Is there such a thing as a lab constant current power supply?

[Blitzschnitzel]

I experiment a lot with lighting and it would be great to have a programable constant current power supply, that can cover a wide range of currents and voltages.
I read somewhere that the current limit of a regular lab power supply is too slow and unusable as a LED driver.

[tszaboo]

Sure there is. You can use an SMU for that.
Keithley also made several current sources, you can pick up some quite cheap if you live in the US.
For less money, a there is also the Advantest R6142.
How much is your budget?

[calzap]

I've used HP and Instek power supplies many times for a variety of LEDs with no problem.            Mike

[wraper]

I read somewhere that the current limit of a regular lab power supply is too slow and unusable as a LED driver.

It's only an issue if you enable the output and attach a relatively weak LED. Charge stored in output capacitor may kill it. The safe way doing this is setting the output voltage to the level where inrush current will be limited by Vf to a safe level. You can rise the voltage afterwards if you need.

[Blitzschnitzel]

Sure there is. You can use an SMU for that.
Keithley also made several current sources, you can pick up some quite cheap if you live in the US.
For less money, a there is also the Advantest R6142.
How much is your budget?

I had never heard of an SMU I'll have to read up on that. Thank you.
The current sources seem to be precision instruments and max out at 100mA. I would need something like 0.1A-3A and 2-200V. As much range as possible but a current variance of +-5%would be absolutely acceptable.

I've used HP and Instek power supplies many times for a variety of LEDs with no problem.

Do you mean just at a fixed voltage or in a current limited mode?

It's only an issue if you enable the output and attach a relatively weak LED. Charge stored in output capacitor may kill it. The safe way doing this is setting the output voltage to the level where inrush current will be limited by Vf to a safe level. You can rise the voltage afterwards if you need.

Interesting. How do I know the safe levels? What would that look like in practice?

[wraper]

It's only an issue if you enable the output and attach a relatively weak LED. Charge stored in output capacitor may kill it. The safe way doing this is setting the output voltage to the level where inrush current will be limited by Vf to a safe level. You can rise the voltage afterwards if you need.

Interesting. How do I know the safe levels? What would that look like in practice?

If you do not know actual Vf, you can set the voltage to 0, set the required current, enable the output and then rise the voltage. Then you set the voltage 5-10% higher than Vf at required current and use it with no fear of killing LEDs.

[BillyO]

Charge stored in output capacitor may kill it.

That would be a very poorly designed power supply indeed.

I experiment a lot with lighting and it would be great to have a programable constant current power supply, that can cover a wide range of currents and voltages.
I read somewhere that the current limit of a regular lab power supply is too slow and unusable as a LED driver.

Any decent moderately priced bench supply will allow you to set current limits from a milli-amp or so to several amps, and let you set voltage limits, over current protection and over voltage protection.

Why don't you give us a budget and we can throw some suggestions your way?  Then you can go investigate them.

[Gyro]

Charge stored in output capacitor may kill it.

That would be a very poorly designed power supply indeed.
...

Many (I might say most) bench PSUs, linear and SMPS, include a capacitor in the 10s to 100s of uF range, on their outputs to ensure stability. As wraper says, pre-setting the output voltage before connecting a sensitive load to avoid a sudden significant voltage step is the way around this. Either that or set the desired current limit and then wind the voltage setting up from zero.

Either way, ensure that the connections to the LED are secure and can't become intermittent.

[Zero999]

Charge stored in output capacitor may kill it.

That would be a very poorly designed power supply indeed.

It depends on what you mean by poorly designed. A bench supply is typically designed to be as gooder voltage source as possible. The current limit is just that, a limit, rather than a good current source. It's difficult to design a power supply which can flip from being a good voltage source, to current source, based on the output impedance presented to it. The problem is a current source requires a theoretically high output impedance and a voltage source needs the complete opposite; a very low output impedance.

A power supply designed to be a good constant current source, would be a poor voltage source. One way of fixing this would be to add a switch to select between better voltage or current regulation; I've toyed with the idea of designing such a bench power supply, but have never needed one, so haven't bothered.

[BillyO]

It might surprise you to know I am quite familiar with power supply design.

The output capacitor should not be placed after the current sensing.  A decently designed and built power supply would specify a current error amplifier with a decent bandwidth so that the current can be limited within a us or so.

I have one right here I designed myself.  I set the voltage to 12V and the current to 15mA and I can connect LEDs till the cows come home without killing a singe one.

If you have a power supply that will kill an LED under those conditions, throw it out now!

[BillyO]

I don't think OP is looking for something with absolute precision current control.  He's mucking about with lighting, not building a 8.5 digit DMM or a heart-lung machine.

This happens all to often here on eevblog.  Things get kooky fast.

Here is that afore mentioned supply I designed and built myself - not blowing up LEDs.

[Zero999]

It might surprise you to know I am quite familiar with power supply design.

The output capacitor should not be placed after the current sensing.  A decently designed and built power supply would specify a current error amplifier with a decent bandwidth so that the current can be limited within a us or so.

How can that produce decent voltage regulation?

And you need an inductor, not a capacitor for good current regulation, at high frequencies.

