Announcement

Collapse
No announcement yet.

Limitations of spice modelling for PI designs?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Limitations of spice modelling for PI designs?

    I am considering learning to use a flavor of Spice (5Spice, http://www.5spice.com/) and I was wondering if Spice models are appropriate for modelling the circuit of PI designs. Can it handle coils pulses (power into the coil and estimating flyback voltages and duration out)? Can Spice address issues associated with preamp saturation and recovery and give reasonably accurate delays? Thanks for any input you can provide!

    Vince

  • #2
    Originally posted by VinceC View Post
    I am considering learning to use a flavor of Spice (5Spice, http://www.5spice.com/) and I was wondering if Spice models are appropriate for modelling the circuit of PI designs. Can it handle coils pulses (power into the coil and estimating flyback voltages and duration out)? Can Spice address issues associated with preamp saturation and recovery and give reasonably accurate delays? Thanks for any input you can provide!

    Vince
    It depends on the model. Not everything is included, and some models are purposely kept simple to increase simulation speed. You would be advised to create a test circuit to investigate whether the characteristics you're interested in are contained in the model or not. If not, you can sometimes add extra components to produce the desired effect.

    SPICE simulation can be extremely useful for exploring new ideas, and you can "blow-up" the circuit as many times as you like without releasing the magic smoke. The problem(s) come when you blindly believe the simulation results without actually building anything. After satisfying yourself with the simulation, you should build the circuit and compare the results. There are often discrepancies that you must feed back into the simulation to make the two match, and in doing so you can improve your understanding.

    I can highly recommend Paul Tuinenga's book - SPICE (Guide to Circuit Simulation & Analysis Using PSpice)
    Even though it references PSpice, most other SPICE simulators follow the same syntax.
    ISBN: 0-213-158775-7

    Comment


    • #3
      The most people in DIY use LTspice. There is a vast community, and a proper forum where you may find some missing models. As all flavours of spice run on just a few not too different engines, your mileage will remain just about the same whatever flavour you pick.

      As far as PI simulation goes, yes, it works. Too perfect because not all effects are encountered, but it works.

      Comment


      • #4
        Originally posted by Qiaozhi View Post
        It depends on the model. Not everything is included, and some models are purposely kept simple to increase simulation speed. You would be advised to create a test circuit to investigate whether the characteristics you're interested in are contained in the model or not. If not, you can sometimes add extra components to produce the desired effect.

        SPICE simulation can be extremely useful for exploring new ideas, and you can "blow-up" the circuit as many times as you like without releasing the magic smoke. The problem(s) come when you blindly believe the simulation results without actually building anything. After satisfying yourself with the simulation, you should build the circuit and compare the results. There are often discrepancies that you must feed back into the simulation to make the two match, and in doing so you can improve your understanding.

        I can highly recommend Paul Tuinenga's book - SPICE (Guide to Circuit Simulation & Analysis Using PSpice)
        Even though it references PSpice, most other SPICE simulators follow the same syntax.
        ISBN: 0-213-158775-7
        I am limited because I am not an EE, but I am technically trained in another field and have a long career of doing both theory as well as "getting my hands dirty". I felt that I could get further faster if I am able to run simulations to address some of the more basic issues that would probably not concern a knowledgeable EE.

        Thanks for the book recommendation. As I am the rankest of rank amateurs in this area, I can use ALL the help starting that I can get!

        Vince

        Comment


        • #5
          Originally posted by Davor View Post
          The most people in DIY use LTspice. There is a vast community, and a proper forum where you may find some missing models. As all flavours of spice run on just a few not too different engines, your mileage will remain just about the same whatever flavour you pick.

          As far as PI simulation goes, yes, it works. Too perfect because not all effects are encountered, but it works.
          Thanks! I will take a look @ LTspice before I commit to a program. I have been doing some exploration on a number of Manufacturer's web sites and it appears that all supply Spice models for components that they build. I suspect that the difficult issue will be proper models for the "home built" components, such as the coil.

          Vince

          Comment


          • #6
            Don't get me wrong, but seeking a realistic model seldom pays for the effort in a learning stage. E.g. coils are not ideal, and they'll always have some resistance and some capacitance, but more often than not their shortcomings are exceeded by some other components' parasitics. You'll learn far more by observing an idealised model, and only after you grasp it well you'll be able to appreciate the real world parasitics and their extent. In most cases the parasite effects account for a negligible difference, and spice is a perfect platform for such comparison.

