3

The circuit and ocilloscope screenshots

I constructed this circuit in the simulator Multisim11. It should switch current 15A trough resistor R1 for PWM regulation. I need to make it work for frequency 2kHz. The function generator XFG1 on the left top simulate output from TTL technology, which control switching power on R1. All active components are faster than 4 MHz, so it should work, but it is not working for that frequency 2kHz.

As you can see on oscilloscope there are two curves. The blue one is the signal from TTL and the red one is the voltage on resistor R1. As you can see it is switched on whole time. The blue curve have minimal effect for the red one.

I tried to set lower frequency (500Hz). Here you can see the problem. Rising edge looks good there is not any delay, but there is big delay with falling edge. It keeps switched on about 0.5ms after TTL signal fell.

Circuit for Multisim11

So this is problem, does somebody know how to solve it in this circuit?

W5VO
  • 18,303
  • 7
  • 63
  • 94
MasakerCZ
  • 33
  • 3

2 Answers2

9

As-is, your first problem is that your circuit will draw crazy base currents from the BC369 PNP guy, as well as the 2N6488s. The 2N6488s certainly need a lot of base current here (100s of milliamps), but it shouldn't be as "unlimited" as your current circuit allows it to be. That's probably breaking the simulator's models, as it implies a tremendous amount of saturation charge which has to be cleared before the transistor can switch back off.

Here's your circuit in CircuitLab:

bjt pwm multi stage

I've added a few resistors (R3, R4) to help control the currents a bit. Click through to the circuit page, run the DC and time-domain simulation, and you'll see closer to tens-of-microseconds switching times -- not the 500us you're seeing. Hope that helps!

EDIT: R4 isn't super important because there's already "feedback" from the rising base voltage. But R3 really is, because it sits between Q1 and Q2, both of which have their emitters tied to opposing rails, so they don't have that same feedback mechanism.

compumike
  • 3,624
  • 23
  • 20
  • 2
    I've never seen CircuitLab before, but it looks sweet! You might consider posting it as an answer to [Good Tools for Drawing Schematics](http://electronics.stackexchange.com/questions/1024/good-tools-for-drawing-schematics). As you've probably noticed, most of our schematics are simply screenshots of desktop applications, and are therefore hard to edit. This could be a great alternative; I'll look into it more! – Kevin Vermeer Mar 28 '12 at 20:17
  • @KevinVermeer Check out the meta question about embedding schematics [here](http://meta.electronics.stackexchange.com/a/1050/319) – Kellenjb Mar 28 '12 at 20:29
  • 3
    By the way, I am really happy with how compumike is approaching the advertising of this product. I fully support advertising your product when you are actually providing good detailed answers and don't seem like a half fast attempt to slip in some links. – Kellenjb Mar 28 '12 at 20:33
  • @Kellenjb - Thanks for the pointer, I'd missed that post. I was just going to write up a meta post about this. And yes, that is an awesome way to advertise. – Kevin Vermeer Mar 28 '12 at 20:40
3

In addition to limiting currents to prevent saturation, I would suggest adding a resistor between the base of the PNP transistor and the positive rail, as well as one between the base of the NPN transistors and the negative rail. The smaller the resistors, the better the turn-off performance, though if they're too small things won't turn "on" properly. Performance might be enhanced further by adding a biasing resistor between the base of the PNP and the negative rail, and then reducing the value of the base resistor to the positive rail so that when there's no current flowing through the opto, the base will sit about 0.5 volts below the positive rail.

supercat
  • 45,939
  • 2
  • 84
  • 143