TPS61165: How to suppress the 1.2 MHz & harmonics in an SDR product

Hi Robin,

Could you attach your sch&layout for further check? Also your waveform of  problem you mentioned is needed.

Thanks,

Colin

Here is the board layout

Here is the circuit: 

As to the waveform, the SDR is picking up the 1.2 MHz drive frequency of the 61165, and it's harmonics.  It has a bandwidth of about 3-4 KHz, sounds a bit like the wind whistling. I did put the scope onto the LEDA line but it seemed to destabilize the chip and it burned out.  We have hundreds of these units and they drive the LCD as expected, all with the same 1.2 MHz signal emission at around -100 dB when no signal is at the antenna

Hi Robin,

The pouring of GND in layout is not completed and it will cause large current loop which cause EMI.

First of all, i think you can debug from current board, please add 1 2.2uF to input Cap and 4.7uF to output cap like above shown.

Could you attach SW waveforms as well because you mentioned that there is sounds a bit like wind whistling.

Thanks,

Colin

Sorry I should have poted he PCB with just the ground plane

I added the capacitors but there was no change to the EMI.
I will try to capture the SW signal and post shortly.

I had understood perhaps wrongly from the datasheet that ground should not be under the inductor to avoid ground pickup.
What is the recommendation?

Here are the SW signal scope shots: 

Hi Robin,

Our expert Colin will get back to you early next week. Thank you!

BRs

Lucia

Thank you Colin for the feedback.  Would it be best to add the 0.1uF to ground decoupers on LEDA and LEDKA next to the FPC connector, at the en of the tracks?
The inductor is the Sumida CDRH4D22HPNP-100MC Datasheet

Should we keep the ground plane-free region belo the inductor?

Hi Robin,

Yes, you can do it.

Keep the ground plane-free region below the inductor is useless to switching frequency band noise. I more suggested you pour the full GND in IC bottom.

Thanks,

COlin

Below the IC there is a full ground plane.
Only below the inductor is there NO ground plane.

So to confirm, I should put a full ground plane below the inductor?

Colin, I assume you are a TI employee, expert on this particular TPS61165 IC? Your profile doesn't say

I added 0.1uF ceramics 50v from LEDA and LEDK to ground and there was no difference to the EMI level.  I also added a 0.1uF betweeb LEDA and LEDK.

Hi Robin,

Got you. Did you mount the EMI filter on your board?

How about C30 value? It is suggested to make C30=2.2uFx2.

Can you try again?

Thanks,

Colin 

Which EMI filter do you mean?
L3 was added but there was no change to the level of EMI (harmonics of 1.2MHz) detected by our SDR receiver, which uses a Tayloe detector and high gain op-amp.  We also tried running the op-amp off the 5V rail, which is heavily decoupled, but there was not difference.
C30 is 100nF ceramic decoupler, 50V X7R.

I've reduced the SW trace from 0.8 to 0.4mm width to reduce the area further, and I've added the ground plane under the inductor and added more ground vias.

Hi Robin,

Sounds good, but you should add more cap into C30 position. Changing C30 from 100nF to 2.2uFx2+100nF. Only 100nf doesn't work for noise filtering.

Thanks,

COlin

Thanks Colin, to the left of C30 shown highlighted, there is a 22uF X7R decoupler, this will filter down to quite a low frequency. It's 4.5mm from C30, is that okay?  Also please let me know your role, I have no idea if you are a TI employee or someone like me on this same forum.

Hi Robin，

This cap is for stabilizing the input and it doesn't work for the noise filtering.

The cap must be placed at C30 position that in front of bead.

Yes, i am a employee in TI which cover this part support work.

Thanks,

Colin

So 2 x 2.2uF at C30, lower ESR and less inductance in the path, how about a TDK C2012X7R1V225KT000N ?

Datasheet

Or the Murata GRM21BR71E225KA73L

I've placed two TDK 2.2uF on both sides of the PCB, C34 and C37 as you can see here: 

3D view. Behind C34 is the other TDK 0805 X7R 2.2uF 35v cap 

Do you think 4.7uF 35v for C3 is the ideal value rather than 1uF from the TPS61165 datasheet?

Before the LC input filter, L3 + C31?  Or before and at L1 too?

Before L3 like marker above picture.

Thanks,

Colin

Okay, thanks, I will make that change for a batch of 50 boards, is it worth leaving the 2x2.2uF right at the inductor?

Thanks Colin, hoping this is going to drop the EMI 10+dB.

Last question, the output capacitor, bump it up from 1 to 4.7 uF ?

Hi Robin,

Yes, it is better.

Thanks,

COlin

Okay, I'll let you know next month the results! Thanks for your help