Delta Electronics GPS-350CN-100 need repair

[goodpsusearch]

Here in Greece we don't see Good brand power supplies often. I was excited when I realized the psu I picked up some days ago was a Delta.

Fortunately, the guy who threw it away took the fan only. Could have been worse!

Don't expect too much though! It's a classic half bridge design with 2X 13009 transistors. In TO-3P package though!

The primary caps are 2x 680uF 200V Capxon. The transformer seems bigger than ERL-35, What do you think? Is it ERL-39?

The input filter is there, including MOV, but there is no NTC inrush current limiter. There is no PFC either, active or passive.

The bridge rectifier is T8KB 80, but with no heatsink attached to it.

The 5vsb mosfet is K3767.

Output rectifiers:
MBR3045PT @ 3.3V
MBR3045PT @ 5V
2x STPR1620CT =32A @ 12V

SG6105D1Z is used as Supervisor + Regulator + PWM

While every rail has a coil and 2 caps (-12V has coil and 1 cap), 12V has no coil and just one filtering cap.

All the output capacitors are Ltec and appear fine except the one used in 12V which is Capxon GL 2200uF 16V d12.5mm and is bad.

I replaced the bad cap, put a fan on it and tested the psu: 5vsb is there, but when you short green with ground, the psu tries to start for fractions of second and shuts down immediately.

Any help? What should I look for?

[Wester547]

Re: Delta Electronics GPS-350CN-100 need repair

Delta using run in the mill half-bridge topology and a two-transitor +5VSB circuit in a 2006 PSU.... (I thought Delta was more "forward" thinking than that. No pun intended...).

And only one capacitor and no ferrite coil on +12V? They could have least used two in parallel.... and of course the sole failed cap in the unit is a CapXon....

No NTC thermistor? Contrast that to those older Dell 250W Newtons which often had two NTC thermistors on the primary...

That aside, I recommend replacing all the small capacitors on the primary. One of the small capacitors biasing the BJTs (or the critical +5VSB cap) could have gone bad and those smaller LTECs like to go open circuit without showing it. A full recap excepting the primaries might be in order too (though, since they're CapXon, it couldn't hurt to replace them, and there appears to be one more tiny CapXon on the secondary). I'd also check the secondary rectifiers and bridge rectifier for shorts. I do see some of that yellow glue but it doesn't appear to have gone conductive (brown) as yet.

And that transformer is definitely size 39, written right on the transformer. Half-bridge topology will need it as the slow switching speed of the BJTs will require larger bulk storage and a greater transformer with a good turns ratio. As far as I know, Delta's forward PSUs do employ a higher switching frequency for the SMPS transformer (they run at around 100KHz) which means they can get away with a smaller transformer for higher wattage PSUs.

[kaboom]

Re: Delta Electronics GPS-350CN-100 need repair

While every rail has a coil and 2 caps (-12V has coil and 1 cap), 12V has no coil and just one filtering cap.

Typical of older designs, where +12 wasn't taken seriously. Or continued for the sake of cheapness.

All the output capacitors are Ltec and appear fine except the one used in 12V which is Capxon GL 2200uF 16V d12.5mm and is bad.

Capxon Garbage 'Lytic- no surprise there. I'd replace the ltecs, too, while it's apart; save yourself the grief.

I replaced the bad cap, put a fan on it and tested the psu: 5vsb is there, but when you short green with ground, the psu tries to start for fractions of second and shuts down immediately.

Any help? What should I look for?

Is it sqealing when it tries to start?


Check for shorted rectifiers on +12, +5, and +3.3, in that order. If okay, also check small rectifiers for -5 and -12.

Assymetric base drive to the switchers can put net DC into TX primary, in which case base drive ckt needs to be rebuilt. There are two 4.7u 50v lytics that go bad.

It's probably the +12 rectifier, since its cap was open, leading to the output inductor stressting that rectifier, so check that first.

[kaboom]

Re: Delta Electronics GPS-350CN-100 need repair

Half-bridge topology will need it as the slow switching speed of the BJTs will require larger bulk storage and a greater transformer with a good turns ratio. As far as I know, Delta's forward PSUs do employ a higher switching frequency for the SMPS transformer (they run at around 100KHz) which means they can get away with a smaller transformer for higher wattage PSUs.

Don't confuse switching frequency with rise and fall times.

MOSFETS can turn off (or on) slowly as well!

Lack of gate current supply or sinking will do it every time.

BJTs can also switch off too slowly- like in a TV or monitor horiz out, where long off-times would kill the device, and sometimes destroy the power supply (sonys often did this).

[Wester547]

Re: Delta Electronics GPS-350CN-100 need repair

Don't confuse switching frequency with rise and fall times.

MOSFETS can turn off (or on) slowly as well!

Lack of gate current supply or sinking will do it every time.

BJTs can also switch off too slowly- like in a TV or monitor horiz out, where long off-times would kill the device, and sometimes destroy the power supply (sonys often did this).

Well, it was to my knowledge that BJTs generally have much slower switching speeds than MOSFETs. Regular old BJTs, I mean, not IGBTs. At least that's what the datasheets would have you believe but it's true that they don't tell the whole story. As to switching frequency, I meant the frequency the main transformer was running at. I know that if MOSFETs (and BJTs) aren't properly driven that things can go amuck. And full bridge topology is clearly very capable (1KW+ I believe) along with half-bridge LLC resonant topology so I'm not saying BJTs are useless by any stretch, though it was also to my knowledge that plain old BJTs operate best at switching frequencies below 70KHz, and that their power dissipation was not as good as that of a MOSFET (though I think they might have an advantage as far as their equivalent of ON resistance goes).

And it does look to me that it was meant to be a +12V heavy unit - the label would at least have one believe that with dual +12V rails (10A and 15A) for a combined rating of 17.5A (210W) which dominates the 130W combined rating of the +3.3V and +5V rails. I just think this is a low-end Delta unit, and you can tell the difference between the low and high end units for sure (using -52 material on the +3.3V toroid and -26 material on the +5V/+12V toroid is an odd choice). It's definitely possible that the CapXon capacitor is the culprit (could have failed shorted or open) with respect to the shorted +12V rectifiers (if one, or the other, or both, is shorted).

[goodpsusearch]

Re: Delta Electronics GPS-350CN-100 need repair

Checked all caps with esr-micro meter. Every cap appears to be perfectly fine.

Replaced the 2 12V rectifiers with a single KNOWN GOOD 12V rectifier but nothing changed.

This psu has 2 mag-amp coils for 3.3V generation. One of them looks slightly discolored. Could it be the reason it doesn't start? How can I measure the coils?

[Wester547]

Re: Delta Electronics GPS-350CN-100 need repair

I find it funny that with accordance to the datasheet, "CapXon GL" (10mm and higher) is supposed to be rated for 6,000 hours @ 105*C endurance... yet it is the one failed capacitor in the unit....

...

Yes, this PSU uses two toroids to derive the +3.3V pins from the +5V pins. Any chance that one of them is shorting to the secondary heatsink (touching it)? Slight discoloration could mean the aforementioned one is bad. Might help to just replace it. One of the other secondary rectifiers or primary rectifiers could still be shorted. In order to test the toroid in question thoroughly, you will need an LC meter.

[goodpsusearch]

Re: Delta Electronics GPS-350CN-100 need repair

I checked all the components marked with green color in the following photos and found them ok. I had to unsolder several of them because when soldered on PCB they tricked my multimeter and gave false values.

https://www.badcaps.net/forum/attach...1&d=1431043164
https://www.badcaps.net/forum/attach...1&d=1431043164

The only thing that is definitely bad is a B647 PNP transistor, marked with red color in the photos attached.

