Help needed - Shunt Regulator - Doing labs in sim due to social distancing

Thread Starter

Wolframore

Joined Jan 21, 2019
2,609
School has closed and we are now meeting online and worst part is doing labs in Multisim and LTSpice. Results are ideal at best... Having issues with this one - using Ti 741 model (http://www.ti.com/product/UA741/) results are similar with other opamps. LM308 gets a little closer but it oscillates even with compensating cap in sim.

Is this a sim/model issue or actual circuit issue? Sometimes the textbook circuits don't always work. Been struggling with this one for a while. It looks like it should work.

I can work out the problems (my answers in RED) by just using the formula but I would feel better if I can get the circuit to work in sim or actually...
Using TIP31 for the NPN. I don't like the diagram for number of reasons (electron flow and zener biasing is a little low)... but it should sort of work.

1586018926920.png


1586018818368.png
result:
1586018845593.png
 

JoeJester

Joined Apr 26, 2005
4,390
The question was the change in current through the transistor,

Add a few more decimal places to see the change in load current.
 
Last edited:

OBW0549

Joined Mar 2, 2015
3,566
In your circuit with the 741 and the TIP31, you've got the op amp inputs reversed-- the overall feedback is positive instead of negative so the circuit latches up with the op amp output stuck at low.

Furthermore, you're operating the op amp outside its input common-mode range so it won't function properly; in order for it to work, the inputs must be no closer to either supply rail than 3 volts (see the data sheet, section 6.4, parameter Vicr, on page 6).

Also, the 741's output is not guaranteed to swing any closer to the negative supply rail than 3 volts, meaning that it may not be able to fully turn off the TIP31 (see the data sheet, same section, Vom parameter).

I would suggest switching to an op amp whose input common-mode range includes ground (i.e., the negative supply terminal voltage), such as an LM324 or LM358. And instead of connecting the op amp output directly to Q1's base, connect them through a voltage divider; try 2.2 kΩ and 1.0 kΩ, for starters; that should allow the op amp to turn Q1 fully off rather than leaving it in an indeterminate, partially-on state.
 

Thread Starter

Wolframore

Joined Jan 21, 2019
2,609
Ahhh thanks @OBW0549!!! that makes sense - I did try the 308 and 358 they work better but even then start to oscillate after a while.

Oops I've posted a version where I tried to switch ref and err...
 

Thread Starter

Wolframore

Joined Jan 21, 2019
2,609
Added 5V negative bias to cheat the rail issue

then added suggested divider at base - improvement - any idea about the oscillation? I get the same results when I use the 308 or 358 without the bias at Q1 base.

1586023990220.png

1586024001442.png
 

Thread Starter

Wolframore

Joined Jan 21, 2019
2,609
Got it to work... but it's touchy... not sure why - I plan on breadboarding it to see if it's this sensitive in real life - removed pull up since our issue is turning Q1 off and left pull down. Also increase zener current to try to get 5.1V. then the higher impedance control voltage helps.

1586024551689.png

1586024571677.png
 

crutschow

Joined Mar 14, 2008
34,280
it's touchy... not sure why - I plan on breadboarding it to see if it's this sensitive in real life:
It likely will be.

The transistor adds open-loop gain, which is not compensated by the op amp internal compensation.
You might try adding a small emitter resistor to reduce the gain, and a small capacitor from the emitter to base, with a 1k ohm resistor in series with the op amp output..

To check stability, run an AC simulation over the 100Hz to 10Mhz range to generate a Bode plot of the output gain and phase.
For that, add a voltage source in series with the op amp (-) input with an AC voltage of 1v.
The resulting plot should have no significant gain peaks or large changes in phase over that frequency range.
If it does than you need to tweak the feedback compensation, until it looks okay.

You can then check the transient response by having the added source generate a 10ms, 1mV pulse.
The output pulse can have some overshoot but should have no more than 1 cycle of ringing.
 

MrAl

Joined Jun 17, 2014
11,389
Hello,

What does the voltage divider do on the base of the transistor? You need a series resistor there.

You have what amounts to something that is probably only conditionally stable unless maybe you can ensure unity gain. It is hard to compensate something like this because the output load is undefined and that means the transfer function is somewhat unknown. Maybe if you can specify resistive load only we would have something concrete to go on. If you look up commercial regulators like this there is always a range of output loads given.
 

Thread Starter

Wolframore

Joined Jan 21, 2019
2,609
You have what amounts to something that is probably only conditionally stable unless maybe you can ensure unity gain. It is hard to compensate something like this because the output load is undefined and that means the transfer function is somewhat unknown. Maybe if you can specify resistive load only we would have something concrete to go on. If you look up commercial regulators like this there is always a range of output loads given.
I'm just trying to simulate my homework, we've had issues in the past with the circuits from the books not quite working like it should. Yes, makes sense to limit the output current and load resistance and would be done in practice.. this one bugs me because there are about 5 questions associated with this circuit where they make changes to values and unfortunately does not simulate as drawn.
 

ci139

Joined Jul 11, 2016
1,898
Your choice of the switch in the #6 suggest using LM301 , RC4558 , NE5534 , MC33072 or other higher output current op amp . . . the max R.base is usually 4.7kΩ for a fast responce . . . to avoid ringing try trivial neg. capacitive feedback . . . or insert (100kΩ||5pF) in series with the op amp inputs - the reason behind is similar to j-Fet input op-amps (not to "overdrive" the inputs . . . ?reduce DC gain?)
 

Attachments

Thread Starter

Wolframore

Joined Jan 21, 2019
2,609
thanks @ci139, I will add those opamps to my collection. Changing R(base) has the most impact in taming the response curve that and the R(emitter). Thanks to OBW and cruts, I think I have a better understanding. as the gain goes up, it turns into an oscillator as the bode plots look similar to the band pass filters we worked on previously. The calculations might be a little challenging as it's the internal capacitance that's causing the phase inversions and it's 2 poles, the high gain causes an asymptote at the oscillation frequency which appear to be about 2 MHz.

I used the uA741 because that's what we've been using in class. It has many issue including the fact that it's not rail to rail and it's very slow. I'm sure there are reasons why our actual lab was only the series pass and not the shunt.

Next week we are moving on to simple switching regulators based on an LM311.

I'm just having fun trying to understand things a little better. I really appreciate all the insights.
 

crutschow

Joined Mar 14, 2008
34,280
I see no useful purpose for R5.

Also 1pF is not a practical capacitor value.
The stray capacitance is likely larger than that.

Also check the transient response, as I suggested.
 

Thread Starter

Wolframore

Joined Jan 21, 2019
2,609
I know I keep playing with it... but if I take R5 out, it ruins the gain structure... if I increase C1 it does the same... transient response is great no more ringing!

removing R5:
1586035425424.png

R5 back C1 increase to 30 pF
1586035479755.png

You're right about C1 - not needed - it helped before all the changes.
1586036161455.png
 

Attachments

Last edited:

Thread Starter

Wolframore

Joined Jan 21, 2019
2,609
Here's the transient response - this is a cool test BTW. It was oscillating from just internal noise before...
Tested circuit:
1586036684410.png

RL = 500
1586036539742.png

RL=1K
1586036566351.png

RL=5K
1586036643344.png
 
Last edited:
Top