What kinds of things can cause a power supply to cut out?

Question

We've been using an HBB5-3-OV-A+G power supply in our rack mount products. One output powers a small fan, and the other powers our product which is an assortment of digital and rf boards and an LCD. It's a 3 amp supply, and our boards and display draw about 1 amp total.

Recently we've been having a problem. Occasionally the output that powers our boards cuts out. When we disconnect and reconnect the AC line it comes back on. The side that powers the fan stays on.

This sometimes happens even when the rack chassis cover is off, and the unit hasn't been on for long, so it doesn't seem like a thermal protection thing. Some units never do this, some do it several times a day.

The 1 amp current measurement is based on the readout of a bench supply, so it wouldn't indicate any short current spikes. I'll have to find some way to monitor the current on a scope. I can't think of any reason why it would spike, though. There are some relays in the device, but they stay in the same position except when the user enters a command, and it doesn't cut out when they are switched.

So, is there anything nonobvious that can trip a supply's overcurrent or thermal protection?

Answer 1

The specification for the power supply states that the current limiting is "Automatic current limit/foldback". Foldback current limited supplies can cause problems with non-linear loads, especially those that draw substantially constant current. This is because there can be two or more stable operating points, one with the supply in constant voltage mode and one or more with the supply in current limit mode. A brief current spike could 'knock' the supply from one mode to another.

But this is somewhat speculative since we don't know the design of the power supply nor the IV characteristics of your product. It would however be worth trying a constant current limit supply instead.

This is an example of what can happen

Answer 2

Problems like these are so hard to trace because they are transient and elusive. but its a great question for me a former Test Engineer.

If you have some sort of Functional Automated Test for your unit, I wonder if it monitors peak current consumption. If not, then the best tool is a digital scope and a current probe or a milliohm level shunt resistor with two balanced probes on V+, and look at A-B for current and A for voltage. When the effect is triggered in a one shot mode you can see which side of your probes lies your fault.. The supply with noisy cables and remote sensing? or your load with current surges from some abnormal condition.

I remember a PC with a bus mounted WiFi adapter card. Every time the PC boot to windows, and the drivers for the RF would load, the PC would shut down from standard UVP sensing. Since I had bought 2 identical PC's for my brother here, but only 1 WiFi card at the time (8 yrs ago?) . I swapped power supplies and the problem went away. IN this case the unexpected load transient on the PS bus caused a dip, enough to trip it. It may have been one of those faulty PSU's with caps that had poor ESR from a big company who is no longer in this business and had many recalls over this issue.

My 2nd inclination is to add a large tantalum or LOW ESR Cap like 20mΩ to the load after I verified my hunch. My 3rd is to verify my hunch that it worked and did not cause another problem with resonance.