just like in my screenshot above.
    My problem is that I also get 0,6V output with 0V input, with 1mV or with 10mV. In this case Rg=200 ohm.

    When I used Rg=100k ohm for testing purposes I got correct outputs (by correct I mean x2) with 1V and 2V inputs.

    Furthermore I would like to explain that I am using a single supply (0-5V) with Vref=0 because I will only excite my load cell in one direction. I believe my Vcm is correctly set at 2,5V and I don't think this is the problem with my system.
    Maybe I did something wrong when assembling the components together. The only thing I could imagine as possible problems are:
    - The power supply used is the same for the load cell and for the measuring circuit. That mean they will share ground.
    - All the components connected to ground on TINA are converging on my protoboard to the same point and that would be the ground of the power supply.
  • Hello Bruno,

    So it sounds to me that the issue you are describing occurs when the bridge isn't connected in the INA333 circuit, the 100 K resistors are connected and providing the bias current returns, and a differential voltage is applied across the INA333 set to a gain of 501 V/V? If that is the case, then I have an explanation why the output remains at about 600 mV despite the input differential voltage being anywhere from 0 mV, to 10 mV.  

    In that case, it is a matter of the common-mode voltage level. The INA333 TINA circuit shown below would be the equivalent of what I believe you are describing. Here the circuit is set for a gain of 501 V/V (RG = 200 Ohms) and a 1 mV differential voltage is applied across the inputs. Since the common-mode voltage is 0 V, the differential voltage is centered about that level. Therefore, the 1 mV is divided just about +/-500 uV about the common-mode 0 V level.

    There are a couple of things wrong with this; the INA333 input common-mode voltage lower limit is +100 mV above the negative supply voltage V- (0 V), and the inverting input is being driven to a level below ground. The INA333 cannot operate correctly under these conditions. The INA333 model's output is showing it is getting close to 0 V, but in reality it is going to hit about 600 mV and stop there. Lowering the gain to 2 V/V should not correct for the common-mode input violation so I have no explanation why you would receive correct operation.

    Once the bridge is connected into the circuit the common-mode voltage is raised to about +2.5 V and all is operating within the linear operating region of the INA333. The other things you are doing with ground and the components should be okay.

    Regards, Thomas

    Precision Amplifiers Applications Engineering

  •  Hi Thomas,

    I should be able to buy you a beer for all your help. :)

    The common-mode voltage was one of the things I struggled most about specifying this circuit. But my understanding is that I configured it correctly at Vcm=2,5V. I understand that under single-supply I need to work with Vcm=2,5V, because otherwise I would limit the max. voltage output drastically. And yeah, if my Vcm=0 I would get some residual voltage as constant signal output.

    The circuit you built does not represent what I have built here. I drew now an exactly replica of how my circuit is connected. The TINA model still works as it should and the output signal is amplified accordingly.

    The wheatstone bridge represents my load cell with their actual resistances

  • Even if my Vcm was wrong I should not be getting Vout=0,7V constant. According to the Vcm vs Vout tool I should get values that vary from 0,05 to a very small Vout max.

    Playing with Vref also did not bring any results. I always get Vout=Vref+0,7V. Even with no input I get Vout=0,7V. As long as the INA333 is supplyed it will produce a Vout=0,7.

    If I remove resistors and capacitors leaving just the INA333 I still get the same thing: Vout=Vref+0,7V.

    I believe I have to assume that the component is bad or I damaged it when soldering at the MSOP8 adapter.

  • Hello Bruno,

    The one thing we didn't really consider is the possibility of the INA333 being damaged. If you don't have another one on hand you should be able to get some samples via the product webpage.

    The real circuit results should reflect what theory and the simulation results indicate they should be!

    Regards, Thomas
    Precision AmMplifiers Applications Engineering
  • Hi Bruno,

    The INA333 TINA circuit looks good and the real one should perform very similar to it.

    Regards, Thomas
    Precision Amplifiers Applications Engineering
  • Hi Thomas,

    I have a second component to try. But maybe you can give me some advice about the best method to do the welding.
    I am using the MSOP8 adapter from TI, but I have little experience with welding/soldering such small components. I went looking for tutorials on youtube and that is what I am working with. But the INA333 is just too small and I am not comfortable with such fine precision soldering.

    I am using soldering "honey" (flux paste) together with regular soldering tin. I also used de-soldering wick/braid to remove excess tin.
    But the positioning of the component alone gives me headaches.


    I just want to get this thing working. It has been several weeks now and the amplifying of my sensor signal is just the begining of my project. There is a lot of work to be done and I am way behind schedule. Thanks for all the support.
  • Hello Bruno,

    Properly soldering small outline packages such as the MSOP can be a challenge if you don't have proper equipment. Here are some hints that may help you with the hand soldering process when attaching an INA333 to your PC board.

    1. Use a magnifier to view the INA333 and the PC board area where it will be soldered.
    2. Make sure that the soldering iron/station tip is small and doesn't dwarf the pins that need to be soldered.
    3. Select solder that has a small diameter that is comparable to the pins being soldered. Solder having a water soluble flux is receommended. 
    4. Make sure the PC board is secure so that it doesn't move when you are soldering.
    5. Use a flux pen, or brush flux on the PC board pads applying flux before positioning the INA333 on the board. Water soluble flux is recommended because it is easier to clean off using water. Other flux types will require a solvent that is compatible with them.
    6. Locate a corner pin pad on the PC board and heat it with the soldering iron. Add a small amount of solder such that the pad is tinned by the solder.
    7. Grasp the INA333 with a pair of tweezers, or a suction pen, and locate it over the PC pattern where it is to be mounted.
    8. Note the pad that has been tinned and the lead that corresponds with it.
    9. Hold the INA333 in place, against the board, and press the iron to the lead and board. The solder should flow and melt around the lead.
    10. If the INA333 is positioned as needed against the board, remove the iron and allow the solder to cool and solidify.
    11. Inspect the alignment of the INA333 with the board pattern and if satisfactory solder the opposite lead, diagonal to the one that has been soldered.
    12. Solder the remaining leads to the board.
    13. Inspect the soldering and make sure that the leads are secure without any solder bridges between them.
    14. If the soldering is satisfactory clean the flux from the pins and board using a short, stiff brush and deionized water.
    15. Blow the board dry with compressed air.
    16. If you have the ability to ultrasonic clean the board that can be beneficial for low current circuits.

    I hope this helps.

    Regards, Thomas

    Precision Amplifiers Applications Engineering

  • Hi Thomas,

    I just want to give you some feedback. Thank you for all your help.
    The amplifier is finally working.

    There was nothing wrong with my TINA circuits and all. The problem was on the component soldering.
    I solved it using a different adapter other then the one from Texas Instruments. The one from TI has very short legs and I was getting short circuits. I used a different MSOP-8 adapter and it worked like it should, right from start. Your soldering tips were also great!
  • Hello Bruno,

    Well - that is good news! I am glad I could help you get your INA333 circuit working properly.

    Regards, Thomas

    Precision Amplifiers Applications Engineering