Problem with first lab power supply

[Eldi4]

Hello there,

Recently i've tried to search some simple power supply design with cc/cv design, after some hour of searching i found this design
https://hackaday.io/project/20791-mini-bench-psu/details
It is 20V 1A linear power supply with adjustable output voltage and current and CC/CV feature, you know, some standard thing for every electronic lab supply, and since my trafo is 18V 5A (25VDC), i will maximize the capability of the trafo.

And after creating the simulation on falstad, and understanding how it work, i modified the circuit with some minor tweaking on the circuit to fit my need, and after simulating the circuit and everything looked fine, i gather the components and build one from perfboard.

After completing the circuit i noticed 2 problem there,

1. When there is no load, the output voltage can only increase, when decreased, it decrease very slowly (100mV/s), and instanly when increased (by adjusting the POT2).
2. When there is load, the output setted voltage will drop dependent on the load's current(/power?), and also dropping the maximum output voltage.

Probably there are another problems, but i don't really noticed it.

If you have the solution for the problem or suggestion on any aspect of my build, please share it there, thank you so much.

Regards Eldi4,

EDIT : the 5V rail actually came from vcc to 7805, forgot to add that on schematic.

[rstofer]

When you reduce the output voltage, that 2200 ufd capacitor holds the voltage up and there is no circuitry that will discharge it.

[IanMacdonald]

Issues I can see:

The output will remain at the last max voltage because there is no discharge path for C5/C6. Simple solution is to use a smaller value and a bleed resistor. 2200u is probably larger than needed anyway.

Your reference voltage to POT2 seems to be fed from an unregulated rail. That will not only mean varying output but it will also inject all kinds of hum and noise.  The way to avoid this is to give IC1C a gain of (say) 2, and use a zener or voltage reference IC giving half Vcc on the pot. 

Your current limiting is likely to be too slow (4.7ms time constant from R10/C1) to prevent destruction of the series pass transistor if the output is shorted. Since this is also in the voltage f/b loop it could also cause damaging voltage overshoots on load reduction.

I would be inclined to use a simpler arrangement, measuring current in the  -ve rail so no level translation needed, and a single opamp driving T1 directly.

HTH.

[rstofer]

The op amp power supply is also unregulated and I don't see the source for +5V used for the pot and the LED.

[Eldi4]

When you reduce the output voltage, that 2200 ufd capacitor holds the voltage up and there is no circuitry that will discharge it.

Issues I can see:

The output will remain at the last max voltage because there is no discharge path for C5/C6. Simple solution is to use a smaller value and a bleed resistor. 2200u is probably larger than needed anyway.

Ah i see, i removed the cap and it work, that's really small problem.
And i think it will not really affecting the overall performance of the power supply if i removed em.

Your reference voltage to POT2 seems to be fed from an unregulated rail. That will not only mean varying output but it will also inject all kinds of hum and noise.  The way to avoid this is to give IC1C a gain of (say) 2, and use a zener or voltage reference IC giving half Vcc on the pot. 

Yep, after some measuring, my transformator voltage are dropped from 18V to 17V at 1A, so it will mess up the reference voltage of POT2, probably i will use some reference voltage IC to make it more stable rather than zener or use the +5V rail instead?.

emm, did my transformator voltage drop are normal?, i mean, it dropped 1V on 1A load, even the VCC dropped from 25V to 20V!
this is my first time dealing with linear transformator by the way, so i don't really know what's happening, because i bought it 6$ from random store that i never visited before.

Your current limiting is likely to be too slow (4.7ms time constant from R10/C1) to prevent destruction of the series pass transistor if the output is shorted. Since this is also in the voltage f/b loop it could also cause damaging voltage overshoots on load reduction.

cause voltage overshoot?
But this is what Tron9000 (guy from hackaday who created this project) suggest, according to his investigation, adding cap in there will reduce the overshoot voltage and some damping in the voltage amp feedback.

But i actually did'nt really understand what's the overall effect to the performance.

And he added 10mA regulation current on the output, which i don't add it to my circuit.

I would be inclined to use a simpler arrangement, measuring current in the  -ve rail so no level translation needed, and a single opamp driving T1 directly.

Measuring the current in the emittor section?, did'nt that will do the same thing as measuring on high-side or as i did?,
I don't understand about "no level translation needed",
Did'nt i already use only single op-amp to drive T1?, or what you mean is to use another IC?

