TI has new spice models for LM3886 LM3875 et all

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Do not waste your time to simulate THD with these models. The nonlinearity is not modeled in these simple models. If it is modeled as good as the chips, anyone understand it can reverse engineer it with ease. The model needs every transistor and parasitic components to fully model the nonlinearity, more than full schematic, it is likely kept as secret. These models can only do small signal AC response, nothing else.

I do not think we talked about THD measurements, all Spice models are a crude approximation, for both discrete components, even more for analog IC, it's just a way to have a general idea of how it tends to change the behavior in amplitude, gain, frequency, phase,
or estimate the behavior of the power supply according to the current delivered and its internal impedance etc..
I agree, the only reliable way is real measure, especially if you want to be sure that the circuit is stable, but not everyone has a suitable equipment, at this point, the only way to have a generic idea is the simulation.
If not, what is the usefulness of the manufacturers to provide spice models?
 
Hello

To post it here you rename your schematic .TSC file into .txt

Thank

Bye

Gaetan


Looking inside the TSC files, they seems to me compyled and not simply text files.
This is running very basic schematic for LM4780/3886, whitout any compensation pr decopuler cap, and ideal supply.
There are only mute,input and feed forward resistors
You can start from here and add mod.
 

Attachments

  • SimplyLM3886.zip
    3 KB · Views: 362
Last edited:
To me there does not seem to be much of a point simulating the LM3886. The standard cookbook circuit is pretty much guaranteed to work. A simulation will confirm this, but is it woth the effort? Well, maybe you are building an active speaker and you need to check the frequency response before melting any solder.

On the other hand, if one is trying something out of the ordinary the world is your oyster. How about combining a whole lot of them in series/parallel to build a 1kW amp? Or using it in a class H monster? How about a tube/silicon Frankenstein?
 
To me there does not seem to be much of a point simulating the LM3886. The standard cookbook circuit is pretty much guaranteed to work. A simulation will confirm this, but is it woth the effort? Well, maybe you are building an active speaker and you need to check the frequency response before melting any solder.

On the other hand, if one is trying something out of the ordinary the world is your oyster. How about combining a whole lot of them in series/parallel to build a 1kW amp? Or using it in a class H monster? How about a tube/silicon Frankenstein?

I think if you dont use standard datasheet configuration the result is not garanted, simulation can be the first step to analize the responce.
Power is limited by voltage/impedance relation, to erogate the power you say is needed a more high supply working chipamp, or use speaker array together overture IC array.
 
Last edited:
I have confirmed the mute behavior, when pulled down to -Vee, the output is muted. When left floating, the output is enabled.

I have attached a zip with an LTSpice "asy" file which contains a schematic symbol for LTSpice, as well as an example schematic. You will need to download the pspice model from TI and put the LM3886.lib file in the same directory as the .asc and .asy files.

The distortion (refer to simulation error log) seems a little low at 0.000053%. Don't forget to turn of 1st/2nd order compression and ensure the simulation runs for an exact multiple of the input waveform period when testing distortion. The input waveform distortion is also tested; this is used to ensure that the simulation is working well enough. If the distortion of the sine source signal is not nearly zero, then your sim is broken need to tweak something.
Hey,

I've done step by step what you mentioned so I could have the LM3886 model in ltspice. But whenever I try to run your example simulation I get an error. It says Port(pin) count mismatch between the definition of subcircuit "lm3886" and instance: "xu1". The instance has more connection terminals than the definition. Anyway you think you could help me out?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.