Cascaded offline SMPS's cause electrical noise?

[eem2am]

smps noise filter

Hello,

I am making an offline LED driver at 20Watts.
It feeds off UK mains 230VAC.

It consisists of two simple SMPS's in cascade:-

First after the mains diode bridge there is a Continuous conduction mode, current mode flyback with a 40V DC output.
This stage switches at 100KHz.

Then this 40V is the input to a current mode controlled Buck converter which powers the LEDs.
This stage also switches at 100KHz.

There is no synchronisation circuitry to synchronize these two converters

I am told that i have done this wrong , and that i should not cascade two SMPS's in this way since it creates a terrible problem with electronic noise.

Is this correct ?. Will i get terrible electronic noise problems ? Also, will it be common-mode or differential-mode noise?

 

[umery2k75]

offline led

There is no synchronisation circuitry to synchronize these two converters

Why would you even want to synchronize the two converter, for what good reason?

The second current mode controlled buck converter is seeing only a high current DC link of 40V. Even if you provide this link with a external power supply. The second converter wouldn't care.

You must worry about the layout. I advise you to put faraday shield in your transformer winding if you haven't, that will help to reduce EMI.High frequency high-voltage switching waveforms can be capactively couple to the ground plane. Care about the heat sink of switching transistors that could be attach to the chassis of the SMPS, capcacitive coupling of high frequency signal could occur.You need to care about transformer, most radiation is emitted through the air GAP, you must do screening of it and for switching elements too.
These are the things you need to work on. Also see, if it's necessary to put a choke/filter at the output of the first converter.

I am told that i have done this wrong , and that i should not cascade two SMPS's in this way since it creates a terrible problem with electronic noise.

If your way is wrong. I hope he/she might have advise you.Share his/her advise too.

 

[eem2am]

two smps

thankyou umery2k75,

"Why would you even want to synchronize the two converter, for what good reason? "

I dont know but i think its like you ensure that the buck is conducting when the flyback secondary is conducting so that the ripple current in the flyback output caps is less........

....but also, suppose that due to oscillator tolerances they are actually say 101KHz for flyback and 100KHz for buck....

...then i am told that what you get is like "beating" between the two SMPS stages....and a low frequency noise which the input EMI filter does not adequately deal with. ...?

Its like with these chips that do Boost PFC and follow on PWM stage all controlled from the one chip....and there is synchronization between the PFC and PWM stages..

eg the ML4803

http://www.datasheetcatalog.org/datasheet/MicroLinearCorporation/mXrqyzu.pdf

This has synchronized switching stages, and i believe it reduces noise.

 

[FvM]

smps noise

Synchronizing SMPS that are cascaded or connected in parallel creates a better defined situation of course and can avaoid special problems as subharmonics generated by downmixing of switching frequencies. But I agree with umery2k75's view, that don't have to fear these effects with a good layout (and some necessary filtering).

On the other hand, not synchronizing the SMPS has most likely an advantage in avoiding high emission peaks at the harmonics of the common (synchronized) switching frequency. It has a similar effect as a spread spectrum clock utilized in PCs.

 

[eem2am]

smps shield winding

FvM,

i agree entirely with your views on the need for synchronizing SMPS's in parallel.

Take your car...........if all your lights are LED lights, you WILL find that the direction_indicators, brake lights , reverse lights and tail lights are ALL done with LINEAR REGULATORS .

This is because having these done with separate SMPS's would mean a bad EMC situation caused by the close-together SMPS's interfereing with one another.

SMPS's do get used in some cases but if so, then the four SMPS's are done with staggered synchronozation, in other words the four clocks of each of the four SMPS is phased apart by 90 degrees to reduce bad EMC effects.

....you then have the difficulty that the synchronizing clock is often at 500KHz plus and this is an awkward signal to easily route around without very good layout.