Electronics > Projects, Designs, and Technical Stuff
Oscillations! Linear variable PSU
Datguy123:
Hi,
Hopefully someone can help me! I've been stuck on this problem for a week.
I'm designing a linear variable PSU which can output 45V@6A. It was put together with whatever I know about electronics (I'm a rookie) and it managed to work until I started load testing with heavier loads (from a 100 ohm load to a 6 ohm load). What happened was that high-frequency oscillations (around 1 to 5Mhz) started appearing as I increased the output voltage and consequently the current. This is a prototype circuit and I have built it on perfboard and have long wires everywhere (between the reservoir capacitors and the output transistors, output transistors and feedback resistors...) presumably causing the oscillation problems. The initial oscillation was periodic and has an amplitude of around hundreds of millivolts to several Vpp which increases and starts merging (the envelopes joins together to form a constant oscillation) with the output current/voltage. The oscillation is not sinusoidal like and looks "chaotic". Furthermore, the oscillation seems to be "polluting" the entire circuit, as upon probing I realized was superimposed to all signals in the circuit, making it hard to locate its origin.
I suspected it was the output transistors causing the oscillation, so I added 1nF capacitors between the collectors of the output pairs and the heatsink (C15 and C16) as well as 4.7 ohm base stopper resistors (R2 and R6). However, the addition of the capacitors caused the oscillations to increase instead. By experimentation, I left 1nF C15 in because it reduces the oscillations (C16 does the contrary). This pushed the voltage before the oscillation starts higher, but in no way removes the oscillation. Grounding the heatsink reduces the "chaotic" behavior of the oscillation.
I started to probe around the circuit and realized that whenever I probe the collector of driver darlington transistors (either one, Q1 and Q13), the oscillations will magically go away. My initial thought was that the capacitance of the oscilloscope probe was bypassing the oscillations to ground, so I did just that by using 12pF for C13 and 1nF with 12pF for C14, which pushed the point where oscillation starts higher. It is weird that if I use 1nF for C13, the output oscillations will increase.
I've tried other things as well, listed below:
1. Smaller bypass capacitors (100pF) for C15 and C16, only reduces the amplitude of the oscillations
2. 4.7k base stopping resistors for Q1 and Q13, but resulted in way bigger and weirder oscillations (in the whereabouts of 5 to 6 Vpp at 11 volts output!). With this, oscillations are found to start from the base of Q1 and Q13 and not from the differential pair circuit.
3. 100pF C19 and C20, but does not remove the oscillation.
I am lost as I am unsure if the problem is because of the circuit design or layout. I am also unsure why adding slightly bigger capacitors would cause bigger oscillations (for C13). I hope my description of what I have done is not too messy and any help would be appreciated! Thank you for your patience because this is a rather long post.
Please note!
1. The LtSpice schematic is ONLY used to visualize the circuit
2. The operational amplifiers in the circuit are in fact LM358 (not OP747). U7,U8 are in a single package, similarily for U5 and U6
3. The circuit works by splitting the voltage drop equally between 2 pairs of transistors (X volts across Q2, Q12 and around X volts across Q3, Q11). Ideally, at 0V output the emitter voltage of Q2 and Q12 is 25V
4. -15V, +12V, and +2.78V rails are powered from the same transformer but from a separate winding.
ogden:
Where's schematics of actual circuit? Sorry, but your "distributed rats nest PSU" is full of feedback antennas, no wonder it oscillates. I won't even consider working on frequency/stability compensation for such circuit & layout unless error amp, power amp and output transistors are on compact perfboard w/o wires flapping around.
Datguy123:
--- Quote from: ogden on May 17, 2020, 05:10:40 am ---Where's schematics of actual circuit? Sorry, but your "distributed rats nest PSU" is full of feedback antennas, no wonder it oscillates. I won't even consider working on frequency/stability compensation for such circuit & layout unless error amp, power amp and output transistors are on compact perfboard w/o wires flapping around.
--- End quote ---
My bad, forgot to attach the circuit. Thanks for your advice, I'll try to clean up the layout but do you think there's anything apparent in my circuit that could attribute to the oscillation?
xavier60:
When I was rats nesting my project I kept the loop consisting of the pass transistors and bypass capacitors confined to a small area at the heatsink.
https://www.eevblog.com/forum/projects/linear-lab-power-supply/msg2028295/#msg2028295
gbaddeley:
U5 & U7 are open loop at DC, too much gain! The driver transistors should have a base stopper 22 - 100 ohms. Get rid of c10. Why are you using the 2 x transistor diff stage? There are easier ways to do level / current shifting.
Navigation
[0] Message Index
[#] Next page
Go to full version