Designing linear power supply with regulated current and voltage

Thread Starter

kgrozdanovski

Joined Nov 22, 2020
27
Hello friends! Recently I have been working on my own linear power supply design as part of my requirements for lab equipment and the need to improve/refresh my knowledge.

The goal of this supply is to provide voltage regulation, current limiting and short-circuit protection in the ranges 1.5-30V, 0-5A. The requirements are not strict however I intend to implement every best practice and the best affordable components to make this power supply as good as I can.

My design revolves the 2n3055 as a power transistor and the LM317 as the basis for the control circuitry.

1606080333834.png
This part of the circuit is fairly straightforward and common - transformer, Graetz/diode bridge and capacitors to smooth out the output.

In the primary circuit, shown above, I intend to use a KBPC5010 diode bridge rectifier and Nichicon BY-series capacitors with minimal ESR.

1606080531872.png
This is the series voltage/current regulation circuit containing the LM317 and 2n3055 components.

The above circuit contains two potentiometers through which we can set the voltage and current limits. This is based on textbook examples for current regulation using two transistors and a feedback loop (read more here).

The way I understand it - R4 is connected to the output of the LM317 (configured as a variable voltage regulator) and acts as a voltage divider, piping current into Q2. The collector of Q2 is connected to the ADJ pin of the LM317 and draws current away from the ADJ pin (effectively reducing the current provided to the base of Q1). Depending the current drawn by the load (shown in figure 1) and the position of R4, Q2 may draw more or less current from the ADJ pin, effectively regulating the output current by Q1.
The R4 potentiometer is basically acting as a Rsense / current sensing probe.

I realize Q2 may be a much smaller transistor, so lets leave that out of focus for the moment.

This circuit seems to work and operate smoothly and predictably when simulated in NI Multisim, however I would really appreciate any suggestions for improvements/criticism/mistakes pointed out. What can we do to improve load and line regulation? Should an NTC be included for better thermal handling?
At first sight I would say another 2n3055 power transistor parallel to Q1 with resistors should be added, in order to increase the maximum current capability or at least split the thermal stress. I also bumped into many circuits using operation amplifiers instead of a transistor (Q2) in the current feedback stage - what would the benefits of the additional complexity be? Would it be maybe better to instead preserve the existing current regulation and just replace the Zener diode (D7) with an integrated voltage reference? I am also uncertain if there is a need to implement output overvoltage protection, in case Q1 fails (this was pointed out on some forum somewhere)?

I am grateful for any feedback you may provide.
 

dl324

Joined Mar 30, 2015
13,145
Welcome to AAC!
This circuit seems to work and operate smoothly and predictably when simulated in NI Multisim, however I would really appreciate any suggestions for improvements/criticism/mistakes pointed out. What can we do to improve load and line regulation?
The first thing you should do is get rid of the base emitter junction on the output. The voltage drop will be dependent on load current and will vary too much over the 0-5A range.

It would be better to use a PNP pass transistor as shown in the datasheet. If you don't have a power PNP, you can use a PNP general purpose transistor with a power PNP.

EDIT:
The way I understand it - R4 is connected to the output of the LM317 (configured as a variable voltage regulator) and acts as a voltage divider, piping current into Q2.
That isn't current limiting. It's based on sampling the output voltage which has nothing to do with output current.
 

AlbertHall

Joined Jun 4, 2014
11,506
I don't see how R4/Q2 can work as a current limit. For instance if the output voltage is 5V or less then Q2 can never conduct because of D7, so Q2 can do nothing.
 

Thread Starter

kgrozdanovski

Joined Nov 22, 2020
27
Hello and thank you for your prompt replies.

It would be better to use a PNP pass transistor as shown in the datasheet. If you don't have a power PNP, you can use a PNP general purpose transistor with a power PNP.
I am not sure which datasheet are you referring to?

I don't see how R4/Q2 can work as a current limit. For instance if the output voltage is 5V or less then Q2 can never conduct because of D7, so Q2 can do nothing.
You are correct. The diode value should be lowered, it was arbitrarily placed and it's value not considered for the sake of testing.

