10 Amps power supply with LM723

[mike_mike]

Hello,

I am planning to build a new power supply using LM723 and NPN power transistors. I saw that there are a lot of schematics on the internet, but I am wondering which of them is a good one ?
So my questions are:
1. Can you recommend a good 10 Amps power supply with LM723 ?
2. Is there the possibility to appear oscillation in this power supply that you recommended ? How can I prevent the oscillations ?
3. What are the layout guidelines for such a power supply ?
4. What other problems can appear in the circuit ?

[Wolfgang]

The ARRL (American Radio Relay League) has some 30V/10A supply suggestions.

Important things are:
- adequate cooling (fan)
- bridge rectifier with sufficient current loading capability (20A min)
- 4 x 2N3055
- frequency compensation (see ARRL schematics)
- overtemp protection
- output crowbar to protect your load if something goes wrong.

http://www.dd1us.de/Downloads/an%20rf%20proof%2030%20amps%20power%20supply%201_0.pdf

[mike_mike]

Thanks for the document. It is very interesting.
I built the power supply from the link: http://www.electronics-lab.com/project/3-30-v2-5-a-stabilized-power-supply/
The single difference was that I used 0.33 R in parallel with 0.18 R for R5.
I tested the power supply using a 24V ac / 100VA transformer.
The output voltage was 19.97Vdc without load and with load it was 19.94Vdc. The load was a resistive load of 2.90A.
It is normal the difference of 0.03V between the output voltage with and without load ?
The output voltage was measured on the divider formed by R4 and R3.

[iMo]

I would add a small resistor, like 100ohm, into TR1's base.
30mV change with 2.90A load? -> the 723 NS datasheet says the load regulation with an external NPN pass transistor is 15mV per 1A change. With thin wires and larger currents it could be even more..
PS:

10 Amps power supply with LM723

with your 0.18ohm || 0.33ohm the over-current protection trips at something like 5.5Amperes..

[mike_mike]

Can you be more explicit ?
I saw in the datasheet that load regulation is maximum 0.6% of Vout, so in my case it is about 0.11V (0.6% * 19.97V). Is that correct ?

[iMo]

Read above datasheet. Figure n4.

[mike_mike]

I don't know which one is fig n4. The figures does not have a number.
Are you talking about the attached one ?

Thank you.

[iMo]

Figure 4 means Figure 4

[mike_mike]

I have a question regarding the LM723 power supply. I built this power supply on a PCB that was made in China.
I verified the output voltage drop with different load currents at different output voltages. BUT, before making this verification, somebody tried to take out the PCB from the power supply case. The PCB was glued inside the power supply case using hot glue. He tried and managed to take out the PCB using a screw driver that he put underneath the power supply PCB and then he lifted up the PCB. I checked the PCB visually, with the naked eye and I did not found any problems.

My questions are:
1. It is possible for this circuit to work correctly if it was damaged by the screwdriver action ?
2. It is possible that if I will build another power supply (using exactly same schematic and a exact copy of the PCB) the circuit to work bad ?
3. It is possible for this circuit to work good because if it was damaged and to work bad if it was not damaged ?

[iMo]

Specify your voltage drops at different loads.

[mike_mike]

V without load = 19.84V, V with load = 19.80V, load = 2.90A
V without load = 25.0V, V with load = 25.0V, load = 2.20A.
for the tests, I used a 0.33R resistor in parallel with the 0.18R.

So the circuit PCB was forced out of the power supply case, using the screwdriver, but on the PCB I did not found any problems while I was inspecting the PCB using the naked eye.

[iMo]

What is your problem then? It seems it works fine.

[mike_mike]

Hello, quick question about the attached schematic: somebody said to me that the current limit (current sense) is temperature dependent. Can you explain why is this happening ?

[iMo]

In 723 the current sense is done via an NPN transistor on the chip (like your Fast CC). With temperature the trigger shifts -2mV/degC. Calculate what current do 2mVolts represent on the shunt resistor and you get the dependency in mA/degC.