I also attach a photo of the 2 mag amp coils but I can't tell if the one on the left is toasty or not...

https://www.badcaps.net/forum/attach...1&d=1431043164

[momaka]

Re: Delta Electronics GPS-350CN-100 need repair

The only thing that is definitely bad is a B647 PNP transistor, marked with red color in the photos attached.

Yep, that looks like the transistor for the 3.3V rail mag-amp circuit. Need to replace it. Considering the original is described as "low frequency power amplifier", that means you can use just about any PNP transistor. Just match the current rating and type (PNP). I really doubt you need a voltage rating as high as the original. And perhaps avoid transistors with a very low current gain (low hFE). On that note, the hFE doesn't need to match perfectly - just need to get it in the ballpark more or less. The transistors is used for switching here, so the hFE only needs to be large enough to enable the transistor to fully turn ON.
Something like TIP42 might work here (although, be advised that TIP42 is TO-220, so you might have to put it else where and run wires to it). Or B772. Just see what your local shops have, and let us know. It really just needs to be PNP type and TO-92MOD case or larger. I would go for TO-126 like the B772 transistor - it's smaller than TO-220, but larger than TO-92MOD. Or try SOT-223 if you want a ghetto-looking repair .

I also attach a photo of the 2 mag amp coils but I can't tell if the one on the left is toasty or not...

https://www.badcaps.net/forum/attach...1&d=1431043164

It does look a bit toasty. Make sure it is not touching the secondary heatsink or the other coil. Perhaps it was shorting on something. Or maybe someone just overloaded the 3.3V rail on this PSU. Who knows. Fix what's broken first and we can go from there. Maybe get two PNP transistors for the mag-amp, just in case .

[momaka]

Re: Delta Electronics GPS-350CN-100 need repair

Well, it was to my knowledge that BJTs generally have much slower switching speeds than MOSFETs.

Not necessarily. It all depends on the device what it is used for.

For example, at really high power applications, the devices will also tend to be very big. For MOSFETs, this usually means large gate capacitances, so that can become a limiting factor in how fast the device can be turned on.

Early in the days of semiconductors (think back to 74- TTL logic and earlier), BJTs used to be regarded as the faster switching devices. But that's because CMOS (MOSFET) technology had not yet fully developed. The problem with early MOSFETs was that they had rather sensitive gates (especially to static) and couldn't be turned ON/OFF very quickly, even if you could provide them with large enough gate currents. But FET technology has evolved much since then, and MOSFETs have caught up and surpassed BJTs in many aspects. However, BJTs are still tougher in general. That's because a BJT is essentially like two diodes on one die.

though it was also to my knowledge that plain old BJTs operate best at switching frequencies below 70KHz, and that their power dissipation was not as good as that of a MOSFET (though I think they might have an advantage as far as their equivalent of ON resistance goes).

BJT's don't have an "ON" resistance. Instead, when fully saturated, they have a constant voltage drop between their C-E terminals, which is usually at least as high as the base-emitter voltage (so upwards of 0.6V). Therefore, the power dissipation of the BJT device increases almost linearly as the current though the BJT device increases (Pd = Vce * Ic).

With MOSFETs, it's a difference story. The power dissipation of the FET depends on the Rds_on resistance and current going through the FET. Or:
Pd = Rds_on * Id^2. (Id = drain current)

And of course, the way the two devices function is completely different. BJTs act like current-controlled current source (push/pull a small current in the base of the BJT, and you get a large current out/in the emitter - all related by the current gain, hFE). With FETs, they are more similar to an electronic switch... or a voltage controlled resistor if you run them in their triode region.
---------------
Woops, sorry for double-posting. I thought I had integrated this into a single post.

[goodpsusearch]

Re: Delta Electronics GPS-350CN-100 need repair

Yep, that looks like the transistor for the 3.3V rail mag-amp circuit. Need to replace it. Considering the original is described as "low frequency power amplifier", that means you can use just about any PNP transistor. Just match the current rating and type (PNP). I really doubt you need a voltage rating as high as the original. And perhaps avoid transistors with a very low current gain (low hFE). On that note, the hFE doesn't need to match perfectly - just need to get it in the ballpark more or less. The transistors is used for switching here, so the hFE only needs to be large enough to enable the transistor to fully turn ON.
Something like TIP42 might work here (although, be advised that TIP42 is TO-220, so you might have to put it else where and run wires to it). Or B772. Just see what your local shops have, and let us know. It really just needs to be PNP type and TO-92MOD case or larger. I would go for TO-126 like the B772 transistor - it's smaller than TO-220, but larger than TO-92MOD. Or try SOT-223 if you want a ghetto-looking repair .


It does look a bit toasty. Make sure it is not touching the secondary heatsink or the other coil. Perhaps it was shorting on something. Or maybe someone just overloaded the 3.3V rail on this PSU. Who knows. Fix what's broken first and we can go from there. Maybe get two PNP transistors for the mag-amp, just in case .

I did that and now the psu works!

I used A1013, bought 3 just to be sure if the first blows, I will have more chances

The 2 mag amp coils were glued together. This is the good type of glue (well, I hate glue in psus, but that's another story) that doesn't get brown with age/heat. Still, I removed the glue and made sure the coils don't touch the heatsink and each other.

The psu works, but the voltage regulation with no load / DVD-Rom / Laptop HDD is disappointing. The psu I got is a stripped down version. There is no PPFC although the primary heatsink is designed to leave space for a PPFC coil. There is not daughterboard for second 12V rail support etc. This psu is a single rail unit.

I found a photo online of how this psu should be:
http://arch.pconline.com.cn/pingce/2...wer_359_07.jpg

Find the differences

[Wester547]

Re: Delta Electronics GPS-350CN-100 need repair

Not necessarily. It all depends on the device what it is used for.

For example, at really high power applications, the devices will also tend to be very big. For MOSFETs, this usually means large gate capacitances, so that can become a limiting factor in how fast the device can be turned on.

Early in the days of semiconductors (think back to 74- TTL logic and earlier), BJTs used to be regarded as the faster switching devices. But that's because CMOS (MOSFET) technology had not yet fully developed. The problem with early MOSFETs was that they had rather sensitive gates (especially to static) and couldn't be turned ON/OFF very quickly, even if you could provide them with large enough gate currents. But FET technology has evolved much since then, and MOSFETs have caught up and surpassed BJTs in many aspects. However, BJTs are still tougher in general. That's because a BJT is essentially like two diodes on one die.

Diodes have been less tough than FETs, in my experience. And as far as I know, BJTs are very fast as linear switches, but as a saturated switch, their switching speed is slow compared to modern MOSFETs.

BJT's don't have an "ON" resistance. Instead, when fully saturated, they have a constant voltage drop between their C-E terminals, which is usually at least as high as the base-emitter voltage (so upwards of 0.6V). Therefore, the power dissipation of the BJT device increases almost linearly as the current though the BJT device increases (Pd = Vce * Ic).

That's why I said their "equivalent" of ON resistance. As I understand it, they do have some sort of "load" resistance during on and off switching times, though. And as far as I know, yes, BJTs have a negative temperature coefficient in that sense. AFAIK, MOSFETs are voltage driven and BJTs current driven as you correctly pointed out.

With MOSFETs, it's a difference story. The power dissipation of the FET depends on the Rds_on resistance and current going through the FET. Or:
Pd = Rds_on * Id^2. (Id = drain current)

And of course, the way the two devices function is completely different. BJTs act like current-controlled current source (push/pull a small current in the base of the BJT, and you get a large current out/in the emitter - all related by the current gain, hFE). With FETs, they are more similar to an electronic switch... or a voltage controlled resistor if you run them in their triode region.

Well, for MOSFETs, I would take into account ON resistance, gate resistance, load resistance, power dissipation, current rating, etc....

And I figured it was a bad rectifier or transistor (in this case transistor). Good work goodpsusearch.

