[SOLVED] Dual Adjustable Power Supply with tracking - oscillates

Hi All,

The image I attach is from a 40V power supply that was published by (now defunct) Electronics Australia and it was sold by Dick Smith Electronics in Australia as kit K-3206 (Versatile 40V/3A Lab Power Supply).
It is voltage adjustable 0 to 40V and also has adjustable current limiting 0 to 3A, has over-temperature protection and ripple warning. Also, it is a dual tracking power supply.
I built the power supply some years ago, and it appeared to work fine, when doing measurements with a voltmeter. At that time, I didn't have an oscilloscope.
The PCB is home made and it is not something I can be proud of. I followed the original layout design.
What I found after using it: When turning the power supply ON/OFF, there is a large spike of voltage, with an amplitude up to 30V or so. It appears especially about 1 second after turning the power off. If the power supply is set on 5V and supplies a PIC microcontroller, when turning it off it will almost always blow the microcontroller.
Another issue is that the output has a tendency to autooscillate even for purely resistive loads. When using an oscilloscope on the outputs under load, I can see oscillation of a few MHz, especially when connecting the load between + and - (no common zero used).
In order to identify the area that generates oscillations, I connected anode of D1 (point A) at +12V, effectively disabling the current limiting function. It appears to me that IC1a+Q1+Q2 oscillates, as well as IC2a+Q3+Q4.
I also added some hundreds of pF capacitors between the inverting input and the output of IC1a, IC2a, but all I can obtain is sometimes changing the amplitude and frequency of the oscillation. I also tried various values for C4. I concentrated on the positive rail power supply (built around IC1 and Q1), which oscillates when it has a resistive load between 0 and +OUT.
One problem I solved was that IC6 was also oscillating, because the designers didn't bother to put some capacitors from the outputs to ground for the 7812, 7912 regulators.
I would build another PCB for this power supply, provided the circuit diagram has capability to work properly. Ideally, I would like to tame the oscillations on the present PCB.
I hope somebody with more experience than me would advise if this diagram is worth building.



Regards,
Nicolae

Ok, I spent some time paying close attention to the circuit and I greatly improved the way it works. First issue, related to the oscillation of the outputs while current limiting function was disabled (by supplying anode of diode D1 from 12V output of IC5). I had a 16 ohm load and I supplied about 4V. All I had to do is to add between each output and common a 22uF/63V (probably it works with lower values as well, but this is what I had available). You may notice there are no capacitors on the outputs in the diagram, and there should be. When I enabled the current limiting function, it started again to oscillate quite badly.
I found out there is a PCB design error in the published PCB (and probably on all the kits sold by Dick Smith). The RC group R36-C16 is connected between pin 7 and pin 7, instead of pin 6 and pin 7. Also, the diagram I uploaded has an error, it shows a short between pin 6 and pin 7 of IC4a, and it should most likely not be there. By adding the integrator group R36-C16 correctly in circuit, the output of IC4a became much smoother. I also reduced C4 and C5 to about a quarter than what the diagram shows (although I am not sure what the best values should be).
There was another problem. When I set a current limit (let's say 1A on 5ohm) and I turn the knob on the voltage pot to set over 5V, in theory the voltage should remain stable at 5V. In reality, I was getting an oscillation with an amplitude proportional with the setting of RV1. I cured this by adding a 10nF capacitor from cathode of D1 to ground. Next, I dealt with the overvoltage appearing at the outputs when I turn off the power supply. Using an oscilloscope I found the output going down toward zero volt, and a few seconds later, the output swings back to higher voltage than set (potentially damaging a sensitive load). I added a relay with two pairs of contacts, supplied from a 12V winding of the transformer, via a 56 ohm resistor in series with one diode (half rectification) and a 47uF filter capacitor. The only problem left unsolved is the occasional overvoltage appearing at the outputs when I turn on the power supply.

Regards,
Nicolae

I assembled one these kits when I lived in Australia circa 2000. When I returned to the United States, I modified it to use the 120 VAC mains in North America. I encountered the same problem you have--on both continents: Oscillation, typically when connected to a load that exhibits significant parallel capacitance.

The root cause is that the supply’s output circuit is built around an op amp (the TL072), and op amps are notorious for oscillating when driving capacitive loads. A more robust design may have been possible if a linear regulator had been used (e.g. the LM317), but even these can oscillate because they use an internal op amp to compare the output to a reference voltage. To give some credit to the anonymous Dick Smith power supply designers, it appears they attempted to mitigate oscillation using lead compensation in the feedback loop, evident by the parallel combination of R2 and C4 on the positive supply circuit. In any case, the bottom line is that any practical, general purpose feedback system is vulnerable to oscillation.

Having gone through this myself, I can offer a few tricks to tame your power supply:

Connect the supply output to the load through a small resistance. One Ohm might be enough. A variation of this technique is to use a power diode, such as a 1N4007. (The dynamic resistance of the diode provides the needed small resistance.) I prefer this approach because the forward voltage of the diode--about 0.7 Volts--is practically independent of the current. Thus, whatever the supply output voltage is, you know the load sees 0.7 Volts less than that.

For oscillation above, say, 100 KHz, I’ve had success installing a ferrite bead on the wire connecting the supply to the load. I’ve used small beads made from “43 material” that accommodate four turns of insulated 22 gauge wire. These are designed for VHF frequencies, but they’ve worked for me.

Why have these tricks worked? Because they’re essentially based on an op amp stabilization technique called out-of-loop compensation. With that said, I can’t guarantee success every time.

I know years have passed since your original post, but if you're still active in this news group, I'd be curious if you found other ways to handle the oscillation problem.

-Dave Drumheller