Hardware Insights Forum

First off, this is probably the nicest Sun Pro PSU you will ever see. Sun Pro made a double forward?! The primary uses two Fairchild 9N90C FET's on the primary. The label claims 450W, which seems pretty honest. It looks to have two 12V rails with OCP. Controller on the secondary is PS222S http://www.siti.com.tw/product/spec/Power/PS222S.pdf

There is a Texas Instruments TL3843P on the primary. Uses an 80mm fan on the rear and 120mm for exhaust. All caps were Canicon/CS/JEE/GL. Only one CS cap and two GL caps. The 2200µF 6.3V CS cap failed on the 3.3V after the coil. I think for two reasons: It was physically touching the -12V output rectifier, and it was 8mmx20mm. Seems small, must be GP (No series printed on it) The only other caps that failed was the 100µF 10V JEE on the -5V rail, and both 4700µF 10V Canicon on the 5V filtering. I was blown away that botth 3300µF 16V JEE on the 12V filtering were okay. They tested 3367µF 0.02Ω and 3412µF 0.02Ω. Those caps I put in are 30mm tall, so look how big that those toroids are! The heatsinks are huge too.

So on the 5VSB, it looks like they used two minimum load resistors, and one is on the bottom. Anyone know why they would do this? Can I just remove the one on the bottom? I wrote down what's what so that you can get a better reference. The 5VSB works, but it runs very hot. Let me know if you guys want to see more pictures of the unit.

On a side note, what do you guys think the PSU is capable of? I think it looks pretty solid. The soldering is horrendous though, I'm working on it.

I take that you meant a 120mm intake fan?

Clearly JEE are hopeless if they can't even last on the output of a linear regulator. (I don't see any load resistor there and drawing enough −5V current to heat the regulator appreciably seems highly dubious.)

Pentium wrote:So on the 5VSB, it looks like they used two minimum load resistors, and one is on the bottom. Anyone know why they would do this? Can I just remove the one on the bottom?

I don't know why they'd do it, either. What are their values (I can't tell clearly from the images)? My suggested replacement would be something like a 100Ω 0.5W, or a 51Ω 1W (I don't think a 2W resistor would fit there). Of course, if a lower load is adequate, go ahead and use a higher value. (In some other Sun Pro units they used a 51Ω 0.5W resistor, running right at its rating…)

That aside, I'm amazed to see a Sun Pro with proper Y capacitors, a ceramic fuse in a holder (instead of a cheap, occasionally explosive glass fuse soldered directly to the board), and an ERL-39 transformer.

By the way, what are you going to replace C45 and C46 with? And what are the values of the Nichicon PWs you used (as well as the value and series of that Rubycon you put after the +5V pi coil)? Any MOVs in the unit? And any plans on replacing those Canicon primaries?

The fan configuration is like this:

Clearly JEE are hopeless if they can't even last on the output of a linear regulator. (I don't see any load resistor there and drawing enough −5V current to heat the regulator appreciably seems highly dubious.)

Yes, they are pretty useless. I was surprised they survived on the 12V though. I know that the higher voltage helps the oxide layer in the electrolyte, but these are 9 year old JEE caps we're talking about

I don't know why they'd do it, either. What are their values (I can't tell clearly from the images)? My suggested replacement would be something like a 100Ω 0.5W, or a 51Ω 1W (I don't think a 2W resistor would fit there). Of course, if a lower load is adequate, go ahead and use a higher value. (In some other Sun Pro units they used a 51Ω 0.5W resistor, running right at its rating…)

I'll check their values. The one on the bottom side looks like 21Ω 5% which is pretty low. Not sure what the top one is. Is there any reason this thing would run with one and not the other? I know Sun Pro does strange things some times...This one also doesn't have a fan control thermistor, so I just hooked the ground on the fan wire to 5V. That should be good enough for airflow/noise.

That aside, I'm amazed to see a Sun Pro with proper Y capacitors, a ceramic fuse in a holder (instead of a cheap, occasionally explosive glass fuse soldered directly to the board), and an ERL-39 transformer.

I was too, especially with the fuse. I've seen a 39 transformer before in a Sun Pro but it was half bridge. It looks like they did a good job minus the caps with this PSU. Even both the Te Bao fans were well oiled with oil (not grease). There's also more input filtering on the AC receptacle.

By the way, what are you going to replace C45 and C46 with? And what are the values of the Nichicon PWs you used (as well as the value and series of that Rubycon you put after the +5V pi coil)? Any MOVs in the unit? And any plans on replacing those Canicon primaries?

I replaced C46 with nichicon PW 1000µF 10V 8mmx20mm. Since C45 is a small space I will probably use the same PW cap there as well. The C45 cap there right now is actually still the same JEE cap (With the Rubycon vent) It's 220µF 16V. The unit has two MOV's. I think the Canicon primaries will be okay, however, they read ~1000µF so they were probably never 1200µF in the first place. The only other suitable options I have would be 1200µF JEE (That actually read 1200) and some brand new 1000µF 200V Xicon .

Pentium wrote:The one on the bottom side looks like 21Ω 5% which is pretty low.

