"POWER PC" ML915 350W power supply

[PeteS in CA]

Re: "POWER PC" ML915 350W power supply

One thing to remember about DC fans. That nice number on the label is the average current. If you look at the actual current using a current probe and an oscilloscope, you'll see that the current is pulses that are much higher than the average rating.

[c_hegge]

Re: "POWER PC" ML915 350W power supply

Nice improvements. I often run into the sky 550W (like the one I mentioned) with bloated KDC primary caps. I would probably replace them with some good 105*C rated ones.

EDIT: You could also bend the fins of the heat sinks apart a bit more. They won't get much air between the fins as they are now.

[Shocker]

Re: "POWER PC" ML915 350W power supply

not a big fan of mospec

Why not???

[Th3_uN1Qu3]

Re: "POWER PC" ML915 350W power supply

Good work. You're right, 5v and 3.3v aren't so critical in new computers anymore. Good job on the recap - that filtering toroid seems to have few turns on it, the larger caps are surely going to help getting more stable output voltage and less ripple.

If you want, you can install a thermal fan circuit too, the circuit is simple and the parts are likely to be found in other scrap PSUs. But if the new fan isn't too noisy, i wouldn't bother.

[Newbie2]

Re: "POWER PC" ML915 350W power supply

A before and after post (nothing drastic):

Before:


After:

[Newbie2]

Re: "POWER PC" ML915 350W power supply

Here's the results from the rebuild. While I did not do as much to this PSU as I did to my Delta Power 450W, I believe that it is now a bit better than before.

Here's what I did:
1. Replaced the four 3A 1N5408 diodes on the primary side with a "D4KB" 4A 800V bridge rectifier (from dead Bestec ATX-250-12E PSU)
2. Placed L and N wires in proper place along with the thermistor, and installed a fuse from the dead Bestec into the proper fuse location (I won't bother touching the 5A 250VAC fuse already in it on the "J1" position of the PCB)
3. Added transient filtering components (X cap, two Y caps, and coil); X cap and coil came from dead Bestec while Y caps came from dead Enermax EG301P-VE PSU
4. Replaced secondary side capacitors (see capacitor list below)
5. Replaced MOSPEC F16C20C 16A rectifier on +12V rail with STMicroelectronics STTH2002CT 20A rectifier (on advice from Th3_uN1Qu3)
6. Replaced "Legend" 80mm exhaust fan with Dynatron/Dynaeon Top Motor DF1208BC 80mm double ball bearing fan

Capacitor list (original cap -> new cap):
2x BH 1000uf 16V 10mm -> 2x Panasonic FM 2700uf 16V 12.5mm
2x BH 1000uf 16V 8mm -> 2x Panasonic FC 1000uf 16V 10mm
4x BH 1000uf 10V 10mm -> 2x Nichicon HE 2700uf 6.3V 10mm (on +5V rail) and 2x Panasonic FM 1500uf 10V 10mm (on +3.3V rail)
1x BH 470uf 16V 8mm -> 1x Panasonic FC 470uf 35V 10mm

I did not replace the two "KDC" 560uf 200V primary caps, the small caps (less than 470uf), or the two MOSPEC S20C45C rectifiers on the +5V and +3.3V rails as I did not want to invest more in this power supply and to use only what I had available. I feel those parts are not as critical to the reliability and quality of the PSU as the transient filtering stage components, bridge rectifier, secondary caps, +12V rail power rectifier, and new double ball bearing exhaust fan.


Here are the pictures:


The upgraded "POWER PC" 350W PSU.


Overhead picture of the internals, with the upgraded components visible.



The transient filtering stage components on the primary side.


The D4KB 4A 800V bridge rectifier is clearly visible in this picture.


The secondary side power rectifiers, with the new STTH2002CT being the middle one.






Five pictures of the new secondary side output capacitors.

Pictures from my smartphone (not the best quality):

The new STTH2002CT 20A power rectifier on the heatsink (it's the part in the middle).


Secondary side picture of this power supply. (my old oscilloscope is in the blurred background)



My simple "load test" of this PSU after the component upgrades, showing 11.62V on the +12V rail shown when five fans are connected to the PSU. One of the 92mm fans shown is connected to the +5V rail to provide a little load there.


Here are the voltages I measured with my multimeter on my five fan load test of this PSU (shown in the two pictures above) after the rebuild:
+12V = 11.62-11.64V
+5V = 5.10V
+3.3V = 3.29V
+5VSB = 5.01V

When this power supply had the MOSPEC F16C20C 16A power rectifier, the +12V rail went right down to 11.62V-11.64V with my one Vantec Tornado 92mm fan running off it. When I replaced that with the STTH2002CT 20A power rectifier, running that one Tornado 92mm fan on this PSU brought the voltage down to 11.70-11.72V, which is higher than the same load with the F16C20C rectifier.

Feel free to comment on my work.

[Newbie2]

Re: "POWER PC" ML915 350W power supply

The STTH2002CT has significantly lower forward voltage which means higher efficiency - less heat.

That's good enough reason for me to spend the effort on replacing the F16C20C rectifier. Less heat is a plus when the heatsink is wimpy, and especially when the +12V rail is the most demanded from modern PCs.

I'll do the replacement later.

[Th3_uN1Qu3]

Re: "POWER PC" ML915 350W power supply

The STTH2002CT has significantly lower forward voltage which means higher efficiency - less heat.

[Newbie2]

Re: "POWER PC" ML915 350W power supply

I will do more with this PSU tomorrow in terms of testing and post pictures of the upgrade; I'm busy with some work to do at the moment and need to finish that up.

