More eBay specials: this time a Gigabyte GP-P450b 450W ATX PSU for $8 with free S&H. Listed “For part/not working”, as usual. Showed up in a beat up EVGA box without any padding… but that's understandable given the total price. PSU made it out fine without damage. It's a non-modular (yay!), pretty modern-day looking PSU with a black paint and all black cables (meh, not my favorite, but figure it's good to keep a few of these in case I do need one for a more modern PC build, just so my old gray boxes don't look too odd in there.)
And a shot of the label:
Apparently its single 12V rail is supposedly capable of 36 Amps (432W), while the whole unit is rated for 450W. 3V3 and 5V have a combined power of 105W, so that suggests group regulation. UL number listed is E313928, which seems to go back to… Gigabyte??
Strange! I didn't know Gigabyte made their own PSUs… and I was pretty sure they didn't. I was that hoping the UL number would reveal a little more information about the real manufacturer, but no. Otherwise, looking up the model number online leads to Gigabyte's page here:
https://www.gigabyte.com/Power-Supply/GP-P450B
Anyways, so I plugged the PSU in and… indeed it was DOA as the listing stated. According to the seller, the PSU came DOA to him due to shipping damage. I don't see any indications on the PSU's case to reaffirm this. But nevertheless, PSU was dead as promised, so I awarded the seller 5 stars for accurate description.
So, time to crack it open and see what's bad.
But wait! The warranty sticker is still intact! 
Given this is a pretty new PSU, it probably still has warranty. I even amused the idea of sending it in for RMA for a few moments
… but then, if Gigabyte really sent me a new PSU, this one would probably get land-filled, which doesn't help anyone or the environment. And besides, what's the fun in getting a working PSU anyways?! 
So, here it goes, warranty VOID-ed and top removed with no regrets whatsoever.
Well, there's your Gigabyte 450W PSU. Honestly, I was a bit surprised at how mediocre it looks. What particularly struck me first are the two toroids on the output, confirming that this is a regular group-regulated PSU. And with that, I have a hard time believing the 12V rail output can be loaded all the way to 36 Amps (432W) without the 5V rail going out of whack with a low/no load. If it can, then I bet the 5V/12V/-12V common-mode output inductor would be running hell-a “over-clocked” to make up for the cross-loading.
Before looking for the problem, I decided to finish the photo-session and do a build quality “summary” here. So, let's start with the primary side:
Nothing crazy to see ^ there – the input filtering is complete (and I have to say, that 2nd common-mode choke looks quite interesting); bridge rectifier is “naked” without a heatsink, but looks capable for the rating of the PSU; APFC boost coil also looks interesting and well-made; primary cap is a *yawn* Teapo 400V, 270 uF, 85°C part, but that's still better than some 3rd -grade dubious cap brand; all silicon on the primary heatsink consists of TO-220 parts (2x MOSFETs for APFC, 1x diode, 2x MOSFETs for main PS 2-transistor forward); and the MOV, thermistor, and fuse are all present, as they should be.
Next, the secondary side:
Ugh. This is getting too mediocre… maybe even bad. I would be surprised if the output filtering is enough to suppress ripple/noise adequately. Perhaps the 3.3V and 5V rails could be fine, as they aren't rated very high and each one is filtered by a 2200 uF and a 1000 uF cap with a small PI coil in between. The 12V rail is the one I'm worried about, as it's also filtered only by a 2200 uF and a 1000 uF cap (again, with a small PI coil). For 36 Amps worth of current, I'm skeptical it will hold up… but maybe it can? I haven't looked up any reviews on this unit yet, so this is pure speculation based on visual observation on my part. On the same note, the main (5V/12V) output inductor is only a T106-45 part… which seems a bit on the small side, especially considering that a -45 core is actually inferior in terms of losses compared to a -52 core (at least at higher frequencies.) On the other hand, -45 material offers higher inductance per turn, and my guess is the manufacturer did this in order to keep the 5V and 12V rails balanced with the higher 12V rail cross-loading modern PSUs are expected to handle. Note, however, that the 3.3V rail still uses a -26 material core material. So this suggests the overall PS switching frequency is likely under 50 KHz.
