Radiated EMC failure due to beat frequencies between different SMPS's?

[treez]

Hi,
A customer has sent us in a battery charging kit which has failed their radiated EMC. This kit comprises two earthed metal enclosures. (each about 20cm by 20cm by 10cm) One enclosure (A) contains a radio transceiver and a 18W offline flyback smps and a 12w offline flyback SMPS. An isolated 12V cable comes out of this enclosure and goes into another enclosure (B). Here the 12V is stepped down by a 1MHz Buck converter to 5V. This 5V is then used to charge a battery at 2A, by a switch mode charger. (this charger is the only bought in SMPS.)

The peaks of the failure frequencies are approx. 60MHz, 80MHz, 93Mhz , 111MHz and 184 MHz.

In order to find out how to fix this failure, I believe we must use an EMC sniffer (SA1002A by Laplace instruments), and a B field (near field) probe connected to this EMC analyser via coaxial cable. We will use a Laplace instruments SA1020A (Wide band amplifier) to amplify the signal from the (near) B field probe on its way to the EMC analyser.

We will take out each SMPS in turn and simply run it about 6cm below the B field probe. (the B field probe is a ring [loop antenna] about 7cm in diameter). We will see which SMPS gives out the failure frequencies. We will then see if we can do any component changes on the PCB (we are not allowed PCB respins) to solve the failure. We may just have to add cable ferrites etc on the cable interlinking the two units. Or add common mode filters at inputs and outputs to the SMPS’s.

Do you think this is a good ploy? I mean, I presume that the failure frequencies are not due to any beating effects between the various SMPS’s?

SA1002A EMC analyser:

https://www.reliantemc.com/download/laplace/Laplace-Instruments-SA1002-1GHz-EMC-emissions-Analyser.pdf

 

[stenzer]

Hi,

before investing in expensive equipment you might use a DIY EMC probe as shown in the video linked below, to narrow down possible causes. Have a further look at the EEVblog videos, I think there are more videos dealing with this DIY probe.
https://www.eevblog.com/2019/02/13/eevblog-1178-build-a-10-diy-emc-probe/
BR

 

[treez]

Thanks, thats essentially what we are doing.
Please allow me to explain here the overall system...

Please help our kiddies post-lockdown pager system game kit pass radiated EMC? I am doing this with a company in South Africa for charity. Its as attached.

It is an earthed metal enclosure A, which contains a 20W offline SMPS and a 12V battery. This enclosure feeds a 12V cable to enclosure B which contains a 10W , 12V to 5V SMPS. This 5V is then fed to each of four removable pagers, (so they get charged up) which the kiddies can remove and talk to each other. (they will be in a “hospitals and doctors” game)

It fails EMC at the moment. (Radiated emissions failure at frequencies from 80MHz to 190MHz) We wish to sell it into the EU, where they need some post-Covid relief as much as everyone else.

Should the cable between A and B be earth screened and the earth screen connected to both enclosure A and B?

The enclosure B is open at the front, so is it therefore a waste of time to earth enclosure B? As such, is it also a waste of time to take earthed 12V cable from A to B?

The wires running from the 12v_to_5v SMPS to each pager, should they be loomed as close as possible to the walls of enclosure B?

The enclosure B has to be open at the front so that kiddies can grab the pagers.

We are thinking we may need a common mode choke at the point in enclosure A where the 12V cable leaves A to go to B? Also, we could put a common mode choke at the input and output of the 12v_to_5v SMPS. (NiZn torroid based common mode chokes with just a few turns) We could also put a flux band round the offline flyback transformer in A. We could also increase the Y capacitor across the flyback transformer in A. At the moment its only 100pF. We don’t want PCB respins if poss as the customers don’t want to pay much.

 

[stenzer]

Hi,

Also, we could put a common mode choke at the input and output of the 12v_to_5v SMPS. (NiZn torroid based common mode chokes with just a few turns)

I would start with the easiest approach by means of a Cable-Ferrite which is but arround your "12 V DC in Cable". This approach does not require a redesign of your system. But this would only prevent the spreading of the "noise" from one enclosure to the other one. But of course it only supresses common mode "noise". Btw, how long is the cable? As your mentioned freqencies are pretty high, your cable might already act as an antenna (\[\frac{\lambda}{4}\left(@~ 80 ~\mathrm{MHz} \right) \approx ~1 ~\mathrm{m}\]). So next thing I would try is to connect the shield of your cable. First only on one side, and than maybe at both sides. Have a look if your individual grounds of enclosure A and B are swinging against each other, and if, how it is affected by the connection of the cable shield.

We could also increase the Y capacitor across the flyback transformer in A. At the moment its only 100pF.

I assume with Y capacitor you are reffering to the capacitor which is placed between the primary and secondary side of your transformer. I assume you might see a beneficial effect by increasing its value at least to a couple of nano Farads.

The thing I'm most curious about is, which one of the two systems/enclousers is causing the troubles. Have you got an EMC report which indicates this?

BR

 

[treez]

The cable between enclosure 1 and 2 (of the attached, simpler diagram than above) can be anywhere between 0.5m and 3m.

It is well known that passing radiated emissions to EU standard with an SMPS that is not surrounded in a metal enclosure is pretty well impossible. Even if it is in a metal enclosure, it will still fail radiated emissions if a cable leaves the enclosure to go somewhere else. (as in the attached). In that case, a common mode choke and Y caps will be needed in order to stop the cable from radiating. Would you agree?

 

[stenzer]

Hi,

yes, a Y-capacitor should suppress the high frequency common mode. By using a common mode choke right before your output should supress common mode noise even further, and especially high frequencies which are more likely to be radiated by your cable.

BR

 

[stenzer]

Hi,

that might be usefull

http://www.epqu.agh.edu.pl/archives/journal/v15i2/Jha[2].pdf

BR

 

[treez]

I am thinking that we must use some shielding. Do you agree? Also, i believe we will need to re-do the DCDC SMPS (inside B) with common mode chokes on its input and output...and have Y caps on the DCDC's pcb to the pillars which screw into the earthed chassis B. Would you agree.?
Any better measures?
Also, Our cheapo radiated emissions test, is to torroid-wrap a near field B probe of diameter ~9cm with 18 turns of twisted pair wire (evenly spread round the loop)...........we then solder this twisted pair "wrap" into the cable between the two units A and B in order to gauge just how much radiation at each frequency is in the cable. (the B field probe ultimately gets BNC'd into an SA1002 spectrum analyser) Our problem frequencies are from 57MHz to 184MHz, with a highest peak at 96MHz.

I am wondering if another trick for us will be to take multiple different lengths of cable to the Radiated emissions lab, and keep swapping them in and out until we get a a pass..then we use that length on the product? (the cable between A and B)

 

[stenzer]

Hi,

The thing I'm most curious about is, which one of the two systems/enclousers is causing the troubles. Have you got an EMC report which indicates this?

as you are using two SMPS, one in each housing, my main question is which one is causing the troubles. Have you got an EMC report which indicates/narrows it down?

A possible process of elimination would be:

(i) Replace the system residing in enclosure 2 by an appropriate load, by means of the same power dissipation, in combination with your cable.
(ii) Replace the system residing in enclosure 1 by a (low noise) power benchtop sypply, or even better a battery. Also while utilizing the cable.

By recording the (emitted) noise along the cable during the two tests you might be able to point out which of the two systems is troubling problems. Of course the interaction between the two systems is not modeled by the two measurements, which might also yourse troubles. At this point we are only considering the noise along the cable, and have not addressed the radiation "escaping" the enclosure at an other location.

BR