[goodpsusearch]

Re: Delta Electronics GPS-350CN-100 need repair

This took a lot of time to repair/rebuild. I hope it's worth the time and effort!

Added an NTC thermistor.

Replaced the Capxon 680uF(real c: 620uF & 630uF) primary caps with Panasonic 800uF 200V 105C.

I Removed the glue around the thermistor and attached it at secondary heatsink with thermal compound for better thermal contact and used a plastic case from a dead Dell psu made by Huntkey.

I chose a Top Motor ball bearing fan from an exploded Antec Smart Power psu to put in place of the original that was missing when the psu was found. The ball bearing fan has a mb connector for RPM sensing.

I modded the fan controller because the psu was very loud without getting even warm. This was a quite challenging task.

I assumed the single 3300uF capacitor for 12V without any pi coil would be insufficient in ripple suppression at high loads, so I moved some 12V cables to make space for a second capacitor for 12V. I used 2x2200uF instead of the single 3300uF capacitor. I tried not to go too low esr and make the psu oscillate.

I recapped the psu using GOOD caps.

3.3V:
1xChemicon KY 2200uF 10V
1xChemicon LXZ 2200uF 16V
5V:
1xChemicon KY 2200uF 10V
1xChemicon LXZ 2200uF 16V
12V:
2xChemicon LXZ 2200uF 16V
5vsb:
1xChemicon LXZ 2200uF 16V
1x Rubycon YXG 1000uF 6.3V
-12V:
1xLtec LTG 470uF 16V (left alone)

Exception: Fan Capacitor
1x Capxon KM 47uF 25V--> Koshin 47uF (anything is better than Capxon! )

So, what do you think?

[c_hegge]

Re: Delta Electronics GPS-350CN-100 need repair

Nice, although I don't think Koshin are any better than CapXon. I don't think I've ever seen a non-failed one that was more than a couple of years old, although for the fan, it probably doesn't matter that much if it fails, as fans don't usually care much even if they get subjected to a couple of volts of ripple.

For a Delta, though, it's definitely worth using only good Jap caps to repair it. It should just about last forever now.

[Pentium4]

Re: Delta Electronics GPS-350CN-100 need repair

Nice fix! I have a few deltas that do the same thing. I will see if they have the same issue. Thanks for sharing!

Even when Delta makes a basic half bridge unit, they still do a really good job. I don't like that the 5VSB switcher is on its own heatsink though.

[Wester547]

Re: Delta Electronics GPS-350CN-100 need repair

Nice fix! I have a few deltas that do the same thing. I will see if they have the same issue. Thanks for sharing!

Even when Delta makes a basic half bridge unit, they still do a really good job. I don't like that the 5VSB switcher is on its own heatsink though.

That's not as bad as some other Newton or older Delta units where the +5VSB switcher has no heatsink at all. On the other hand, even without a heatsink, those FETs didn't get very hot, so Delta seem to know what they're doing for the most part.

[goodpsusearch]

Re: Delta Electronics GPS-350CN-100 need repair

Nice fix! I have a few deltas that do the same thing. I will see if they have the same issue. Thanks for sharing!

Even when Delta makes a basic half bridge unit, they still do a really good job. I don't like that the 5VSB switcher is on its own heatsink though.

It doesn't overheat No discoloration on PCB too!

[CapSwapper]

Re: Delta Electronics GPS-350CN-100 need repair

I see people enjoy repairing these old psu's but if I had it it would in a landfill

[momaka]

Re: Delta Electronics GPS-350CN-100 need repair

^ But then you learn nothing .

Also, you'd be surprised how many useful components there are in even the crappiest of the crappiest PSUs. At the very least, you will get two 13007 or similar NPN BJTs, which can be pretty useful for many projects.

And you also get many other smaller transistors and resistors. Now that Radio Shack is gone, having a small stash of "junk" parts is even more useful if you need something ASAP.

[CapSwapper]

Re: Delta Electronics GPS-350CN-100 need repair

I yank out the heat sinks before tossing them. but thats about it.