Also visible on this picture is the likely PSU manufacturer: MEIC. Various info sources online suggest their website disappeared when Gigabyte had (has?) the scandal with their bundled Newegg PSUs (the ones that were blowing up or DOA frequently.) On that note, the reviews for this particular model on Newegg also show a disproportionately high number of DOAs. With that stated, I'm including the PCB m/n on mine, in case this happens to apply only to certain batch of PCBs / models. It's P4502119-be04 and dated 2020/05/13. So yes, the PSU is definitely still under warranty!
Finally, here's a shot of the solder side:
The soldering is relatively good overall. I did notice a number of loose solder balls rolling around in the PSU case and a few stuck with flux on the PCB. But they were all really small and didn't seem like they could cause any shorting. Nonetheless, if there are small solder balls, this means bigger ones can exist too. I probably just didn't happen to run into any on my unit. But really MEIC just needs to tweak the temperature settings on their wave solder machine, it seems. If anyone is interested in high-resolution pictures of the solder side (say, to look up a resistor value for a blown unit), see the attached zip file at the end of the post.
Last but not least, the fan:
Hmmm, now where have I seen that fan brand before?
Oh yes, that crappy KDMPower MIPC MI-X8775CD ATX PSU I showed a while back had the same BOK (or as I like to pronounce it: “Be OK?”
) fan brand. Not exactly inspiring any confidence here, is it? I didn't open the fan to see if it really has a rifle-bearing, which the Gigabyte product page proudly advertises. But then again, it's a budget, “cheap” PSU (or maybe not, given the $49.99 price tag on Newegg?), so I couldn't care too much either way.
With that said, let me just finish the build summary here so that we can jump to the troubleshooting part afterwards:
Primary Side:
* EMI/RFI filtering: two 102M + three 222M(?) Y2-class caps; two 0.47 uF X2-class caps; two CM chokes
* Protection: T10AH250V fuse; NTC thermistor; MOV between L&N (not heatshrinked!)
* Other: 8-10 Amp (?) bridge rectifier (not heatsinked); 1N540x (1N5406?) APFC bypass diode
* APFC: 105j450V P.P. cap after B.R.; “MS30EQ-X02” APFC boost coil; 2x JCS18N50f MOSFETs (TO-220FP) + 1x SU860 diode (TO-220)
* Caps: 1x Teapo LH, 400V, 270 uF, 25x42 mm; 1x YC LE, 50V, 10 uF, 5x11 mm (5VSB startup?); 1x ChnCap TM, 50V, 47 uF, 6.3x13 mm (pri. aux. filter); 1x YC TH, 50V, 4.7 uF, 5x11 mm (main PS run/timing?)
* Main PS (2TF topology): 2x JCS18N50f MOSFETs (TO-220FP) + 2x SR308 diodes
* Transformers: “MS35ERL-X06” (35 mm) for main PS; “MS19ES-X30” (19 mm) for 5VSB; “MS10ES-X17” (10 mm) driver for main PS MOSFETs
ICs:
*PR6249 (PDIP-8) for 5VSB; CM6805bg (SSOP-10) for PWM+PFC; WT7525Y (PDIP-14) for supervision, PS-ON, and PG.
Secondary Side:
*Output Inductors: T106-45 common-mode toroid for 5V/12/-12V; T80-26 (20 mm dia.) for 3.3V rail
* 5VSB
*** 1x ChnCap TP, 10/16(?)V, 1500(?) uF, 8x20 mm cap (in 10 mm dia. spot) before PI coil
*** 1x ChnCap TP, 16V, 1000 uF, 8x16 mm cap after PI coil
*** SR5T45 (5 Amp, 45V) Schottky rectifier
*** PI coil: 14-turn, 20 AWG, 2.5x16 mm rod core
* 12V Rail
*** 1x YC LE, 16V, 2200 uF, 10x20 mm cap before PI coil
*** 1x YC LE, 16V, 1000 uF, 10x16 mm cap after PI coil
*** 2x SBP30V60ct (30 Amp, 60V, TO-220) Schottky rectifiers in parallel
*** PI coil: 4.5-turn, 16 AWG, 4x20 mm rod core
*** 2x 1-KOhm SMD load resistors in parallel
* 5V Rail
*** 1x YC LE, 10V, 2200 uF, 10x20 mm cap before PI coil
*** 1x ChnCap TP, 16V, 1000 uF, 8x16 mm cap (in 10 mm dia. spot) after PI coil
*** 1x SR3045Lct (30 Amp, 45V, TO-220) Schottky rectifier
*** PI coil: 4.5-turn, 18 AWG, 4x20 mm rod core
*** 2x 30-Ohm SMD load resistors in parallel
* 3.3V Rail
*** same exact arrangement as 5V rail above
* -12V Rail
*** 1x ChnCap TP, 16V, 680 uF, 8x14 mm cap (in 10 mm dia. spot) & no PI coil
*** 2x 1.5 Amp (?) diodes
And a shot of the label:
Apparently its single 12V rail is supposedly capable of 36 Amps (432W), while the whole unit is rated for 450W. 3V3 and 5V have a combined power of 105W, so that suggests group regulation. UL number listed is E313928, which seems to go back to… Gigabyte??

