Killing HF RFI on a cheap ATX power supply

[KA2PTE]

Back in June my NZXT PP-800 desktop supply failed and I bought a cheap
chinese replacement, which works but its putting out some intense RFI, mostly 20
to 30 over birdies on the 160m band and some on 80m.

https://www.ebay.com/itm/334616637902

I bought some RF beads online:

2673000101
Fair-Rite ferrite shield bead. 73 material, 25 ohm @ 10 MHz, 35 ohm @ 25
MHz. 0.137" (3.5mm) O.D. x 0.051" (1.3mm) I.D. x 0.127" (3.25mm).


and opened the PS and aimed to install them on the leeds of the switching transistors
as I had seen this done in other designs I have worked on in the past.

This supply used (2) of these:
https://pdf1.alldatasheet.com/datasheet-pdf/view/1140733/JINGDAO/P13009.html

So I went and put a bead on the base of 1, and the collector of the other hoping to hit
a bullseye. After its all back up and running, still have the intense birdies, freq shifted a little.

The only other components I could apply them to seem to be T0-220 packages with I believe dual diodes
in them, but not sure if I have the best type of bead, os would it be more properly placed somewhere else
in the design?

The NZXT supply never produced any birdies or RFI, but its way more packed with more complex circuits
and I guess a more expensive PS would be worth a try but I know ferrite can probably cure this.

[W1ITT]

On 160 meters, a single bead doesn't amount to much inductance.  I would suspect that the solution will have to involve brute force filters, probably Pi sections with lots of L and C.
Either that or a ballistic trajectory.
73 de Norm W1ITT

[K8DI]

Given that NZXT power supplies are in the $200 range, and the original one was quiet, rather than spending a lot of time effort and repeated bits of money on bandaids, buy the right PSU. And use the Chinese POS as a bullseye…

Buy once, cry once.

Ed

[KA3EKH]

think a lot of the crap comes out the AC power input. have had lots of issues trying to quiet down switching supplies. they always have a habit of producing noise right on the frequencies that I want to use them on.
Be watching to see what others have as answer.

[Tom WA3KLR]

Before you throw in the towel - a hail Mary pass here - add filters caps from each side of the line at the power supply input to ground, 0.01 uF to 0.1 uF. if none that large there. There are line-rated X/Y caps but see what you have and for 160 Meter problems, probably the closer to 0.1 uF, the better.  And then check ground from p.s. to the IEC connector and on thru the line cord to the male ground pin, and outlet ground confirmation if possible.

Is this the same outlet as the quiet p.s. was operating from?

GL OM.

[KA2PTE]

Yes same outlet, its going to a power center like this one:
https://www.ebay.com/itm/266577734304

I do not think they have any RFI filtering.

I am familiar with the X-Y approach you mention. I wonder if this would just slide right in to the
current IEC opening

https://www.ebay.com/itm/382231085266

I already have some new old stock PI filtering inductors that were used in flat screen TV designs many years ago
as shown in the photo attached so maybe I could use them and try some caps as you mention.

Before you throw in the towel - a hail Mary pass here - add filters caps from each side of the line at the power supply input to ground, 0.01 uF to 0.1 uF. if none there. There are line-rated X/Y caps but see what you have and for 160 Meter problems, probably the closer to 0.1 uF, the better.  And then check ground from p.s. to the IEC connector and on thru the line cord to the male ground pin, and outlet ground confirmation if possible.

Is this the same outlet as the quiet p.s. was operating from?

GL OM.

[KD6VXI]

Get some decent 2.4 inch torroids and wrap every wire going in and out of the power supply, as close to the supply as possible.

Also, make SURE the ground pin has a GOOD connection.  I've seen them riveted and the rivets where not tight!

One bead is meaningless. Impedance is the square of the turns, so you'll end up with 4 times the impedance by wrapping a turn through it.  That would be considered two turns.

Another way to look at it is....  It takes about 50 ferrite beads to equal the choking impedance of a single 2.4 inch torroids with 6 or so turns through it.

--Shane
WP2ASS / ex KD6VXI