But 21 isn't a standard E24 series value. (The E24 series goes: 1.0, 1.1, 1.2, 1.3, 1.5, 1.6, 1.8, 2.0, 2.2, 2.4, 2.7, 3.0, 3.3, 3.6, 3.9, 4.3, 4.7, 5.1, 5.6, 6.2, 6.8, 7.5, 8.2, 9.1)

And if it is a 21Ω resistor, it would be dissipating 5^2 / 21 = 1.19W!!! Going by its size, it's probably rated for 125mW, but might just be 250mW (there are "miniaturised" versions of axial resistors that have the same power rating as the next size up in the "standard" models, although you can't tell which is which by visual inspection) — so in that case, it ought to run damn hot.

For what it's worth, metal-oxide film resistors (grey coated) are probably the best type for the application, but observing the proximity of R18 to C45, the endurance rating of the cap probably matters more. So I would suggest Chemi-con KY/Nichicon HE in either 330µF 6.3V or 220µF 10V.

I pulled it and it read 22.1Ω. The one on the top read 33.2Ω. They both lost some of their coating from getting so hot. The PI coil also got so hot that when I went to move it, one of the legs fell off! I replaced it from one pulled from an Allied PSU. The JEE cap by the way was 330µF 10V. And it bulged out the bottom! It read open circuit and 28Ω ESR. I removed the bottom resistor, and replaced the top one with 100Ω. It obviously didn't like this, because the 5VSB was running at 10.30V! And 8.45V plugged into a motherboard. The 5VSB circuit was making no audible noise whatsoever. I had replaced the 330µF JEE cap with a 1000µF 10V nichicon PW.

Pentium wrote:I removed the bottom resistor, and replaced the top one with 100Ω. It obviously didn't like this, because the 5VSB was running at 10.30V! And 8.45V plugged into a motherboard. The 5VSB circuit was making no audible noise whatsoever. I had replaced the 330µF JEE cap with a 1000µF 10V nichicon PW.

Well, that's quite a marvellous fuck-up, to take some ridiculous pre-load to maintain regulation on the standby output, isn't it? Way to go, Sun Pro! You build a unit that's so close to decent, only to ruin it by royally screwing up the standby supply? I do at least give Macron credit for not making design blunders like this.

(By the way, both carbon film and metal film resistors have a temperature rating of 155°C — although it's not said for how long that is. The power rating of most small resistors is given at 70°C ambient.)

LongRunner wrote:

Pentium wrote:I removed the bottom resistor, and replaced the top one with 100Ω. It obviously didn't like this, because the 5VSB was running at 10.30V! And 8.45V plugged into a motherboard. The 5VSB circuit was making no audible noise whatsoever. I had replaced the 330µF JEE cap with a 1000µF 10V nichicon PW.

Well, that's quite a marvellous fuck-up, to take some ridiculous pre-load to maintain regulation on the standby output, isn't it? Way to go, Sun Pro! You build a unit that's so close to decent, only to ruin it by royally screwing up the standby supply? I do at least give Macron credit for not making design blunders like this.

(By the way, both carbon film and metal film resistors have a temperature rating of 155°C — although it's not said for how long that is. The power rating of most small resistors is given at 70°C ambient.)

I know, it's annoying that it was so close to a great design with poor build quality. This 5VSB circuit does not have a critical cap, so I know it isn't that. I changed out the optocoupler, and a 431 shunt regulator, and it's still doing the same thing. There is also a 945 transistor that I have not checked yet. So, I did something that I knew probably wouldn't work, but I wanted to see if it would bring it down to a good voltage. So I installed just a 5Ω resistor, and this brought the 5VSB down to 5.06V, but the resistor melted the heatshrink around itself in about 20 seconds. What does this tell us?? Also, thank you for the temperature info on resistors, I didn't know that

Pentium wrote:So I installed just a 5Ω resistor, and this brought the 5VSB down to 5.06V, but the resistor melted the heatshrink around itself in about 20 seconds. What does this tell us??

What power rating and type of resistor was that?

Metal-oxide film resistors BTW can withstand up to 200°C, sometimes more. I suppose you could check out every other part of the standby supply, if you have nothing better to do…

Pentium wrote:I'll check their values. The one on the bottom side looks like 21Ω 5% which is pretty low. Not sure what the top one is. Is there any reason this thing would run with one and not the other? I know Sun Pro does strange things some times...This one also doesn't have a fan control thermistor, so I just hooked the ground on the fan wire to 5V. That should be good enough for airflow/noise.

Probably because they thought they need to draw higher current to make it stable, but lower resistor than 21 ohm would vaporize, so they used two in parallel. I've recently read about flyback controllers and it seems they can work with no feedback from secondary, being controlled just from extra primary winding. But than you need continuous minimum load on output so it can detect magnetic flow. Interesting fact, maybe that applies for more topologies and that explains the often need for big minimum load resistors?

With that fan, it may not be a good idea if there is not high enough zero-load draw from +5 V. Because if not, than basically current will flow oposite direction (instead of from the rail to the rail from +12 V), it can make the driving circuits go crazy. Better is to add a resistor in series with the fan.