Where should I look for a Spice Mextram model for commercial transistors, such as the FMMT415?


Should I try a more advanced analog simulation software, incorporating Mextram or similar transistor modelling capabilities?


Since several weeks I have been looking for a good Avalanche Transistor Model, able to simulate an Avalanche Oscillator, for a standard Spice Simulator.

The purpose of the simulator, rather than the electrical behavior that can be readily tested by breadboarding, is to test some impedance and charge physical models over an unknown net (to be tested on field), hence the need of having a reasonably good (not perfect) knowledge of how the transistor can operate.

I found the www.diodes.com FMMT415 Avalanche Transistor Spice model, which I never was able to reproduce for the specific the avalanche effect, perhaps because:

  • The model simply do not implement the avalanche | breakdown, despite of their several switches,
  • The effect is different to what I expected (the shown curve vs the simulated one),
  • I was probably unable to reproduce it in the simulator, by putting a controlled \$V_{CB}\$ voltage source and a controled \$I_E\$ current source). Because a controlled Current Source do not allow me to put a Voltage source sharing a node simultaneously (remembering the current source carry a non-controlled voltage),
  • I should made a sweep in another variable (which one?),
  • The simulation do not converge for modelling the avalanche. Which I think is false, because neither the breakdown is observed,
  • Another totally different reason.

VCE-IE Curve for the Avalanche Effect Transistor Model Transistor Curves

The next advanced model in the scale should be the so called Mextram 503 + 504 model series, which brought to me a totally new family of simulators in answer for the Verilog-AMS models for Mextram available in Auburn.edu, at least, new for me:

At this moment I have not been able to model any of these softwares, but, I am in the right way? Should I seriously start trying these software if I care about the avalanche?

For sure, I am clear on my current need about simulating with a reasonable accuracy, just good and average but not perfect or astronomical, allowing me to have some way to predict what is going to happen in the board, and perhaps make some RF analysis or others, if I would like, in the future.

  • 1
    \$\begingroup\$ In what context are you trying to model the avalanche behaviour? Do you need it to relate to a real device or can it be a generic model? Would the VBIC model suffice? Have you considered starting with an avalanching diode first? \$\endgroup\$ – loudnoises Jun 23 '18 at 5:50
  • 2
    \$\begingroup\$ For the convergence problems did you try to add series resistance and shunt capacitance? \$\endgroup\$ – HKOB Jun 23 '18 at 5:58
  • 1
    \$\begingroup\$ LTspice claims to have MEXTRAM 504: "The model parameter "level" can be used to specify another type of BJT in LTspice. Set Level=504 to use the MEXTRAM 504 transistor due to NXP(Philips)." But the documentation is lacking. \$\endgroup\$ – HKOB Jun 23 '18 at 7:33
  • 1
    \$\begingroup\$ I am curious if this is possible within a standard spice engine. From my (limited) understanding of the algorithms used, I don't think it can use any "state" to remember if the transistor is past avalanche or not, and when it linearizes the device around the operating point it is now, it just chooses one of the two. Don't quote me on this though - I am far from a spice guru \$\endgroup\$ – Joren Vaes Jun 23 '18 at 9:22
  • 1
    \$\begingroup\$ @hyprfrcb Probably easiest, you should just write Mike Engelhardt (I can provide a direct address or you can just write to ltspice@linear.com) and ask for his advice. Second, you should contact researchers who specialize in this area (Mike probably can provide some thoughts, there.) Third, there are texts on the subject. You will then require full 3D details and associated production variations of same, including doping and annealing details that manufacturers will NOT want to provide you as it provides insights into processing details they'd like not to disclose. Best wishes. \$\endgroup\$ – jonk Jun 23 '18 at 14:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.