Simple DIY Linear PSU

[panoss]

I want to build a relatively simple Linear PSU having the ability of transformer taps' switching (I have a transformer with output of 30V with a center tap, so it's 0-15 and 15-30).
I 'm in the process of building this power supply. Relatively simple, has tap switching and works ok (according to the writer).
What do you think?
Can you see anything wrong with the schematic?
Maybe someone has built it?

[Zero999]

It looks reasonable to me, although the transformer will multiple secondary windings, with odd voltages might be difficult to source.

I would probably have used a comparator such as the LM393, rather than the LM324 op-amp (seems odd they're using a quad, rather than the dual LM358, as only two are used) and some PNP transistors to drive the relays.

[panoss]

the transformer will multiple secondary windings, with odd voltages might be difficult to source.

What do you mean by 'odd voltages'?
Should I stick to the exact voltages (of the transformer) shown in the schematic?
I 'm planning to use the transformer I mentioned, a 0V-15V-30V.    (that is a 30V with center tap)
And a second small transformer for the 21.7V that feeds the 7818.

[kallek]

I am not very familiar with lm723, but I still want to mention what I think. I think that even with center tap transformer, these transistors have too much heat to dissipate.

You have few options.
- Add thermal protection, the simplest solution is mechanical thermal switch.
- Use big heatsink and more transistors, and maybe transistors with lower junction-to-case thermal resistance.
- Reduce maximum current.

I would myself add a fuse to the secondary side also. Often with big transformers your primary fuse have to be quite high value so it does not blow from inrush current.

That 0.18 ohm 5W resistor have to be current sensing resistor. It will be very hot with max current and therefore not very accurate. For example, three 0.56 ohm 5W resistors in parallel would be better solution.

Edit. I would also add protection diode over pass transistors.

[coromonadalix]

This psu design is good, not perfect,   many companies use switched xformer secondaries to limit thermal dissipation at low voltage output ,,, you have some models with 5 secondaries taps ...

The lm 723 design(s) are proven, you can add some trick(s) to get it down to zero volts  etc ....  this psu as a floating supply design for the regulation section,  it give better stability.


You could help the cooling with an temperature controlled fan on top of the heatsink, or add an small temperature controlled fan driver board ...

[panoss]

I 'm trying to simulate it in Proteus 8.
I have ommited the part with LM324, for simplicity.

I 've replaced some parts that I didn't find in Proteus:
Instead of KSA940 I used ZTX753.
Instead of 2SD667 I used MPSA20.
Instead of 2SC9014 I used 2SC2547.

But it doesn't go lower than 7V.
Can you check if I do the simulation correctly? I must have done something wrong with the groundings...
I have attached the Proteus project, just change it from .pdf to .pdsprj and it will become again a Proteus project (the platform does not allow .pdsprj files to be uploaded).

[panoss]

I tested it again, it looks like working ok...!
I increased the primary inductance of main transformer (TR2) to 50H, and secondary inductance to 9H and Secondary DC resistance to 15Ohm and looks better now.
It goes down to 0V!
i don't know if this is very realistic...

[Zero999]

the transformer will multiple secondary windings, with odd voltages might be difficult to source.

What do you mean by 'odd voltages'?
Should I stick to the exact voltages (of the transformer) shown in the schematic?
I 'm planning to use the transformer I mentioned, a 0V-15V-30V.    (that is a 30V with center tap)
And a second small transformer for the 21.7V that feeds the 7818.

That seems reasonable.

I tested it again, it looks like working ok...!
I increased the primary inductance of main transformer (TR2) to 50H, and secondary inductance to 9H and Secondary DC resistance to 15Ohm and looks better now.
It goes down to 0V!
i don't know if this is very realistic...

Don't bother with the transformer. No one simulates mains transformers. Just use an AC voltage source, with a series resistor, equal to the impedance of the transformer.

Look at the data sheets for the transformers you plan to use. There will be a regulation specification, which tells you how much higher the unloaded voltage will be and can help you work out the impedance. Suppose it's 10% and is rated for 3A and the voltage is 30V. The unloaded voltage will be 10% than the rating, so 30*1.1 = 33V, a difference of 3V from the unloaded voltage, therefore R = V/I = 3/3 = 1Ohm.

[mariush]

You may want to look into transformers that use 12v secondary windings ... a 24v AC transformer with 2 x 12v windings should be easy to source.

You could get two of those and switch between 0 - 12 - 24 - 36-48v or maybe have two independent outputs up to 24v rectified

Looking at tme.eu in the 80..160va range, the most common windings are 11.5v, 12v, 18v and 24v

toroidal : really long tme.eu link

regular transformers : another long tme.eu link

example 
100va 2 x 12v toroidal for 17$ : https://www.tme.eu/en/details/tst100w_2x12v/toroidal-transformers/indel/tst-100-004/
160va 2 x 12v for 24$ : https://www.tme.eu/en/details/tst150w_2x12v/toroidal-transformers/indel/tst-160-002/

I've posted some schematics in the past, which may give you some ideas :

https://www.eevblog.com/forum/beginners/secondary-current-rating-of-a-center-tap-transformer/msg1958257/#msg1958257
https://www.eevblog.com/forum/projects/design-power-supply-30v-5a/msg1789325/#msg1789325

[panoss]

Has anybody tried the simulation?
Unfortunately, it doesn't seem to work very well.
The output voltage is not stable, for example it fluctuates from 14V to 16.4V.
I have replaced the main transformer with a Voltage Source, see the attachment (don't forget to replace the .pdf with .pdsprj)

[exe]

Do you really need a 30V supply? I mostly use voltages below 5V.  Occasionally I use 9V, 12V and rarely 15V to provide supply for devices.

