An improvement of this is to take the feedback for the zener diode directly from the output of the opamp, and that is a common circuit for 30+ years or so. This way, the zener diode has a constant current (through a resistor from a constant voltage) and thus the zener voltage is independent of the input voltage of the circuit.

Thanks for your improvement. Show me a example circuit with you improvement.

I am not sure this will work.

The zener diode provides a stable -6.2V reference voltage in order to set an adjustable output voltage of 0 - 9.99 V.

The "Hiland Power supply" uses feedback from the output to generate a steady 10V reference.

https://duckduckgo.com/?q=hiland+power+supply+schematic&t=h_&iax=images&ia=images

If you want a variable output voltage, then you have to use another part of the circuit for a voltage divider. The "Hiland" design uses a potentiometer for that, but a few thumbwheel switches or a DAC can of course also be used. Zener voltage can be very stable, but only when fed with a constant current. As a test to verify voltage change with current, put a zener diode with a resistor on a breadboard, and then vary the current through the resistor (with an adjustable voltage source).

The "Hiland" is a copy of a schematic posted on an electronics magazine somewhere in the '70-ies, and you can buy it as a cheap kit from China these days. It is also a quite an educational project you can learn from.

Thanks for the circuit of the Hiland Power Supply.

The solution with the 10 V reference is a useful circuit.

I use the 6V2 zener diode as a reference voltage. 6.2 zenerdiodes, i.e. the 1N821, have a very low temperature coefficient.

My circuit is designed for a 12V power supply.

Yes, zener diodes around 6V have a low temperature coefficient, but they are still just diodes, and the voltage over a diode varies with the current.

Have a look at the graph on page 4 of this datasheet:

https://www.microsemi.com/document-portal/doc_view/125462-lds-0220-datasheet

At 2mA, the Zener voltage is 80mV below nominal, and at 20mA, the zener voltage is 70mV above nominal and that is a significant change.

Your ICL7662 is also just a charge pump with an unregulated output voltage. This voltage changes both with it's input voltage and with output load current.

Zener diodes only have a stable output voltage if the zener current is constant, and you have no provision in your circuit to stabilize the zener current.

Why not use a newer, more capable op amp that doesn't have the various issues that a 741 style op amp has (e.g. high saturation output of 2.5V or more, poor common mode performance, and other issues)? Then why not use a DAC to control the output? Many have built in EEPROM to maintain the settings. Then have a few commonly used settings by adding a small micro controller. You can get or make a negative supply generator to get +/-12 or 15V for better performance. You have a good beginning, but don't stop there.

This project will show you how you can use an old style opamp in a simple and useful application.

Since this is a OPAMP challenge my goal was to use an analogue circuit without any digital components.

Of course you can create a voltage reference with a microcontroller and an DAC.

BTW: there is a power supply circuit on the board in order to get +/- 12V. I used a ICL7662 charge pump IC

Wow ☺ , that's an interesting circuit to use a pushwheel switches ! from this i got an Idea ,we can use this to set voltages in a home made bench power supply to dial voltage instead of using rotory switches or potentiometer.

Your idea is perfect. 

This voltage reference project is an ideal tool for your project in order to provide a set signal.

The voltage reference is analog electronics with no microcontroller ;-)

>I used pushwheel switches in industrial systems (temperature controlling systems).

A temperature controlled system would have high inertia. I'm not sure I would risk it for a bench power supply where you might get a voltage spike in between steps.

What happens though if the switches break before make, wouldn't that cause discontinuities in the voltage going from one step to the next when the reference voltage is indeterminate during the step? A potentiometer wouldn't have this problem as it's continuous.

hmmm ,you are right Ken Yep. but iam sure there's some workaround 555techlab might be working on

An undefined range may be possible between the switching processes. But you can dampen that with a low-pass filter at the output. I used pushwheel switches in industrial systems (temperature controlling systems). I never had problems. After setting the number, they always keep the right position.

You can use a potentiometer. But standard potentiometers do not have a scale and small steps are difficult.

The pushwheel switches I used switch without any problems from one step to the other. I will prepare a screenshot from my scope.

This is a gorgeous idea and execution. I love it! Enkelt och genialt as the Swedes would say :)

Thanks :-)