77Views15Replies

Author Options:

NPN BJT strange behavior? Answered

My linear power supply design has been using a PNP MJE2955 pass transistor wired to a 2N4401 as a complementary darlington configuration, but this was not an inherently stable design, and would oscillate when adding output smoothing capacitors, particularly MLCC types (with very low ESR).

So I decided to try the far more common darlington arrangement, although I am not fond of the higher voltage drop. However the output was oscillating at approximately 3KHz with 1uF output capacitance, with NO input to darlington. I removed the 2N4401 so all I had left was just a 2N3055 transistor with the emitter connected to the output and the collector connected to a 20V supply. The output was ringing like crazy. The oscillation frequency could be changed by adding/removing output capacitance.

Why is this happening? How should I go about stopping this oscillation? I originally had a 4.7k resistive load on the output but it did not significantly affect the output oscillation.

Tags:NPN

Discussions

0
None
steveastrouk

Best Answer 2 years ago

Sounds like Miller effect feedback.

0
None
-max-steveastrouk

Answer 2 years ago

What is particularly strange is that the NPN is in the emitter follower configuration, so the input to the base is in phase with the emitter, I would not expect to run into the miller effect in this situation. In fact, I have never seen such behavior from such a familiar circuit before! I tested both a TIP31A and a MJE3055, simalar results.

0
None
Downunder35m-max-

Answer 2 years ago

Might not help you at all but I give it shot anyway:
On my old and analog lab supply I had similar problems that I only noticed once I started to play with my oscilloscope.
The "overlays" were never a real issue for anything I had connected so I never bothered checking that thing LOL.
One day the lab supply gave up on me.
While still showing the correct voltage the output was way off and the added noise on the DC was just terrible.
In my supply there were small capacitors from the gate of the darlington driver to ground and also a small inductivity in series with the gate of the driver.
Did not measure it, was a short, 5mm ferrite rod with 8 windings on it.

I suspect this weird behaviour might be based on the capacitance of the power transistors.
Meaning that it causes a feedback loop voltage back to the opamp.

0
None
-max-Downunder35m

Answer 2 years ago

I totally appreciate the story! :)

I would first suspect dry or cracked solder joints on the output smoothing capacitors, or caps that dried out over a long period of time. Strange they had to bother with so PID tuning the gate of the darlington. In my experience that is a pretty stable circuit to deal with. Just apply a fixed voltage to the gate and the output resistance will be more-or-less constant and clean.

Complimentary darlingtons are the same way until you introduce an op-amp, which increases the propagation delay (this type of phase shift is EVIL!) and gain (of the transfer function before feedback) just enough to introduce all sorts of problems! Quite a bit harder to stabilize without more advanced knowledge.

-

I have already realized my dumb mistake, I reversed the emitter and collector, and what I think was happening was emitter-base junction (subjected to the full 20V supply) would incur an avalanche or zener breakdown, causing the voltage to spike at the base, turning the transistor on, causing the voltage at the output to rise sharply, and due to output capacitance and a small load, that would take awhile to discharge, then the process repeats. It's actually a pretty good/strong oscillation so long as the input voltage is high enough. Not sure how long the transistor would last like that though.

0
None
-max-steveastrouk

Answer 2 years ago

I should mention that the oscillation is a ramp wave. It shoots high, then steadily linearly falls a few volts then shoots up again fairly fast.

0
None
steveastrouk-max-

Answer 2 years ago

What's your power rail look like ? That sounds like a sort of relaxation oscillation.

0
None
-max-steveastrouk

Answer 2 years ago

That was my suspicion too, considering the long leads running to the supply and a 200mA current limit. It is mostly stable and the circuit is not drawing a large about of current.

I later noticed a constant current load circuit I forgot to remove, upon removing it the oscillation stopped... until I added a 100K load resistor, and it came back with vengeance at 20Hz. :O The oscillation can be set by the load resistor and the 'smoothing' capacitor. Any ideas?

0
None
steveastrouk-max-

Answer 2 years ago

You're getting feedback somehow ! Are we in real circuit or LTspice ?

0
None
-max-steveastrouk

Answer 2 years ago

No, I cannot replicate this behavior in LTspice. I messed with adding parasitic capacitors between different points to see if I could achieve behavior like this. I cannot. I might have connected the E and C backwards, I think E is the center pin, correct?

0
None
steveastrouk-max-

Answer 2 years ago

I've only ever seen 2n3055 in TO3 cans. There isn't a middle pin.

0
None
-max-steveastrouk

Answer 2 years ago

MJE3055** It is in a TO220 package. I also used to have a TIP3055 in the oldscool TO218 package.

0
None
iceng-max-

Answer 2 years ago

The center TO-220 NPN pin is the Collector

TO-220 pinout.gif
0
None
-max-steveastrouk

Answer 2 years ago

OK, I goofed, the center pin is actually collector [facepalm!] That's what I get for being lazy and using the skeleton for the original complementary darlington wiring as a skeleton for the darlington. However this oscillation fascinates me. Maybe I made a relaxation oscillator due to low reverse breakdown voltage of the emitter-base junction (???) This would explain why I get oscillation only at 18V or higher...

0
None
-max-steveastrouk

Answer 2 years ago

So now I have a nice stable supply, however, I cannot seem to get the output voltage to remain stable. The equivalent output impedance is unacceptable, the voltage sags by over 20% with a moderate load. I used a brute 12v (15.7v unloaded) power supply so I doubt that was an issue. I cannot measure the same voltage dip on the inverting input of the opamp, so it's doing it's job as far as I can tell, and I am using a 10k pot for the feedback path. Any ideas? Is the input of my opamp actually drawing current significant enough to screw with the output like this?