[rstofer]

And he added 10mA regulation current on the output, which i don't add it to my circuit.

If this 10 mA current was on the output, likely it was used to discharge the output capacitors when the voltage level was to be reduced.

It's not unusual for power supplies to need some small current.

ETA:  Now that I have seen the original schematic, that's exactly what that load current was used for.  That and the output capacitor was a good deal smaller.

[Eldi4]

And he added 10mA regulation current on the output, which i don't add it to my circuit.

If this 10 mA current was on the output, likely it was used to discharge the output capacitors when the voltage level was to be reduced.

It's not unusual for power supplies to need some small current.

ETA:  Now that I have seen the original schematic, that's exactly what that load current was used for.  That and the output capacitor was a good deal smaller.

So then, not using it while there are no caps won't affecting the circuit right?

[exe]

Please do not remove output cap unless you 100% sure what you are doing.

Most regulators need one for stability, it place major role in feedback. It also makes output level much more stable if load suddenly changes as power supply don't have much bandwidth. It also works as a bypass capacitor (although, it's better to put such caps close to the load).

I'd just add a small current sink. For this purposes I used lm334, simple BJT current sources, and jfets.

[not1xor1]

Hello there,

Recently i've tried to search some simple power supply design with cc/cv design, after some hour of searching i found this design
https://hackaday.io/project/20791-mini-bench-psu/details
It is 20V 1A linear power supply with adjustable output voltage and current and CC/CV feature, you know, some standard thing for every electronic lab supply, and since my trafo is 18V 5A (25VDC), i will maximize the capability of the trafo.

Forget that schematic. It looks like yet another random assemblage of electronics components.
Apart that, real components need derating, and LM324 is not a rail to rail opamp.

For instance:
-1) 1N4004 diodes are unsuitable for 1A DC output (diode manufacturers advice to derate diodes at least 20% on capacitive loads)
-2) A 18V transformer cannot provide 20VDC out as you should take into account electric line fluctuations of +/-10%, rectifier bridge voltage drop, ripple at maximum output current, drop through the current sense resistor, voltage drop trhough the pass transistor
-3) even if the TIP3055 had an hfe of 100, an LM324 would need a  stable supply voltage of at least 24V to drive its base up to about 21V and even real rail to rail opamps get close to the supply voltage rails just with low current loads.
-4) 4 220µF capacitors on the output is ridicolous
-5) there is no real need to use an LM334 as down programmer. A couple of BJTs or a LED+BJT current sink would work as well at a fraction of the cost and may be easily choosed to suit any output voltage range.
... and so on

[Eldi4]

Hello there,

Recently i've tried to search some simple power supply design with cc/cv design, after some hour of searching i found this design
https://hackaday.io/project/20791-mini-bench-psu/details
It is 20V 1A linear power supply with adjustable output voltage and current and CC/CV feature, you know, some standard thing for every electronic lab supply, and since my trafo is 18V 5A (25VDC), i will maximize the capability of the trafo.

Forget that schematic. It looks like yet another random assemblage of electronics components.
Apart that, real components need derating, and LM324 is not a rail to rail opamp.

For instance:
-1) 1N4004 diodes are unsuitable for 1A DC output (diode manufacturers advice to derate diodes at least 20% on capacitive loads)
-2) A 18V transformer cannot provide 20VDC out as you should take into account electric line fluctuations of +/-10%, rectifier bridge voltage drop, ripple at maximum output current, drop through the current sense resistor, voltage drop trhough the pass transistor
-3) even if the TIP3055 had an hfe of 100, an LM324 would need a  stable supply voltage of at least 24V to drive its base up to about 21V and even real rail to rail opamps get close to the supply voltage rails just with low current loads.
-4) 4 220µF capacitors on the output is ridicolous
-5) there is no real need to use an LM334 as down programmer. A couple of BJTs or a LED+BJT current sink would work as well at a fraction of the cost and may be easily choosed to suit any output voltage range.
... and so on

I'm sorry, but the circuit actually perform well when i power it with my 25V SMPS, and no problem really noticable, it's output voltage can reach 23.5V and the current limiting are pretty precise, so i will use SMPS instead, because my transformator seems to have some problem (it's output voltage drop is really high even on 1 amp?), i don't think i will need really low noise power supply, as i will just power some simple circuit, testing led, and some un-sensitive load.
I don't really have precise measuring instrument to perfect-ing the circuit, i don't have oscilloscope (planning to buy soon!), my multimeter is 10 year old manual ranging multimeter, but it pretty fullfil my little need.

