• State TI Thinks Resolved
  • Locked Locked
  • Replies 17 replies
  • Answers 4 answers
  • Subscribers 53 subscribers
  • Views 609 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

Original question:

OPA2837: Unencrypted model

OPA2837: OPA837.cir (spice model)

Joselito Go
Expert 4020 points
Part Number: OPA2837
Other Parts Discussed in Thread: OPA837, TINA-TI

Hi Experts,

Good day.

This is related to E2E: https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/1202385/opa2837-unencrypted-model

The customer comes back and mentioned below.
I get some weird DC results. According to the datasheet, the Ibias is around 700nA out of the input pins. However, in simulation, the current is 6.2uA into the input pins!
I have verified that the input levels are within the specified Vin range. I run the simulation with a single 3V supply but have also tried with +/-3V with the same result.

I will try to add some current sources to cancel out the incorrect bias currents, however, I start wondering what else is wrong.
It is essential for us to be able to design in spice, designing by building the circuits is simply not efficient enough when building nonstandard circuits.

Please investigate and advise.

Regards,

Josel

  • 0
    Guru 140200 points

    Hi Josel,

    when you open the Spice model you will see what everything is modelled and what is not:

    As you can see, the input impedance of OPA2837 is not modelled.

    Keep in mind that there's no Spice model in the world which models all parameters of an OPAmp. So, you always have to restrict yourself to a limited set of parameters. Also keep in mind that a Spice simuation is never telling the whole truth. You always need to carry out real world measurements during the development phase.

    Kai

  • 0 in reply to kai klaas69
    TI__Genius 15661 points

    Hello Josel,

    Kai's response is correct; the SPICE models are intended to be functional models which provide a starting point for evaluating devices in their application.  For further information or information not modelled in the SPICE model, the EVMs and lab measurements become absolutely imperative. 

    While it may be efficient and small in scope, the use of SPICE modeling as the sole evaluation of a device is not recommended.  I can assist you with ordering the OPA2837DGK EVM if you encounter issue doing so.  Hands-on interaction with the EVM will provide many answers to questions you would likely address in series as you went down a checklist of evaluating the device in SPICE.  The input impedance data on the datasheet is not from a SPICE sim; this parameter should be investigated in lab.

    Best,

    Alec

  • 0 in reply to Alec Saebeler
    TI__Genius 15349 points

    Hi Kai and Alec,

    Josel is out and we would like to ask further support on this post:

    Seems that it misunderstood the query here.

    The customer simulations show a bias input current that is 0.3uA to 13uA for input DC voltage 0 to 1.8V with supply 3.0V according to the datasheet it should be below 800nA.

    Hoping for your clarification and guidance. Thank you.

    Regards,
    Archie / Josel

    opa837_bias.zip

  • 0 in reply to ArTzy
    TI__Genius 15661 points

    Hello Archie,

    I have been out of office; I am looking over your sim files now.  Just to clarify, the new point of concern is the model showcasing a higher input bias current than specified on the device datasheet?

    Best,

    Alec

  • 0 in reply to Alec Saebeler
    TI__Expert 4020 points

    Hi Alec,

    The customer said that the previous response seems misunderstood him. The simulations show a biased input current that is 0.3uA to 13uA for an input DC voltage of 0 to 1.8V with supply 3.0V according to the datasheet it should be below 800nA.

    Please advise.

    Regards,

    Josel

  • 0 in reply to Joselito Go
    TI__Genius 15661 points

    Hello Josel,

    I am looking into how the simulations perform for me on my workstation.  If I have additional questions, I will reply.

    Best,

    Alec

  • 0 in reply to Joselito Go
    TI__Expert 8020 points

    Hi Joselito, 

    We looked into the model and the bias current is incorrect for the current revision. We are working on releasing a new version that will have the correct values. In the mean-time he can add a series current source on the inputs to reduce the effective bias current to the proper value. If you'd like me to send over an example please let me know. 

    Regards, 

    Jacob 

  • 0 in reply to Jacob Freet
    Guru 48395 points

    I think Jacob meant a shunt ideal current source, not series, to cancel the erroneous part of the input bias curren, be sure to check the other input as well for error in the bias current. 