[lycralout]

A nice summary of design rules to note when using the 723 is contained in https://electronicprojectsforfun.wordpress.com/power-supplies/a-collection-of-proper-design-practices-using-the-lm723-ic-regulator/

[mike_mike]

Hello. I build the power supply from the attached schematic, but I found that there is a problem regarding the current limit control.
If I short between the pins 2 and 3, the output voltage varies from 1V to about 30V. But if I connect pins 2 and 3 to the circuit, the voltage varies up to about 12V. I also tried to use an external NPN small power transistor (BC548) and I found the same problem is happening again. I used only POT1, POT2 and POT3 from the original schematic. I did not used any Ammeter, I just connected a wire. The BE voltage of the small power NPN transistor is always about 0.55-0.6V.
I am uploading the original schematic: original_schematic.png
And I am uploading the schematic when I used the external small power NPN transistor: schematic_img.png.

More details about the problem:
1. Input voltage 33Vdc, from 24V/160VA transformer.
2. Voltage on pin 12: about 33Vdc.
3. Voltage on pin 11: about 32.5Vdc.

I tried the following solutions:
1. I changed the potentiometers with some new ones, but the problem is the same with the new pots.
2. I checked the PCB for bridges and it is OK.

What should I do to make this circuit work correctly ?

[iMo]

In the original schematics if you short 2 and 3 the current limit is off.
If 2 and 3 is wired into the circuit and it triggers at 12V it means at 12V output the current through the shunt 0.15ohm is high enough to create 0.6V voltage drop on the shunt and open the the current limit transistor.
Probably a problem with your wiring or shunt value.

You do not use 723's pins 2 and 3 in your schematics (eagle). The external transistor T4 is doing the limiting. So let 2 and 3 shorted (and connect to COMP maybe).

Double check the voltage drop at the 0.15ohm shunt while varying the output voltage.
Doublecheck the value of the 0.15ohm shunt!

[mike_mike]

I changed the 0.15R resistor with a new one and I checked the voltage on the 0.15R resistor and it is about 2.8 miliV, when the output voltage is 12V.
The BE voltage of the transistor is still 0.55-0.6V.
I connected the pin 2 and pin 3 together but the situation is the same, the output voltage is still about 12V, and sometimes it is varying around 12V.

Edit: When I powered again the power supply, the voltage was about 17V and not 12V as it was when I powered it initially.
When I set the output voltage at about 2.1V, the voltage on the 0.15R resistor is 2.1mV.
I also changed LM723 with a new one, but without succes.

[iMo]

I changed the 0.15R resistor with a new one and I checked the voltage on the 0.15R resistor and it is about 2.8 miliV, when the output voltage is 12V.
The BE voltage of the transistor is still 0.55-0.6V.

How you can have BE=0.6V when the voltage at the shunt is 2.8mV??
It has to be the same when CL(base) and CS(emitter) are wired to the shunt..

[mike_mike]

I found that the 0.5-0.6V voltage drop at BE junction is because there is 0.5-0.6V voltage drop on the 120R resistor ...

I changed the 120R resistor with a new one, but it did not changed the situation.
Somewhere on the PCB is a component that makes the BE jonction to be at 0.6V... i think.

[iMo]

When the pot's wiper is at the lower side the BE voltage cannot be higher than 0.16V.
The Vref is say 7.2V and the divider is 120/(4k7+500+120)*7.2=0.16V.

[mike_mike]

Yes. When the pot is at lower side and the output voltage is at minimum, then the BE voltage is 0.165V.
But if I rotate the Voltage pot, then the BE voltage increases slowly from 0.165V to 0.595V, even if the current pot is at the lower side.

[iMo]

There is a wiring problem then.
When increasing the output voltage, AND the current through the shunt is low (say 10mA), the BE voltage should not change much.
With set constant 0.165V at the current limit pot you would need 3Amps through the 0.15ohm shunt to get 0.6V at the base-emitter.

