Looking for a schematic/advice for 0-20V, 3A bench PSU

[John_Edward]

I'm in a need of a variable PSU, after getting frustrated with an ATX PSU and a 3$ regulator thingamajig from china.
Don't really have money to buy one, but I happen to have availability to a ton of free components and a PCB mill courtesy of my school.
The problem is that I don't have any experience designing circuits.
I have found these two schematics while researching this, but don't know if they are any good, or even if they provide what I need...

http://diyfan.blogspot.fi/2012/02/adjustable-lab-power-supply.html
http://www.thebackshed.com/windmill/articles/VariablePowerSupply/TheBackShed723Reg.gif


The requirements would be 0V to 20V, around 3A and variable current control.
I like the simplicity of potentiometers, so analogue Coarse&Fine adjustment pots are good.

For displays, we have some basic voltage ones at school, but for current, I've been thinking about this amazing thing:
So very precise current control would be pretty neat.
If possible, I would also like if the current limit could be seen and set beforehand, with a slightly more advanced way than "just short the leads."
Maybe a button that shorts the output for you, high tech!
Those Constant Voltage and Constant Current LEDs could also be nice, but not really required.

As playing with mains AC doesn't sound very appealing, I would power it with a 24V "90W" laptop power brick.
If the brick is what it says, would the 4V and 0.75A be enough for the supply ?

Does anyone here happen to have made one like this, or know a good schematic for one ?

[Thor-Arne]

http://diyfan.blogspot.fi/2012/02/adjustable-lab-power-supply.html

I've been building something quite similar to this one, but I haven't had time to debug it yet.
Seems like there is some issues in the original schematic.

For displays, we have some basic voltage ones at school, but for current, I've been thinking about this amazing thing

A couple of things with these displays:

• You need a 5V rail for those.
• And they often doesn't have floating ground so you need the same ground as the output.

So very precise current control would be pretty neat.
If possible, I would also like if the current limit could be seen and set beforehand, with a slightly more advanced way than "just short the leads."
Maybe a button that shorts the output for you, high tech!

Usually one uses a mcu for presetting this without the load, and that's a different design.

As playing with mains AC doesn't sound very appealing, I would power it with a 24V "90W" laptop power brick.

If the brick is what it says, would the 4V and 0.75A be enough for the supply ?

Does anyone here happen to have made one like this, or know a good schematic for one ?

You really should get a proper transformer, there is no way you'll get 0-20V/3A from that brick.
The mains part isn't complicated in this case, just use the proper fuses for the transformer.

As said; I've been building one, but there's some testing to be done before I can say it's good or not.

The design with the 723 should be the simplest, and might be the cheapest one to build.
I've use 723's before and they are nice little chips.

[John_Edward]

A couple of things with these displays:

• You need a 5V rail for those.
• And they often doesn't have floating ground so you need the same ground as the output.

If i understood the (confusing) reference image correctly,

be the correct way to connect them to the LM723 circuit ?
Adding a 5V regulator would probably be a good idea, but anyway.

Usually one uses a mcu for presetting this without the load, and that's a different design.

If I would short the output, then wouldn't the current cap at the limit ?
I don't need the limit to be always visible, just when setting it.
Adding a 6.7ohm 60W resistor bank would be more healthy for the circuit though..?

You really should get a proper transformer, there is no way you'll get 0-20V/3A from that brick.
The mains part isn't complicated in this case, just use the proper fuses for the transformer.

20v @ 3A is only 60Watts.
There shouldn't be too big of a voltage drop from the 24V when there is still 30Watts 'left' ?
Or is 24V too low in the first place to get 20V with any power ?

The design with the 723 should be the simplest, and might be the cheapest one to build.
I've use 723's before and they are nice little chips.

The circuit has four transistors, each behind a 5Watt 0.1ohm resistor,But they then go through two 0.47ohm 5W resistors in parallel.
Doesn't that limit the circuit to 10 Watts, or those resistors burn ?

Thanks for the help, and sorry for all the questions

[Thor-Arne]

If i understood the (confusing) reference image correctly,

be the correct way to connect them to the LM723 circuit ?
Adding a 5V regulator would probably be a good idea, but anyway.

As far as I can see, yes, assuming the displays really can handle the supply voltage.
Normally these doesn't have a internal regulator and requires 5V supply.
Make sure they have, or drop inn a 5V regulator.

