• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 2 answers
  • Subscribers 47 subscribers
  • Views 3837 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

TL974: Simple voltage follower, not following at 0V but follows just fine at 5V

Steve D9
Prodigy 20 points
Part Number: TL974
Other Parts Discussed in Thread: TL971,

I am stumped. 

This is a quad opamp wired up with a single supply (0,5VDC), feeding into the positive is the output from a DAC(no resistor, straight connection). The output is fed back into the non-inverting with no caps or resistor. It is the simplest form of a voltage follower, just like Figure 17 on the datasheet although it is not a 12V rail rather a 5V rail.

When DAC output is 4.9V, (input to ch1+) voltage follower output is 4.9V (expected), when DAC output is 0V, opamp output is 1.7V, (not expected)

I have debugged the circuit and now only have exactly as described. There are no other signals ensured though a liberal cutting of traces to isolate the problem.

To be double sure, the +input of the opamp was hard grounded and still results in a 1.7V output.

I'm at a bit of loss as to what is going on.  The wiring does not get any simpler than this, power is solid 5V, ground is a solid 0V, DC signals.

Double checked wiring and PCB layout, double checked that bug is in correctly, This problem occurs on all 4 channels as well as a second PCB. At least it is consistent.

Any suggestion on what could be happening here?

  • 0
    TI__Guru 73181 points

    Hello Steve,

    This op Amp is not RRI. The common mode voltage is limited to Vcc- + 1.15V. By setting the input to GND you are violating the input common mode minimum recommended by TI. 

    Simulation:

    The input I set up is generating a 1k sine wave 4Vpp with a DC offset of 2V. Ideally we expect to see 4Vpp however due to the non-ideal characteristics of the transistor that make up the op amp, this does not occur.

    Solution:

    To remedy your problem you can apply a negative voltage to Vcc minus (I would try -2V) or choose a different op amp with the capability of RRIO.

    EDIT: Pictures did not save.

  • 0 in reply to BOBBY
    Prodigy 20 points

    Hello and thanks for the response. Its been a while since I had to dive into the wonderful world of opamps and I think that I have forgotten more than I ever remembered!

    I attempted to recreate the spice model that you had, it turned into a little bit of a frustrating experience trying to simply import a library into Tina so that I could select the correct opamp. Any chance that there is a getting started tutorial out there that shows how to do this? I could then SPICE all by myself. I spent about 1 hour just trying to import the part! (rrrrrrrrr)

    I notice on the simulation that you have a 5V input waveform, resulting in a clipped output signal at 4V and another one at 1.something. 

    Actual hardware does not clip at 4V and goes pretty much to the rail. Spice models are ok but I find that they do not always work.

    I had thought that rail to rail means that the opamp will go rail to rail. Based your response Vicm  also plays a factor - this I did not know until now. 

    So when searching for an opamp for this application I would have to find a rail to rail opamp with a very low Vicm as well? Its all very confusing and compounded by the application circuit on the data sheet indicating a voltage follower. 

    Some additional guidance would be appreciated if you have some time.

    Cheers,


    Steve

  • 0 in reply to Steve D9
    TI__Guru 73181 points

    "I attempted to recreate the spice model that you had, it turned into a little bit of a frustrating experience trying to simply import a library into Tina so that I could select the correct opamp. Any chance that there is a getting started tutorial out there that shows how to do this? I could then SPICE all by myself. I spent about 1 hour just trying to import the part! (rrrrrrrrr)"


    TI does have a document which walks through the importing process found here: 

    I personally try to avoid importing models if I can because it can sometimes be a hassle. The TL974 did not have a model in it's software section but I saw that the TL971 was the same device but single channel and went to that page for the spice model. Instead of importing it, I just opened the reference design and copy and pasted the design into file. Below: just download the reference design and open with TI TINA and done.Do note the model does not look like the triangle model we normally see but as the real world device in TI TINA. I had a colleague modify it to be the triangle to help explain what was going on in the previous answer.

    "Actual hardware does not clip at 4V and goes pretty much to the rail. Spice models are ok but I find that they do not always work."

    I actually had quite a bit of trouble working with this specific model and found it to have convergence issues. I have submitted it for rework. I agree with you on the fact spice models may not always work however can be useful tools to verify work before building.

    "I had thought that rail to rail means that the opamp will go rail to rail."

    RRIO will accept close to and/or above and below supply rails and output voltages close to its rails never past, for the output it will never reach the upper supply rail due to voltage drops across the output transistor. RRI will only accept voltages close to it's supply rails and in some topologies allow for above and below supply rails but not output to the supply rails. RRO allow for the vice versa.

    Great material on this can be found here:

    "So when searching for an opamp for this application I would have to find a rail to rail opamp with a very low Vicm as well? Its all very confusing and compounded by the application circuit on the data sheet indicating a voltage follower. "


    Yes, if you did not want to make changes to your circuit then choosing a RRIO op amp would be necessary. I have submitted the issue on the datasheet being misleading/incorrect on the application section to the system so hopefully it should be updated and no one else will have issues on this in the future.

    "Some additional guidance would be appreciated if you have some time."


    For your application because you have a PCB made already I think switching out the op amp would be the easiest approach to the problem.

    The link below can be used to search for an op amp.  Be sure to check off the boxes below (in to V- and out):

    From here you would have to filter on what parameters are important for your application. With a DAC with large signal output you may care about Slew Rate and GBWP which will depend on your max switching frequency. With high frequency application, you may be concerned about broadband/white noise and may want a low noise op amp.

    Alternatively, you can try to use the TL974 and apply a bias voltage to the input  but this will cause the output to shift up by that bias voltage and you have to limit your input voltage to be lower than 5V as I believe the op amp is actually clipping at the output and is saturated/in overdrive. In some applications this may be important to avoid due to the time it takes to recover from overdrive.