[mike_mike]

I built another schematic, and I hoped that it will work. But it doesn't and I do not know why.
When I power up the circuit, the 500R trimmer gets hot and the output voltage is always about 32.5-33Vdc. I changed the LM723 with a new one, I changed the 10K pot with a new one, I changed the power transistor (for the tests I used only one power transistor) with a new one. I also checked the layout and the PCB and there are no problems. 
This schematic is 100% functional, it was build by someone from my country and it works.
What should I do ?

[iMo]

I've done a simulation of your original schematics above and it works.
FYI:
CL POT           Current LIMIT (14.6V Vout)
5ohm             3.17A
250ohm         1.00A
300ohm         0.55A
330ohm         0.29A
350ohm         0.12A
               

[mike_mike]

I know that it works.
But my circuit does not, and I am not capable of finding were is the problem...
Probably I could find someone in my city for help.

[iMo]

Be careful with grounds.
There are 2 "grounds" - in my simulation the left hand side of the R_Shunt is COM and the right hand side is GND. I often connect the input filter capacitor to GND (wrong).
Doublecheck how did you wire the stuff.
When the simulation works it should in HW too..

Enclosed the simulation source - try it and compare it carefully with your kludge

[mike_mike]

Thank you for your help. I will try to find the problem.
Meanwhile, a quick question: does the attached schematic have current limit down to 0A ?

[Wolfgang]

... in fact yes, but with low precision. The 723 is not very well suited to lab grade current limiting.
An obvious problem here is the Vbe tempco of the pass transistors when running under high power.

[mike_mike]

I finally managed to make the schematic from post #23 to work, but I found that the current limiting pot does not regulate linear the output current (I don't know how to explain exactly), so if I set the pot at half of its way, then the output current is about 0.5A, while the maximum current is 3A. The pot is a linear one, B500R.
If the pot is at half, then the output current should be 1.5A ?
Is that true ?

[iMo]

1. What was the problem, could you tell us?
2. See my previous post with 500ohm pot's wiper value (measured from its low side) vs. current limit. In the mid position it shows 1A limit. It is a simulation only..

[mike_mike]

The problem was the PNP transistor. There was a bridge between the C and E of this transistor, and I did not saw it when I made the inspection...

Later Edit: If I use 2 220R resistor in series instead of the 500R pot, then the output current is about 1A. The connection between the 2 200R resistor is connected to the point were the pot wiper was connected before.
I do not have any other 500R pot or trimmer to check if the first one was faulty, can there be used 1K pot instead of 500R pot ?


I checked again with 1k pot for current limit, and at the half of the pot, the current is 0A. Only on the other half the current starts to rise from 0A to 3A.

[iMo]

In case of 1k pot you have to increase the 4k7 resistor.
I would put, say, a 22k trimmer instead of the 4k7, set the 1k pot's wiper to the top position and set the minimal current limit with the 22k trimmer. Replace the 22k trimmer with a close valued resistor finally.

For example a simulation with 10k the top resistor, 1k pot, 120ohm bottom, AND a 10k resistor from pot's wiper to gnd
POT's wiper       CLimit
top                   23mA
mid                  1.8A
down                3.8A

[mike_mike]

I checked using a 22k trimmer, the 1k pot, and I found that the proper value is around 12.7K for the trimmer.
The current limit starts from 0A and goes up to 3.31A.
I found that when I rotate the pot, it starts from 0A, then if I rotate it a very little then the current suddenly increase to more than 200mA, and there is a very little "space" on the pot for 1mA or 5mA. Is that normal ?

Later Edit: only now I have saw your edit.

[iMo]

The pot setting is highly none-linear. It follows the Vbe characteristics.
The 10k (or similar) from wiper to gnd could make it "more linear"..

PS: or use 1k "logarithmic" potentiometer.