If I would short the output, then wouldn't the current cap at the limit ?
I don't need the limit to be always visible, just when setting it.
Adding a 6.7ohm 60W resistor bank would be more healthy for the circuit though..?

It should be ok to short the output if the design is good, but don't take my word for it. 

20v @ 3A is only 60Watts.
There shouldn't be too big of a voltage drop from the 24V when there is still 30Watts 'left' ?
Or is 24V too low in the first place to get 20V with any power ?

Depending on how the voltage is measured 24V might be enough.
Like on transformers the output voltage can be measured with or without load, if it's  24V at full load then it might be enough.
Only way to make sure is to try or read the datasheets.
Trying wouldn't harm the circuit.

The circuit has four transistors, each behind a 5Watt 0.1ohm resistor,But they then go through two 0.47ohm 5W resistors in parallel.
Doesn't that limit the circuit to 10 Watts, or those resistors burn ?

The 0,1 ohm resistors are for balancing the differences in the transistors, and the 0,47 ohm resistors is the current sense resistors.

It's a little late to calculate ohm's law right now, as far as I can see 24V/2A should dissipate 1,88W in the 0,47 ohm resistor and 4W in the 0,1 ohm resistor.
I'm sure someone will correct me if I'm wrong on this. 

Thanks for the help, and sorry for all the questions

You're welcome 

[amspire]

I wouldn't use the Lm723 chip. The Fairchild uA723 was amongst the earliest analog ICs - it iwas released almost 50 years ago and the performance is not great. It is the same vintage as the uA709 opamp and be very greatful you will never have to use a uA709.

The 723 gain is limited so it is not a great regulator, and you cannot adjust down below 2v. If you are going to build a power supply, it is far too much effort to end up with a fairly poor supply that cannot regulate to 0V.

The 723 current limit is very crude and it will drift noticeably as the supply warms up. The use of a power supply as a stable current source is a very useful feature.

The DIY circuit is fat superior. I would absolutely go with it and forget the 732 circuit.

It takes a really good heatsink to dissipate the power in the 2N3055 and so I would recommend multiple 2N3055 devices with 0.1 ohm emitter resistors (as in the 723 circuit) to replace a single 2N3055. You will need either a big passive heatsink, or a good CPU cooler can handle the power as well. The driver transistor needs to be on a heatsink as well.

The DIY design does go down to 0V, it does have a good quality current limit, it has a much higher gain regulator so you will get better DC regulation, it looks like it has proper frequency compensation. It looks like power on transients has been considered. I think it is probably not a bad design at all.

You could replace Ic1, D8, R4, R5 and R5 with a modern 5V voltage reference for better stability.

Richard

[Thor-Arne]

I wouldn't use the Lm723 chip. The Fairchild uA723 was amongst the earliest analog ICs - it iwas released almost 50 years ago and the performance is not great. It is the same vintage as the uA709 opamp and be very greatful you will never have to use a uA709.

So it's that bad?

As I see it it's an easy way to make something useful, that has a very simple schematic and board layout.

But anyway, I'll take your word for it.

I didn't go for the 723 solution myself because I wanted something I could probe around in and actually learn a few things.

[amspire]

It would be useable, but the diy design is probably so much better that it is worth the effort. The 723 circuit looks simpler, but building a case, heatsink, mounting a big transformer, meters, switches, potentiometers,  etc is a far bigger task then a few extra components. To put all that work into a supply that couldn't be used to power a 1.5v device  is nuts.

[Thor-Arne]

It would be useable, but the diy design is probably so much better that it is worth the effort. The 723 circuit looks simpler, but building a case, heatsink, mounting a big transformer, meters, switches, potentiometers,  etc is a far bigger task then a few extra components. To put all that work into a supply that couldn't be used to power a 1.5v device  is nuts.

I agree to this.
I spent like forever on making a case for the one I'm working on since I bought a proper steel psu case. 

However, making a pcb can be a challenge for someone that haven't done this before.

[mariush]

(wrote this before the forum was reinstalled... thanks to Lazarus extension i can post it again)

For a few of their linear adjustable power supplies, Circuit Specialists has the circuit datasheets posted directly on the page with the units.

For example, I'm attaching the datasheets they had for single output, 1-5 a power supplies and one with 2 outputs... i've saved them to do my own research on the designs.

They look ok to my "untrained' eye, just slightly  more complex on the control side due to using buttons on the front coupled with a digital potentiometer.  If you tweak the design to use a standard ten turn pot and replace the custom lcd with your own mcu or another multimeter chip with 7 segment display (they use icl7106 which has lcd display driver, icl7107 is available and drives seven segment digits directly and there's even ready made boards on ebay) so it should be easy to make them.

( The 1x30v 5A is probably the most detailed and easy to follow, so I'd suggest starting with that one...  )

[amspire]

It would be useable, but the diy design is probably so much better that it is worth the effort. The 723 circuit looks simpler, but building a case, heatsink, mounting a big transformer, meters, switches, potentiometers,  etc is a far bigger task then a few extra components. To put all that work into a supply that couldn't be used to power a 1.5v device  is nuts.

I agree to this.
I spent like forever on making a case for the one I'm working on since I bought a proper steel psu case. 

However, making a pcb can be a challenge for someone that haven't done this before.

If making a PCB slows you down, don't. You can build a perfectly good controller on veroboard/stripboard and at any time in the future, you can make a neater PCB regulator board to replace it. A stripboard controller though can be completely reliable and functional and you can put it together easily in a day.

There are actually a number of pretty good stripboard/breadboard designer software (like Fritzing, VeeCad).

The key will be the case, getting an adequate heatsink (you probably need one costing $20-$30 at least). The thing about power supplies is if you make it decently, there is no reason you wont still be using it in 30 years. Make it bad and it will be on the shelf in a year when you get your hands on something better. The transformer will need to be 80VA-100VA so you are talking about a fairly well built case.

I went through the exact same situation. I made a 1A 30V supply when I was at school (early 70's)  with a 723. Build the case myself from aluminium sheet. Made it look as good as I could.  I was sick of the  '723 regulator board after about about 2 years, so I rebuilt it with some better opamps that were becoming available (CA3130's). I rewound the transformer secondary to push it up to 40V at 1.1A, added a crude negative volts rail (a TO92 regulator IC attached to a pot) so I could power dual rail opamp circuits. It is still a great supply today - use it all the time.

The parts for the controller board cost well under $10 so you can replace them anytime. The transformer, meter, heatsink, switches, potentiometers and case are about the same for all power supply designs and do cost real money.

Richard.

[Thor-Arne]

John_Edward: Sorry for the hijack, I'm done now. 

If making a PCB slows you down, don't. You can build a perfectly good controller on veroboard/stripboard and at any time in the future, you can make a neater PCB regulator board to replace it. A stripboard controller though can be completely reliable and functional and you can put it together easily in a day.

Only thing that slows me down is to actually etch the board, I don't have a setup for that so I need to order my board from a fab house or use stripboard.

There are actually a number of pretty good stripboard/breadboard designer software (like Fritzing, VeeCad).

I didn't like Fritzing, I'll check out VeeCad next.

The key will be the case, getting an adequate heatsink (you probably need one costing $20-$30 at least). The thing about power supplies is if you make it decently, there is no reason you wont still be using it in 30 years. Make it bad and it will be on the shelf in a year when you get your hands on something better. The transformer will need to be 80VA-100VA so you are talking about a fairly well built case.

Yes, I've got a decent heatsink (Fischer SK47/150/SA, 0.51k/w) and a 160VA toroid transformer.

I went through the exact same situation. I made a 1A 30V supply when I was at school (early 70's)  with a 723. Build the case myself from aluminium sheet. Made it look as good as I could.  I was sick of the  '723 regulator board after about about 2 years, so I rebuilt it with some better opamps that were becoming available (CA3130's). I rewound the transformer secondary to push it up to 40V at 1.1A, added a crude negative volts rail (a TO92 regulator IC attached to a pot) so I could power dual rail opamp circuits. It is still a great supply today - use it all the time.

The parts for the controller board cost well under $10 so you can replace them anytime. The transformer, meter, heatsink, switches, potentiometers and case are about the same for all power supply designs and do cost real money.

Richard.

I have already put together a psu similar to the op-amp based one posted here, not sure when I have the time to do some testing.
There's a few changes made on this one so I'll take a closer look at it later.

[John_Edward]

Thor-Arne, don't worry about it, you are talking about the same design I'm probably going to do as well.
So the more you talk, the more info I have too

And yeah, I replicated the schematic with LiveWire, and lets just say that I don't even know where to start with the

:p
I've done quite a few stripboard layouts, gotta see how that would go.

Oh well, if nothing else, my 'into to PCB layout designing' classes are scheduled for next year, so that should help in the final board

[Thor-Arne]

Thor-Arne, don't worry about it, you are talking about the same design I'm probably going to do as well.
So the more you talk, the more info I have too

There's several other threads about psu designs in the forum, worth to check out.

And yeah, I replicated the schematic with LiveWire, and lets just say that I don't even know where to start with the

:p
I've done quite a few stripboard layouts, gotta see how that would go.

I'm not familiar with LiveWire, I'm a KiCad user.
A good starting point could be to group the parts around the IC's they is connected to.
This goes for both the pcb layout and the stipboard designs.

I noticed you have the wrong footprint for the BD139 transistor (you used TO92 and it should be TO126), check the datasheets for the correct footprints.
Also check the electrolytic capacitors, they are the wrong size.

Oh well, if nothing else, my 'into to PCB layout designing' classes are scheduled for next year, so that should help in the final board

Hehe, can you wait until next year? 

[John_Edward]

I'm not familiar with LiveWire, I'm a KiCad user.
A good starting point could be to group the parts around the IC's they is connected to.
This goes for both the pcb layout and the stipboard designs.

LiveWire isn't too good, but I got it for free years ago.
It has some basic simulation, and the schematics have a pretty nice graphical design.

I'm not even sure what our school teaches, but our computers do have KiCad installed, among others.
If it is free and even remotely popular, they got it installed. Great school

I noticed you have the wrong footprint for the BD139 transistor (you used TO92 and it should be TO126), check the datasheets for the correct footprints.
Also check the electrolytic capacitors, they are the wrong size.

Yeah, I just went with the defaults; if I get it to any kind of plausible arrangement, I can always edit it afterwards.
As I'm working with free parts, getting all of the parts and working with a paper print of the PCB would come before actually producing it.

Hehe, can you wait until next year? 

Probably not

[Thor-Arne]

Yeah, sounds like a great school.

You should probably check what they teach and start playing with i as soon as possible.
I've found that switching between different EDA software isn't a good idea, as many of them are very different.

[seb1982]

Just popped in to say that I'm just starting planning a bench power supply along these lines, so it's been really useful and interesting to read this thread.

Thanks, guys! 

I'm going to try and keep a running record of everything I do so I can post it up here when I'm done.

[Thor-Arne]

There's several other threads about lab psu's here, you should check out these also.

And, yes, please post your progress/result.

[seb1982]

Apologies for hijacking the thread - I promise I'll go back to lurking!  I just thought this might be the best place to ask a question, as the people here are already familiar with the schematic for the DIY PSU.

I've seen a few PSUs based on this design (another link here for convenience:http://diyfan.blogspot.fi/2012/02/adjustable-lab-power-supply.html] [url]http://diyfan.blogspot.fi/2012/02/adjustable-lab-power-supply.html)[/url], and they all have seem to have 3 potentiometers on the front panel, but I can only ever see 2 in the schematics.  Does anyone know where people are adding the third one (which I presume is a finer voltage control, but may well be wrong!)

[Thor-Arne]

That should be a 1k pot in series with the 10k one.

However, a 10k multiturn would be a better but a more expensive solution.

[Thor-Arne]

I finally got around to test a little bit, the results are a little disappointing at this point.

I built it from the schematic mentioned by LEECH666 here.

As mentioned in this thread I use a transformer with two secondary windings and two bridge rectifiers.

There's a few differences. I used the TLE2141IP OpAmps and I used 0.47 ohm resistors as I had these already.

The problems so far is:

• The voltage varies a lot with the load.
  I guess this is caused by the supply voltage for U2.
  The schematic here has a zener to compensate for this variation.

• The output voltage drops to almost zero with high load, above 3.2 amps.
  As the transformer can supply 2*3.2A @ 25V I'm a bit confused about the reason for this.

I have checked the Vref and the negative supply, both are stable, and there's 31.8V on the +V rail at full load.

So... I will do a little more testing to figure out that full-load problem and probably redo the circuit with the modifications mentioned here.

Did anyone else build this ting?

[ecat]

I looked at this about two weeks ago but ended up buying a circa 1984 psu from ebay, well it cost less than the component costs of this design.

Using the TLE2141IP, did you use a 5k pot for RV1 and add the required R10 (1k) as per http://www.electronics-lab.com/forum/index.php?topic=7317.msg1011835#msg1011835 ?

As for the OP voltage drop, is this just the current limit doing its job?

Reading through the electronics-lab thread linked above most user feedback appears positive but in the kingdom of the blind the voice of reason falls on deaf ears, as they say .

The diyfan spin on this is a little strange: Lovely layout and his changes as described make sense. But he kept Q2 from an older design which is apparently no longer needed, so, I don't know. When I see modifications of modifications of something that originally didn't work very well I'm always sceptical.

[Thor-Arne]

I looked at this about two weeks ago but ended up buying a circa 1984 psu from ebay, well it cost less than the component costs of this design.

I considered buying one, but I decided that I needed the experience. 
I've made quite a few 723 based ones, but that's a completely different case compared to using OpAmps.

Using the TLE2141IP, did you use a 5k pot for RV1 and add the required R10 (1k) as per http://www.electronics-lab.com/forum/index.php?topic=7317.msg1011835#msg1011835 ?

No, I didn't see any such information. Doesn't have an account there either.
What I've got hooked up at the moment is exactly what was suggested by LEECH666 in that other post with the changes I mentioned.

As for the OP voltage drop, is this just the current limit doing its job?

I don't think so, I've checked the voltage on the limit circuit and it's more or less the same with low or high load. When the limiting is active this voltage is below 2 volts.

Reading through the electronics-lab thread linked above most user feedback appears positive but in the kingdom of the blind the voice of reason falls on deaf ears, as they say .

The diyfan spin on this is a little strange: Lovely layout and his changes as described make sense. But he kept Q2 from an older design which is apparently no longer needed, so, I don't know. When I see modifications of modifications of something that originally didn't work very well I'm always sceptical.

I've seen countless times that negative comments is being deleted, so all positive comments doesn't mean anything.

I'll look into this design a little more, I might just ditch the whole ting and do something else.
Perhaps the one suggested by amspire.

EDIT:

Which Q2 are you referring to, the BD139 or BC546B. There's two different schematics for this thing with different annotations.

And, if there's a new schematic pleas post it here for completeness. 

[seb1982]

I'm currently going through the (104 page  ) forum relating to this design over on the Electronics Lab page too - I'm hoping to have a play with it in the next few days myself - I'll post anything useful on here.

Like you, I've come to the conclusion that the experience is worth the expense of doing this!  It's looking like it might require 3 high-voltage (expensive) op-amps, doubling or trebling of the 2N3055 with a 0R1 or 1R 5W resistor on the emitters and replacing both 2N2219 and Q1 with TIP31 as well as having a 30V 4A transformer (in addition to all the changes in DIYfan's schematic) in order to get the claimed performance.

I can't deny that it would be a nice sense of achievement to make one of these blo*dy things actually work!  It seems there are no simple, easy, working decent PSU schematics anywhere on the web, which is somewhat surprising!

Anyway, I'll keep you posted!

[ecat]

Experience:
I'm currently reverse engineering the schematic of the one I bought, experience comes in many forms

Anyway, attached is my Eagle rendition of the Electronics Lab psu, though I would much prefer you to join their forum if you want to use their design. Still, the schematic (may or may not be correct) and board (use at own risk) is my work so I guess there's no real harm. At least it shows the mods needed for the TLE part, even though it was a late addition for me too and is not tracked on the board.


Yay, my 2nd useful post

[Thor-Arne]

I'm currently going through the (104 page  ) forum relating to this design over on the Electronics Lab page too - I'm hoping to have a play with it in the next few days myself - I'll post anything useful on here.

Please do, can't have to much references. 
But I think there's a better chance to accomplish something good from the designs started here.

Like you, I've come to the conclusion that the experience is worth the expense of doing this!  It's looking like it might require 3 high-voltage (expensive) op-amps, doubling or trebling of the 2N3055 with a 0R1 or 1R 5W resistor on the emitters and replacing both 2N2219 and Q1 with TIP31 as well as having a 30V 4A transformer (in addition to all the changes in DIYfan's schematic) in order to get the claimed performance.

Yes, there's definitly some issues with dissipating the heat in this design.
I'm currently using 2* 2N3055 TO-3, and they get really hot. I was thinking on TIP31 to, perhaps a lot of them. 

As mentioned my transformer is rated at 2* 3.2 amps at full load, which means that the secondary voltage is a little above 40VAC with no load. So in my case I need to have some high-voltage OpAmps.

I can't deny that it would be a nice sense of achievement to make one of these blo*dy things actually work!  It seems there are no simple, easy, working decent PSU schematics anywhere on the web, which is somewhat surprising!

Check out amspire's posts, he has some good things going. But I think none of them is finished yet.

EDIT: Amspire's General Purpose Power Supply Design.