1

I've encountered a problem that I don't understand. Let's take two MOSFETs of the same model and the same batch (one with lower on-resistance (V1.T1) and the other with higher on-resistance (V1.T5)) tested on the same V_GS, V_DS, I_D.

  1. Why is the turn-on slower than the turn-off in general?
  2. Why does the MOSFET with a higher on-resistance turn on slower and turn off faster than the one with lower on-resistance?

enter image description here

Daniel Spinu
  • 19
  • 3
  • @TimWilliams All I can say they are identical SiC power MOSFETs. But this applies to basically all types of MOSFETs apparently – Daniel Spinu Feb 25 '23 at 13:52
  • The first one: compare Vgs(th). The second: I know plenty of transistors with lower Rds(on) and switching times than higher Rds(on) ones, there is no overall correlation here. – Tim Williams Feb 25 '23 at 13:56
  • @TimWilliams I forgot to mention they are identical transistors tested on a DPT with identical V_GS, I_D, V_DS – Daniel Spinu Feb 25 '23 at 13:57
  • 2
    @TimWilliams Double Pulse Test :) – Daniel Spinu Feb 25 '23 at 14:02
  • Huh. So this is some fab test or something? Beats me, ask the fab for the variance matrix? – Tim Williams Feb 25 '23 at 14:30
  • 1
    @TimWilliams It is a testing method used to evaluate the switching characteristics of transistors. What does a "fab test" mean though? – Daniel Spinu Feb 25 '23 at 14:41
  • I don't understand why you'd be testing parts, with an aim to compare fractional nanoseconds between them, unless you're characterizing new ones, i.e. the fab that designed (and intends to sell) them? ...What are you really trying to do here? Is this just testing parts for other purposes and the question is a mere curiosity about a variance? – Tim Williams Feb 25 '23 at 14:44
  • 1
    I'm writing a paper on the variance of static characteristics (R_DS(ON)) of SiC MOSFETs and if those have an impact on their dynamic performance – Daniel Spinu Feb 25 '23 at 14:52
  • 1
    Good for statistics I guess? An explanation probably isn't very meaningful without buy-in from the fab behind the parts -- the ultimate underlying cause is geometry and doping tolerances. – Tim Williams Feb 25 '23 at 15:09
  • When you say "identical" do you mean the same exact part number and batch? – Polynomial Feb 25 '23 at 19:07
  • @Polynomial The same model from the same batch – Daniel Spinu Feb 26 '23 at 14:46
  • How was the switching time measured? Were both devices tested on the exact same test fixture? Can the variance be explained by one part being plugged into the socket a bit further, leading to lower lead inductance? Were some wires positioned slightly differently, leading to different parasitic capacitances? When you're talking fractional nanoseconds, a lot of tiny details matter. – Hearth Feb 26 '23 at 16:25
  • SiC FETs do undergo a significant amount of bias temperature instability, which might be a factor here. You could also be running into subthreshold hysteresis, but that depends on what your \$V_{ds}\$ set point is for determining the 10/90 transient timings. – Polynomial Feb 26 '23 at 19:08
  • @Polynomial The chassis temperature is at room temperature. The devices did not heat a more than 2 oC due to the short (30us) of testing. The waveforms had indeed high freq oscillations and it was hard to determine the 10/90 points. However, I'm just looking to finding the theoretical reason to why the on-resistance has an impact on the switching behaviour – Daniel Spinu Feb 27 '23 at 09:39
  • @DanielSpinu BTI is about self-heating, not ambient temperature. – Polynomial Feb 27 '23 at 18:37

0 Answers0