Linear Technology LT8614 “SilentSwitcher”

The Linear Technology LT8614 “SilentSwitcher” claims to reduce EMI/EMC emissions to well below the CISPR 25 Class 5 limit. It also seems a well specified step-down switching regulator with 2.5µA quiescent current, up to 42V input and 4A output current. It also has a high efficiency of 96% at 1MHz (it is a synchronous regulator). It does not seem a cheap IC and while it is listed as $6.07 in 1-99 quantites on Linear Technologies’ web site, if you go to Digikey or Newark to buy it you will be paying $9.07 for the LT8614EUDC#PBF not $6.07.

The data sheets and application notes do not give much detail on how the low EMI is achieved but their application note AN144 gives some clues with a comparison of the switching node voltage with the LT8610:

Here is can be seen that the amount of ringing on the switching node is considerably reduced compared to the LT8610. Other devices rely on snubbers to reduce this ringing to meet CISPR 25. One example is from an application note SLVA619 from Texas Instruments for the TPS54160-Q1. This application note concerns the same issue of meeting CISPR 25 radiated emissions. As well as some input voltage filtering they add a 10W 1.5nF snubber on the switching node. The Texas Instruments SLVA619 is rather economical with words and doesn’t actually say what the effect of the snubber is e.g. does the circuit only pass CISPR 25 with the snubber? The effect of (presumably) the lower ringing of the switching node is shown in the comparison of radiated emissions measurements from the Linear Technology application note AN144 for the LT8610 and LT8614:

However, you might ask why the LT8614 is necessary when the LT8610 non-SilientSwitcher meets the specification anyway? I don’t know the answer to that but could guess at the need to keep the EMI low because the switcher will not be the only device generating EMI. Another reason could be that the bigger the margin between the limits and the likely EMI, the better chance you have of meeting the EMI requirements in a “real” application.

Good PCB layout is required to get good EMI results and while you might assume that you should always be able to have a good PCB layout, constraints such as cost and size may make the PCB layout less than ideal. The LT8614 demo board uses a 4 layer PCB – you may want to use 2 layers to save cost. The constraints of PCB size, shape and the need to put other components on the PCB could all further restrict the PCB layout and make it less than ideal. A design that is inherently better for EMI will therefore presumably fare better than one that is inherently noisy.

The LT8614 datasheet and applications note seem to mostly be concerned about the radiated rather than conducted emissions. I am not sure if this is because the radiated emissions are much improved on the LT8614 (by around 20dB) but the conducted emissions aren’t, or if it is because the conducted emissions are highly dependent on what supplies the input power. However they do give an “Ultralow EMI” schematic in the data sheet and it is included on the demo board and can optionally be used. It would have been good if they had actually given a comparison of the benefits of the additional input filtering as well as the comparison with the LT8610 for the radiated emissions.