| jackinnj |
I thought I would post this -- it's just a simulation -- the standard TL431 model doesn't incorporate all of the reactive componenets in the device -- but does have the temperature dependent resistances fitted in -- some of the TI folks had pointed this out in 2005, so here's what the bode plot should look like -- note I put in 6dB gain block to lift the x-axis -- the graph on the left is the usual TI model, that on the left mirrors reality a bit more.
 |
|
|
| Conrad Hoffman |
| Don't ya hate it when you post something cool and nobody says anything? Anyway, I don't use the 431 much, but do have a little bag of 'em to play with, and thought your improved model was more enlightening than the factory version :D |
|
|
| andy_c |
Hi Jack,
This forum tends to be a little sleepy sometimes, partly because it's not linked from the main page I think. I missed your post earlier.
Anyway, here's the schematic of another TL431 macromodel, by analogspiceman of the LTSpice user's group. Also, Helmut Sennewald has done a device-level subcircuit based on the schematics from the datasheet (which have component values). I'll post the subcircuit files of those next.
I originally had a bunch of convergence problems with the primitive TI macromodel. I've had good success with convergence using the macromodel from analogspiceman. I've also compared the frequency domain data of open-loop gain and so forth, and the macromodel seems a bit closer to the datasheet than the device-level model. If you use these along with the LTSpice loop gain probe, you can see stability issues with various capacitive loads. Kind of neat. I've found them useful anyway. |
|
|
| andy_c |
Okay, here are the models themselves. Included in the attached zip file are:
1) Macromodel subcircuit (TL431_a)
2) Device-level subcircuit (TL431A)
3) LTSpice symbol file for the above (TL431.asy) |
|
|
|
