Help with Radiated emmisions

Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

cross mob
DaHu_285096
Level 5
Level 5
10 likes received 250 replies posted 100 replies posted

I am using a 5LP device in a product that is being tested for FCC  and CE certification. The response is slightly out of limits at 80MHz and I suspect this is the controller. I need to reduce Radiated emission by 2uV at 80MHz.

   

The controller is being run at 3.3V. 

   

Are there some relatively easy to implement techniques to lower the RE? (ie. A clock parameter change or something I would turn off)?

   

The hardware has short traces, plenty of decoupling close by the controller and laid out with large ground plane with lots of vias stitching top and bottom layers. (it is a 2 layer PCB, 1.6m thick with 1oz copper).

   

My project uses a 1ms timer, SPI, RS485 and a generated 125 kHz Square wave signal from PWM.

   

 

   

Thanks

0 Likes
1 Solution
Bob_Marlowe
Level 10
Level 10
First like given 50 questions asked 10 questions asked

Well, when thinking of frequencies you ought to keep in mind that the rise and fall (no, not the house of Usher 😉 of a signal introduces very high frequencies.

   

As you probably now you can compose (theoretically) any waveform by summing sines of different frequencies. For a square wave you will need high frequencies to accomplish the sudden changes. Reducing any slew rate will decrease that frequencies.

   

If it is not too complicated for you to measure, give it at least a try.

   

I'm not quite used to measuring 80MHz radiation, but can you use a probe that picks up that noise to see from where it is emitted?

   

 

   

Bob

View solution in original post

0 Likes
14 Replies
Bob_Marlowe
Level 10
Level 10
First like given 50 questions asked 10 questions asked

Is the PSoC running at 80MHz? IMO or external clock? Any Crystal?

   

Are any signals leaving the chip? Can those pins be configured with "slow slew rate" (at pins output configuration) which will reduce straying rf.

   

 

   

Bob

0 Likes
lock attach
Attachments are accessible only for community members.
DaHu_285096
Level 5
Level 5
10 likes received 250 replies posted 100 replies posted

Bob,

   

I have attached the Clock settings. The system is not using external crystal.

0 Likes
DaHu_285096
Level 5
Level 5
10 likes received 250 replies posted 100 replies posted

RE tests show response on first test at 80.644 MHz and second test at 79.243MHz.

   

I am not sure if the micro is the issue but the other clock sources on PCb are all much lower (115k RS485, 250KHz switching PSU, 125KKz and SPI bus).

   

If not the controller radiated emissions then maybe harmonic of the SPI?

0 Likes
Bob_Marlowe
Level 10
Level 10
First like given 50 questions asked 10 questions asked

So one signal leaving your board is the UART. Are you using a level converter on hte board as Max232?

   

Can you reduce rf stray by setting the Tx pin to slow slew rate while keeping your rs485 functional?

   

 

   

Bob

0 Likes
DaHu_285096
Level 5
Level 5
10 likes received 250 replies posted 100 replies posted

I am using equivalent of Max485 for level shifting.

   

The frequency of the 485 (around 65khz) is miles away from the 80MHz signal being emitted (1230th harmonic)

0 Likes
Bob_Marlowe
Level 10
Level 10
First like given 50 questions asked 10 questions asked

Well, when thinking of frequencies you ought to keep in mind that the rise and fall (no, not the house of Usher 😉 of a signal introduces very high frequencies.

   

As you probably now you can compose (theoretically) any waveform by summing sines of different frequencies. For a square wave you will need high frequencies to accomplish the sudden changes. Reducing any slew rate will decrease that frequencies.

   

If it is not too complicated for you to measure, give it at least a try.

   

I'm not quite used to measuring 80MHz radiation, but can you use a probe that picks up that noise to see from where it is emitted?

   

 

   

Bob

0 Likes
DaHu_285096
Level 5
Level 5
10 likes received 250 replies posted 100 replies posted

Bob,

   

Thanks again for your help. I agree, it could be rise times (very high frequency component due to fast transition).

   

I initially suspected controller main clock due to the response for PSOC in top graph (extracted from the reducing emissions manual) and the response of the failed FCC test on my product below it.

   

0 Likes
DaHu_285096
Level 5
Level 5
10 likes received 250 replies posted 100 replies posted

In fact, the response just over 100MHz also shows up in both grapghs.

0 Likes
DaHu_285096
Level 5
Level 5
10 likes received 250 replies posted 100 replies posted

why can't we spell it grafs ? English is such a pain, wish I knew another language.

0 Likes

I could serve you with German, but even here it is spelled "Graph", the "Graf" is the Duke or Count 😉 Do you write "graffitti" or "Gaphitty"?

   

 

   

Bob

0 Likes
odissey1
Level 9
Level 9
First comment on KBA 1000 replies posted 750 replies posted

To measure radiation from PSoC I use RTL-SDR dongle. It works well in 10MHz-1.6GHz range and available at about $6 shipped on e.ay. The dongle is very sensitive,  numerous harmonics form PSoC are observable even in >400MHz range. What you see at 80-130MHz range are some overtones. With help of RTL-SDR I would use ~10pf capacitor to go over all traces on the board to see which one affecting 80 MHz band. To observe spectrum use  SDR# , few other wide-band spectrum analyzer software available (but they are slow). 

   

There should be no misunderstanding: one side of probing capacitor goes to the ground (not RTL-SDR antenna), and RTL-SDR antenna is just hanging nearby (can use a piece of wire instead of antenna). 

   

P.S. Does PSoC5 have a spread spectrum option?

   

   

0 Likes
DaHu_285096
Level 5
Level 5
10 likes received 250 replies posted 100 replies posted

Probably moving well off topic now but we need a referendum on the spelling of Yatch. We have a "Batch" process and a "Patch" of grass but we sail a "YOT".  I can live with "Toe May Toe" and "Toe Mar Toe", but not Yot.

   

Anyways, back to EMI now. Will play with some Digital IO slew rates today. I not the IMO has a fundamental frequency of 3MHz and internal PLL mixes up to the 24MHz with around +/- 1% accuracy. I wonder if there is a component of the mixing product of the 3MHz and 24MHz producing a third harmonic. ie. 3 x 27 MHz = 81 MHz?

0 Likes
Bob_Marlowe
Level 10
Level 10
First like given 50 questions asked 10 questions asked

Staying off-topic, it is spelled "Yacht"  and there is no "pacht", or are you talking about "pot"?

   

EMI: CapSense oscillations? the caps are comparably small.

   

How are unused pins configured? Hi-Z? Try (will cost power) to set them resistive to ground.

   

 

   

Bob

0 Likes
DaHu_285096
Level 5
Level 5
10 likes received 250 replies posted 100 replies posted

arhh, my spelling is ar tro shus. Must be the pot.

0 Likes