Trying to make a precision power supply with current limiting

[Dj1312]

Hello everybody !

I write my first message on this beautiful problem because I have a problem where I am really stuck since few days..... 

What I want to make ?
For lab demonstrations, I need to made few precision power supply (voltage supply with current limiting). In the best case, current limit will be adjustable but if not, it is ok : this not my priority in this design but I will be great.
All this supply will work for small Laser Diode or LED (actually processed in the Lab).
In fact, I need 3 of them :
+4,5V /+6V with a limit of 100mA
  +2V / +5V with a limit of 100mA
  +2V / +5V with a limit of 500uA
I plane to just use battery as alimentation. This 3 power supply will be on 3 separate PCBs but I try to made the same topology, so I can only made 1 PCB.

What is my problems ?
I did not find how I can make the voltage regulation : I need a LDO because I want decrease the voltange and avoid any peaks (no buck/boost). If I use a LDO and connect my LED, the current is directly linked with voltage and I did not know what I can put between the LDO and the LED to compensate the voltage difference (in the case of the current limiting is activated)...
For the current limiting, I have no idea where I need to put my 2N3904 on my circuit (I know that in my schematic, it is on a wrong place but where is the right place... no idea).

What about the components ?
Can you say me what you think about this components ?
I plan to use a LT6650 as a 400mV reference (represent as a simple voltage supply on my schematic) to compare with the current measure on a resistor. This resistor will be measured with a AD8276. As a LDO, I will take the LT3014B for the supply with a current limit of 500uA and the two others with a current limit of 100mA. As a general op amp to drive the 2N3904, I use a simple LT1097.

All answers are welcome ! 
Thanks a lot,
Dj1312

PS : Sorry if there is a few english mistakes : I am not a native speaker.

[Zero999]

At the moment Q1 will just short the supply. It should be in series with the regulator, but there are other errors and even if you do manage to stick it together, it's not the correct way to do it.

Four 1.5V, presumably alkaline?, batteries doesn't give any headroom, so the 6V supply won't be regulated at all.

You should use the LT3081, which can do both current and voltage limiting.
http://cds.linear.com/docs/en/datasheet/3081fc.pdf

500µA is a very low current limit and thus an odd requirement: why?

[Dj1312]

Hi !
Thanks for your answer 

I am very flexible and I did not buy any part, so if needed, I can rebuilt a totally new topology !

If I put Q1 in series with the regulator, I will impose the current. But I want a variable voltage supply (for example, which varies with the ratio between R1 and R2) with a current limiter (not a set current).

I did not show that in the schematic but I will also add many capacitors and filters (and also another LDO for the supply of all op amp). Is it what I need to solve the problem you call "batteries doesn't give any headroom"?

LEDs directly made in the lab are very specific and need a maximum current of 500uA. For demonstrations, we did not want to bring the Keithley 2400 all around and this is why I try to design this specific supply.

[Zero999]

Hi !
Thanks for your answer 

I am very flexible and I did not buy any part, so if needed, I can rebuilt a totally new topology !

If I put Q1 in series with the regulator, I will impose the current. But I want a variable voltage supply (for example, which varies with the ratio between R1 and R2) with a current limiter (not a set current).

Actually it's not possible to simultaneously control both the current and voltage. The regulator will have two modes of operation: constant current (the load resistance is too low to allow the set voltage across it, without exceeding the set current) and constant voltage (the load resistance is too high to allow the set current through it, without exceeding the set voltage)

Simply adding a current limiter in series with the voltage regulator would accomplish the above goals. When the current is below the set limit, Q1 would turn on hard, allowing the U1 to regulate the voltage. Then the current limit is exceeded, the current limiter will begin to turn off Q1, to reduce the current and since the same current flows through U1, it will limit the current to the load.

The main downside to this topology is it requires a large voltage difference between the input and output of the regulator.

I did not show that in the schematic but I will also add many capacitors and filters (and also another LDO for the supply of all op amp). Is it what I need to solve the problem you call "batteries doesn't give any headroom"?

No, I was referring to the drop-out voltage. The output voltage of a linear regulator, must be below the input voltage, in order for it to regulate properly. You need a higher voltage battery to give a regulated 6V supply.

[Dj1312]

No, I was referring to the drop-out voltage. The output voltage of a linear regulator, must be below the input voltage, in order for it to regulate properly. You need a higher voltage battery to give a regulated 6V supply.

Ok I see ! So, I will just add another battery in this case or maybe consider directly one with a bigger voltage.

The main downside to this topology is it requires a large voltage difference between the input and output of the regulator.

So, did you have an idea about another topology ?

[daddylonglegs]

  The voltages and currents you need are quite low, could you use an amplifier based design? Something along the lines of the circuit described as "Voltage/Current Regulator" on page 12 of the datasheet for the LT1010 buffer:

http://www.linear.com/docs/1683

  Note:
To get to 0V you would need a negative voltage rail and your positive rail needs to be at least a couple of volts above your maximum output voltage.
The control input for the current limit needs to be low impedance (if you use a voltage reference and a potentiometer you need to add a buffer).

  This won't be efficient as a power supply but can be made fast and accurate.

[Kalvin]

LT1970a is another solution:

http://www.linear.com/product/LT1970A

Check the maximum power capacity and stability. Add external buffering if needed. Quite easy to control and has nice status outputs.

[Dj1312]

Hi all !
Thanks for your answer.

Today, I see one of my old electronic teacher and he advise me to try with a LM723. He did not have one in stock but he give me a non-used L146CB to try. I will make a test and give you a feedback 

[Zero999]

So, did you have an idea about another topology ?

Yes, use one pass element, controlled by both current and voltage error amplifiers, rather than two in series.

Hi all !
Thanks for your answer.

Today, I see one of my old electronic teacher and he advise me to try with a LM723. He did not have one in stock but he give me a non-used L146CB to try. I will make a test and give you a feedback 

Those old ICs are fine as long as you don't expect an accurate constant current mode. The current limiting feature is approximate and is there as a protection measure, rather than to be used as a constant current power supply. The LT1970A, as mentioned above and LT3081, have an accurate constant current mode.

[Dj1312]

Those old ICs are fine as long as you don't expect an accurate constant current mode. The current limiting feature is approximate and is there as a protection measure, rather than to be used as a constant current power supply. The LT1970A, as mentioned above and LT3081, have an accurate constant current mode.

Ok ! So, I will try also with this in this case.
Thanks