How to filter spikes from the 5V rail in a PSU?

[mayu_wakui]

Hi all..I have a PSU outputting 5Vdc..but when there's interruption in the 240Vac mains, i get spikes on the 5V rail..the spikes can be around 12-16Vpp and sometime burst to 40Vpp..the spikes occurs at random frequencies..could anyone suggest me on how to filter these spikes..? i manage to somewhat reduce the spikes by putting a choke at incoming neutral..but since the gnd plane of the board is really bad (ancient design), so i believe that there are still EMI noise travelling around..

 

[Davood Amerion]

Filter Design

Use one 100uF/16V and one zener diode (5.6-6.2 Volt/1W) at 5 Volt rail (output).

 

[Kral]

Re: Filter Design

mayu_wakui,
I'm puzzled by 2 aspects of your design:
*
You only show a regulator for the positive side. If you don't need a negative output, then eliminate the 47uF capacitor, and don't use the lower 2 diodes of the bridge.
~
The connection from the mains to the secondary should be removed. The spikes are probably coming from capacitive coupling between the mains neutral and the PSU ground.
~
One way to reduce EMI from this type supply is to connect a series RC network across the transformer. You need to experiment, but a good staring point is 2.2k, .001uF.
Regards,
Kral

 

[Lucifre]

Re: Filter Design

Yeah i agree with kral... you should go with the standard center tapped transformer power supply layout ( assuming you only need one sided dc voltage supply)
and remove the wire from mains to secondary.... i haven't seen it in all the power supply circuits i looked at ( is there a special purpose for this connection?)

 

[mayu_wakui]

Re: Filter Design

Use one 100uF/16V and one zener diode (5.6-6.2 Volt/1W) at 5 Volt rail (output)

i can only imagine this..could u upload the diagram please..?

You only show a regulator for the positive side. If you don't need a negative output, then eliminate the 47uF capacitor, and don't use the lower 2 diodes of the bridge

i do need the -ve side..this circuit is only part of the whole design

The connection from the mains to the secondary should be removed. The spikes are probably coming from capacitive coupling between the mains neutral and the PSU ground

i did isolate the mains to secondary..the spikes is somewhat reduced by amplitude and frequency..

One way to reduce EMI from this type supply is to connect a series RC network across the transformer. You need to experiment, but a good staring point is 2.2k, .001uF

thanks..i did tried this after the regulator..but the RC filter drained to much current..so my op was halved..but i think my calculation was wrong

i haven't seen it in all the power supply circuits i looked at ( is there a special purpose for this connection?)

its called an 'auto configuration'..used when u need extra boost of current from the mains..

anyway, here's a capture of the spike..well, its very random though..so if anybody might suggest me any filter technique/design..i'll be much abliged..thanks

 

[Davood Amerion]

Re: Filter Design

mayu_wakui,

i can only imagine this..could u upload the diagram please..?

see circuit below

Regards,
Davood

 

[mayu_wakui]

Re: Filter Design

hi..thanks davood..i did tried this config b4..the idea of clipping the spike while retaining 5V was the simplest idea i could think of..but i put the zener before C4..and the result was the zener was draining almost all of the current..so i didn't get any output..FYI the current demand at point +ve of C4 is 10mA..anyway, i'll try ur config also..i'm so desperate, so every penny is worth a pound..

 

[mayu_wakui]

Filter Design

hi..davood, it didnt work..

 

[ptmsl]

Re: Filter Design

Use trancient voltage suppressor (like SMBJ5.0A) instead of zener diode they have much better clamping characteristics also use high frequency decoupling both at the input and output of your regulator with a small capacitor designed for high frequency. If you still have problems try using chokes at the input or even ferrite clamps at the mains cables.

 

[mayu_wakui]

Re: Filter Design

Use trancient voltage suppressor

i do have some JV0603ML050A..(5.5V Vdc 15.5 V max)..should i put it parallel to the op of regulator?

If you still have problems try using chokes at the input or even ferrite clamps at the mains cables

already did that..did manage to reduce the spikes at 70%..these ferrite clamps, are they the same as ferrite beads?

 

[Davood Amerion]

Filter Design

hi mayu_wakui
with you discussions, it seems, it is because of bad grounding. check and strengthen ground plane (wiring).

 

[mayu_wakui]

Re: Filter Design

yes..im very aware of the bad gnd..but i need to salvage the board with less changes to the board..i do have another board with better gnd design though..but that is when all hopes are lost..hehe..anyway ptmsl, the TVS seemed to be working..i would say 70% of the spikes are reduced..perhaps i should get a higher rating..? seems that higher amplitude spikes are still around..still trying to get my hands on hi freq bypass caps..anyway, here's an upload of the old board..yeah it sucks..haha

 

[ptmsl]

Re: Filter Design

I dont know this type of suppressor but you can try using it. Connect it like a zener observing the polarity if it is unidirectional. The ferrite clamps work like ferrite beads on cables. Since ferrite clamps seem to work try passing your mains cables (both of them) a few times through a ferrite torroid to form a common mode choke at the input of the circuit or connect a common mode choke before the mains transformer. There are many common mode chokes available for different frequencies and currents from companies like Pulse Engineering, VAC and others.

 

[mayu_wakui]

Re: Filter Design

The ferrite clamps work like ferrite beads on cables

would u say ferrite clamps are the same as ferrite cores..? bcoz i do have a 20MHz cores lying around..its a doughnut shape thingy that could fit to the cable..

Since ferrite clamps seem to work try passing your mains cables (both of them) a few times through a ferrite torroid to form a common mode choke at the input of the circuit or connect a common mode choke before the mains transformer

used a 4MHz common mode choke at both the live and neutral..helped a bit..but the choke has some effects on the operation of the board..so it was scrapped..

i'm renting a spectrum analyzer (since we don't have one)..perhaps we could see the frequency of these noise..then determine the correct value of chokes and ferrites..any comments?

 

[ptmsl]

Re: Filter Design

What was the ill effect of the common mode chokes? Perhaps their current rating was not sufficient? Try different types of ferrites, the 20MHz cores look promising. Experiment with different number of turns. Also try another type of TVS in case the type you used was not fast enough. If your power rating allows it put small resistors before the diode bridge to improve the filtering effect of the bulk capacitor and put small capacitors 100nF and 100pF in parallel with it and after the regulator. Also try to examine the board and find where the spikes have their largest amplitude in case EMI has found another path and try local decoupling. Use short ground leads for your osciloscope idealy only a small wire from the edge of the probe connected to ground close to the tip in case the spikes are picked up by the loop of the probe ground wire acting as an antenna.

 

[mayu_wakui]

Re: Filter Design

What was the ill effect of the common mode chokes?

we have a measurement IC onboard..the chokes induced in creeping..a phenomenon when the IC measures incoming current when there is actually none..

Try different types of ferrites, the 20MHz cores look promising

i heard there's different types of material used for higher rated chokes..care to explain?