https://www.gigabyte.com/Power-Supply/GP-P450B
Anyways, so I plugged the PSU in and… indeed it was DOA as the listing stated. According to the seller, the PSU came DOA to him due to shipping damage. I don't see any indications on the PSU's case to reaffirm this. But nevertheless, PSU was dead as promised, so I awarded the seller 5 stars for accurate description.



Given this is a pretty new PSU, it probably still has warranty. I even amused the idea of sending it in for RMA for a few moments


So, here it goes, warranty VOID-ed and top removed with no regrets whatsoever.

Well, there's your Gigabyte 450W PSU. Honestly, I was a bit surprised at how mediocre it looks. What particularly struck me first are the two toroids on the output, confirming that this is a regular group-regulated PSU. And with that, I have a hard time believing the 12V rail output can be loaded all the way to 36 Amps (432W) without the 5V rail going out of whack with a low/no load. If it can, then I bet the 5V/12V/-12V common-mode output inductor would be running hell-a “over-clocked” to make up for the cross-loading.
Before looking for the problem, I decided to finish the photo-session and do a build quality “summary” here. So, let's start with the primary side:
Nothing crazy to see ^ there – the input filtering is complete (and I have to say, that 2nd common-mode choke looks quite interesting); bridge rectifier is “naked” without a heatsink, but looks capable for the rating of the PSU; APFC boost coil also looks interesting and well-made; primary cap is a *yawn* Teapo 400V, 270 uF, 85°C part, but that's still better than some 3rd -grade dubious cap brand; all silicon on the primary heatsink consists of TO-220 parts (2x MOSFETs for APFC, 1x diode, 2x MOSFETs for main PS 2-transistor forward); and the MOV, thermistor, and fuse are all present, as they should be.
Next, the secondary side:
Ugh. This is getting too mediocre… maybe even bad. I would be surprised if the output filtering is enough to suppress ripple/noise adequately. Perhaps the 3.3V and 5V rails could be fine, as they aren't rated very high and each one is filtered by a 2200 uF and a 1000 uF cap with a small PI coil in between. The 12V rail is the one I'm worried about, as it's also filtered only by a 2200 uF and a 1000 uF cap (again, with a small PI coil). For 36 Amps worth of current, I'm skeptical it will hold up… but maybe it can? I haven't looked up any reviews on this unit yet, so this is pure speculation based on visual observation on my part. On the same note, the main (5V/12V) output inductor is only a T106-45 part… which seems a bit on the small side, especially considering that a -45 core is actually inferior in terms of losses compared to a -52 core (at least at higher frequencies.) On the other hand, -45 material offers higher inductance per turn, and my guess is the manufacturer did this in order to keep the 5V and 12V rails balanced with the higher 12V rail cross-loading modern PSUs are expected to handle. Note, however, that the 3.3V rail still uses a -26 material core material. So this suggests the overall PS switching frequency is likely under 50 KHz.
Also visible on this picture is the likely PSU manufacturer: MEIC. Various info sources online suggest their website disappeared when Gigabyte had (has?) the scandal with their bundled Newegg PSUs (the ones that were blowing up or DOA frequently.) On that note, the reviews for this particular model on Newegg also show a disproportionately high number of DOAs. With that stated, I'm including the PCB m/n on mine, in case this happens to apply only to certain batch of PCBs / models. It's P4502119-be04 and dated 2020/05/13. So yes, the PSU is definitely still under warranty!
Finally, here's a shot of the solder side:
The soldering is relatively good overall. I did notice a number of loose solder balls rolling around in the PSU case and a few stuck with flux on the PCB. But they were all really small and didn't seem like they could cause any shorting. Nonetheless, if there are small solder balls, this means bigger ones can exist too. I probably just didn't happen to run into any on my unit. But really MEIC just needs to tweak the temperature settings on their wave solder machine, it seems. If anyone is interested in high-resolution pictures of the solder side (say, to look up a resistor value for a blown unit), see the attached zip file at the end of the post.
Last but not least, the fan:
Hmmm, now where have I seen that fan brand before?