That isn't current limiting. It's based on sampling the output voltage which has nothing to do with output current.
I suppose I only observed what I was hoping to observe. Would you perhaps suggest that I instead use a an operational amplifier with a low-value Rsense resistor in series with the load? I couldn't figure out how to integrate such a circuit with the existing LM317 regulator - most examples showed a direct connection with the series pass transistor (Q1 in this case).
 

dl324

Joined Mar 30, 2015
13,145
I am not sure which datasheet are you referring to?
From TI's LM317 datasheet:
clipimage.jpg
LM195 is a blowout proof power transistor, but any power NPN can be used. You can't parallel power transistors without ballast resistors. LM195 is a special case and it incorporates ballast resistors in it's design.
I suppose I only observed what I was hoping to observe. Would you perhaps suggest that I instead use a an operational amplifier with a low-value Rsense resistor in series with the load? I couldn't figure out how to integrate such a circuit with the existing LM317 regulator - most examples showed a direct connection with the series pass transistor (Q1 in this case).
Since you have a regulator and an external pass transistor, you need to put current limiting on the input side. How to do that depends on how you implement the pass transistor and current sharing with the voltage regulator.
 

dl324

Joined Mar 30, 2015
13,145
If I was going to use an external pass transistor, I'd design it for a specific current sharing ratio so the safe area protection in the voltage regulator can be used to protect the pass transistor:

From NatSemi Voltage Regulator Handbook:
clipimage.jpg
 

Thread Starter

kgrozdanovski

Joined Nov 22, 2020
27
If I was going to use an external pass transistor, I'd design it for a specific current sharing ratio so the safe area protection in the voltage regulator can be used to protect the pass transistor...
So basically, in the case of an overload we would "route" it from the pass transistor towards the regulator (in this case the lm317) which will use it's internal safety mechanisms and handle the problem?

On another note, I modified the design to incorporate an MJ2955 (PNP complementary to the 2N3055, however I will likely use the MJ4502 in the final design). What are the benefits of using a PNP power transistor in this specific case? Are we maybe hoping to limit the current on the input side instead of sinking it as per my original schematic, or is that a separate topic?

I am am still confused as how to implement current limiting to this design, however I will continue researching.

1606088533412.png

P.S. I am very grateful for all the input you guys have provided!
 

dl324

Joined Mar 30, 2015
13,145
So basically, in the case of an overload we would "route" it from the pass transistor towards the regulator (in this case the lm317) which will use it's internal safety mechanisms and handle the problem?
That won't be the cause with your circuit. To have the regulator protection protect the pass transistor, you need a specific current division such that at your maximum output current, the regulator is at its maximum current. Any current above that would cause the regulator to self protect. That in turn would protect the pass transistor. For this to work, the pass transistor and regulator need to be on the same heat sink.
I modified the design to incorporate an MJ2955 (PNP complementary to the 2N3055, however I will likely use the MJ4502 in the final design). What are the benefits of using a PNP power transistor in this specific case?
Using a power PNP used to be a disadvantage because they cost more than a power NPN.
I am am still confused as how to implement current limiting to this design
There are a couple approaches used with LM317 regulators. One is to clamp the voltage to 1.2V under overload conditions. The other is to disconnect the input voltage. The latter will cause the current limiting switch to oscillate.
 

Thread Starter

kgrozdanovski

Joined Nov 22, 2020
27
That won't be the cause with your circuit. To have the regulator protection protect the pass transistor, you need a specific current division such that at your maximum output current, the regulator is at its maximum current. Any current above that would cause the regulator to self protect. That in turn would protect the pass transistor. For this to work, the pass transistor and regulator need to be on the same heat sink.
Yes, but I am considering implementing such a mechanism. However, wouldn't it work even if they are not on the same heatsink? I assume they would be on the same heatsink if we wanted thermal overload protection, however the LM317 also has internal current limiting if I am not mistaken, which would act as an absolute maximum limit, no?

There are a couple approaches used with LM317 regulators. One is to clamp the voltage to 1.2V under overload conditions. The other is to disconnect the input voltage. The latter will cause the current limiting switch to oscillate.
The idea is to be able to regulate the current limit for cases where constant current would be required and not only for short-circuit protection. Is that applicable to the two examples you provided?
Wouldn't a current sense resistor in series with the load, coupled with an OpAmp also be an option for current limiting? If so, how would one incorporate that into the LM317 circuit?
 

dl324

Joined Mar 30, 2015
13,145
LM317 also has internal current limiting if I am not mistaken, which would act as an absolute maximum limit, no?
That won't protect the pass transistor from thermal issues.
Is that applicable to the two examples you provided?
Nothing I posted included current limiting.
Wouldn't a current sense resistor in series with the load, coupled with an OpAmp also be an option for current limiting?
I'd put the sense resistor on the input side of the regulator and pass transistor so regulation isn't affected by current draw. That would also ease restrictions on the voltage drop across the sense resistor.
The idea is to be able to regulate the current limit for cases where constant current would be required and not only for short-circuit protection.
It would work for both. The current limiting switch would switch rapidly enough that the load would think it's getting a constant current.
 

