Oscillation in CC mode of programmable power supply

[SHARANYA]

Hi experts
I was designing a programmable power supply 0-30v/0-4A.The control part is based totally on analog electronics.By far,my power supply is working as expected.But the main problem is the current control loop.Whenever the control loop is limiting current,I can hear humming sound from the Q1(In my schematic,it is TIP41 but in i am using MJE15032).I have captured the waveform of the Current control op-amp's output & the main output.Also the main output is fluctuating between 1V window.(For say,if i have set the output to 5v,then it is going to 6v).
The op-amp i am using is LT1013.The IC2(LM317) is powering the op-amps at 33v.The main power transistor(T1) is TIP35.Please take a look at the schematics.Is the oscillation normal?If not,then how to clear it??

[fourtytwo42]

Just to give you some clues, your integrators have uncontrolled gain AND what an earth is C9 doing ?
Do you have any simulation package like LTspice ?
I think you would find it far easier to experiment with such a tool rather than burning your fingers all the time

[SHARANYA]

I am not a specialist about analog electronics!I used C2 & C4 by trial & error method(I used to know them as compensation caps!).I got good result with these values currently used.I used c9 to smoothing the output.Should I remove it?I am working on this project during last 4 years(off & on) and in a real hurry to accomplish it!Please help....

[fourtytwo42]

Hmm then maybe you need to post in beginners but also you may be better to ask if someone has a circuit that already works you could use, control theory is not that simple.

[SHARANYA]

Please give some hints about how to modify this circuit.

[Kleinstein]

Removing C9 would be a first step. There likely also needs to be a resistor between the output of the current setting DAC and the regulating OP. There also a big problem, that the transistors adds quite some nonlinear gain to the current regulating loop. So it might be better to just use a diode to turn down the set voltage. However this will usually need a negative supply of the current regulating OP.

With these changes, the circuit slowly approaches the circuit you can find as cheap 30V/3A LNG kit - though that kit is neither good for 3 A, nor 30 V, but at least working up to about 15 V and 1-2 A.

To look a the details is really is a good idea to use something like LTspice.

[SHARANYA]

There likely also needs to be a resistor between the output of the current setting DAC and the regulating OP

Is the resistor for current limit?What value would be a good choice?

So it might be better to just use a diode to turn down the set voltage.

This part is not cleared!Where to install the diode?
I removed the cap(c9) from the system but the oscillation is getting much worse!The spikes are getting over 1v.

[Zero999]

You have far too much gain on the current limit.

I agree with the diode comment. It will need a negative supply, unless the op-amp can go to within 0.8V of the negative rail, even with a Schottky diode.

[SHARANYA]

adding a 2k resistor after the current controller dac output solved my problem!Please  now check if i have to change any other part's value!Although I blew another TIP35 during testing!!

[David Hess]

T2 adds non-linear uncontrolled voltage gain within the current control loop making frequency compensation more difficult.

The hall effect sensor is slow adding phase delay making frequency compensation of the current control loop more difficult.

The voltage control loop provides the load for the current control loop which is probably not the best configuration.

The series RC networks to ground at the output of each operational amplifier are suitable for controlling local oscillations which should not be present anyway but not appropriate for frequency compensation of a voltage output device like the LT1013.

I would start by driving the Darlington output stage with a current source from the unregulated supply and configuring both operational amplifiers to pull the base of the Darlington down to control the output.  That way a pair of diodes can be used to isolate the amplifier outputs from each other.  Then the frequency compensation of each may be adjusted; the current control loop will need to be much slower because of the hall effect sensor.  Note that removing T2 will reverse the operation of IC1G1.

If you do not want to change that much, start by adding a resistor in series with T2's emitter to control its transconductance.

[VEGETA]

Have you don't any AC analysis on the loop?

For my PSU design I didn't do one completely so I am not the right one to teach you that, I just hope someone makes a good tutorial for it. Especially how to determine compensation caps+resistors values which I did by testing values.

I recommend LTSpice because it has everything you need.

[magtux]

You will also need to test the Power Supply at all the load corners. The oscillations may occur for specific load conditions. I think you should mock it up in LTSpice and do a bode plot of the current loop to see the phase and gain margins you have.

[SHARANYA]

I would start by driving the Darlington output stage with a current source from the unregulated supply and configuring both operational amplifiers to pull the base of the Darlington down to control the output.  That way a pair of diodes can be used to isolate the amplifier outputs from each other.  Then the frequency compensation of each may be adjusted; the current control loop will need to be much slower because of the hall effect sensor.  Note that removing T2 will reverse the operation of IC1G1.

As per this info,i found a circuit on the internet.Please take a look!But the problem is this circuit is totally new to me and not my design!But anyway.....

[Kleinstein]

We did not like C9 in the old circuit, so we will not like C1 in the new circuit either. I have some doubt it will work well - not really worth a more detailed analysis. 

It's usually a bad idea to sense the current at the collector side of the power transistor.

[SHARANYA]

The hall effect sensor is slow adding phase delay making frequency compensation of the current control loop more difficult.

Previously I was using INA169 as the current sensor.But i burnt it during testing.Today I am ordering it once again from Mouser.

I have some doubt it will work well - not really worth a more detailed analysis.

I am sticking to my own design then.....

[pe6o991]

 Try searching for circuits based on LM317 with CC control circuitry.I think those are the easiest to build and to understand.Most of them will not go down to 0v cuz of the internal 1.25 REF but its easily overcome by pulling down the adj pin to -1.25(or whatever neg voltage that makes it go to 0v on the output).If you still insist on designing it yourself I've attached a image showing the simplified version of a CC CV lab power supply.You can start from there and choose your components based on your spec.

[jaycee]

As others have said, what you pretty much want is the topology the Mastech power supplies and others use.  ELV 22532 is a good schematic to look at. The basic topology is this: