LM723 PSU with 0V lowest voltage

[earckens]

Here is a circuit diagram for a very simple but also very effective design for a PSU using LM723 and with an effective output of 0-30V as well as current limitation (0 to whatever maximum current you have the unit designed for)..

Output at full load (here 3A) remains at the maximum voltage rating due to the 30V AC transformer.

I recently dug up a PSU that I made during my student years and am now refurbishing: instead of an analog display of current and voltage I am incorporating a digital LCD display and a ATmega168 controller for both display and control functions (short circuit and temperature control).

I had a look around on the net to see if there were any similar designs but all I found were complicated designs using all sorts of zeners, transistors etc to try to get 0V output.

Therefor I decided to draw this circuit from the PCB on which I currently have it (I do not have the original drawing any more) and publish it here for your review.

So if you feel like it, please let me know yours opinions and comments.

 

[MikeMl]

If I was going to put an Arduino in a supply like this, I would get a transformer with several secondary taps/windings, and use it to automate switching of the unregulated voltage upsteam of the regulator to minimize the heating produced. As it is, the heatsinks will have to be huge... (~140w)

 

[Nigel Goodwin]

If you're fancying an AVR based PSU, there's a nice cheap kit one at Bangood:

https://www.banggood.com/0-28V-0_01...it-Current-Limiting-Protection-p-1060253.html

The actual regulation is done via software, with the PSU part controlled via an R2R D2A from an AVR, voltage control and current limiting are set via rotary encoders.

I've currently got mine running on test, feeding a 21 ohm WW resistor with 12V - after about 5 minutes the cooling fan kicked in. I've got to get a transformer for it, and build it in a case - but the transformer I'm testing it with is much too large.

If nothing else it might give you some ideas?.

 

[earckens]

If I was going to put an Arduino in a supply like this, I would get a transformer with several secondary taps/windings, and use it to automate switching of the unregulated voltage upsteam of the regulator to minimize the heating produced. As it is, the heatsinks will have to be huge... (~140w)

The only heat in the housing is from the transformer, and it is a toroidal type so dissipation is not too bad. For the ATmega chip I use a stabilised supply based on a LM4040 reference diode.

What do you mean "switching of the unregulated voltage upstream"?

The heatsinks: 2x 1,6 degrees per Watt so not toot bad, at max 0,8x140=112 degrees, and the power transistors can handle that.

 

[earckens]

If you're fancying an AVR based PSU, there's a nice cheap kit one at Bangood:

https://www.banggood.com/0-28V-0_01...it-Current-Limiting-Protection-p-1060253.html

The actual regulation is done via software, with the PSU part controlled via an R2R D2A from an AVR, voltage control and current limiting are set via rotary encoders.

I've currently got mine running on test, feeding a 21 ohm WW resistor with 12V - after about 5 minutes the cooling fan kicked in. I've got to get a transformer for it, and build it in a case - but the transformer I'm testing it with is much too large.

If nothing else it might give you some ideas?.

Nice to know. But for me the challenge is to build something from scratch

 

[Nigel Goodwin]

Nice to know. But for me the challenge is to build something from scratch

Hence my suggestion for it 'giving you some ideas'

The tapped transformer scheme is a commonly used idea - switching to lower taps from the transformer for lower output voltages, this dramatically reduces heat dissipation in the regulators.

For a simple example, imagine two taps on the transformer, full and half - so say 40V DC and 20V DC on the reservoir capacitor. So for up to 15V output you select the lower tap, for more than 15V you select the full voltage. Obviously the more taps you have the lower the dissipation, and the less heat generated and smaller heatsinks required.

My commercial Velleman PSU appears to use tapped transformers, as you hear relays click in and out as you adjust the voltage. You don't require a processor to do it though, just simple comparators will do.

 

[earckens]

Transformer taps: you would need some kind of switching device, either mechanical or electrical which means you introduce potential extra wear and tear. Also the extra cost for a tapped transformer needs to be taken into account. Versus the cost of dissipated heat which only comes into play at higher current levels. I wonder if it is worth the effort?

 

[Nigel Goodwin]

Transformer taps: you would need some kind of switching device, either mechanical or electrical which means you introduce potential extra wear and tear. Also the extra cost for a tapped transformer needs to be taken into account. Versus the cost of dissipated heat which only comes into play at higher current levels. I wonder if it is worth the effort?

Very much so, the cost savings on the heatsinking required is probably worth the cost, plus the greatly improved reliability due to it's much lower running temperatures - particularly true in your case, with fairly high voltages and currents, as your maximum dissipation could exceed 120W.

The taps are normally switched using relays.

 

[spec]

Hy earckens,

LM723 with 14DIL package pin out (just information):
(on your schematic of post #1, '- Comp' should read pin 4, not pin 5. Pin 8 should be open circuit)

 

[earckens]

The BC181 used as driver for the power transistors is obsolete. Can anyone suggest some alternatives, preferably in TO220 or TO39 packages (the one I have is in TO39)? Max ratings: Ic=600mA, Vce=100V, hFE=30, P=0,5W (but a heat sink is necessary, it gets hot). I have a 2N2907a alternative but it is in the smaller TO18 package, too small for my heatsink.

 

[spec]

A 2N2905A would make an excellent replacement for a BC181: https://www.onsemi.com/pub_link/Collateral/2N2905A-D.PDF

But I don't think that a TO39 transistor will be able to dissipate enough power in your power supply application. While I suspect that your circuit diagram is not correct, especially around the PNP driver transistor and the current limit, I would expect a transistor driving a couple of common collector power transistor, say 2N3055, to give 0V to 30V @ 0A to 3A, to have a maximum dissipation of around 6.6 Watts. So a TO220 transistor, say, would be required. For example, a TIP42C on a heatsink should be up to the job: https://www.onsemi.com/pub_link/Collateral/TIP41A-D.PDF

spec

 

[spec]

The schematic below is what I suspect your circuit should be.

See post #22 for a revised schematic 2016_08_29 OBSOLETE: see post #69

There are some points:
(1) Without transformer switching to reduce power, as the other members have mentioned , the dissipation in the output transistors would be 132 Watts total. Thus, for example, four 2N3055 transistors would be required, and they would need to be mounted on a heatsink of 2 degrees C/W or less.
(2) The minimum output current would be around 20 mA worst case with four 2N3055 output transistors. This could be negated by embodying a negative current sink.
(3) The current control would not be precise, because the current control relies on the VBE of a transitor, it includes some commutation current, and it includes the current consumed by the power supply circuits.
(4) There is no decoupling

spec

 

[earckens]

Thank you very much spec! Your input is very valuable to me.

However, I do not fully understand your statement in line 2: "..minimum output current would be 20mA worst case.." and "..negated embodying a negative current sink.". Also line 4: "..no decoupling."?

Thank you too for the schematic.

 

[earckens]

Hy earckens,

LM723 with 14DIL package pin out (just information):
(on your schematic of post #1, '- Comp' should read pin 4, not pin 5. Pin 8 should be open circuit)

View attachment 100952​

Thank you spec, I corrected this on my drawing, I will edit my post! On pin 8, is it imperative to leave it unconnected?

 

[bountyhunter]

Transformer taps: you would need some kind of switching device, either mechanical or electrical which means you introduce potential extra wear and tear. Also the extra cost for a tapped transformer needs to be taken into account. Versus the cost of dissipated heat which only comes into play at higher current levels. I wonder if it is worth the effort?

You need a relay. No it's not worht the effort, I would just bulid a switcher that regulates about 2V higher than VOUT and use a 2V linear supply to drop it down. You get low noise and fast transient response of a linear but high efficiency near switcher level.

No offense but the LM723 is a 70's era technology, I wouldn't build a design based on one. My lab supply is a linear using discreete circuitry. The main advantage of the LM723 was it had an on board reference and current limiting neither of which are decent performance level.