I have one right here I designed myself.  I set the voltage to 12V and the current to 15mA and I can connect LEDs till the cows come home without killing a singe one.

If you have a power supply that will kill an LED under those conditions, throw it out now!

Please post a schematic.

[BillyO]

I have no idea what you are talking about.  I was commenting on someone else's remark about output capacitors blowing up LEDs.  I was not talking about voltage regulation at all but fast response in the current limiting circuit.

My power supply design uses no inductor in the current limiting circuitry and as you can see from the video, get's the job of not blowing up LED's done just fine.

As to a schematic, this would not be the place for it, would it?  This would be the place to discuss what the OP needs for his lighting experiments, and I stand by my suggestion that a decent bench power supply would be ideal.

I get the feeling you're trying to pick a fight.  Let's not do that.

[tszaboo]

It might surprise you to know I am quite familiar with power supply design.

The output capacitor should not be placed after the current sensing.  A decently designed and built power supply would specify a current error amplifier with a decent bandwidth so that the current can be limited within a us or so.

How can that produce decent voltage regulation?

And you need an inductor, not a capacitor for good current regulation, at high frequencies.

I have one right here I designed myself.  I set the voltage to 12V and the current to 15mA and I can connect LEDs till the cows come home without killing a singe one.

If you have a power supply that will kill an LED under those conditions, throw it out now!

Please post a schematic.

You don't really need an inductor for that. When I was making battery tester for a living (CC), we had a difference amplifier across a shunt, providing feedback to the error amplifier. This for CC control. CV was software. If you want hardware control, for CV, you have an amplifier measuring the output voltage. Make the error signals a current drain, then whichever is smaller will control the output, and in a one quadrant supply one of them will always be positive. Because it's either CC or CV, so one error will be zero, the other positive. There is one extra step, preventing the error amplifier from going into saturation, eg. diode in the feedback, and you have a quick CC CV controlled supply.

[Zero999]

I have no idea what you are talking about.  I was commenting on someone else's remark about output capacitors blowing up LEDs.  I was not talking about voltage regulation at all but fast response in the current limiting circuit.

My power supply design uses no inductor in the current limiting circuitry and as you can see from the video, get's the job of not blowing up LED's done just fine.

As to a schematic, this would not be the place for it, would it?  This would be the place to discuss what the OP needs for his lighting experiments, and I stand by my suggestion that a decent bench power supply would be ideal.

I get the feeling you're trying to pick a fight.  Let's not do that.

The OP is talking about a lab PSU. I was merely stating that lab PSUs are primarily designed for constant output voltage, rather than current. It might work, but one has to be careful about how they use it. The suggestion of setting the voltage to zero first, is a good idea and will help to avoid damaging the LED.

Why not post the circuit here? It's very much on-topic.

I have no intention of picking a fight. What gives you that idea?

It might surprise you to know I am quite familiar with power supply design.

The output capacitor should not be placed after the current sensing.  A decently designed and built power supply would specify a current error amplifier with a decent bandwidth so that the current can be limited within a us or so.

How can that produce decent voltage regulation?

And you need an inductor, not a capacitor for good current regulation, at high frequencies.

I have one right here I designed myself.  I set the voltage to 12V and the current to 15mA and I can connect LEDs till the cows come home without killing a singe one.

If you have a power supply that will kill an LED under those conditions, throw it out now!

Please post a schematic.

You don't really need an inductor for that. When I was making battery tester for a living (CC), we had a difference amplifier across a shunt, providing feedback to the error amplifier. This for CC control. CV was software. If you want hardware control, for CV, you have an amplifier measuring the output voltage. Make the error signals a current drain, then whichever is smaller will control the output, and in a one quadrant supply one of them will always be positive. Because it's either CC or CV, so one error will be zero, the other positive. There is one extra step, preventing the error amplifier from going into saturation, eg. diode in the feedback, and you have a quick CC CV controlled supply.

A battery charger isn't a generic lab power supply.

You don't need an inductor for using a constant current source to drive LEDs and the like, but it's needed at higher frequencies, just as a voltage source doesn't need an output capacitor for many applications, but it becomes important at high frequencies. Anyway, I suppose that was off-topic, since this thread is about LEDs.

The problem is when CC is being switched to CV and vice versa, it results in current/voltage spikes. An output capacitor will help to minimise the voltage spikes, at the expense of increasing the current spikes. An inductor will reduce the current spikes, at the expense of increasing the voltage spikes.

[pqass]

I experiment a lot with lighting and it would be great to have a programable constant current power supply, that can cover a wide range of currents and voltages.
I read somewhere that the current limit of a regular lab power supply is too slow and unusable as a LED driver.

...
I had never heard of an SMU I'll have to read up on that. Thank you.
The current sources seem to be precision instruments and max out at 100mA. I would need something like 0.1A-3A and 2-200V. As much range as possible but a current variance of +-5%would be absolutely acceptable.

