TPS54824: conducted and/or radiated EMI data

Hello Ryan,

Thanks for the reply. Yes, I understand that the EMI is very layout dependant, but some manufacturers are starting to give EVM EMI data as an indication of how the part performs.

It seems that quite a few are getting on the symmetrical pin-out bandwagon (e.g. LT8640, MPQ8636). But it makes sense to keep the circulating currents balanced so they cancel out, and hence low EMI. Some of these parts also have spread spectrum which is nice (e.g. LT8640, your LM53635). What would be nice for my design is something that can do 5-12Vin to 1V1out @ 10A, symmetrical pin-out, and spread spectrum. Something for the wish list…

Best regards, Ken

Hi Ryan,

Yes I have been looking at the TPS56C215 and it is high on my list. I had not thought of the DCAP-3 control being SS, but that's a good point. Thanks for the tip! A bit of a shame that it can't sync to an external clock.

I'm also looking at the TPS54020 (and its module version LMZ31710 which shows EMI data) which can sync to an external clock, so I may add an external SS generator, although the pin-out of the TPS54020 is not symmetrical.

And finally I’m considering the TPS54A20. A dual phase 2MHz part with very small layout, but unfortunately it only goes down to 8Vin. I need down to 5Vin for auto cold-start.

Thanks, Ken

Hi Ryan,

Just thought I would give you an update on my progress...

Got hold of several evaluation boards so that I could estimate radiated noise. Used a near field loop probe and a Cleverscope CS328 oscilloscope which has a spectrum analyser built in. OK, it’s not as good as a proper EMI chamber, but at least it can give me an indication. Here are the results:

Test Board	Type	RMS noise	Spectral Peak	F peak	Vin Range	Part cost US$ (100)	Efficiency (data sheet) 12Vin, 1V1, 6.4A out
LMZ31710EVM	Module	95mV	-26dB	312kHz	3-17V	11.68	85
TPS54020EVM	Chip	117mV	-22dB	500kHz	4.5-17V	5.11	?
TPSM84A21EVM	Module	49mV	-33dB	1995kHz	8-14V	12.75	85
TPS54A20EVM	Chip	38mV	-35dB	3928kHz	8-14V	5.35	86
MPM3682	Module	78mV	-26dB	436kHz	4.5-18V	9.80	90
MPQ8636	Chip	94mV	-25dB	300kHz	4.5-18V	3.13	90
TPS56C215EVM	Chip	62mV	-29dB	1243kHz	4.5-17V	3.95	83

 

The TPS54A20 & TPSM84A21 are dual phase 2MHz parts and give the best performance for radiated noise, but they do not operate below 8Vin. There is an internal 7.65V UVLO which shuts the part down. So these parts cannot be used for the automotive start condition (ISO 16750 normal cold start 4.5V for 19ms, rising to 6.5V over 50ms).

Next best part which could keep running through cold start is the TPS56C215. It has symmetrical pin-out so good for EMI. So will use the TPS56C215 for the design.

Then I started experimenting…

The idea of these symmetrical pin-out parts (e.g. LT8640, LM53635, TPS56C215, MPQ8636, etc) is that current flows in balanced loops to cancel out the generated EMI:

All good, except there’s only one inductor so it can’t be symmetrical and cancel out. So try with 2 inductors…

The LM53635 board has the most symmetrical layout of the ones I have, so modify it with 2x 4.7uH inductors in parallel. Placed them with “pin 1” in opposite directions so that flux will cancel; e.g. with current flowing from left to right, one inductor will have flux going into board and the other will have flux coming out of board.

LM53635 board: Original single 2.2uH:

LM53635 board: Modified with 2x 4.7uH, pin 1 opposite directions:

 

Frequency display: Dark trace 2.2uH original inductor, light trace 2x 4.7uH (peak about 6dB better):

 

Time display: Loop centred over inductor(s). Dark trace with 2x 4.7uH, light trace with original 2.2uH. For the 2 inductor case the flux is cancelled directly above the centre of them.

 

Using dual inductors seems to improve the peak radiated noise (at least for near field) by about 6dB, and directly above the inductors the flux is cancelled. This also has the advantage that the current is half in each inductor, so flux density and RMS losses will lower.

We will implement parallel inductors in our design.

 

Hello Ken,

Thank you for sharing your findings!

A lot of good information that will be helpful other people to learn from, I know I learned from your post.

Let me know if there is anything else that I can do to help as your design moves forward.

Hi Ryan,

I don't have the luxury of a proper EMI test facility to hand, hence the near field loop probe. It would be interesting if you, or your colleagues (perhaps in the LM53635 group) could test this idea with real EMI measurements. Just a thought…

Best regards, Ken