Benchtop power supplies: linear vs switching?

Get whatever meets your needs for voltage, current, readouts, size, price, etc. Don't worry about whether it is a switcher or linear.

In general, linears are less efficient. However, this matters little to a bench supply. The few watts or even 10s of watts it might occasionally draw more than the equivalent switcher is irrelevant. It will get hotter, but presumably since you are buying a whole box this has been designed in. Unless perhaps you have a very specific physical spot in mind for this box and there is little room for ventillation, the extra heat of a linear won't matter.

Switchers will have some switching noise on their output. Again, this shouldn't matter. Check the ripple spec, but the ripple of any finished-box commercial lab supply really shouldn't be that high, a few 10s of mV at most.

What exactly is the problem with ripple? Not much in a bench setting. Things like relays, motors, LEDs and even the occasional LEB (light emitting bulb) aren't going to care. But the most important point is that a well designed circuit should be fairly immune to power supply ripple. If your circuit can't handle a few 10s of mV of supply ripple, what's it going to do when it gets off the bench? In the few cases where supply ripple might matter, you should get into the habit of adding the appropriate filters anyway. For example, to power the opamp for the sensitive input circuit of a microphone preamp, put a ferrite chip inductor in series followed by maybe a 10 µF cap to ground feeding the opamp power pin. Other places may need a bit of filtering too, but that's something you should be doing anyway. Using the microphone amp as a example again, the final stage may draw enough power so that it makes its own "ripple" on your local supply, whether the original power supply was perfectly clean or not. This is just normal design practise.

So all this is a long way of saying don't worry about it. There are even hybrid types where a switcher does most of the work with a linear post-regulator that only drops half a volt or so to clean up the noise or let you get down to low currents and voltages nicely (which some switchers have a hard time with). Again though, you are buying the overall box. Look at what it does as a black box and don't worry how exactly all the specs were accomplished.

Linear power supplies are less efficient, and waste more power as heat, than switching power supplies. This also results in switching supplies being able to provide higher current for a given cooling arrangement of the power supply design, whether passive, fan-cooled, or something more exotic.

A well-designed switch mode bench power supply ought not have higher ripple or noise than a linear supply. When operated within the load specifications of the bench power supply, the output voltage quality should not be a concern.

Given that a linear power supply uses simpler electronic design principles than switching supplies, but on the other hand has heavier cooling needs, an assumption that an unspecified bench power supply is one type or the other is not justified, contacting the manufacturer would be your best option.

Lab supplies that meet your specifications are very commonly used in industry and should be easy to find. HP/Agilent, BK Precision, GW Instek and many others make good bench supplies.

A purely linear supply will be large but will be the quietest solution. A switcher with good linear post-regulation (the linear stage will smooth out the switching noise and take care of the volts and amps regulation) will be almost as good and certainly adequate for the vast majority of bench work.

Most of the bench supplies I've played with are of these two types, and in practice they both produce DC that's cleaner than the typical flyback converter housekeeping supply used in many power supply designs. You should be fine.

A pure switcher, even with adjustable output, isn't the best choice for this sort of work. You won't get very clean output if you try to adjust the voltage very low (you may even get into cycle-skipping mode with larger-than-expected output ripple).