I'm looking at (the millions of) buck regulators available for a project where the board will be a host board for a Gumstix Dual Cortex A9. This can pull quite a bit of current at maximum so I was figuring 2.5A for a good bit of leeway. However it will also be battery powered so, assuming I can get the Gumstix and the rest of the equipment to sleep efficiently, the quiescent current of the regulator becomes important.

I was thinking that would mean I should use a non-synchronous regulator as, traditionally, the synchronous have been poor in low current modes. The non-sync can operate in Discontinous Current Mode due to the freewheeling diode, which the sync can't do. However, there seems to be synchronous regulators that can do it all. Looking at the LT8610 from Linear, it has high efficiency at high power and high efficiency at low power. It achieves this by using "Burst Mode" at low currents. Apparantly down to 2.5uA which is quite something.

My concern is that there may be a catch. I believe various Burst Modes can lead to chirping - that irritating squeak that comes from the bursting on-off frequency causing the capacitors/inductors to vibrate at audible frequencies. I used to hear that quite a lot, mainly from laptops, and I don't really want the same thing in my project.

The other catch of course is that the LT8610 is nearly twice the cost of a simple non-synchronous regulator like a TPS5420 and is harder to work with (16 pin MSOP vs 8 pin SOIC). I can deal with that if it's has nice features.

It is somewhat mystifying how systems like the Kindle manage to be in sleep mode to the extent that they are. Perhaps they use linear regulators... or some sort of combination

  • \$\begingroup\$ A switcher that delivers high efficiency across a wide dynamic range of current, such as this requirement seems to be, would not fit into (the millions of) buck regulators category. A price premium is expected for a niche part. Besides, Linear isn't exactly the low-cost option in its pricing strategy. A hybrid regulator approach such as your link, or an integrated version thereof, seems to be used in some hand-held devices, possibly for this reason. \$\endgroup\$ – Anindo Ghosh Aug 1 '13 at 10:43
  • \$\begingroup\$ True. There are many many regulators out there but once you start looking for all dancing chips, the choices narrow up somewhat. And I do expect the premium. I only saw the hybrid approach at the last moment and I should investigage this more, especially as I expect a PIC in there somewhere as a helper device (maybe display "Starting" or "Charging" messages while powering on the GumStix). I'd use a linear regulator for that. \$\endgroup\$ – carveone Aug 1 '13 at 10:54
  • \$\begingroup\$ Actually, it's possible I'm being dim here. I'm wanting low quiescent in low current without asking "why"! :-). Say I don't want to lose more than 1Ah over, say, 200 hours. That's still 5mA. Then a monitor could shutdown the processor and turn off the regulator (that's what my laptop does anyway - goes to hibernate). So maybe it doesn't matter that much. Given that a, say, TPS54260 also does pulse skipping, I'd still love to know does that result in audible chirping. Thanks all. \$\endgroup\$ – carveone Aug 1 '13 at 15:42
  • \$\begingroup\$ Sample the new Texas Instruments TPS65290, it seems to be suitable for all your requirements. No, it won't chirp. \$\endgroup\$ – Anindo Ghosh Aug 1 '13 at 19:43

I'll try and answer this myself and see what others think. The answer appears to be "it depends".

In general, a smps will not generate audible noise operating at their normal switching frequency (say, 500kHz). However, under light load conditions, a synchronous regulator may enter into a burst switching mode into order to reduce standby consumption (some can be prevented from doing this). When the frequency of the switching bundles enters the range of human hearing, audible noise may be generated:

Normal vs Burst mode operation

The amount of noise emitted can be mitigated using careful design. Transformers/inductors produce electromagnetic fields which cause forces between the coil elements, setting up vibrations in the component. This can be minimised by varnishing or using adhesive. Ceramic capacitors exhibit piezoelectric properties which depend on the types of materials used in their manufacture (barium-titanate is particularly bad). Using a Class 1 capacitor or a film type instead of ceramic type can reduce audible noise.

Other than that, some manufacturers may have IC specific advice in their datasheets. For example, the LT8610 can use Burst mode or pulse skipping mode depending on the logic on the sync pin (I don't know what effect that would have though!).


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.