So for once, I can't be made fun of for working on an old ATX 1.3 half bridge unit because it is being used in a Pentium II machine
I found this unit at the local PC recycling store for $4. It's a decent looking ATNG unit. Keep in mind, this thing had never had power applied to it! It was made in early 2009. This will give us an idea of the quality of Teapo caps compared to Japanese parts. You probably quickly noticed the bulged 2200uF 10V Teapo SC. It read 2020uF 0.72Ω ESR. Some of the other caps were out of spec as well without bulging. Such as one of the 1000uF caps on the 5VSB that read 1359uF.
What you also probably noticed is those huge resistors, especially on the 5V rail. 7.5Ω, what's up with that? 10Ω on the 3.3V rail. 270Ω on the 12V. So the first thing I did was plug in the unit with a paperclip to make sure it worked, and to see how much power it burned just to be running. Well, it was pulling 22W AC! That's the highest I've seen in any unit. That 7.5Ω resistor got really hot pretty quickly, imagine that It's making what, over 3W? If this thing had actually been put in service, I can only imagine how fast that resistor would have cooked the caps on the -12V and -5V rail, even the way the fan is mounted. The fan is placed in the front of the PSU, blowing towards the back of the unit, but the fan spins pretty slow. Since I was going to use this unit in a 5V based machine, I replaced the resistor on the 5V with a 240Ω resistor. I replaced the 10Ω resistor on the 3.3V with 24Ω. It still burns a fair amount but a lot less than the original. I left the 270Ω resistor on the 12V since it wasn't going to have much load on it at all.
I replaced all the electrolytic capacitors except for the 680uF CS caps on the input. I'm not sure if they were actually 680uF at one point, because they would have degraded since then sitting for 6 years. They read 583uF and 597uF. Interestingly, the Teapo SEK seemed to hold up better than the Teapo SC. Is this because the SEK are GP, so they don't have as good of specs to hold in the first place? The only thing that was somewhat sketchy was replacing the cap before the coil on the 12V rail. They originally had a 2200uF 16V 10x30mm Teapo SC. 1780 ripple current and 0.032Ω ESR. Since they had suspended the cap to fit in a 8mm slot, I used a 1000uF 16V 8x20mm 1560 ripple current and 0.030Ω ESR. I'm not worried so much about the RC or ESR, but the capacitance. I also figured that it wouldn't matter that much since this unit will hardly be using any 12V at all.
The bridge rectifier is 6A, and this unit surprisingly uses beefy 2SC2625 switchers.
I figured that the rectifiers were a little underspecc'd. 30A on the 5V and 20A on the 3.3V. So I put the 30A schottky from the 5V on the 3.3V, and put a 40A schottky on the 5V rail. I left the 16A ultra fast on the 12V because I know that with such a low load on the 12V, a schottky or higher rated ultra fast would have just shot up the 12V rail. It sits at a nice 12.08V in this system so I'm happy. I think it's odd how much bigger the 3.3V toroid is compared to the other one. I guess it's a true 1.3 unit
The soldering isn't bad at all, but there were some leads that definitely needed to be cut. I redid a few of the joints. The sleeve bearing Globe Fan had a really dry bearing so I oiled it.
After replacing the capacitors, rectifiers and minimum load resistors, the PSU then pulled 12W from the wall just using a paper clip to turn it on. Quite a difference!
The system it's powering is a Pentium II 300MHz, 128MB 100MHz SDRAM, 4x AGP video card, 3ware 8006-2LP RAID card running a RAID 1 with two Western Digital WD800JD. It also has an old Intel + Realtek 10/100 network cards. (That PSU pictured was a test Sun Pro PSU). I installed the hardware into an actual desktop style chassis, and put a brand new Pentium II sticker on it
You should have seen the look on the dude's face when I asked if they had any recycled Pentium II equipment. "A Pentium two?! Did I hear you right?" I got the board and proc for $7. What it's doing is it's a router/firewall/VPN at my brother's apartment so that now we can connect our networks.