Oh yes, that crappy KDMPower MIPC MI-X8775CD ATX PSU I showed a while back had the same BOK (or as I like to pronounce it: “Be OK?”

With that said, let me just finish the build summary here so that we can jump to the troubleshooting part afterwards:

Primary Side:
* EMI/RFI filtering: two 102M + three 222M(?) Y2-class caps; two 0.47 uF X2-class caps; two CM chokes
* Protection: T10AH250V fuse; NTC thermistor; MOV between L&N (not heatshrinked!)
* Other: 8-10 Amp (?) bridge rectifier (not heatsinked); 1N540x (1N5406?) APFC bypass diode
* APFC: 105j450V P.P. cap after B.R.; “MS30EQ-X02” APFC boost coil; 2x JCS18N50f MOSFETs (TO-220FP) + 1x SU860 diode (TO-220)
* Caps: 1x Teapo LH, 400V, 270 uF, 25x42 mm; 1x YC LE, 50V, 10 uF, 5x11 mm (5VSB startup?); 1x ChnCap TM, 50V, 47 uF, 6.3x13 mm (pri. aux. filter); 1x YC TH, 50V, 4.7 uF, 5x11 mm (main PS run/timing?)
* Main PS (2TF topology): 2x JCS18N50f MOSFETs (TO-220FP) + 2x SR308 diodes
* Transformers: “MS35ERL-X06” (35 mm) for main PS; “MS19ES-X30” (19 mm) for 5VSB; “MS10ES-X17” (10 mm) driver for main PS MOSFETs
ICs:
*PR6249 (PDIP-8) for 5VSB; CM6805bg (SSOP-10) for PWM+PFC; WT7525Y (PDIP-14) for supervision, PS-ON, and PG.
Secondary Side:
*Output Inductors: T106-45 common-mode toroid for 5V/12/-12V; T80-26 (20 mm dia.) for 3.3V rail
* 5VSB
*** 1x ChnCap TP, 10/16(?)V, 1500(?) uF, 8x20 mm cap (in 10 mm dia. spot) before PI coil
*** 1x ChnCap TP, 16V, 1000 uF, 8x16 mm cap after PI coil
*** SR5T45 (5 Amp, 45V) Schottky rectifier
*** PI coil: 14-turn, 20 AWG, 2.5x16 mm rod core
* 12V Rail
*** 1x YC LE, 16V, 2200 uF, 10x20 mm cap before PI coil
*** 1x YC LE, 16V, 1000 uF, 10x16 mm cap after PI coil
*** 2x SBP30V60ct (30 Amp, 60V, TO-220) Schottky rectifiers in parallel
*** PI coil: 4.5-turn, 16 AWG, 4x20 mm rod core
*** 2x 1-KOhm SMD load resistors in parallel
* 5V Rail
*** 1x YC LE, 10V, 2200 uF, 10x20 mm cap before PI coil
*** 1x ChnCap TP, 16V, 1000 uF, 8x16 mm cap (in 10 mm dia. spot) after PI coil
*** 1x SR3045Lct (30 Amp, 45V, TO-220) Schottky rectifier
*** PI coil: 4.5-turn, 18 AWG, 4x20 mm rod core
*** 2x 30-Ohm SMD load resistors in parallel
* 3.3V Rail
*** same exact arrangement as 5V rail above
* -12V Rail
*** 1x ChnCap TP, 16V, 680 uF, 8x14 mm cap (in 10 mm dia. spot) & no PI coil
*** 2x 1.5 Amp (?) diodes
Comment