Hi Guys,

I've got a circuit working on the bench and yet the same circuit in ltspice won't run.

I've gone over this many time but simply don't see the problem.

R1 is in reality an bank of 56 LEDS but the 49K as shown in the schematic works as well.


Any help would be appreciated.

 

I don't see any source for base voltage to start Q2.

How did you arrive at the parameters for L1 and L2?

 

You've missed the important connection to the winding secondary and also the polarity of secondary winding reversed.

 

The source voltage for the base of Q2 is through the transformer coupling.

The parameters for L1/L2 were not really arrived at by design.
I just used a E-Core/Bobbin that I had and filled the first layer with #30 / 93 turns and wound a primary on top of the with #24. / 8 turns.

Works find on the bench - but now I just wanted to do a little optimizing it using LTspice.

 

You've missed the important connection to the winding secondary and also the polarity of secondary winding reversed.

L.Chung,

Thanks for your input but this is not the way it is wired and running on the bench. The sense of the winding must be reversed as I've shown in the schematic.

Edit;
The sense of the coils needs to be reversed so that Q2 has a positive polarity applied to its base as seen in the attached.

In your sim you have forced the polarity of to top of L2 negative with the additional connection you've made.

 

Found a couple things.

There is no path for the Q2 base current from the transformer secondary. I connected the top of the secondary to ground to provide a complete path.

Oscillations in simulators often don't start by themselves since there is no inherent noise to initiate them. I added a current source at the base of Q1 to inject a single 0.1ms, 1mA pulse to start the oscillations. Since a current source is an infinite impedance, it has no effect on the circuit once the single pulse is generated.

See attachment.

 

Found a couple things.

There is no path for the Q2 base current from the transformer secondary. I connected the top of the secondary to ground to provide a complete path.

Is this a spice oddity?
The circuit running on the bench has no such connection.

Oscillations in simulators often don't start by themselves since there is no inherent noise to initiate them. I added a current source at the base of Q1 to inject a single 0.1ms, 1mA pulse to start the oscillations. Since a current source is an infinite impedance, it has no effect on the circuit once the single pulse is generated.

Thanks for pointing this out. I had forgotten about that. I suppose that could also be accomplished by giving the inductor an initial value

The current waveform looks ok but the voltage waveform is well into the KV range. That would fry Q2.

Thank you for the feedback.

 

Is this a spice oddity?

No, it's an oddity of physics. All current needs a return path.

The circuit running on the bench has no such connection.

Well, that is indeed magical, since there is no path for the current to ground.

So the resistor or LEDs aren't connected to ground?

 

No, it's an oddity of physics. All current needs a return path.


Well, that is indeed magical, since there is no path for the current to ground.

So the resistor or LEDs aren't connected to ground?

That is correct.
The circuit runs just as posted.

 

That is correct.
The circuit runs just as posted.

I agree with Carl. There must be a path for the base current of Q2.

 

Guys,
I don't know what else to tell you.
It functions as drawn.

Attached is a pic of the show the transformer connections with the start & finish windings labeled. The 56 bank LED are lite though it a little had to see that in the pic.

There is no ground on the secondary side...

 

The distributed capacitive coupling between L1 and L2 in the real circuit likely provides the return path to ground for the fast rising pulses in the signal. This is not a reliable path to use. Add the ground as shown in the sim and a base resistor to limit the current.

Regards,
Ifixit

 

Yes, I had though about distributed capacitance and tried placing a capacitor at various places in the circuit but could not get the results I liked.

In case anyone is interested:

Here is what I've come up with. I've placed a centre tap on L2 and split the inductance/resistance across the halves.

I've also kept the current source at the base of Q1 to inject a single 0.1ms, 1mA pulse to start the oscillations as suggested by crutschow.
Giving L2/L3 and initial charge did not do the trick. Oscillation would start but would ring down.

This is not how the circuit was implemented on the bench but maybe this model will help in optimizing.

Thanks guys for all your input.

 

Yes, I had though about distributed capacitance and tried placing a capacitor at various places in the circuit but could not get the results I liked.

In case anyone is interested:

Here is what I've come up with. I've placed a centre tap on L2 and split the inductance/resistance across the halves.

I've also kept the current source at the base of Q1 to inject a single 0.1ms, 1mA pulse to start the oscillations as suggested by crutschow.
Giving L2/L3 and initial charge did not do the trick. Oscillation would start but would ring down.

This is not how the circuit was implemented on the bench but maybe this model will help in optimizing.

Thanks guys for all your input.

You need to add L3 to your mutual coupling:
k1 L1 L2 L3 .99

Also, change the polarity of L1. It oscillates in simulation when you do these two things.
You can remove the current source and set .ic I(L1)=1m.
Be aware that Q1 will suffer Vce breakdown if you actually have a 49k load on the secondary.

 

Did you try my modification which connected the top of the transformer to ground?