Other than that, I decided not to replace the MOSPEC F16C20C 16A rectifier on the +12V rail with my STTH2002CT 20A rectifier, as I felt the small difference in rating wasn't worth the effort to remove the whole secondary heatsink to replace it. Would the STTH2002CT make any difference compared to the F16C20C in terms of performance and/or reliability?

Links to data sheets for easy reference:
ST STTH2002CT: https://cdn.badcaps-static.com/pdfs/...29d9a8155e.pdf
MOSPEC F16C20C: https://cdn.badcaps-static.com/pdfs/...6b56a0265d.pdf

[Scenic]

Re: "POWER PC" ML915 350W power supply

Might help adding some load resistors internally?! From what you posted, it doesn't need to be much of a load if one fan already brings it down to more sane levels.

Just thought that might help. I imagine there could be some sort of spike on the outputs when that PSU is in a system and you apply AC power to it.
Th3_uN1Qu3 might shed some light on this.. I'm not a PSU guru

[shovenose]

Re: "POWER PC" ML915 350W power supply

Yep, a few fans brings this PSU into acceptable voltages on the +12V and +5V rails.

The +5V rail went to a safe 5.17-5.18V in my tests, and here are the results from the +12V rail:
Two ADDA 92mm fans: 11.72-11.74V
Two ADDA 92mm fans + One Sunon 92mm fan: 11.69V-11.70V
One Vantec Tornado 92mm fan: 11.62-11.64V
Two ADDA 92mm fans + One Sunon 92mm fan + Vantec Tornado 92mm: 11.58-11.59V

These are within ATX specification, however to me those voltages are a bit low seeing how the minimum allowed for ATX spec is 11.40V, and these fans are a very small load. Should I be concerned?

I would honestly just try it on a real PC. IMO those voltages are fine, given that the 3.3V rail is not being loaded, and a few other factors.

[Th3_uN1Qu3]

Re: "POWER PC" ML915 350W power supply

This power supply is 5v-heavy, so it needs more 5v load for 12v to be stable.

[Newbie2]

Re: "POWER PC" ML915 350W power supply

Yep, a few fans brings this PSU into acceptable voltages on the +12V and +5V rails.

The +5V rail went to a safe 5.17-5.18V in my tests, and here are the results from the +12V rail:
Two ADDA 92mm fans: 11.72-11.74V
Two ADDA 92mm fans + One Sunon 92mm fan: 11.69V-11.70V
One Vantec Tornado 92mm fan: 11.62-11.64V
Two ADDA 92mm fans + One Sunon 92mm fan + Vantec Tornado 92mm: 11.58-11.59V

These are within ATX specification, however to me those voltages are a bit low seeing how the minimum allowed for ATX spec is 11.40V, and these fans are a very small load. Should I be concerned?

[shovenose]

Re: "POWER PC" ML915 350W power supply

a few fans should do the trick...

[Newbie2]

Re: "POWER PC" ML915 350W power supply

They likely made the feedback loop so that it keeps normal voltages with those crap parts they put in, so it sits on the high side with good ones. Put an old hard drive and a few fans on it and see what it does with that.

Could the new caps on the +12V and +5V rails that I put in be too high in capacitance?

+12V rail
Original caps: 2 x BH 1000uf 16V 10mm
New caps: 2 x Panasonic FM 2700uf 16V 12.5mm

+5V rail
Original caps: 2 x BH 1000uf 10V 10mm
New caps: 2x Nichicon HE 2700uf 6.3V 10mm

Unfortunately I do not currently have any hard drives laying around that I want to risk damaging. I also don't have any old computers either that I could be fine with frying the motherboard.

[Th3_uN1Qu3]

Re: "POWER PC" ML915 350W power supply

They likely made the feedback loop so that it keeps normal voltages with those crap parts they put in, so it sits on the high side with good ones. Put an old hard drive and a few fans on it and see what it does with that.

[Newbie2]

Re: "POWER PC" ML915 350W power supply

I've just rebuilt the PSU, however when jump-starting it with no load the +12V rail is too high at 13.50V, along with the +5V rail at 5.73V. The 3.3V rail is fine at 3.35V. I unfortunately forgot to get pre-rebuild voltage values, so I cannot compare these no-load voltages to what the power supply would've originally had.

I know that switching power supplies should not be run without a load and may not function properly without a load, so I want to ask would the voltages go down to within ATX specification (maximum 12.60V for +12V and 5.25V for +5V) if it were properly loaded?

[Th3_uN1Qu3]

Re: "POWER PC" ML915 350W power supply

My father, an electronic engineer, does agree with what Hardware Secrets says about that.

If you're really rigurous then yes, the 3A diodes can be considered inadequate for the job. However, the reality is that they do just fine. I've tested some of them with a current source and they perform exactly like i said they do - almost double the rating.

Same goes for the "diodes-on-a-bracket" on 12v. They are 3A each so in theory the 12v rail would only do 6A, but they do 10A for years. What going with oversized rectifiers does improve is efficiency (lower voltage drop), but if you got the heatsink to cool them the marginal rectifiers they put in will work just fine. I do however recommend replacing them with better ones, as higher efficiency also means the primary transistors are stressed less and you get higher maximum power. 400W is possible from a pair of TO-220 13009s if you can get efficiency above 80%.

[momaka]

Re: "POWER PC" ML915 350W power supply

How does solder location save them money?

If they install the wires at their proper locations, then they would have to install additional jumpers where the coil is and maybe a few other places. This takes time and money (even if it's not much, it adds up when you make 1000 of these PSUs).

[shovenose]

Re: "POWER PC" ML915 350W power supply

It saves 1c

How does solder location save them money?