 

[bountyhunter]

Here is a generic lab supply schematic that will give much higher performance.

 

[earckens]

Here is a generic lab supply schematic that will give much higher performance.

Interesting circuit. What is the approx. current draw at the 15-20V input? What do you connect to IN A (transformer, rectifier and capacitors?)? The values of capacitors without unit is nF? Some resistor values are missing: can you give the values? What is the approx. rated output current here? What is the rated max output voltage? 1x 2N6124 seems a bit minimal for a 150VA transformer at the input?

 

[spec]

Thank you too for the schematic.

No probs earckens,

I used the 723 a lot in the past. It was a revolution when it first came out and was pretty expensive too.
I would just use the Eagle schematic of post #12 from now on to avoid confusion. The schematic is drawn with data flow principals, voltage cascading, and minimum cockle which makes the schematic easier to modify and comprehend, not that you didn't do a lot of hard work transposing from printed circuit to sketch (known as cartoon by draftsmen).

Please note that I have not calculated the values shown on the schematic of post #12 and have only transposed the values and interpreted the circuit from your sketch.

I do not fully understand your statement in line 2: "..minimum output current would be 20mA worst case.."
and "..negated embodying a negative current sink.".

This is a classic problem with power supplies. If you take the LM78xx series of three terminal regulators for example, the minimum output load current is 10mA. If you take less than 10mA the output voltage is liable to rise to the input voltage and there will be no regulation.

The reason for this is that the pass elements have a leakage current, which for the 2N3055 can be 5mA. So if you had four 2N3055 transistors in parallel the total leakage current could be 4 * 5mA = 20mA. Note that the 2N3055 specification says that the leakage current could be as high as 5mA at a junction temperature of 25 deg C. In fact, for any individual 2N3055, the leakage current could be a lot less. But the leakage current increases with junction temperature and the junction temperature of the 2N3055 transistors in your power supply is liable to be near the maximum allowable of 200 Deg C when the power supply is outputting a low voltage and 3A.

The way to negate the leakage current is to generate a simple negative supply line, from the existing mains transformer of -5V say and have a constant current generator sinking 30mA say, down to the -5V line. The brings other advantages too: better frequency stability and better voltage regulation at low output currents.

Also line 4: "..no decoupling."?

A schematic is an illusion. It does not represent the real world. Even a short piece of wire or printed circuit trace is composed of resistors, inductors, and capacitors.

The net result of this is that a circuit can oscillate or the components will not perform to specification. This is especially the case for high gain circuits, high frequency circuits, and precision circuits. For example, the average operational amplifier (opamp) has a voltage gain around one million and the 723 contains an opamp.

To negate the effect of parasitics (unwanted circuit components and couplings) you add decoupling capacitors and isolating resistors and inductors and also optimize the layout for maximum performance. That is why on a production equipment you often find all sorts of odd components splattered all over the place.

The theory of decoupling etc is quite complex but there are rules of thumb that can be used. I will modify the schematic of post #12 to show suggested decoupling.

spec

 

[earckens]

I will modify the schematic of post #12 to show suggested decoupling.

spec

Great!

Do you have an alternative as per the post of #16 of bountyhunter (I miss some component values there for it to be usefull for building)? I will refurbish my old PSU but may be building another one with technology using current wisdom.

 

[spec]

Great!

Do you have an alternative as per the post of #16 of bountyhunter (I miss some component values there for it to be usefull for building)? I will refurbish my old PSU but may be building another one with technology using current wisdom.

Are you saying that the component values are not on Bounty's PSU circuit or are you saying that you have not/cannot get the parts.

Apart from the pass transistors shown, which can be substituted for modern versions, all parts are cheap and freely available.

The points that I made about the pass transistor leakage current still applies as does the high power dissipation of the power transistors. This is down to fundamental physics rather than a particular power supply design.

Perhaps bounty can comment on all this in relation to the power supply of post #16.

spec