Precision Power Supply / Elektor, December, 1982

Question

Since this PSU was published by Elektor I assume that some of you might heard or read about this project.

http://www.retro.co.za/zs1ke/projects/PrecisionPowerSupply/PrecisionPSU-Elektor-Dec-1982.pdf

This PSU is capable of delivering from 0-33V at 3A. Its voltage is stabilized, can be regulated and has current limiting control + short-circuit protection.

This is also my final school project. I made two of these PSU's in one housing. First one works perfectly fine. But second one causes a problem on the output that couldn't been solved yet.

Me and my mentor (professor) did all kinds of measuring on the not-working PSU - measuring connections, short circuit possibilities, voltage measurement on specified components and with oscilloscope too.

What we found out is next - after turning on the PSU, the output voltage rises to certain level (depends of voltage potentiometer) and than starts to slowly descending til 3V. Then the voltage changes slowly from 3V-5V. When we measured the Op-amps and LM723, on certain pins there was voltage also slowly changing. And if I connect load on the output the voltage slowly descends till 0V (everything seems like there should be a capacitor making these problems)

I changed Op-amp, LM723 and a few capacitors (every of them but smoothing capacitor - my professor said that it wouldn't be making such problems).

*Capacitors are older ones (but not used yet), all other components are new (smoothing capacitor too - the big one).

I also invested a lot of money and time into it and I don't want to end like this (with only half of it working - I made two in one with separated transformer stages so I can get negative voltage from +33V to -33V).

I hope someone has any ideas or has worked on similar project so I can solve my problem (which stayed unsolved till now).

All of these measurements were measured to the common GND of circuit in that part of circuit (stabilizer stage). This stage has a transformer source of 10V/0/10V (TR1). TR2 is power stage of 26V.

NOT-WORKING PCB

C1= 10,7V (it should be more(Uin times square root of two))

C2= 10,7V (-||-)

C3= fluctuates around 9V

C4(IC1/pin 13)= fluctuates around 10V

C5= fluctuates around 9V

C6(R7)= fluctuates around 5V

C7(IC2/pin 6)= fluctuates around 8V

C8(R14)= fluctuates around 8V

C9= fluctuates around 8V

C10(*to GND of power stage)= 35,5V

C11(R23)= fluctuates around 9V

IC1(stabilizer):

1= 0V

2= 9,6V-9,8V

3= same

4= same

5= same

6= same

7= 8,8V-9V

8= 0V

9= 4,7V-4,8V

10= 9,6V-9,8V

11= 10,7V

12= same

13= 10,6V-10,8V

14= 0V

IC2(op-amp):

1= 10,8V

2= 7,2V-7,5V

3= 5,3V-6V

4= 10,8V

5= same

6= 7V-9V

7= 10,6V-10,8V

8= 0V

IC3(op-amp):

1= 10,8V

2= 8,8V-9V

3= same

4= 10,8V

5= same

6= 10,3V

7= 10,8V

8= 0V

WORKING PCB

C1= 13V

C2= 13V

C3= 7,2V

C4(IC1/pin 13)=7,3V

C5= 7,3V

C6(R7)= 35V

C7(IC2/pin 6)= 12,5V

C8(R14)= 0,15V

C9= 0,15V

C10(*to GND of power stage)= 35,5V

C11(R23)= 0,15V

IC1(stabilizer):

1= 0V

2= 7,3V

3= same

4= same

5= same

6= same

7= 0V

8= 0V

9= 1,2V

10= 7,3V

11= 13V

12= same

13= 8,6V

14= 0V

IC2(op-amp):

1= 13,1V

2= 0,04V

3= 0,43V

4= 13,3V

5= 13,1V

6= 12,5V

7= 13,1V

8= 0V

IC3(op-amp):

1= 0V

2= 0,15V

3= 0,5V

4= 13,1V

5= same

6= 12,4V

7= 13,1V

8= 0V

Answer 1

@Keno - "Yes, heart of the problem is 100% "the changing of voltage" Right. So stop beating around the bush and tell us what the voltages ARE. And if the 723 voltages are changing, then isolate the 723 section by removing the 3 base drive components (R9, D4 and D5), and find out why the 723 is acting up. Come on, this isn't hard. Isolate the problem section, determine the error, and fix it. Stop beating around the bush with generalities.

