# Is a switching power supply likely to fatigue or burn out early if turned on or off quickly?

I'm working on a design for growing biomass where I need high power LEDs to switch on and off somewhere in the neighborhood of 1-30 times per second, so that they provide light to grow but don't oversaturate the organism's photosynthetic chemicals.

While I would guess that a better approach to powering these would be to directly power them with transistors or something could be guaranteed not to mind switching at that rate, for a quick and dirty prototype, I've got a switching power supply on the way that will provide appropriate DC voltage and amperage to the LEDs.

The question is: if I switch that power supply itself on or off at 1-30 Hz, is it likely to burn anything in it out prematurely?

I assume that this is less an issue with a switching supply than a solid state one since it won't be heating and cooling as much, but I don't know if there are other failure modes I should be looking out for.

• Switching supplies are solid state ... Do you mean linear ? – brhans Aug 17 '16 at 0:42
• Have you considered using a proper LED driver with PWM input? – Ignacio Vazquez-Abrams Aug 17 '16 at 0:59
• Are you switching the supply on and off to generate an average brightness? Or is there some biological reason for applying light pulses rather than constantly? – user4574 Aug 17 '16 at 1:18
• @brhans, yes, I said solid state, meant linear. – Aaron Altman Aug 17 '16 at 18:43
• @user96037, yes, there's a biological reason called photoinhibition. – Aaron Altman Aug 17 '16 at 18:43

Only a general answer is possible with the information you've provided.

In general, this sort of thing wouldn't be a great idea. If you had a mechanism for switching the (I'm assuming mains power) on and off that fast I'd be inclined to use it to switch the output instead of the input.

Power-on is often a stressful time for a switch-mode power supply (SMPS) due to inrush current. There would likely be less repetitive inrush current than the "cold turn-on" inrush, but the likely inrush limiting mechanism would also likely become disabled after a short run time, so that's the first risk.

The second thing I can think of is that there's no guarantee the output goes to zero immediately after input power is withdrawn. There are power storage elements in there that will generally keep things going for a while. So you might not get the effect you want without an extended off-time, and that would lead to closer to cold turn-on effects with inrush current limiting disabled. So that's thing two.

Further, depending on the supply, it might not be terrifically well behaved during start-up and shut down. There could be voltage spikes and so on which might end up damaging your LED array or whatever load you're running.

So in short, I'd say not a great idea. I'd be thinking to work on an output power switching setup right from the get-go.

If you don't mind burning up a set-up or two you could experiment to see how it reacts and maybe take some scope measurements of the voltage and current. It's possible it might work.

Are you feeling lucky? :)

• He shouldn't be switching the output either as many SMPS have a minimum load. The only sane way to do this is a proper AC->DC followed by a switched current supply, or ideally, just use a dedicated LED controller. – jbarlow Aug 17 '16 at 5:50
• @jbarlow is quite right that some SMPS's have a minimum load, emphasizing the point that a working solution would consider many other factors, details of the load (voltage, current, etc.) and details of the selected supply. So these remarks should be taken as general guidelines. You could post new questions to solicit guidance on particulars, in which you'll need to provide all these specifics folks need to see the problem clearly. Although SMPS's seem like black boxes (literally :) when connected to our devices, there's a lot of work to matching them to the application. – scanny Aug 17 '16 at 19:36
• Thanks for the comments. It sounds like a dedicated (PWM?) controller is probably more what I'm looking for. Two things I'm unclear on: 1) Are there reasonably priced (let's say, less than $15 each) controllers that handle >= 3W loads? 2) Do any of those controllers do PWM as slow as I'm looking for, in the range of 1-30Hz (important for the biological portion of the process, see "photoinhibition") – Aaron Altman Aug 17 '16 at 19:45 • @AaronAltman That would take us into the realm of so-called shopping questions, which are strongly discouraged here on EE.SE. I'd say Google would be your friend on that one. If you come up with a candidate that looks good but you're not sure, you could post a question about that specific one against your requirements. You probably want to work up a schematic (maybe block level) so you can communicate your proposed design in future questions. And don't forget to pick a "correct" answer. That's how your respondents get "paid" and thereby keen to answer your next ones :) – scanny Aug 17 '16 at 19:51 • I'm asking whether such things exist, not for specific recommendations :) – Aaron Altman Aug 17 '16 at 21:59 When a power supply starts there is typically some inrush current associated with charging its input and output capacitors. That inrush consumes a certain amount of energy. And if you were to repeat that process it would amount to a certain wattage. WATTAGE = ENERGY x FREQUENCY. So yes rapidly switching a supply on and off will generate some extra heat in the supply If you want to pulse the output of your supply then you would be better off keeping the supply on constantly and then putting a relay on the output of the supply (rather than on the input). The relay would connect the supply output to the LEDs. And you could switch the relay 30 times a second if desired to control the average LED brightness. Unless you have some specific reason for pulsing the LEDs rather than providing a constant brightness, you would be better off getting a supply that provides the correct brightness to begin with, or one that is adjustable. To pulse the LEDs you could use a 555 timer chip such as the NA555P to generate 30Hz pulses... http://www.ti.com/lit/ds/symlink/sa555.pdf Which can be purchased on Digikey for$0.42.
http://www.digikey.com/product-detail/en/texas-instruments/NA555P/296-21753-5-ND/1571935

And feed those pulses into a MOSFET gate driver such as the MCP1407-E/P...
which can be purchased on Digikey for $0.98 http://www.digikey.com/product-detail/en/microchip-technology/MCP1407-E-P/MCP1407-E-P-ND/1228640 And then feed the gate driver output into the gate of a power MOSFET such as the PSMN027-100XS,127 http://www.nxp.com/documents/data_sheet/PSMN027-100XS.pdf which can be purchased on Digikey for$1.08