    In the meantime, here is an earlier version of the OPA837 model, this is a 2017 model where i modified the Ib internally to be more correct, I must have this file as this came up before? Also, the original 2017 model might have had a better open loop Zo emulation. 

    0385.OPA837prelimRev3 modified Ib July2020.TSC

  • 0 in reply to Michael Steffes
    TI__Expert 4020 points

    Hi Michael,

    I am happy that you have found the present model is defective. I already added behavioral current sources to reduce the bias current to something in the right range. Behavioral and with an RC filter on the voltage input to as the current correction must change with the DC input voltage to be useable.
    I will try the old model sometime next week after the present easter vacation.

    Regards,

    Josel

  • 0 in reply to Joselito Go
    TI__Expert 4020 points

    Hi Michael,

    You have sent me a Tina file? I don’t know Tina, I don’t know to extract the old model from it, and I don’t want to spend time on Tina as it can’t do what I need to do.
    Please send only the old Spice model file.
    Any progress in correcting the latest model file?

    Regards,

    Josel

  • 0 in reply to Joselito Go
    TI__Mastermind 24965 points

    Hello Josel,

      Attached below is the working bias current model that Michael shared, but as a .CIR file which you may import into Spice. 

    OPA837.CIR

    Thank you,
    Sima

  • 0 in reply to SimaJalaleddine
    Guru 140200 points

    Hi Joselito,

    here the TINA-TI simulation of Michael's model of the OPA2837:

    Sima's model shows the same input bias currents.

    Kai

  • 0 in reply to kai klaas69
    TI__Expert 4020 points

    Hi All,

    Good day.

    The customer responds below.

    I finally did some tests this afternoon.
    I can not use that model, it does not work with a single 3V supply, it seems to start working when I supply it with +/- 3.3V, at supplies any lower than that it is completely dead, it even seems to short the input.

    Please go to the manager of the Opamp spice models and tell him/her that Teledyne (a huge customer at TI) needs a better model.

    Please advise.

    Regards,

    Josel

  • 0 in reply to Joselito Go
    Guru 48395 points

    Hey Joselito, before we impugn the model to much any more, can you attach your .lib file to make sure there is not some other problem

    Either Kai or I can work with that. 

  • 0 in reply to Joselito Go
    Guru 140200 points

    Hi Joselito,

    Joselito Go said:
    I can not use that model, it does not work with a single 3V supply, it seems to start working when I supply it with +/- 3.3V, at supplies any lower than that it is completely dead, it even seems to short the input.

    I cannot confirm that the model is "completely dead" at any supplies lower than +/-3.3V. For me the model works with a single supply voltage of 3V in TINA-TI. The input currents are becoming higher again, though. Maybe there's another issue with the circuit itself rather than with the SPICE model?

    As Michael already mentioned, the customer may show us his schematic, the simulation file and the .lib file.

    Kai

  • 0 in reply to kai klaas69
    TI__Expert 4020 points

    Hi All,

    Good day.

    The customer replied.

    Do you ask for my .lib? You have sent me the .cir file and the other one is downloaded from the web?
    Do you mean the design files? I am not allowed to send out design material so I can not do that.
    I do not need help doing model workarounds, I have done workarounds for the DC bias errors in the web downloaded .cir file and got it to work. But I am worried what else could be wrong when a HUGE error like that is passed to customers. There seems to be ZERO quality control of the spice models.

    However what I want is to press TI to make better models, correct models are essential to make quality design, it is too time-consuming if you have to do design the old way spending weeks at the lab bench measuring to find design errors that could have been avoided by proper simulation.

    Please advise

    Regards,

    Josel

  • 0 in reply to Joselito Go
    TI__Mastermind 24965 points

    Hello Josel,

      Either .lib or .cir file would work for us to import into Tina-TI. Instead of full design files, are you able to share a snippet of the circuit simulation or the PSpice files?

      We apologize for the inconsistency you have been facing. Starting from a few years back, we have been improving our PSpice and Tina models by modeling our new and converting our old popular models using a different architecture (from transistor based to blocks matching device specs). This is due to our devices becoming too complex for simple transistor modeling, and block based modeling will take care to matching datasheet figures. The OPA837 is one of the devices on the list to be converted/updated. 

    Thank you,
    Sima