Answer 2

I am from Brazil, I am also mounting this source (3 outputs: 2 x 0 to 36Vac >> 47Vcc @ 4A, 1x 8vca >> 9vcc @ 3A) I am also facing problems, but in the part of IC2 (voltage adjustment) already I did a lot of testing and switching components that are new, I believe the LM741 I bought is not true, but I have not yet found another reliable supplier.

My LM723 regulator works well with stable 7.1V as described in the Elektor magazine article, and by the way its not working properly.

I saw through your measurements that some voltages are way out of the way ... I would tell you Check some things:

• Possible batch of fake LM723?
• High ESR of electrolytic capacitors (did you check this?).
• Possible connection (US + and US- inverted sensor cables)?
• Instability due to the use of lower voltage supply in the 10.7V regulator circuit?

If you can put more pictures, it may be easier for someone to see something wrong and find the defect. Be sure to comment on finding the defect. Forgive me if I wrote some nonsense (I do not speak English and I use google translator). Good luck!

Answer 3

There are a couple of issues with this circuit a bit beyond the issues described by Keno. These are along the sense lines. It looks a bit like those would have been added later without much thought.

1) I'd like the instrument to show the voltage across the sense wires. Hence the V measurement (M2) needs to connect to Us- and Us+, not to U+ and U-. This would show the actual "effective" voltage on the powered object/DUT.

2) Common of the controller circuit (IC1 pin 7) is tied to Us+ which is reasonable to regulate voltage. So the actual voltage measurement is between Us+ / Us-, which is correct. The current is measured via R21 which rises above common ground (IC1 pin 7) relative to the effective current. For example: this is 0.66V for 3A. Lets assume there is a difference of 0.3V between U+ and Us+ cause by a voltage drop in the power cable and other losses. This leads to a voltage of 0.3V across R23 and 0.66V across R21. Hence R14 is at 0.96V. From that it is clear that a variation of delta Us+ and U+ leads to a different current, which contradicts most likely the design idea. I'd call that an unplausible behaviour.

There is an option to connect P2 to U+. Then IC2 with R13 / R14 compare the "right" voltages. But as P2 floats over common ground the current through R15 and P2 will vary. This will to a changing voltage to R13 too. Hence this is not a remedy.

I can only imagine a second voltage souce that considers U+ as ground as a proper solution. Between the regulated rail (IC1 pin3) and common ground should be enough "room" to implement R15 and P2 floating over U+ with ~1..1.5V of maximum sense drop. That would a proper bodge. Assuming that this should be "precision" as the circuit name imples, this fix is relativly substancial.

Maybe someone else sees a better solution?

Answer 4

Recall yesterday I commented " If LM723 Reg. output oscillates , then you have problems pulling up output. "

Clearly your unregulated supply is overloaded on both sides dropping V from +/- 13 (ok) to +/-10.7 (fail). Thus something is loading it.

This old design uses external compensation for 2nd order effects which in closed loop Op Amps causes loss of phase margin and can cause oscillation is no margin.

A common issue is the thru hole lead inductance of cap and ground connections can cause phase shift and if not compensated properly will oscillate.

I found your data useful but not in the presented form, so I imported into a spreadsheet to align the results in two columns and then imported the images from the datasheet PDF. You can show this on your thesis to demonstrate real world issues. All Power Supplies must be tested for gain-phase margin until proven stable. Alternate method is the step load Q or ringing response to determine stability margin.

Recommendations.

• Increase the value of C4 until stable up to 10x.
• Warning this will affect loop bandwidth and step load response time.