And i use darlington transistor as series pass transistor, because my maximum output current will be 5A, which will overload the op-amp if i'm not using darlington transistor as series pass transistor.

The only problem that is left now is just tingling when i touch the power supply, probably earthing the ground will fix this problem, not a real problem.

And if you really have some another simple circuit suggestion please let me know, or suggestion to this circuit.

[Kleinstein]

The circuit does not look like it is a good idea. A agree in forgetting about that circuit, though for different reasons:
The biggest problem is that chances are high that the current regulation can, likely will oscillate with some loads. Stability could be a problem in CV mode too.

The current limit would not work below some 20 mV - so a dead short could damage the regulator. The NPN transistor does have a small residual CE voltage.

There is a realtively cheap kit that is quite often discussed.
https://www.banggood.com/0-30V-2mA-3A-Adjustable-DC-Regulated-Power-Supply-DIY-Kit-p-958308.html?cur_warehouse=UK

It is not perfect and has some issues, however with a few modifications that are discussed in several threads here and other forums, the circuit is useful. The main point is to reduce the transformer voltage to some 18-20 V - thus it would fit the 18 V transformer. With a few additions one might be able to get some 20 V at low current and with nominal grid voltage. At 1 A it is likely more like 16 V max.

[not1xor1]

I'm sorry, but the circuit actually perform well when i power it with my 25V SMPS, and no problem really noticable, it's output voltage can reach 23.5V and the current limiting are pretty precise, so i will use SMPS instead, because my transformator seems to have some problem (it's output voltage drop is really high even on 1 amp?), i don't think i will need really low noise power supply, as i will just power some simple circuit, testing led, and some un-sensitive load.
I don't really have precise measuring instrument to perfect-ing the circuit, i don't have oscilloscope (planning to buy soon!), my multimeter is 10 year old manual ranging multimeter, but it pretty fullfil my little need.

And i use darlington transistor as series pass transistor, because my maximum output current will be 5A, which will overload the op-amp if i'm not using darlington transistor as series pass transistor.

The only problem that is left now is just tingling when i touch the power supply, probably earthing the ground will fix this problem, not a real problem.

And if you really have some another simple circuit suggestion please let me know, or suggestion to this circuit.

I guess you just got a wrong feeling it works.
Typically an LM324 with 3mA of load can't get closer than 3V to the positive rail so while you might be lucky and get 23.5V without load, I bet you would not be able to get more than 21V (with really poor regulation) at the maximum load.
Besides that you can't tell anything about oscillations, transient stability, PARD, etc without an oscilloscope.

[exe]

PARD

What is PARD? Partnership on Religion and Sustainable Development?

[schmitt trigger]

Periodic And Random Noise.

[not1xor1]

The circuit does not look like it is a good idea. A agree in forgetting about that circuit, though for different reasons:
The biggest problem is that chances are high that the current regulation can, likely will oscillate with some loads. Stability could be a problem in CV mode too.

I just preferred to focus on easier to understand issues (and did not even mention all them)

There is a realtively cheap kit that is quite often discussed.
https://www.banggood.com/0-30V-2mA-3A-Adjustable-DC-Regulated-Power-Supply-DIY-Kit-p-958308.html?cur_warehouse=UK

It is not perfect and has some issues, however with a few modifications that are discussed in several threads here and other forums, the circuit is useful. The main point is to reduce the transformer voltage to some 18-20 V - thus it would fit the 18 V transformer. With a few additions one might be able to get some 20 V at low current and with nominal grid voltage. At 1 A it is likely more like 16 V max.

For low voltage values, single winding transformer PSUs, I think that a circuit like the one I quickly drew and tested on ltspice is more suitable (but I would use TL431s rather than zeners and LM385s)

It is just a plain power DC amplifier with a darlington buffer and a basic constant current sink as down programmer.
Voltage doublers are used to get two additional rails for the opamp which can work within the granted input/output limits.

From the transient test it looks like the circuit might even work without a compensation network.
Of course real circuit are always different than simulated ones.
Besides that I sketched it in haste and it is likely that there are a few mistakes.

BTW R6/R7 should be a V reg potentiometer while R9/R10 should be the current reg pot (with additional resistor to reduce the maximum current).

[not1xor1]

Periodic And Random Noise.

AKA ripple + noise