Does switching on a bench power supply cause a voltage spike on its outputs?

Question

In an Adafruit guide, it is claimed that bench power supplies may produce a voltage spike upon turn on:

Source

I cannot find any reference to this. My assumption is that their target audience (hobbiests) may have low quality bench power supplies, although they claim that even reputable brands may produce these voltage spikes.

My questions are:

1. Does this apply to linear or switching power supplies?
2. Does this apply to power supplies with a seperate output switch?
3. Is this caused by user error? Is switching on a power supply with the load attached user error, or is there a seperate issue such as long leads?

Answer 1

It's quite difficult to get a circuit to have a clean behavior at power-on and power-off: filtered reference voltages may take some time to stabilize, opamps don't work properly until they get the required supply voltage, microcontrollers have to boot, etc.

However, it is absolutely possible to make a bench power supply that starts up and shuts down cleanly, without exceeding the user set voltage. Some do that with an output relay that only connects the load when output voltage is stabilized, others do it by design... and others fail and make spikes at the output.

In my opinion, we should encourage manufacturers to fix this by not buying power supplies that spike at turn-on or turn-off.

Then, the most likely way to damage stuff is to set the power supply to a high voltage, forget about it, then the next day plug in a microcontroller, and turn it on.

Another way of destroying stuff occurs with analog supplies: when the potentiometer eventually dies, if it is a good design the output voltage will fall to zero. If it is a bad design, it will max out. There is no way to know without looking at the schematics. I've had this happen, it jumped from +15V to +40V when the pot died, and it fried all the opamps.

Does this apply to linear or switching power supplies?

Both.

Does this apply to power supplies with a seperate output switch?

If the supply powers up to the "off" state and you have to push the button to turn on the output after it has stabilized, that's one way to avoid the problem.

Is this caused by user error (is switching on a power supply with the load attached user error, or is there a seperate issue such as long leads)?

It's only user error if you know it's going to spike and turn it on anyway. Basically, don't buy a model that does it. A bit of search on forums should answer the question.

What I would call "user error" would be to set the voltage to, say, 10V and current to 20mA, then plug a LED into the output. The current limit only begins to work after the output capacitor has discharged into the LED, and at that point it is already fried. This is a case where you must manually ramp up the voltage. Another mistake would be to expect it will be stable with any load capacitance, then use a load with high transient current, which can cause overshoot.

I have a pair of Korad KA3005P, pretty cheap, and they behave well. The UI is nice except for the output relay, which has a quirky UI wich takes a long press on a button. It boots up in the ON state, which makes the relay a bit useless. However it is controllable with USB remote control, which is very useful.

Answer 2

What Adafruit is recommending is powering up the supply, then plugging in the circuit? In other words, hot plugging.

A few years back we did a hot plugging test, powering up a 28 VDC supply then plugging the connector into a circuit. At the input of the circuit we saw spikes of more than 60 V. This was after we had removed the TVS diode on the power line.

The Linear Technology LT3480 DC to DC converter data sheet has a nice illustration of how hot plugging, cable inductance, and load capacitance work together to form a tank circuit, and ways to mitigate it:

I've seen overshoots on the order of 10 %, if you plug in the circuit first, then turn the DC supply (what Adafruit tells you not to do). I suspect it might be tanking as well, but because the supply takes time to power up, it limits rise time of the voltage, and thus reduces the amplitude of the spike.

EDIT

Sorry I didn't fully answer your question. If the spiking behavior that Adafruit is referring to is tanking, then the problem should be independent of supply type (linear or switching) or quality of the supply. The length of the leads and the output impedance of the supply would have a large influence on the spike amplitude. A supply with a separate output switch would have worse spiking problems, assuming the switch allowed voltage to rise quickly.

Answer 3

Yes, of course there can be a glitch.

It does not mean there is a problem with all power supplies.

So that warning may look like absurd but if you have ever fried something and don't know what caused it, you develop procedures that avoid all kinds of problems when turning on equipment or connecting wires.

The instructions to turn on device first, then connect wires, ground wire first makes perfect sense in general, it is not limited to lab supplies. It also prevents you from connecting 5V equipment to lab supply, only to find out it was previously set to 12V and you already turned it on.

1. Yes, both, because it is a generic warning for a generic issue.

2. Unlikely, because the output is disconnected from internal circuits with the output relay or output button when the supply starts up. But the output switch can create another problem. Depending on what kind of wiring you have, connecting output switch creates an impulse current to charge caps and long leads have inductance, so even turning the output on can generate an overvoltage spike that can kill sensitive devices.

3. You don't know how the supply will glitch at start-up. Or if you do, someday it may be old and worn out and develop a feature that voltage overshoots at startup. So it is not an user error but a feature of the power supply. If you do know that it glitches, then it's a preventable user error.

Answer 4

Does this apply to linear or switching power supplies?

A spike can apply to both, but the nature of said spike will be different, there are too many factors to capture perfectly anything that can happen.

Does this apply to power supplies with a seperate output switch?

Having only a switch does not protect you completely from spikes, so it does apply

Is this caused by user error? Is switching on a power supply with the load attached user error, or is there a seperate issue such as long leads?

Switching it on with a load attached isn't really user error by itself, but certain settings can be as someone said before you could have too high of a current limit or you could have too high of a capacitance as a load, as I said before there are many factors that can have an effect in this case.

With all this said, I would not recommend to connect the wires for the LED strip to a live power supply, if you are worried about the spike you can build something that can dampen them and protect your strip every time. Look into an overvoltage protection circuit and you can protect your load without having to connect after the fact.

Main issue with having to connect each time is that you introduce the possibility of someone connecting the GND to the High voltage terminal or other connection mistakes.

Answer 5

Voltage spikes do not usually occur on a good bench supply. They occur on cheap supplies sourced from a certain country that is known for cheap products. The first time I noticed this I was a student and we were plugging some expensive hardware and noticed that we were causing an overvoltage of 2-3x on our main bus line from a cheap supply.

Most of the nice supplies have a ramp and no overvoltage.

Another problem that cheap supplies have is EMC which can cause a scope to trigger on starting the supply.

Does this apply to linear or switching power supplies?

It really depends on the circuit and how it was designed. Most voltage regulators will exibit a ramp. SMPS can have overshoot.

Does this apply to power supplies with a seperate output switch?

Probably, I've seen the cheaper supplys with a master switch exibit this more than - say - a nice tek supply with an on\off button that only turns the rail on and off. If you think about it though, there is a lot of inrush current if the whole supply has to turn on vs just the output

Is this caused by user error? Is switching on a power supply with the load attached user error, or is there a seperate issue such as long leads?

I've mainly seen it in bench supplies with an on off button, but you should check every supply if you are using sensitive electronics.

Answer 6

I remember a horrific incident in the old world order that occurred on a radar test set. At the conclusion of testing, there was a voltage spike that DESTROYED the unit under test. After a long and convoluted investigation, and loss of several test specimens, the responsible engineers (not me, thank goodness) determined that in an effort to obtain the best possible regulation at the unit under test, the sense leads were routed separately through the adapter fixture. Upon shutdown, when the power was disconnected from the unit under test, there was an instant when the voltage sense leads were disconnected and the power supply was running open loop. This was easily fixed in software by having the power disconnect from the unit under test in a separate command prior to shutdown.

The bottom line is this: be sure you understand all of the functions of your power supply, expecially regarding voltage regulation sensing. It would be wise to include resistors between the regulation sense leads and the power output lines near the power supply so that the feedback loop never be open. These will have negligible impact on normal supply regulation via the sense leads. Why didn't the power supply supplier include these internally in the supply?