Is this a bad TVS diode SPICE model?

kellogs · July 28, 2023, 8:24pm

Hi,

New to this spice thing, just made my first simulation today (for a voltage source, a mosfet and a resistor - worked fine).

Added in this big 6.6 kW TVS diode - https://www.vishay.com/docs/89690/sm8s15a.txt

I can’t see any V_clamp hints in that document, or leakage current. Is that model complete ? my simulation of an 150V spike looks just the same with or without the diode in place. I think ngspice does load the model because of the parser lines in the output:


Warning: Model issue on line 0 :
.model xd1:df d ( trs1=0.00397025 t_measured=25 is=7e-09 rs=0.0100007 n= ...
unrecognized parameter (t_measured) - ignored
Warning: Model issue on line 0 :
.model xd1:dbd d ( t_measured=25 is=1e-15 n=1 ) ...
unrecognized parameter (t_measured) - ignored

Am I missing something ?

holger · July 28, 2023, 9:32pm

Please post your (zipped) project here, or at least the ngspice netlist, which is available by
Eeschema–>File–>Export–>Netlist…–>Spice–>Export Netlist

kellogs · July 28, 2023, 11:02pm

Attached, thanks!

min_repro.zip (5.6 KB)

holger · July 29, 2023, 9:15am

At first sight:

You have directly coupled the ngspice voltage sources to the tvs diode. The ngspice voltage sources are very strong (zero series resistance), thus they force their voltage onto the diode, even if it costs MegaAmps of current. You will need to place a resistor between voltage source and diode to see any clamping action (as in real life).

holger · July 29, 2023, 9:26am

And you have to check the pin sequence of the diode (symbol pin 1 is cathode, pin 2 is anode). With the model given the sequence is the opposite:

...
*   SPICE3
*                    anode     cathode
* Reverse direction: node 9 <- node 2
.SUBCKT sm8s15a 9 2
...

First node of the .subckt is anode, second is cathode.

So set the alternate node sequence to 2 1

And you have to check for the datasheet: What is the maximum power dissipation allowed in the diode?

kellogs · July 29, 2023, 12:16pm

Those two things did the trick!

P PPM (10 x 10 000 μs) → 5.2 kW they say in https://www.vishay.com/docs/88387/sm8s.pdf

I went ahead and modeled an automotive load dump, worst case: exp(0 89 5m 1m 0 80m) and Ri = 0.5 ohm. By just eyeballing the situation the diode will not survive it. However, I think it will with Ri = 0.75 ohm, which is in accordance to one of their application notes.

I wonder if I could model for increased ambient temperature. I also wonder why the given V_cl_max = 24.4V is not reached in my simulations: