Separated Area For +5VSB Additional Filtering

[Hitto]

I'm trying to tame the +5VSB ripple onto a F.S.P. power supply.

Despite its capacitors being okay it is quite high: no problems with USB pens or light devices but random disconnections on sensitive devices.

The diameter is quite strict (8mm) and I've already upgraded the only two capacitors (sigh) available for +5VSB filtering (1000 + 1500uF) plus the 47uF one (photo was taken prior the upgrade).

Problems was reduced but still here.

There is an inductance but space is not truly available for placing a bigger one. I have to keep the inductance.


There is, howewer, an area with some free space to solder a pair of capacitors, or at least a big 3300uF capacitor which should provide enough additional filtering.



I would have to bridge both positive and negative though.

Any problems by jumpering (thus extending) the area of both negative and positive rail with one wire for each rail?

This power supply have a very small area for +5VSB.. It looks like it was made this way to counter interferences (plenty of space elsewhere). Everything is very close. Seen a similar design on at least another power supply.


Even fitting 10mm caps is a hell and requires moving everything around to a degree and diodes and inductors have very short legs. Not possible, already tried.

[Hitto]

Re: Separated Area For +5VSB Additional Filtering

Any suggestion about what to do is welcomed.

Some informations:
1) The reusable area is totally disconnected from the rest of the board;
2) I was thinking about placing a huge negative wire from +5VSB ground to secondary ground and maybe in the process solve an existing ground loop problem (there is currently a small wire which connects +5VSB ground to the rest of the board)... I doubt there are design mistakes though - Just capacitors not being "enough" for the ripple produced by the flyback converter.

[redwire]

Re: Separated Area For +5VSB Additional Filtering

What model FSP ? +5VSB is generated by a two transistor oscillator Q4, Q5?
That circuit does not perform well, ever. People upgrade to an IC like KA1H0165R. We could find the FSP schematic. Is the voltage OK, not 4.8V or something low?

I wonder if ripple is really the problem, with a 1A load having 1,500uF there should be fine. Just upsizing the capacitance past that, to 3,300uF should not make much difference.
You could first try the cap outside the power supply, on the +5VSB and GND wires. But I think the circuit may not be working well.

[Hitto]

Re: Separated Area For +5VSB Additional Filtering

I have to re-read the exact model though schematics were not available when I searched for them. It looks like the +5VSB design is not the arcaic one.

Voltage is okay and does not go low when +5VSB is fully loaded.

So adding external filtering will lead to troubles?

[redwire]

Re: Separated Area For +5VSB Additional Filtering

5VSB SMPS run at high frequency 50-100kHz and with 1,500-3,300uF filter cap, ripple voltage is low, not enough to upset anything. Assuming the regulation is working good, it stays near 5V.
You can look for an SMPS IC on the other side of the pc board. I see Q4, Q5 and that looks like a old two-transistor circuit, FSP used sometimes.

I have seen some motherboard's USB ports are to blame - low/weak 5V output. You might check that.

[Hitto]

Re: Separated Area For +5VSB Additional Filtering

With some re-routing I may use a spare inductor as a bridge (+5VSB Positive Rail > Inductor bridging original positive rail to unused area > 3300uF capacitor > +5VSB purple wire connected).

Motherboard has polymers for USB ports and measure correctly.


On a cheaper power supply just upgrading the single +5VSB capacitor, capacitance under 3000uF, solved every unstability.

[PeteS in CA]

Re: Separated Area For +5VSB Additional Filtering

Is the noise spikes? Or something like a square wave. If it is spikes the problem might be the ground lead on your oscilloscope is acting as an antenna, picking up high frequency noise. If is like a square wave then the noise is real.

[Hitto]

Re: Separated Area For +5VSB Additional Filtering

+5VSB extended: added inductor, then a 2200uF capacitor, then the purple wire. Negative was bridged using a 20AWG wire, as short as possible, from +5VSB ground to the new ground (on the lower side of the board).

So, LC filter (original) -> LC filter (added). LCLC (1500uF+Inductor+1000+Inductor+2200uF).

Not absolutely purist but problem was solved.




Going to add some goo glue to block that inductance from moving.

[SoporteHR]

Re: Separated Area For +5VSB Additional Filtering

Good job.

Thank you for sharing the solution, so we can take it into account for similar cases that may arise.

Greetings.

[Hitto]

Re: Separated Area For +5VSB Additional Filtering

If they only cared to avoid cramming so many components in a very restricted space then a couple of 2200uF/10V capacitors in place of the original ones would have been enough to make the situation better without these bridges.

But in 8mm diameter there are only 2200uF/6.3V capacitors which is too low of a voltage to be used in +5VSB (even on uber-cheap ATXs producers do not dare to go lower than 10V in the +5VSB rail and from what I've read the spikes on that line can exceed even 10V).

[budm]

Re: Separated Area For +5VSB Additional Filtering

So just curious as to what is the ripples and spikes level before and after the mod?

[Hitto]

Re: Separated Area For +5VSB Additional Filtering

I was not able to measure it.

It would have been indeed very useful to know as I was very tempted to use 2200uF/6,3V components to avoid doing these bridges.

Any suggestion is welcomed.