I found this unit at the local PC recycling store for $4. It's a decent looking ATNG unit. Keep in mind, this thing had never had power applied to it! It was made in early 2009. This will give us an idea of the quality of Teapo caps compared to Japanese parts. You probably quickly noticed the bulged 2200uF 10V Teapo SC. It read 2020uF 0.72Ω ESR. Some of the other caps were out of spec as well without bulging. Such as one of the 1000uF caps on the 5VSB that read 1359uF.
What you also probably noticed is those huge resistors, especially on the 5V rail. 7.5Ω, what's up with that? 10Ω on the 3.3V rail. 270Ω on the 12V. So the first thing I did was plug in the unit with a paperclip to make sure it worked, and to see how much power it burned just to be running. Well, it was pulling 22W AC! That's the highest I've seen in any unit. That 7.5Ω resistor got really hot pretty quickly, imagine that It's making what, over 3W? If this thing had actually been put in service, I can only imagine how fast that resistor would have cooked the caps on the -12V and -5V rail, even the way the fan is mounted. The fan is placed in the front of the PSU, blowing towards the back of the unit, but the fan spins pretty slow. Since I was going to use this unit in a 5V based machine, I replaced the resistor on the 5V with a 240Ω resistor. I replaced the 10Ω resistor on the 3.3V with 24Ω. It still burns a fair amount but a lot less than the original. I left the 270Ω resistor on the 12V since it wasn't going to have much load on it at all.
I replaced all the electrolytic capacitors except for the 680uF CS caps on the input. I'm not sure if they were actually 680uF at one point, because they would have degraded since then sitting for 6 years. They read 583uF and 597uF. Interestingly, the Teapo SEK seemed to hold up better than the Teapo SC. Is this because the SEK are GP, so they don't have as good of specs to hold in the first place? The only thing that was somewhat sketchy was replacing the cap before the coil on the 12V rail. They originally had a 2200uF 16V 10x30mm Teapo SC. 1780 ripple current and 0.032Ω ESR. Since they had suspended the cap to fit in a 8mm slot, I used a 1000uF 16V 8x20mm 1560 ripple current and 0.030Ω ESR. I'm not worried so much about the RC or ESR, but the capacitance. I also figured that it wouldn't matter that much since this unit will hardly be using any 12V at all.
The bridge rectifier is 6A, and this unit surprisingly uses beefy 2SC2625 switchers.
I figured that the rectifiers were a little underspecc'd. 30A on the 5V and 20A on the 3.3V. So I put the 30A schottky from the 5V on the 3.3V, and put a 40A schottky on the 5V rail. I left the 16A ultra fast on the 12V because I know that with such a low load on the 12V, a schottky or higher rated ultra fast would have just shot up the 12V rail. It sits at a nice 12.08V in this system so I'm happy. I think it's odd how much bigger the 3.3V toroid is compared to the other one. I guess it's a true 1.3 unit
The soldering isn't bad at all, but there were some leads that definitely needed to be cut. I redid a few of the joints. The sleeve bearing Globe Fan had a really dry bearing so I oiled it.
After replacing the capacitors, rectifiers and minimum load resistors, the PSU then pulled 12W from the wall just using a paper clip to turn it on. Quite a difference!
The system it's powering is a Pentium II 300MHz, 128MB 100MHz SDRAM, 4x AGP video card, 3ware 8006-2LP RAID card running a RAID 1 with two Western Digital WD800JD. It also has an old Intel + Realtek 10/100 network cards. (That PSU pictured was a test Sun Pro PSU). I installed the hardware into an actual desktop style chassis, and put a brand new Pentium II sticker on it
You should have seen the look on the dude's face when I asked if they had any recycled Pentium II equipment. "A Pentium two?! Did I hear you right?" I got the board and proc for $7. What it's doing is it's a router/firewall/VPN at my brother's apartment so that now we can connect our networks.
Comment