Very Worn Out HP DC5100 SFF Computer

[Pentium4]

A lady I was doing work for had this desktop that she had been keeping to get fixed up. It was used at a business for a long time and she acquired it when leaving the company. It just turned on and would make a series of beeps, no video. I told her how much work it would take to get it running and she passed, so it's mine now.

It was a quick diagnostic to figure out why. I'll start off by saying that this thing has a total of 51,668 running hours on it. The original HDD tested good too!

As you can judge by my username, I plan on fixing it up

I assume it will come back to life when I replace the capacitors. I was thinking about doing a Polymod to it. Since this thing obviously ran hot enough to bulge some Rubycon MCZ, then every capacitor in it should be replaced?

I think what didn't help was that there are also at least two bad capacitors in the power supply. OST RLP on the 5V rail. And if there are two bulged OST, I suspect some of the others may have also failed without any visible symptoms.

There are not many quality capacitors on the board at all. Of course there are some nice UCC polymers on the VRM out but besides that the only other quality caps are two Panasonic FL caps. The rest are Rubycon MCZ, KZG, G-Luxon LT, and besides the two Luxon, all the small caps are Evercon. Overall, terrible choices for such a chassis with limited airflow. I mean...just check out that exploded Evercon!

The other thing that upsets me that HP did, is they put a Prescott P4 in here. A 541 (3.20GHz) to be exact. Not just a Prescott, but a 1M Prescott...The computer itself was manufactured in September, 2006. Cedar Mill was well established by then, they really should have used one of those. Maybe they were getting rid of their old stock? I also don't think that heatsink was ever adequate for such a CPU

What do you guys think?

[delaware74b]

Re: Very Worn Out HP DC5100 SFF Computer

I recapped one of those DC5100's. I still have it in my basement. These things run HOT. I have the CPU fan defaulted to 75% in the CMOS setup.

Recap that power supply and get rid of every bit of the glue in it. My power supply had shorted out from that cr#p. No problem, new caps and fuse, and scraped all the glue out.

[Topcat]

Re: Very Worn Out HP DC5100 SFF Computer

that 1000uF 6.3v ruby looked a little domed. Yes, fix it....its fun repurposing stuff that others deem junk.

[Pentium4]

Re: Very Worn Out HP DC5100 SFF Computer

I recapped one of those DC5100's. I still have it in my basement. These things run HOT. I have the CPU fan defaulted to 75% in the CMOS setup.

Recap that power supply and get rid of every bit of the glue in it. My power supply had shorted out from that cr#p. No problem, new caps and fuse, and scraped all the glue out.

Oh yeah, that glue has to go! Did you have caps fail on your board as well? How many of them did you replace?

that 1000uF 6.3v ruby looked a little domed. Yes, fix it....its fun repurposing stuff that others deem junk.

Yeah it definitely is. I'm going to check my on hand stock. Might need to order some caps from you while you still have them.

Is it okay to replace the caps for filtering USB power, LAN ports, etc. with regular low ESR caps such as nichicon PW?

[Pentium4]

Re: Very Worn Out HP DC5100 SFF Computer

Okay so this post is pretty much just for Wester547 and his internal database

51,668 running hours since late 2006, office environment. I pulled most of the filtering caps from the PSU to test them. I was wrong with some of the info above so I wanted to post correct info.

12V rail:
Cap before coil: Taicon HG 2200uF, 16V. Read: 2362uF, 0.03Ω
Cap after coil: Taicon HG 1500uF, 16V. Read: 1478uF, 0.03Ω

5V rail:
Cap before coil: OST RLP 2200uF, 10V, 10x25mm. Read: In Circuit/Leaky, 3.2Ω
Cap after coil: OST RLP 2200uF, 10V, 10x25mm. Read: In Circuit/Leaky, 9.0Ω (Didn't bulge)

3.3V rail:
Cap before coil: OST RLS 2200uF, 6.3V, 10x20mm. Read: 1.16uF, 26.3Ω
Cap after coil: Taicon HG 3300uF, 6.3V. Read: 3958uF, 0.00Ω

5VSB rail:
Cap before coil: OST RLP 3300uF, 6.3V, 10x30mm. Read: 3265uF, 0.04Ω
Cap after coil: Ltec LZP 470uF, 10V. Read: 487uF, 0.10Ω

Seems the Taicon HG 3300 6.3 had a similar failure mode as a defective nichicon HN and was tricking my meter. It is not bulged.