[mike_mike]

For example a simulation with 10k the top resistor, 1k pot, 120ohm bottom, AND a 10k resistor from pot's wiper to gnd
POT's wiper       CLimit
top                   23mA
mid                  1.8A
down                3.8A

I used the modification specified above and it works a little bit better, but at mid, the output current is about 1.20A, instead of 1.8A, and the maximum output current is 3.06A. The current starts to rise from only at about 1/4 of the pot course.
For the tests I used only one TIP3055 and only one 4700uF capacitor.v Also, for the tests I did not used D1 and D2.

Is the Rwiper connected correctly ?
Please have a look at the schematic and at the results above and tell me what you think.

[iMo]

I would play with those 3 caps see below.
Also I would make the Voltage setting at the Vref side around the +IN (fixed divider at the output).

[mike_mike]

The role of those caps is to filter the voltage, for example the 10uF polarized capacitors is to filter the the voltage reference for voltage regulation and current regulation ?
The results of the output current pot are good ?
What are the advantages of using the pot at the Vred side ?

[iMo]

Yes.
Yes.

The output divider values contribute to the feedback control loop dynamic parameters.
When changing them the feedback response is changing.
Therefore it is better to have the Voltage regulation at the IN+ side.

For example with fixed 9k/1k output divider (/10) the output voltage will be aprox. 10*V(IN+).

[mike_mike]

I tested the attached schematic, but the minimum output voltage is about 9V, and if I reduce the R3 to 0 ohms, the minimum output voltage is about 7V.
I also tested with R12=8k2, R10=2k2, R1=0R and R3=680, and the minimum output voltage vas 3.5V and the maximum was about 29.5V. The voltage pot has a part where if the voltage is at maximum (29.5V) then the output voltage does not go down until the pot is at 1/4 of the course. I found those values on a schematic on the internet: http://www.electroniq.net/power-supply/lm723-variable-power-supply.html

[iMo]

You wired the C9 10u capacitor at wrong place.
Look at my schematics above.

Try with R1=12k, R2=10k pot, R3=0 and with the divider by 10 at the output.
Wire C5=10u from the bottom of R1 against gnd of 723.

[mike_mike]

I tested the attached schematic but it goes only from about 7V to about 30V.
The voltage on pin 5 is 0V and on pin 4 is 0.74V, while on the output is about 7V.

[mike_mike]

I also tested with 4k7 and 3k9 resistor for R12. The value for R1 was 1k2.
The output voltages were:
with 3k9: from 3.7V to 29.8V, but when the output voltage was 3.7V, and if I started to rotate the pot, the voltage started to rise only when the pot was at 1/4 of the course.

[mike_mike]

I tested with the following values, and the output voltage varied from 3.1V to 25.9V. The maximum that I need is 25V, so 25.9V is OK. The lowest output voltage is also OK.
R1=1.2K, R2=10K pot, R3=680R, R12=6K8 and R10=2K2.
There is a problem that still persists, the 10K pot has a "dead" portion on the low voltage. When the output voltage is abut 3.1V, the voltage starts to rise only when the pot is at about 1/4 of its course.

[mike_mike]

I try to make a simulation of the attached schematic in LTSpice, but the program does not show the .raw file when I run the simulation.
At the beginning I set to DC op pnt, but after I run the simulation, I set it to transient.
Can you please tell me where I am wrong ?

[iMo]

GND is missing in your schematics.

[mike_mike]

Thank you. Now the simulation works.
Regarding the previous posts, are the results good ? Are there any problems ? I mentioned that the voltage pot has a "dead" portion, can it be solved ? It is not such a big problem, if the power supply works correctly...

[iMo]

It is many decades I messed with 723 PSUs. Afaik the minimum output voltage is 2V with 723.

[Wolfgang]

Correct. Below that the 723 inernal error amp does not work properly anymore. Below the emitters there is a current source, and its collector voltage is the error amp input voltage minus Vbe, so when you have 2V at the inputs, its about 1.4V Thats already not too much, so below that the current source is simply not working with a good and high output resistance.