Thread Starter

kgrozdanovski

Joined Nov 22, 2020
27
It would work for both. The current limiting switch would switch rapidly enough that the load would think it's getting a constant current.
Could you perhaps provide an example for this approach or maybe a link to a resource where I can study such a solution? I have been attempting to figure out how to implement adjustable current limiting to the circuit attached in #8 however with no success. I have also been studying the circuits provided as examples in the LM117 and LM317 datasheets from TI but have not been able to base an implementation on them.

The end-goal here is consume the best practices and advice and create a relatively-simple design which makers could build on a variable budget and learn from.
 

dl324

Joined Mar 30, 2015
13,145
Could you perhaps provide an example for this approach or maybe a link to a resource where I can study such a solution? I have been attempting to figure out how to implement adjustable current limiting to the circuit attached in #8 however with no success. I have also been studying the circuits provided as examples in the LM117 and LM317 datasheets from TI but have not been able to base an implementation on them.
What have you tried?
 

Thread Starter

kgrozdanovski

Joined Nov 22, 2020
27
I have attempted to implement the circuit shown in #7, implement a current sense resistor on the input side and feed a current back to the power transistor (MJ2955) through an NPN transistor, and other connections I botched up.

I am obviously no hardware expert however I prefer to learn from example and principles of operation. I suppose I am not quite sure which aspects of this entire current limiting endeavour manage to confuse me.

What I have generally been attempting is to either decrease the Iadj current of the LM317 or decrease the base current on the power transistor.
 

dl324

Joined Mar 30, 2015
13,145
I have attempted to implement the circuit shown in #7, implement a current sense resistor on the input side and feed a current back to the power transistor (MJ2955) through an NPN transistor, and other connections I botched up.
Post a schematic of how you tried to implement current limiting.
 

Thread Starter

kgrozdanovski

Joined Nov 22, 2020
27
I have decided to attempt to implement in reference to an existing circuit found on Stack Exchange, however with problems.

1606176773157.png
The original circuit I found.

However, while the voltage regulation functions (albeit with only ~60% of the desired voltage range - I suspect some resistor values need to be changed), the current regulation does not. Specifically, voltage may be regulated with the Voltage adjust pot in the bottom of the circuit but the Current adjust is simply neglected and has no effect.
1606176811573.png
My implementation for simulation in NI Multisim


Also, is there any benefit to using an operation amplifier as negative feedback as opposed to using a transistor for the current feedback?
 

AlbertHall

Joined Jun 4, 2014
11,506
What load did you connect to check the current mode?
It needs to draw a current above the limit at the output voltage.
You could try a simple short circuit.

[Edit] In your drawing you have the bottom of R11 and C10 connected to Vout. That connection will prevent the current limit working.
 

crutschow

Joined Mar 14, 2008
27,765
Note that the short circuit current (if the output is pulled below the 1.2V minimum output of the LM317) will be limited by the internal short circuit current of the LM317.
The external current limit provide by the LM301 will have no effect on that current.
 

sparky 1

Joined Nov 3, 2018
580
Here is an inexpensive rectifier filter board I bought. It has 4 50V rubycon 4700uF ecaps it uses a center tap transformer
the output is dual polarity. You can regulate it the way you want. Don't compromise the quality for anything less than this.
The 50V filter and good diodes and good bypass caps and good board it can function in different ways. Configuration is snap.


It says on the board :
Electronics-Salon
A-460 Ver: 1.0
IMG_20201123_164904.jpg
 
Last edited:

kaindub

Joined Oct 28, 2019
89
Simple solution. Disconnect the top of R4 from the output of the lm317. The negative output of your regulator now becomes the top of r4. Make d7 a short circuit. Base of Q2 to top of r4. The current passes through the resistor to the real ground. Select r4 to drop about 0.6 v at the current limit.
You can make it adjustable by selecting a larger value of r4 and putting a pot across it. The wiper of the pot goes to base q2.
This type of current limit depends on the base emitter voltage of Q2. The cutoff is very soft. Also if you make the variable limit as I described, the power dissipation in r4 is very high st high currents.
 
Top