8
\$\begingroup\$

I'd like to build a bench power supply for prototyping and development work. Most of the benchtop power supply designs I have seen are based around a transformer, rectifier and linear regulator (eg 7805/LM317).

However, I like the idea of keeping it efficient, so am interested in building something switchmode instead.

My work is mainly microcontroller-based, but I mess around with audio stuff from time-to-time too. I'm thinking that a 3-rail design (±15V to 0V adjustable, +5V, up to 1A) would cover most things.

Would the increased ripple voltage of a switchmode power supply make it unsuitable for general-purpose use?

Or, given that I'll be relying on this power supply for a lot of future work, should I stick with the simplicity of a linear regulator design?

\$\endgroup\$
  • 1
    \$\begingroup\$ Switching supplies don't necessarily have worse ripple, but they are more commonly worse. Designing a well-performing switching converter is something pretty specialized, while throwing together a linear supply is relatively easy. The main thing people are looking for in a benchtop supply is accuracy and low-noise, not efficiency. \$\endgroup\$ – endolith Jun 1 '11 at 15:55
  • 1
    \$\begingroup\$ @endolith: I'm happy to spend a bit of time coming up with a decent design if it's feasible; cranking out another linear supply isn't really a challenge anymore. If it's possible to get all three (accurate, low-noise and efficient), I'd like to have a go at it :) \$\endgroup\$ – Jeremy Kerr Jun 2 '11 at 2:48
5
\$\begingroup\$

If your output voltages are fixed you should be able to design it with linear regulators with a minimum of voltage drop over the regulators. And 1A is not the end of the world.
A switcher would be more appropriate if you have a variable output voltage. A linear regulator which can supply 30V @ 5A, but which is set to 1V out it will have to dissipate 150W.
An SMPS is dimensioned for a certain input voltage, output voltage and current. If you would vary the output voltage its efficiency will be lower. You won't get the 90% figure manufacturers boast about over you full range, maybe more like 75 to 80%.

edit
If you want the +/-15V adjustable I'd go for the SMPS, for the dissipation reason I mentioned. You can attack the ripple with a pi-filter. A high frequency switcher (like 1.5MHz) will allow for a smaller coil.

\$\endgroup\$
  • \$\begingroup\$ It would be nice to have an adjustable output; I'll change the question to reflect this. \$\endgroup\$ – Jeremy Kerr Jun 1 '11 at 8:29
  • \$\begingroup\$ You could make the input to the regulator switchable between different ranges to make it more efficient? Like a rotary switch selects different ranges of output voltages, and changes the tap on a transformer at the same time? \$\endgroup\$ – endolith Jun 1 '11 at 15:59
  • \$\begingroup\$ @endolith - that's a possibility, and it's being done on certain bench supplies. Switching is done automagically by relays based on the output voltage. \$\endgroup\$ – stevenvh Jun 3 '11 at 7:03
2
\$\begingroup\$

Switchers don't necessarily have worse ripple than linear supplies. If that was the case they wouldn't be used in very demanding applications like supplies for FPGAs and other high-performance systems. If you use a Nat Semi Simple Switcher with their on-line design software I think you can specify the amount of ripple as part of the design process.

I use one of these bench supplies which incorporates a switcher. Ripple and noise are 100mV p-p.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.