The main filter capacitor is an OST SPS 150uF 450V. It read 139uF, 0.30Ω. I might replace it anyways. I think I have a similarly sized Ltec that is 180uF.

[Wester547]

Re: Very Worn Out HP DC5100 SFF Computer

My "internal database" is already maxed out by date codes.

Hey, look at the bright side... they didn't use any Teapos in that HP DC5100 SFF. We all know how quickly Teapos would have popped compared to the rest. That HP DC5100 runs hot enough to pop an MBZ. Even the midtower HP DC5100 has poor ventilation. Taicon HGs and HHs are easy to identify - they have bright green sleeves. That Taicon is technically in-spec for capacitance even though said capacitance is rather elevated. The bad Nichicon HMs and HNs were known to fail more catastrophically than that, often reading 2x to 5x the nominal capacitance, which means either the corrosion inhibiting additives in the electrolyte were not good (resulting in the thinning of the anode dielectric until hydrogen gas forms) or the electrolyte was extremely volatile.

None of those "bad" series or brands did particularly badly considering the thing almost clocked 52,000 hours and runs so hot. I like how they used an ADDA hyprobearing fan, which by now is probably drawing close to seizing because it's running dry. At least the system itself has a double ball bearing PWM-controlled Delta fan.

[delaware74b]

Re: Very Worn Out HP DC5100 SFF Computer

I recapped it completely, both mobo and power supply. Full compliment of Nichicons for all caps on the boards. I had done that about 3-4 years ago. It had seen 3 years of 24/7 use on WinXP, used as audio uploader for scanner radio apps. My feed is Kent County Fire and EMS. It is now being uploaded on my Win7 file server.

[Pentium4]

Re: Very Worn Out HP DC5100 SFF Computer

My "internal database" is already maxed out by date codes.

Hey, look at the bright side... they didn't use any Teapos in that HP DC5100 SFF. We all know how quickly Teapos would have popped compared to the rest.

True! I trust those OST RLP/RLS a lot more than Teapo SC.

That HP DC5100 runs hot enough to pop an MBZ.

There are no MBZ in this one. Just MCZ and two small 470 6.3 YXG.

Taicon HGs and HHs are easy to identify - they have bright green sleeves. That Taicon is technically in-spec for capacitance even though said capacitance is rather elevated. The bad Nichicon HMs and HNs were known to fail more catastrophically than that, often reading 2x to 5x the nominal capacitance

Yeah. However, when I first pulled it and was still warm from my iron, it read 4158uF. The other reading was after it cooled down. Maybe the higher one is more accurate? Also, most of those HMs and HNs were older than these HGs. I'm guessing after a few more years the capacitance would have continued to rise.

None of those "bad" series or brands did particularly badly considering the thing almost clocked 52,000 hours and runs so hot.

Yeah, you're right. That is a lot of running hours for a small chassis with a Prescott. Also, the fans in this system spin very slowly. I booted it up just to make sure it worked and I could barely hear the fans spinning even with the top of the case off. That's saying something considering there are two Delta fans

I like how they used an ADDA hyprobearing fan, which by now is probably drawing close to seizing because it's running dry. At least the system itself has a double ball bearing PWM-controlled Delta fan.

Do you think I should just oil it, or replace it? It's a 4 pin fan which seems bizarre for a PSU.

Edit: They used 470mf 16V KZG for all the caps for USB and LAN filtering. Those can't possibly be using 12V? Did they just use those caps with higher voltage for the lower ESR? I have seen Lelon RGA caps in this position before, so I'm thinking nichicon PW would be a good substitute. I was also going to replace the Panasonic FL. They are good caps but with 52k hours in a hot case and being fed by bad caps in the PSU for who knows how long, I think I'm going to retire them

[Pentium4]

Re: Very Worn Out HP DC5100 SFF Computer

I recapped it completely, both mobo and power supply. Full compliment of Nichicons for all caps on the boards. I had done that about 3-4 years ago. It had seen 3 years of 24/7 use on WinXP, used as audio uploader for scanner radio apps. My feed is Kent County Fire and EMS. It is now being uploaded on my Win7 file server.

Very cool! Did you test the caps that you pulled? Were any of the little Evercon caps failed?

[Wester547]

Re: Very Worn Out HP DC5100 SFF Computer

True! I trust those OST RLP/RLS a lot more than Teapo SC.

Yes, despite that OST isn't good, Teapo are really terrible capacitors. Even though they were always crap, they've really gone downhill in the last 10-11 years.

