Announcement

Collapse
No announcement yet.

LM3886 Spice model

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

  • LM3886 Spice model

    Can someone guide me thru using this spice model from TI?

    http://www.ti.com/litv/zip/snam041a

    I unzipped it and tried to open some of the files and can't figure this out. Never done this before. I don't see any schematic or symbol files.

  • #2
    Also, if you see pg 21 in the LM3886 datasheet design notes, in the lower right corner, it states a "input level" at 1v. Is this peak, rms, peak to peak?? All other listed voltages and power say "peak" so they are clear. I'm trying to see what my input voltage needs to be in order to get 60watts at 4ohms with supply rails of +/- 30v.

    Comment


    • #3
      Ok I found some info on this:

      So, if my supply is +/- 30v and I wanna stay below the "limit" threshold, then my max output swing is 27v * 2 = 54v p-p.

      So at 4ohms this is: ((27 * .707)^2)/4 = 91watts RMS ?? Is that right? Seems extreme as the LM3886 datasheet states 68watts at 4ohms.

      Want to get this straight before I move on to calculating the gain.

      Comment


      • #4
        You may have to draw the symbol for the LM3886 yourself. SPICE started out as a text-only program, there's no standard for the graphical side of things.

        I wouldn't bother to model the LM3886, I'd treat it as an ideal gain block.

        The power output won't be 91W because there is some volt drop in the chip's output stage. Also, a 4 ohm load with sag-free 30V rails may be outside of the device's safe operating area, so the power would be limited by protection (or explosion)
        "Enzo, I see that you replied parasitic oscillations. Is that a hypothesis? Or is that your amazing metal band I should check out?"

        Comment


        • #5
          Ok. So my calculation is right though?

          If we were to use 68watts as stated in the datasheet we'd have a voltage output of 16.49Vrms. sqrROOT(68*4). So this is 23Vpeak and 46v p-p. I guess reading between the lines then the datasheet expects that this is safe for clean headroom?

          GAIN
          Assuming 22k and 1k feedback resistors.
          1 + (Rf1/Ri) = 1 +22/1 = 23

          So 16.49/23= 700MVrms or about 1Vpeak input?

          I'm hoping to use a 9v preamp supply so if this is correct I should be fine.

          Please verify my math and correct me here if I'm off.

          Comment


          • #6
            Yes, that seems right.

            When driving a 4 ohm load, the LM3886 starts to clip when the output is about 4V away from the rails, as you can see from the bottom left graph on page 10 of the datasheet. So, to get 23V peak output, you need +/-27V supplies. You'll be using unregulated supplies, so that 27V is the ripple trough voltage under full load. The nominal voltage will be somewhat higher, perhaps a lot higher at idle.

            By definition, when you are operating at the maximum output power before clipping, there is no "clean headroom" left. It's all used up. The headroom is the difference between the signal level you'll actually use in practice, and the signal level that causes clipping.

            The bottom right graph on page 13 shows that you can squeeze almost 100W out of the device, but this leaves no safety margin whatsoever for reactive loads (or abuse by musicians) so it can't be recommended.
            "Enzo, I see that you replied parasitic oscillations. Is that a hypothesis? Or is that your amazing metal band I should check out?"

            Comment


            • #7
              Hey will ya look at that... I did something right. :-)

              Thanks Steve for proofing my math.

              It's probably good that I'm keeping it around the 68watt range since this PA will be fed a signal similar to a 5E3 that's on 11. And not a nice clean signal as per a hi fi preamp.

              Comment


              • #8
                Here is a ready-made asc and sym for LM3886: TI has new spice models for LM3886 LM3875 et all - diyAudio.

                Jaz

                Comment

                Working...
                X