EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: MathWizard on April 29, 2023, 10:53:41 pm
-
I've been skimming a few pdfs on BJT junction capacitance, but from info in a datasheet, what should I be looking at to be able to do transient analysis from cutoff, to saturation, with some ODEs, and then get a reasonable prediction of the over/undershoot......so that then I can get rid of it with calculations.
I found what works good enough with the simulator, but I want some calculation's to go with it, besides just the rise/fall times, etc of the base drive, and some Thevenin equivalent of the base drive RC circuit.
And in the basic small signal model I know, it's all based on a DC operating point, when you get gm and r_pi...but what do u do for all of them changing with time too ?
So what other pdf's should I be reading to do this stuff (from data you can calculate or find easily, without going too deep into the semi-conductor physics). There's a lot of terms in the 1 below I have to learn, like what's tau_p and how does the excess minority carrier charge in the base, Q_B relate to C_pi ?
https://www.ee.columbia.edu/~bbathula/courses/SSDT/lect10.pdf (https://www.ee.columbia.edu/~bbathula/courses/SSDT/lect10.pdf)
-
I've been skimming a few pdfs on BJT junction capacitance, but from info in a datasheet, what should I be looking at to be able to do transient analysis from cutoff, to saturation, with some ODEs, and then get a reasonable prediction of the over/undershoot......so that then I can get rid of it with calculations.
I found what works good enough with the simulator, but I want some calculation's to go with it, besides just the rise/fall times, etc of the base drive, and some Thevenin equivalent of the base drive RC circuit.
And in the basic small signal model I know, it's all based on a DC operating point, when you get gm and r_pi...but what do u do for all of them changing with time too ?
So what other pdf's should I be reading to do this stuff (from data you can calculate or find easily, without going too deep into the semi-conductor physics). There's a lot of terms in the 1 below I have to learn, like what's tau_p and how does the excess minority carrier charge in the base, Q_B relate to C_pi ?
https://www.ee.columbia.edu/~bbathula/courses/SSDT/lect10.pdf (https://www.ee.columbia.edu/~bbathula/courses/SSDT/lect10.pdf)
Hi,
Are you talking about the overshoot that appears on the collector when the transistor turns off? If so, that is due mostly to the load inductance.
-
Like in a CE circuit, feeding a square wave to the base, on the collector, I see a brief negative voltage spike as the base turns off, and brief positive voltage spikes as the base turns on.
As a basic model, I replaced the BJT with a 2.2pF cap as the CB junction, and resistor, so I had 0.7V on the low side of the cap. And that gets me a similar sized plus and minus spike, as the charge on the cap does change, although it lasts about a few times longer than what a real BJT does.
I'm sure the real equations for all this go way deeper than what I was looking for. But I don't see over/under shoot mentioned in a couple of the books I have.
Correcting the over/under shoot was pretty easy in this case, but it won't be so easy for some op-amp circuits that are next. I have all the parts for a Heathkit curve tracer that I'm modifying, but it has a lot of over/under shoot in the sim's.
[attachimg=1][attach=2][attachimg=2]
-
I look forward to an answer. I long ago gave up relying on simulations to accurately model this behavior, and to be fair, it is irrelevant in most circuits.
Correcting the over/under shoot was pretty easy in this case, but it won't be so easy for some op-amp circuits that are next. I have all the parts for a Heathkit curve tracer that I'm modifying, but it has a lot of over/under shoot in the sim's.
Better curve tracers have "loop compensation" or something like that to correct for it, but I think it is dominated by the fixture capacitance rather than the transistor being tested.
-
Well I started writing notes on the semi-conductor physics, so I'll have ways to go before trying the Gummel-Poon model. Hopefully with the values in LTSpice, someone could on paper do something useful with that. Like quai-static models. I don't expect to do what SPICE does though.
In the 2-3 textbooks I have on BJT's/etc, I haven't yet found any real discussion or equation about over/under shoot.
On google, looking up a few things, I don't find an easy answer. IDK, I'll use the DC values at or before saturation, and see what they can do with a simple model of BJT's with Cpi and Cmu. I've tried calculating those C's last year or more, but not being happy with something about it.
-
Like in a CE circuit, feeding a square wave to the base, on the collector, I see a brief negative voltage spike as the base turns off, and brief positive voltage spikes as the base turns on.
As a basic model, I replaced the BJT with a 2.2pF cap as the CB junction, and resistor, so I had 0.7V on the low side of the cap. And that gets me a similar sized plus and minus spike, as the charge on the cap does change, although it lasts about a few times longer than what a real BJT does.
I'm sure the real equations for all this go way deeper than what I was looking for. But I don't see over/under shoot mentioned in a couple of the books I have.
Correcting the over/under shoot was pretty easy in this case, but it won't be so easy for some op-amp circuits that are next. I have all the parts for a Heathkit curve tracer that I'm modifying, but it has a lot of over/under shoot in the sim's.
(Attachment Link) (Attachment Link) (Attachment Link)
Hi,
Taking a quick look, it looks like you should decrease the minimum step size by a factor of maybe 1/2 or less. You should get a more detailed waveform that way.
For now, it looks like the 'real' waveform is at least 2nd order while the 'fake' one is first order.
The first order would be due (might be obvious) to the RC time constant.
The second order could be due to capacitance combined with some inductance. I am wondering if there is anything in the model or the simulator that uses an inductance or a minimum inductance value. You could check for that. It could also be the way the gain works with some capacitance base to collector or similar.
What else you could do is create a sim model yourself using a current controlled current source and some external components. Try to find an energy storage element, which when you add to the model, produces a similar behavior. The idea is to start with an 'ideal' model of some type, then try to modify it until you get a similar waveform. The modification will include either capacitance or inductance or both. You could also work this up mathematically which could be very interesting.
-
OK so you mean the timestep size in LTSPice? Yeah I've never changed that for anything, and sometimes when trying to work out RCL circuits/ equations, the plots would change slope rather abruptly compared to an ideal equation of just an 1st order RC circuit.
Thanks I'll have to try that too. I need to look up what else LTspice usually does for dealing with caps, because sometimes the plot will be off by a percent or 2 compared to the eqn's, for say at time = 1 time constant.
Like here's the circuit I'm going for, for the base drive, and if I find the Thev. eqiv time constant, for just the RC input part, and Vth at the cap or output, the sim doesn't quite match my eqn's, and I've done enough 1st order/2nd eqn's to think it's the sim doing extra stuff with parasitic.
But I could be wrong.
-
OK so you mean the timestep size in LTSPice? Yeah I've never changed that for anything, and sometimes when trying to work out RCL circuits/ equations, the plots would change slope rather abruptly compared to an ideal equation of just an 1st order RC circuit.
Thanks I'll have to try that too. I need to look up what else LTspice usually does for dealing with caps, because sometimes the plot will be off by a percent or 2 compared to the eqn's, for say at time = 1 time constant.
Like here's the circuit I'm going for, for the base drive, and if I find the Thev. eqiv time constant, for just the RC input part, and Vth at the cap or output, the sim doesn't quite match my eqn's, and I've done enough 1st order/2nd eqn's to think it's the sim doing extra stuff with parasitic.
But I could be wrong.
Yes, the minimum step size set in the transient analysis window. Decreasing usually gives a more detailed plot.