How about a DC supply (either regulated or unregulated) of your choice in series with an electronic load, optional bench ammeter, and your LED-under-test.
Start with the LED-under-test shorted. Set the current knob. Then bring down the voltage to its lowest (close to zero) on the PS before removing the short and slowly increasing the voltage until just past the set current is met.  That should be gentle enough on the LED-under-test.

I have a DIY 2A eload (pot+opamp+NFET+0.25ohm shunt). I'd replaced the usual shunt with a 10ohm shunt which allowed me to dial single-digit milliamps into a blue LASER diode without issue.

[tszaboo]

The problem is when CC is being switched to CV and vice versa, it results in current/voltage spikes. An output capacitor will help to minimise the voltage spikes, at the expense of increasing the current spikes. An inductor will reduce the current spikes, at the expense of increasing the voltage spikes.

There is no switching. There are two error amplifiers that drive an integrator, that controls a pass element. You only see spikes if you let the error amplifiers go into saturation, if their amplification is too much. But that only happens if you don't limit the error signals, and if you are trying to make the supply fast reacting the wrong way.

[Zero999]

The problem is when CC is being switched to CV and vice versa, it results in current/voltage spikes. An output capacitor will help to minimise the voltage spikes, at the expense of increasing the current spikes. An inductor will reduce the current spikes, at the expense of increasing the voltage spikes.

There is no switching.

What I meant by switching is the PSU has to change i.e. switch between CV to CC modes very quickly, which is what causes spikes.

There are two error amplifiers that drive an integrator, that controls a pass element. You only see spikes if you let the error amplifiers go into saturation, if their amplification is too much. But that only happens if you don't limit the error signals, and if you are trying to make the supply fast reacting the wrong way.

It's difficult to visualise what you're describing. Please post a schematic.

[IanB]

I experiment a lot with lighting and it would be great to have a programable constant current power supply, that can cover a wide range of currents and voltages.
I read somewhere that the current limit of a regular lab power supply is too slow and unusable as a LED driver.

To get the thread back on track, what do you mean by "too slow"? I interpret "slow" to mean the time to respond to changes in load and regulate accordingly. Why is that an issue for you? Is it because you want to switch LEDs in and out of circuit while powered, thus changing the load? That may or many not be an issue with cheap power supplies, but not with good ones.

In general, set the current limit to the desired value, set the voltage to zero, connect your LED test fixture, and then turn up the voltage until the current limit is reached. Will work fine on any decent lab power supply

[tszaboo]

The problem is when CC is being switched to CV and vice versa, it results in current/voltage spikes. An output capacitor will help to minimise the voltage spikes, at the expense of increasing the current spikes. An inductor will reduce the current spikes, at the expense of increasing the voltage spikes.

There is no switching.

What I meant by switching is the PSU has to change i.e. switch between CV to CC modes very quickly, which is what causes spikes.

There are two error amplifiers that drive an integrator, that controls a pass element. You only see spikes if you let the error amplifiers go into saturation, if their amplification is too much. But that only happens if you don't limit the error signals, and if you are trying to make the supply fast reacting the wrong way.

It's difficult to visualise what you're describing. Please post a schematic.

Like figure 15.
https://www.onsemi.com/download/data-sheet/pdf/ncp4328-d.pdf
The schematic is
Because the two error amplifier can only sink current whichever becomes larger controls the pass element (or the feedback to the primary in this case)

Just think of any lab power supply as a giant PI controller, with two feedback.

[IanB]

Like figure 15.
https://www.onsemi.com/download/data-sheet/pdf/ncp4328-d.pdf
The schematic is
Because the two error amplifier can only sink current whichever becomes larger controls the pass element (or the feedback to the primary in this case)

Just think of any lab power supply as a giant PI controller, with two feedback.

"Whichever becomes larger" selects the maximum of two signals. And then to avoid jitter, there would usually be some kind of hysteresis in the cross-over, so that the maximum signal selector has a dead band before switching modes. This generally means such supplies work best when either strongly in the voltage control region or strongly in the current control region. If you make them hover right at the CV/CC switching threshold the results will not be so good.

[TimFox]

Yes, there are “lab” constant-current supplies, often used for driving electromagnets, that are not designed as constant-voltage supplies with auxiliary current limiting.  Instead, they have an auxiliary control for voltage limit.
I have a couple of classic Keithleys with modest current ratings that switch in an output capacitor when desired, useful for measuring transistor parameters.

[pqass]

FYI: a good description of a typical HP linear bench supply can be found here.

[calzap]

I've used HP and Instek power supplies many times for a variety of LEDs with no problem.

Do you mean just at a fixed voltage or in a current limited mode?

In current limiting mode.            Mike

[Psi]

I read somewhere that the current limit of a regular lab power supply is too slow and unusable as a LED driver.

For tiny LEDs yeah, and laser diodes, but if you are working with power LEDs I don't think it will be much of an issue.
And I assume you are working with power LEDs as you said 0.1A-3A and even 100mA is more than a typical 20mA led.

Just get a proper branded powersupply from a known good brand rather than one from aliexpress.

The output voltage range of  2-200V  is where you may find it challenging to find something.
That is a very wide voltage range and anything non-standard is usually expensive.
Not to mention how dangerous 200V @ 3A is.