Curious: Dimming constant current

[oreo57]

Thinking about some of the lesser drivers out there w/ no dim circuits but run constant current. Can the output be "dimmed" w/ a say MOSFET and of course a control pulse or is there problems with this?

I remember that some Meanwell drivers would pop LEDs if "hot wired" to an active driver..

Chopping the output of a constant current driver would be like hot wiring a string at like 1000x second..

Sounds like a bad idea but I don't know for sure..
Guess is it might be driver design dependent.

 

[Half Vaped]

I think it sounds plausible with some types of drivers, but I think a lot of drivers will have a high initial starting current and can be unpredictable in how they respond to a constantly switching load. Even if it does work, you may be cutting short the life of your driver and LEDs.

I never tried it though. But I agree it sounds like a bad idea.

Why go through the trouble anyway? With the extra parts and labor, it's much more worthwhile investing in a decent PWM driver and controller.

 

[der_wille_zur_macht]

I doubt this would work with any driver. In effect, most drivers increase voltage in the output until they hit their target current. With no load (when your switch is open) the driver basically goes wide open to full voltage since it can't hit its target. Then you connect your load and it is instantly way to high.

If you actually wanted to pursue this a better option might be to switch the input to the driver, not the output. But it would still make me nervous, depending on the how the driver responds and it's own switching frequency compared to the signal you are using. Like half vaped said, it seems like a better idea would be to build or buy a driver that already does what you want versus hacking something that doesn't.

 

[Gorgok]

Without some sort of bypass back to the driver for current regulation it would be as stated. Driver probably maxing out trying to regulate then having to drop rapidly when connected.

With a bypass that is just burning power all the time the driver would just have to switch from light to heavy load, still maybe not a great idea, but probably possible.

The only question is how much you need to bypass for it to work. Most drivers have a low limit on regulation, so you need to hit at least that. And then you can't add more load to it than the combined loads. It would also be a total waste of power, as it would run the bypass all the time no matter what the dimming part was doing, even when you turn them entirely off. At least unless you have a separate switch to then toggle power off to the driver itself.

 

[oreo57]

Without some sort of bypass back to the driver for current regulation it would be as stated. Driver probably maxing out trying to regulate then having to drop rapidly when connected.

With a bypass that is just burning power all the time the driver would just have to switch from light to heavy load, still maybe not a great idea, but probably possible.

The only question is how much you need to bypass for it to work. Most drivers have a low limit on regulation, so you need to hit at least that. And then you can't add more load to it than the combined loads. It would also be a total waste of power, as it would run the bypass all the time no matter what the dimming part was doing, even when you turn them entirely off. At least unless you have a separate switch to then toggle power off to the driver itself.

Trying to get my head around that..

Anyways one small twist..
I've on numerous occasions "hot swapped" LED strings out w/ LDD drivers.
I know for a fact that w/ "no load" they output the 56V of my power supplies..at least according to a VOM.


Even smaller strings don't "pop" (say V(f)(tot) of 11V, deep reds)
and never noticed a surge of brightness in connecting though that IS possible..

Doing it 1000x/sec might not be the same though..

Still, point is it might just be driver design dependent.
Was always told the Meanwell AC/DC (some I suspect) all in one drivers would pop diodes if done the same..

hmmm..

Agree w/ all.. not a wise choice though, no matter..

 

[der_wille_zur_macht]

If you really want to know if or why it's safe in one situation versus another, you're getting way in to the weeds in terms of the style of driver, how it delivers energy for the output stage, and how it stores energy to smooth the output ripple. Driver performance even for a specific brand and model of driver is almost certainly not consistent across it's entire design range either, so a driver that's safe hot-swapping one string of LEDs may pop another string.

Plus as you insinuated, manually hot swapping is a lot different than trying to dim via PWM-chop of the output. Even if you can hot swap "safely" and you proceed with your experiment and find that you can chop the output with a PWM signal "safely" you may find that you don't actually get reliable dimming performance that resembles your signal against your target current. The driver is meant to match a target current in constant-state operation, the first tiny fraction of a second right when it's hot swapped isn't guaranteed to match the desired steady state - you may find that you're basically always operating the driver outside your design goal for output current because you keep turning it on and off again before it's had a chance to regulate and stabilize. Even when a driver IC is designed with PWM dimming in mind there are often situations where the chip just can't cut it and the driver is designed to avoid those situations - ie drivers that cut dimming below a minimum threshold.

Suffice to say, hot-swapping, much less controlling the output like you've described, is pretty much never a design feature and hence is going to be unpredictable at best IMHO.

 

[der_wille_zur_macht]

Trying to get my head around that..

Basically what he's saying is you can put a shunt - a dummy load - in your output, and then even as you switch the LED string in and out of the circuit, the current is always flowing through the dummy load (since it's not in the switched portion of the circuit) so there's always at least some load on the driver. This gives the driver feedback even when the LEDs are off, so it has something against which it can regulate current.

It's a great theory but as Gorgok pointed out, there are a number of reasons why it probably doesn't make sense. First and foremost, you're basically throwing a ton of power out as heat through your dummy load, which kind of defeats one of the main reasons to use LEDs. You also need to over-size your driver and power supply for a given target LED load, since some of the output is being wasted. It's also not guaranteed to actually work, since (once again) we don't have any reasonable assurance that a driver will perform as designed when the load is variable.

 

[theatrus]

There are a number of driver topologies which chop the output by basically shorting the output to ground. All of the "dimmers" you see for LED lighting ropes work this way. There isn't a need for a ballast resistor in theory - that will waste extra power. The driver should regulate down to the same current. However some drivers will think they hit a short circuit and then crowbar off. Instead of a resistor, pick a FET with a shitty Rds(on), but heat sink appropriately, if you need some extra load.


As stated in here, you don't want to go open circuit, as the drivers loop control can be too slow before LED damage occurs, especially if you have miles of wiring (inductance).

 

[oreo57]

Thanks gang. Very informative, as usual..

yea got the shunt part.. Dummy Load would (should) need to closely match the LEd load for one thing so the driver sees, as close as possible, a steady state condition..

Lets just leave it in the realm of BAD idea..

Neither electrically wise nor economical..
Intuition and the fact it is "not" done should be good enough..

 

[Legot]

The easiest and most effective way I've seen and used for linear drivers was too use PWM from the input side, rather than switching the output. I don't recall any switching drivers that aren't capable of direct dimming of the output via PWM.