Back in 2020 i bought a (2nd hand) compact ATX psu to replace my current one. My PC case only fits very short ATX power supplies so i didn't really have the choice of buying a decent 80Plus unit. The current unit is a modded YoungYear unit that i made a thread about a while ago:
A "nice" YoungYear unit? - Badcaps
The "new" unit is a KDM-M6480 480W psu, the 480W number is a typical KDM bullshit claim of course, but the 24A rating on the 12V rail seemed reasonable and the "Active PFC" claim gave me a bit of hope that this was going to be decent-ish. I mean, they even included an input voltage selector switch 🤨
Opening up the unit made clear it wasn't a decent Active-PFC unit and that it wasn't going to hit that 24A spec, 200W in total would've probably been a more realistic expectation...
Low and behold, a very basic and compactly built half-bridge unit... Just a mediocre one. There hasn't been any extreme corner cutting when compared to the real gutless wonders, but definitely not powerful in any way. The transformer and output toroid are pretty small and the PFC is nowhere to be found. Some noteable things were the partially present input filtering (yay), low profile transformer (EI33 but shorter), the small output toroid and the 3.3V rail which is actually tapped from the 5V rail and regulated with a mosfet.
The last couple of years this PSU was one of my side projects which sometimes got a bit of attention. The modded YoungYear unit was lasting better than expected so the urge to complete this project disappeared for a bit. But recently i acquired an Oculus Quest 3 VR system which required a bit more graphical horsepower from my PC. So here i go, ordering a Radeon RX6800 GPU while being fully aware that my current YoungYear unit can't handle it 😅 All hopes are on this KDM unit now 🤞 A couple of years ago i did make the first round of upgrades and posted the results in the hall of shame thread:
https://www.badcaps.net/forum/troubl...75#post1423575
To keep thing simple i'll just post the end result here. Last week i gave it a final round of upgrades, hoping it would do ~300W. It got upgraded quite extensively, the transistors, 12v rectifier, main transformer, output toroidal and heatsinks were upgraded. The whole PSU got Teapo-fied except for the primaries. Passive PFC was also added as well as temperature-based fan control.
Primary:
NO PFC -> Passive PFC
2x Y capacitor
NO X capacitor -> added X capacitor
1x common mode choke -> 1x better wound common mode choke
2x 2.5A diode + 2x 4A diode -> KBL406 rectifier bridge
2x 470uF 200V Ltec primary caps
2x 13007 primary transistors -> 2x D304X primary transistors (still TO220)
1x SVD2N60F mosfet
Swapped primary heatsink
Middleware:
ERL33 low-profile transformer -> ERL35 full-height transformer (swapped the primary snubber circuit with it)
SC6105 switching controller
Added temperature-based fan control
Secondary:
12V: MOSPEC F16C20C 16A TO220 Fast Recovery -> DSSK 40-008B 40A 80V TO-247 Schottky
5V: MOSPEC S30D45CS 30A TO247 Schottky
3.3V: Advanced Power AP40N03GP 40A mosfet (tapped from 5V)
Tiny output toroid -> Way bigger output toroid (with more efficient green/blue core and thicker windings)
Swapped secondary heatsink (modified a bit to make the larger output toroid fit)
Output Filtering:
12V: 1x 1000µF 16V ChengX, PI Coil -> 2x 1500µF 16V Teapo SY, PI Coil
5V: 2x 1000µF 10V ChengX, PI Coil -> 2x 2200µF 6.3V Teapo SC, PI Coil
3.3V: 1x 1000µF 10V ChengX, PI Coil -> 1x 2200µF 6.3V Teapo SC, PI Coil
+5VSB: 1x 470µF 16V ChengX -> 1x 1000µF 6.3V Teapo SY
-12V: 1x 470µF 16V ChengX, NO PI Coil -> 1x 1500µF 16V Teapo SY, added PI Coil
-5V: 1x 470µF 16V ChengX, NO PI Coil -> 1x 1000µF 6.3V Teapo SY
Testing - Power Output:
A heavily overclocked furnace ...uhh.... Pentium D CPU was used to test the constant power output capability. The test was running for about ~20 minutes to make sure heat-soak wouldn't get me in the end. Using my PR10 power meter (wall power) and a current clamp (output power) i got the following numbers:
Input power: ~440W (225V, 0.73 PF)
Output power: ~333W (~76% efficiency, ~24.4A@12V, ~5.2A@5V, ~3.0A@3.3V)
I used an overclocked Radeon HD6870 for testing it with a bit more transient loading. The Pentium D was downclocked a bit to keep power draw withing reasonable limits. The PSU didn't even flinch at this test while drawing 400~460W from the wall.
Testing - voltage regulation:
Voltage regulation seems to be okay-ish, perfectly adequate to power a PC:
3.3V: ~3.4V
5V: ~5.3V
12V: 12.45V unloaded, 12.0V when heavily loaded
Testing - Output Ripple:
Output ripple has been tested with the CPU load test whilst drawing about 450W from the wall. The scope shots are a bit noisy but this is most likely caused by the PC system itself, normally these shots are made when using resistive loads or specialized PSU test equipment. But anyway, it's better than nothing....
12V: ~80mV
5V: ~35mV
3.3V: ~15mV
Switch Frequency: ~55KHz
Conclusion:
All-in-all this isn't a bad result i think, especially being able to output ~325W with it's compact design and 470µF primary caps. Ripple ain't that bad either. The components were getting a bit warm though at this power level, i had a temp-probe attached to the secondary heatsink which was reading between 65~70°C. Not shockingly hot but definately indicating that there isn't much wiggling room left. The passive PFC coil is getting a bit toasty though.
I think it's ready for the real challenge, powering the Radeon RX6800 😁
A "nice" YoungYear unit? - Badcaps
The "new" unit is a KDM-M6480 480W psu, the 480W number is a typical KDM bullshit claim of course, but the 24A rating on the 12V rail seemed reasonable and the "Active PFC" claim gave me a bit of hope that this was going to be decent-ish. I mean, they even included an input voltage selector switch 🤨
Opening up the unit made clear it wasn't a decent Active-PFC unit and that it wasn't going to hit that 24A spec, 200W in total would've probably been a more realistic expectation...
Low and behold, a very basic and compactly built half-bridge unit... Just a mediocre one. There hasn't been any extreme corner cutting when compared to the real gutless wonders, but definitely not powerful in any way. The transformer and output toroid are pretty small and the PFC is nowhere to be found. Some noteable things were the partially present input filtering (yay), low profile transformer (EI33 but shorter), the small output toroid and the 3.3V rail which is actually tapped from the 5V rail and regulated with a mosfet.
The last couple of years this PSU was one of my side projects which sometimes got a bit of attention. The modded YoungYear unit was lasting better than expected so the urge to complete this project disappeared for a bit. But recently i acquired an Oculus Quest 3 VR system which required a bit more graphical horsepower from my PC. So here i go, ordering a Radeon RX6800 GPU while being fully aware that my current YoungYear unit can't handle it 😅 All hopes are on this KDM unit now 🤞 A couple of years ago i did make the first round of upgrades and posted the results in the hall of shame thread:
https://www.badcaps.net/forum/troubl...75#post1423575
To keep thing simple i'll just post the end result here. Last week i gave it a final round of upgrades, hoping it would do ~300W. It got upgraded quite extensively, the transistors, 12v rectifier, main transformer, output toroidal and heatsinks were upgraded. The whole PSU got Teapo-fied except for the primaries. Passive PFC was also added as well as temperature-based fan control.
Primary:
NO PFC -> Passive PFC
2x Y capacitor
NO X capacitor -> added X capacitor
1x common mode choke -> 1x better wound common mode choke
2x 2.5A diode + 2x 4A diode -> KBL406 rectifier bridge
2x 470uF 200V Ltec primary caps
2x 13007 primary transistors -> 2x D304X primary transistors (still TO220)
1x SVD2N60F mosfet
Swapped primary heatsink
Middleware:
ERL33 low-profile transformer -> ERL35 full-height transformer (swapped the primary snubber circuit with it)
SC6105 switching controller
Added temperature-based fan control
Secondary:
12V: MOSPEC F16C20C 16A TO220 Fast Recovery -> DSSK 40-008B 40A 80V TO-247 Schottky
5V: MOSPEC S30D45CS 30A TO247 Schottky
3.3V: Advanced Power AP40N03GP 40A mosfet (tapped from 5V)
Tiny output toroid -> Way bigger output toroid (with more efficient green/blue core and thicker windings)
Swapped secondary heatsink (modified a bit to make the larger output toroid fit)
Output Filtering:
12V: 1x 1000µF 16V ChengX, PI Coil -> 2x 1500µF 16V Teapo SY, PI Coil
5V: 2x 1000µF 10V ChengX, PI Coil -> 2x 2200µF 6.3V Teapo SC, PI Coil
3.3V: 1x 1000µF 10V ChengX, PI Coil -> 1x 2200µF 6.3V Teapo SC, PI Coil
+5VSB: 1x 470µF 16V ChengX -> 1x 1000µF 6.3V Teapo SY
-12V: 1x 470µF 16V ChengX, NO PI Coil -> 1x 1500µF 16V Teapo SY, added PI Coil
-5V: 1x 470µF 16V ChengX, NO PI Coil -> 1x 1000µF 6.3V Teapo SY
Testing - Power Output:
A heavily overclocked furnace ...uhh.... Pentium D CPU was used to test the constant power output capability. The test was running for about ~20 minutes to make sure heat-soak wouldn't get me in the end. Using my PR10 power meter (wall power) and a current clamp (output power) i got the following numbers:
Input power: ~440W (225V, 0.73 PF)
Output power: ~333W (~76% efficiency, ~24.4A@12V, ~5.2A@5V, ~3.0A@3.3V)
I used an overclocked Radeon HD6870 for testing it with a bit more transient loading. The Pentium D was downclocked a bit to keep power draw withing reasonable limits. The PSU didn't even flinch at this test while drawing 400~460W from the wall.
Testing - voltage regulation:
Voltage regulation seems to be okay-ish, perfectly adequate to power a PC:
3.3V: ~3.4V
5V: ~5.3V
12V: 12.45V unloaded, 12.0V when heavily loaded
Testing - Output Ripple:
Output ripple has been tested with the CPU load test whilst drawing about 450W from the wall. The scope shots are a bit noisy but this is most likely caused by the PC system itself, normally these shots are made when using resistive loads or specialized PSU test equipment. But anyway, it's better than nothing....
12V: ~80mV
5V: ~35mV
3.3V: ~15mV
Switch Frequency: ~55KHz
Conclusion:
All-in-all this isn't a bad result i think, especially being able to output ~325W with it's compact design and 470µF primary caps. Ripple ain't that bad either. The components were getting a bit warm though at this power level, i had a temp-probe attached to the secondary heatsink which was reading between 65~70°C. Not shockingly hot but definately indicating that there isn't much wiggling room left. The passive PFC coil is getting a bit toasty though.
I think it's ready for the real challenge, powering the Radeon RX6800 😁
Attached Files
Comment