I bought this 24VDC 60W power supply with PWM output for LED dimming. I however, intended to leave it at 100% duty cycle and use it as a regular 24V DC power supply. The load is a ticket machine which draws ~1.1A peak on startup inrush, ~30mA during normal operation with basically an LED on, and ~2A (for less than 0.1s) when a solenoid physically stamps a ticket.
If I connect the machine to the 24V output of the supply, then plug in the AC side of the supply, it trips either the short circuit or over-voltage protection, and I need to cycle AC power before the supply works again. I took a shot of the startup voltage in this scenario on my oscilloscope. It peaks at about 29V and the datasheet for the supply says over-voltage kicks in at 28V:
I don’t have a current probe so I put a 0.5 Ohm resistor shunt on the 24V line to measure current on oscilloscope, and got a short peak of only about 1.1A with the machine connected and AC power plugged in cold. Well within power supply specs. The power supply protection did not trigger this time however, and the ticket machine ran happily and within all the power supply specs I have read and understood so far.
Long story short, a fair bit of testing shows the power supply starts up reliable Lots of testing shows that it works reliably with this inline resistor. I looked at output voltage with this 0.5 Ohm resistor in series with the ticket machine, and it looked like what I expected, (pretty quickly ramping up to 24V and staying there). Exact same settings as in image above:
So my questions are:
- Did I go wrong in specifying a PWM power supply? If so what would be more appropriate? Constant voltage supplies only? I needed something 24VDC, >=3.5A peak, with “Class 2” power limited output per Canadian standards, and in an enclosure suitable for outdoors/wet location. Ideally something in the same form factor…
- Can anyone explain what is going on? Something on the ticket machine startup is incompatible with the power supply startup, obviously but I do not have internal schematics for the machine. I presume it has a little power supply of it’s own for microcontroller, etc. voltages
- Does anyone see a problem with adding an inline 3W 0.5 Ohm resistor and calling it good? Startup power consumption and peak power consumption are within spec of 3W resistor (can handle 2.5x continuous rating for <5s) and voltage drop is not much. The machine works reliably and the resistor doesn’t even change temperature as far as i can tell. I have a tight deadline on deploying these ticket machines and the client already bought enough PWM-60-24s, so I’d love to be able to use them.
I left a message at MeanWell’s public tech support phone number but have not heard back yet… will post here if I do.
I cannot be certain, but it could be that this device doesn’t much care for inrush current and the high loads that come from it. It can sustain the load within spec, but the rapid cycling may be causing some gremlins as the device was designed to drive LEDs rather than inductive loads. I don’t think the inline resistor should cause significant issues, though I might also look into an ICL. We’ve got some half-ohm ICLs that could potentially help available here: https://www.digikey.com/short/d7m9zvqh
Ok thanks. I am not surprised now that you’ve pointed them out, but I never even thought to look for purpose-built inrush limiters.
My best guess was that an LED string connected to this would have some series resistance to limit current at 100% duty, and my ticket machine must have close to zero series resistance (until a core on it’s power module saturates or something), and this resistor adds just enough to sneak it by the protection mechanism.
Typical AC-DC converters are usually equipped with a fairly beefy output capacitance in order to reduce ripple and improve transient response. That’s not so helpful however in a context where one wants to make a bunch of LEDs blink at a fast enough rate so as to minimize visible flicker while dimming.
Likewise, it’s probable that your powered equipment has some input capacitance, which will require a fair bit of current to charge during power-on. There will be some inductance due to the leads and probably also the transformer within the supply, and connecting an L & C together is a common way of making scope traces look boingy. That added half ohm of R was probably just enough to de-tune the resonant circuit enough to keep whatever protection mechanism was tripping from kicking in.
I’d avoid the inline resistance if possible–while it might “work” it’s rather undignified, and dependent on the transient nature of the load to avoid heating to an unpleasant degree. Resistor power ratings are commonly reckoned as the amount of dissipation that will result in the maximum rated temperature being reached when in free air at room temperature, so if there was ever a situation where the supply delivered its full current for any sustained period, that 3W resistor would likely go north of 150°C.
There are a few IP6X rated converters here that might be worth a look. Note that devices called inrush current limiters are often just a dressed-up NTC thermistor, and will run hot by design. Unless, as in this case, the load doesn’t draw substantial current most of the time. In that case you’ll have a cold NTC that’s not going to get out of the way when you want a quick slurp of current from the supply…
It’s unclear whether you’re doing anything at board level, but folks wanting low-loss reverse-supply protection will often stick a P-channel FET in the high side, such that it’s biased on during normal operation. Add a bit of C in the bias circuit, and an inrush limiting behavior can be had pretty simply.
Thank you, very informative. I have ordered one of these to evaluate but I need more than Digikey has in stock, and unfortunately my client already ordered enough of the PWM supply from Digikey on a guest account (though we might try to return it). Any thoughts on adding my own capacitor across the 24V output, or other better ways to mitigate this with the PWM supply? You said “de-tune” so If I succeed in de-tuning by adding capacitance, that is probably better than the resistor and related heating.
Edit: I’m not doing anything at board level on this project. It is as simple (or not, it turns out) as connecting a 24V device (ticket machine) to a 24V supply.
It may be instructive to take a look at the wikipedia page on RLC circuits excerpted below, particularly the section on damping factor; higher damping means less ringing, and one can get that through more R, more C, or less L.
Added C would thus seem like an approach worth exploring, with the knowledge that it can bring problems of its own.
One more question rick_1976 if you don’t mind: would a constant voltage LED supply like this be more suitable for use as a general purpose 24VDC supply? Or am I going to run into output circuits tailored specifically for LEDs (and likely to trigger some protection under reactive load) with anything labelled an “LED Driver”?