[mike_mike]

Hello,
I have found the attached power supply schematic.
Can you please help me with part numbers for replacement transistors TR3-TR6 (A823) and TR2 (3223) ?
The replacement transistors are not really necessary to be pin to pin compatible, because I can redesign the pcb layout.
I would also like to know the voltage for ZD3, the part number for D4 (1N4148 ?) and the secondary voltage of the transformer.

[xavier60]

Hello,
I have found the attached power supply schematic.
Can you please help me with part numbers for replacement transistors TR3-TR6 (A823) and TR2 (3223) ?
The replacement transistors are not really necessary to be pin to pin compatible, because I can redesign the pcb layout.
I would also like to know the voltage for ZD3, the part number for D4 (1N4148 ?) and the secondary voltage of the transformer.

BC556 will be Ok for TR3-TR6. Even BC557 would do if the unreg rail isn't much higher than 30V. T3 sees very little voltage.
D45H11 for T2.

Quoted from ON Semiconductor,
"
These series of plastic, NPN and PNP power transistors can be used as general purpose power amplification and switching such as output or driver stages in applications such as switching regulators, converters and power amplifiers."
https://www.onsemi.com/products/discretes-drivers/general-purpose-and-low-vcesat-transistors/d45h11

[mike_mike]

Even BC557 would do if the unreg rail isn't much higher than 30V.

I need output voltage up to 50V (as it is written on the schematic). In this case, the unreg rail will be more than 30V. This means that I will need to use transistors with higher collector to emitter voltage...
For example, 2N5401 is a good replacement for TR3-TR6 ?

[xavier60]

Even BC557 would do if the unreg rail isn't much higher than 30V.

I need output voltage up to 50V (as it is written on the schematic). In this case, the unreg rail will be more than 30V. This means that I will need to use transistors with higher collector to emitter voltage...
For example, 2N5401 is a good replacement for TR3-TR6 ?

That should be fine.

[mike_mike]

What replacements (equivalents) do you recommend for BF422 and BF423 from the attached schematic ?
Can BF422 be substitute by 2N5551 and BF423 by 2N5401 ?
They do not need to be pin to pin compatible, I can redesign the pcb layout.

[xavier60]

What replacements (equivalents) do you recommend for BF422 and BF423 from the attached schematic ?
Can BF422 be substitute by 2N5551 and BF423 by 2N5401 ?
They do not need to be pin to pin compatible, I can redesign the pcb layout.

The current ratings of the 2N5551 and 2N5401 are unnecessarily high but should not cause a problem.
That circuit has delayed over current shutdown but no fast limiting.
Which can be fixed by putting C3 across BE of TR2.

More: I'd be curious to know what the current gain measures when you receive the 2N5551 and 2N5401 transistors.

[LaserTazerPhaser]

The ARRL (American Radio Relay League) has some 30V/10A supply suggestions.

Important things are:
- adequate cooling (fan)
- bridge rectifier with sufficient current loading capability (20A min)
- 4 x 2N3055
- frequency compensation (see ARRL schematics)
- overtemp protection
- output crowbar to protect your load if something goes wrong.

http://www.dd1us.de/Downloads/an%20rf%20proof%2030%20amps%20power%20supply%201_0.pdf

Power supply doesen't need stud diode heatsinks, D1 & D2 studs can be replaced with a few ganged high efficiency schottkies dpak or other smt parts.

Aren't electrolytics inappropriate for mains side applications?

[ArthurDent]

LaserTazerPhaser - "Aren't electrolytics inappropriate for mains side applications?"

The capacitors you marked are not polarized. The curved electrode generally indicates outside foil. There is no + or - that you would see on electrolytics. Check the values and you'll see they can't be electrolytics.

[mike_mike]

More: I'd be curious to know what the current gain measures when you receive the 2N5551 and 2N5401 transistors.

hFE for 2N5401 is 180 and for 2N5551 is 163.