Using LM317 or LM338 as adjustable current limiter for ATX Lab Power Supply

[djcrunkmix]

Hello EEV people  ,

I've bought a regular ATX 450W power supply for dirt cheap ($15). I could have gotten it for free but I didnt have an old ATX PSU lying around.

Anyways, I am apprehensive and a little intimidated by the idea of using ATX PSU as a bench top lab power supply. The main concern is massive amounts of current that the PSU can put out. I've seen several examples of using LM317 (max 1.5A) and LM338 (max 5A) voltage regulators to control the output voltage from 1.25V to 12V. But any ideas on using the same regulators for adjusting output current? I'd like to start off at 0A and slowly rotate the knob to reach a desirable output max current. If I have a bunch of LED's hooked up, I don't want to blow them up by supply too much current (if I had forgotten to connect a 330 Ohm resistor).

Any suggestions would be greatly appreciated.

[Simon]

the regulator has a dropout of up to 2V so your 5V output is now 3. you need a unit with built in variable current control. these units usually sample the output from after the current control keeping the output stable and at what you want

[armandas]

The datasheet of LM338 has an example circuit.

[djcrunkmix]

^ yes, I've taken a look at the sample circuits. Infact, I did build a circuit on NI Multisim and tested it out. When using LM338, I can get voltages ranging from 1.25V to 10.25V using a 2K pot. The voltage starts dipping when the load draws current above 5A - which makes sense.

Now, I also added a LM338 in series as an adjustable current limiter (simpler version of the circuit mentioned by armandas), but the fact is that it uses a 0.24Ohm variable pot that is rated for 2W. Where the heck can I find that :-/

The circuit mentioned above also uses a -5V to -10V reference voltage. Any ideas on how to get it using the existing ATX outputs? i.e. -12V (0.8A max), +3.3V, +5V and +12V.

Regards,
DJ

[Zero999]

The LM317 constant current source is worse: it stops regulating when the input-output voltage is between 3.25V and 4.25V.

If you just need overcurrent protection, you're not using the right part. I'd recommend a polyfuse or a circuit breaker which will suddenly trip then overloaded and reset when the overload is removed.

[djcrunkmix]

^ I could always use a fuse. I am looking for a adjustable current limiter just as you'd have in a professional lab power supply. Suggestions?

[Zero999]

It depends on whether you need current limiting i.e. if it's set to 1A the current will be 1A under short circuit conditions or over current protection i.e. when the current exceeds 1A the current will be interrupted and the load won't be reconnected until either the reset button is pressed or the load impedance increases beyond a certain point?

[djcrunkmix]

^ That was hard to read   But, I think what you're asking is whether I want the current to flow or not under short circuit. If the current limiter is set to 1A, under short circuit conditions, I don't want power interruption. I still want it to flow 1A.  

Here is what I mean:

[alm]

Just look at the schematic of most professional linear lab supplies (eg. Agilent E361xA). It's usually a single pass (darlington) transistor (or multiple in parallel), base fed with a constant current source, and two op-amps (one for voltage, one for current) act as current sinks controlled by the difference between actual voltage/current and the voltage/current limits. The op-amps 'outputs' are ORed by two diodes. The op-amps are usually fed with floating symmetrical supplies centered around the positive voltage output. Not a trivial circuit, and you might as well dump the ATX supply and feed it with a transformer/rectifier/caps. There are also circuits based on two LM317's, but they perform inferior (I'm not aware of any commercial design using this), drop more voltage (two regulators in series) and are inelegant in my opinion (you're basically abusing a voltage regulator to do something different). Another option may be the LM723, although it's not great in constant current mode.

[neoone]

Damn, that's certainly "Not a trivial circuit". All this effort for 30W power supply...

[DrGeoff]

^ That was hard to read   But, I think what you're asking is whether I want the current to flow or not under short circuit. If the current limiter is set to 1A, under short circuit conditions, I don't want power interruption. I still want it to flow 1A.  

Here is what I mean:

If it's a short circuit on the output, you are hardly going to get 9V between the shorted terminals are you.

Adding current limiting requires a current sense resistor and amplifier, which changes the voltage on the ADJ pin so that a constant current flows, however the voltage will reduce to zero as the load approaches 0 ohms.

[djcrunkmix]

^ When you connect two LM338's in series (one regulates the voltage and one limits the current), the simulation tells me that the current draw reaches the max (as defined by the 0.24Ohm resistor in the current limiter circuit) even though the output is shorted (0 resistance to GND).

I will upload the simulation results soon.

[djcrunkmix]

Yeah, you're right. The voltage drops like a rock when its shorted. Here is the simulation screenshot (pardon a ton of voltmeters and ammeters):

Load is in the lower right hand corner. 100Ohm pot set at 0% (shorted).

Baseline (without current limiter. Only voltage regulation):



Shorted:

[kind2011]

its a great idea to have an adjustable current limiter...

[Tooms]

Yeah, you're right. The voltage drops like a rock when its shorted. Here is the simulation screenshot (pardon a ton of voltmeters and ammeters):

Load is in the lower right hand corner. 100Ohm pot set at 0% (shorted).

Baseline (without current limiter. Only voltage regulation):



Shorted:

What is the name of this smart program ?

Thanks
Thomas