            Comment


            • #7
              I'm not a EE either and hadn't used spice until I joined this site about 2 years ago. I've been using LT spice and TINA from TI. I've been using LT spice lately for modeling the front end of the PI. Works good and I think I've learned a lot. Tried to model the MPP integrator with LT spice. I'm sure it's probably easy but I wasn't able to. Did it with TINA using a few TLC555's and switches. With all the calculations it's slow, but I think the result is correct. It's been fun and like Qiaozhi said it's smokeless.

              Comment


              • #8
                Originally posted by green View Post
                I'm not a EE either and hadn't used spice until I joined this site about 2 years ago. I've been using LT spice and TINA from TI. I've been using LT spice lately for modeling the front end of the PI. Works good and I think I've learned a lot. Tried to model the MPP integrator with LT spice. I'm sure it's probably easy but I wasn't able to. Did it with TINA using a few TLC555's and switches. With all the calculations it's slow, but I think the result is correct. It's been fun and like Qiaozhi said it's smokeless.
                It was your postings of Spice models of variations of the coil drive circuit that inspired me to consider learning Spice. While I would prefer to build and test the concepts directly I find that the delays procuring components have made modelling the designs a more than reasonable first step for the process.

                Comment


                • #9
                  Originally posted by Davor View Post
                  Don't get me wrong, but seeking a realistic model seldom pays for the effort in a learning stage. E.g. coils are not ideal, and they'll always have some resistance and some capacitance, but more often than not their shortcomings are exceeded by some other components' parasitics. You'll learn far more by observing an idealised model, and only after you grasp it well you'll be able to appreciate the real world parasitics and their extent. In most cases the parasite effects account for a negligible difference, and spice is a perfect platform for such comparison.
                  I suspect that my ignorance of the SOTA of components, and their associated non-linearities and parasitics, will be a worthwhile but frustrating experience. It has been many years since I had to build DAQ systems from scratch (first from discretes and later from modules) and I have been utterly spoiled by the fully integrated systems available off-the-shelf, at wonderful prices no less. In the interim my hands-on knowledge became horribly dated and I am truly starting from ground-zero now.

                  Comment


                  • #10
                    5Spice vs LTSpice IV

                    I have spent many hours examining these two flavors of Spice in the last day. Doing a comparison of both essentially back-to-back has led me to some unfortunate conclusions.

                    1) LTSpice has one of the most annoying, least intuitive UI that I have seen on a modern modelling program in many years. It appears that a crude UI veneer was slapped on top of a early 90's scrip-based modelling engine. What are people thinking! I have spent hours learning and have yet to complete the design and analysis of a simple opamp circuit. Very frustrating! HOWEVER, the underlying modelling engine appears to be completely up-to-date, incorporating all of the keywords used in PSpice (from Cadence), which from my exploration appears to be the gold standard of professional Spice tools. I added a bunch of Spice component models from TI and IR and the process went flawlessly. Impressive!

                    2) The 5Spice author has obviously spent a tremendous amount of time and effort developing a modern, intuitive UI. I went from DLing the program to building and analyzing my first simple opamp circuit in much less than 30 minutes. WOW! HOWEVER, this wonderful UI has been put on top of a limited and dated modelling engine. I attempted to add a number of Spice component models from LT and TI and more that half of them were rejected because the program fails to recognize some keywords that have been added in the last 7 years. Frustrating!

                    So, I am faced with a significant dilemma. I suspect that I will suck-it-up and deal with LTSpice, but I am not comfortable with that decision, so I will mull it over a bit longer.

                    Vince

                    Comment


                    • #11
                      LTSpice simulation Problem

                      This is am example of the issue I having with LTSpice. I built a simple "gain of 100" opamp circuit, but my results are obviously very wrong. I have spent a long time on it and I cannot see the error!
                      Attached Files

                      Comment


                      • #12
                        Originally posted by VinceC View Post
                        This is am example of the issue I having with LTSpice. I built a simple "gain of 100" opamp circuit, but my results are obviously very wrong. I have spent a long time on it and I cannot see the error!
                        You might want to try providing proper supply voltages to the opamp voltage pins.

                        Comment


                        • #13
                          Originally posted by KingJL View Post
                          You might want to try providing proper supply voltages to the opamp voltage pins.
                          Oh! Thanks!!
                          It works, but that is a bit kludgy (at least the way I did it [attached]). Is there an easier way??
                          Attached Files

                          Comment


                          • #14
                            Originally posted by VinceC View Post
                            Oh! Thanks!!
                            It works, but that is a bit kludgy (at least the way I did it [attached]). Is there an easier way??
                            Easier... not really... but less clutter by using Tags (Labels).

                            Comment


                            • #15
                              Originally posted by KingJL View Post
                              Easier... not really... but less clutter by using Tags.
                              What are Tags? There is no entry for tags in LTSpice that I could find and no item in the Help file either.

                              Comment

                              Working...
                              X