I'd say a two-channel 15V psu is much more useful. If I ever need 30V I'll just connect outputs in series.

I suggest have two or four isolated low-voltage channels, rather than one beefy channel. This also helps with heat dissipation.
In terms of heat dissipation, a better approach could be to use a pre-regulator like this one: http://www.bramcam.nl/NA/NA-01-PSU/NA-PSU-SCH-21.png . I successfully build it, I set the drop out to ~3V, and it works really well. The downside is increased complexity.

Finally, make sure your transformer is rated properly. It is commonly suggested to buy a beefier transformer as VA != W.

[panoss]

I need a 30V supply. Ok, most of the times I need 5V, but not rarely I need voltages up to 25V.
I 'll consider your suggestions, of course.

I changed the values of C4 and C3 to 1n and 100n respectively, and looks very stable now!

[xavier60]

Has anybody tried the simulation?
Unfortunately, it doesn't seem to work very well.
The output voltage is not stable, for example it fluctuates from 14V to 16.4V.
I have replaced the main transformer with a Voltage Source, see the attachment (don't forget to replace the .pdf with .pdsprj)

Not saying it's a bad design, the whole auxiliary supply hanging off the LM723's reference pin, sort of, scares me.
This design doesn't need a separate winding for its control supply and can be further simplified,
 https://www.eevblog.com/forum/beginners/lm324-power-supply-with-variable-voltage-and-current/725/

[Zero999]

Has anybody tried the simulation?
Unfortunately, it doesn't seem to work very well.
The output voltage is not stable, for example it fluctuates from 14V to 16.4V.
I have replaced the main transformer with a Voltage Source, see the attachment (don't forget to replace the .pdf with .pdsprj)

I think it should work. Unfortunately, few people here use Proteus and changing the file extension will just confuse people. Zipping it to a .zip file is a more acceptable way to upload file types which are not supported by the forum.

[panoss]

https://www.eevblog.com/forum/beginners/lm324-power-supply-with-variable-voltage-and-current/725/

Have you built it? Does it work ok?

Could you please send me the Eagle files (I have Eagle 7.7.0) and the LTSpice files?
I 'll send you a pm with my email.
EDIT: I tried but you have blocked PMs?

[panoss]

Zipping it to a .zip file is a more acceptable way to upload file types which are not supported by the forum.

When I had zipped a file I was told it 'd be better not to.
Because people avoid zipped files because of viruses.

So, here it is zipped, in the attachment.

[xavier60]

https://www.eevblog.com/forum/beginners/lm324-power-supply-with-variable-voltage-and-current/725/

Have you built it? Does it work ok?

Could you please send me the Eagle files (I have Eagle 7.7.0) and the LTSpice files?
I 'll send you a pm with my email.
EDIT: I tried but you have blocked PMs?

Sorry, I don't use a simulator. The design has been thoroughly tested by the thread's opening poster.
I had built a bench supply with almost identical design which works perfectly well, https://www.eevblog.com/forum/projects/linear-lab-power-supply/375/. But I prefer the other design as it has better Under-Voltage-Lock-Out behavior.
If you decide to go down this path, you need to be prepared to do bread-board testing. I want some changes made.

[panoss]

If you decide to go down this path, you need to be prepared to do bread-board testing. I want some changes made.

By 'this path' you mean mike_mike 's design?
Why do I need bread-board testing? mike_mike has tested it already

[xavier60]

If you decide to go down this path, you need to be prepared to do bread-board testing. I want some changes made.

By 'this path' you mean mike_mike 's design?
Why do I need bread-board testing? mike_mike has tested it already.

That's right, the design that mike_mike settled on. I would like to make some minor changes. There are some parts that aren't needed like the op-amp for the CC LED and a few other bits. I like to test a design on a bread-board before making a PCB.
Testing on a simulator is ok also I guess, I have never tried it.
I can draw a schematic.

[panoss]

Draw it and I 'll try it on Proteus as it 's a lot easier than  the breadboard.
Also add the transistors (for 5A output current) in the 'Power block' as in mike_mike 's design they have been ommited.

[xavier60]

It will work with an LM358.

[panoss]

It will work with an LM358.

It looks really nice! I will simulate it and let you know.
I suppose there is some advantage in using TLC072 instead of LM358?

[xavier60]

It will work with an LM358.

It looks really nice! I will simulate it and let you know.
I suppose there is some advantage in using TLC072 instead of LM358?

Although the TLC072 is much faster, it only makes a worth while improvement to the speed of the fast current  limiting.

[panoss]

Proteus has the TL072 but not TLC072.
I 'll simulate it with LM358 as I already have it (and this is what I 'll use), while the TLC072 will take some time to come if I order it.

[xavier60]

Proteus has the TL072 but not TLC072.
I 'll simulate it with LM358 as I already have it (and this is what I 'll use), while the TLC072 will take some time to come if I order it.

The TLC072 mainly has a much higher output current which might have some affect on the fast current limiting only.