There are no MBZ in this one. Just MCZ and two small 470 6.3 YXG.

Well, they could have used MBZ in place of KZG on other boards. Even though MBZ is a lot more reliable than KZG, MBZs (and especially MCZs) are known to dry up and die in hot-running SFF machines, especially those that run 24/7 for many years.

Yeah. However, when I first pulled it and was still warm from my iron, it read 4158uF. The other reading was after it cooled down. Maybe the higher one is more accurate? Also, most of those HMs and HNs were older than these HGs. I'm guessing after a few more years the capacitance would have continued to rise.

Capacitance is measured at 20*C to 25*C in the datasheets and capacitance increases as the temperature rises. It probably would have continued to increase if they were left in storage.

Yeah, you're right. That is a lot of running hours for a small chassis with a Prescott. Also, the fans in this system spin very slowly. I booted it up just to make sure it worked and I could barely hear the fans spinning even with the top of the case off. That's saying something considering there are two Delta fans

If you don't hear them clicking, they are probably spinning quite a bit lower than 1000RPM.

Do you think I should just oil it, or replace it? It's a 4 pin fan which seems bizarre for a PSU.

ADDA make good and silent fans, so I think it merits oiling. Whether the PSU actually uses the PWM wire is a question to be answered either by the header itself or the underside (or a multimeter).

Edit: They used 470mf 16V KZG for all the caps for USB and LAN filtering. Those can't possibly be using 12V? Did they just use those caps with higher voltage for the lower ESR? I have seen Lelon RGA caps in this position before, so I'm thinking nichicon PW would be a good substitute. I was also going to replace the Panasonic FL. They are good caps but with 52k hours in a hot case and being fed by bad caps in the PSU for who knows how long, I think I'm going to retire them

That Panasonic FL is probably internally dried up. Any electrolytic capacitor with a liquid solvent and solution will eventually evaporate internally, either by way of rubber seal diffusion (epoxy-sealed capacitors dry up considerably slower) or through an electrochemical reaction on the boundary of the aluminum foil. Those KZGs are more likely filtering +5V or +5VSB (possibly +3.3VSB?) outputs on the motherboard, except for one by the PCI-Express slot which is likely directly filtering that +12V output. KZGs were cheaper than the other "top-tier" brand ultra low ESR capacitors of the timeframe and could have been used because their availability was better. It might be safe to use Nichicon PW if those KZGs aren't on the output or input of a buck regulator.

[delaware74b]

Re: Very Worn Out HP DC5100 SFF Computer

Pentium4 - It turns out I did the full recap on a DC510, not the 5100.......(I have slept once or twice since then!) The 510 was my scanner radio uploader. I have recapped a DC5100, replacing everything except the ones around the VRM.

[Pentium4]

Re: Very Worn Out HP DC5100 SFF Computer

Yes, despite that OST isn't good, Teapo are really terrible capacitors. Even though they were always crap, they've really gone downhill in the last 10-11 years.

Do you think that pre-acquisition G-Luxon was better than Teapo is now?

If you don't hear them clicking, they are probably spinning quite a bit lower than 1000RPM.

I could hear a quiet ticking noise. I removed the rubber seal from the Adda fan and it was very dry. I put two drops of oil in there. I like that Adda used a very tight rubber seal. I had to use a pocket knife to get it out, where as I normally use my longest fingernail.

ADDA make good and silent fans, so I think it merits oiling. Whether the PSU actually uses the PWM wire is a question to be answered either by the header itself or the underside (or a multimeter).

It looks like a high quality fan. With that fresh oil and how slow it spins while running, I'm guessing it will last a long time

Pentium4 - It turns out I did the full recap on a DC510, not the 5100.......(I have slept once or twice since then!) The 510 was my scanner radio uploader. I have recapped a DC5100, replacing everything except the ones around the VRM.

Haha, thanks for the info

Okay so I was SO close to being done removing the glue...and then literally on the last component that was smothered in glue...this zener diode cracked in half and I was so careful too. Add solder to one leg, wait a minute....remove solder, wait a minute....repeat on the other leg. Then the stupid thing just breaks in half. It's covered in glue so I can't read anything on it. Is there anything I can do? I have lots of spares, but I've never had to replace a zener before so I don't really know what to do here. I don't want to guess and have something go wrong.



I'll appreciate any advice you can give me!

[Wester547]

Re: Very Worn Out HP DC5100 SFF Computer

Do you think that pre-acquisition G-Luxon was better than Teapo is now?

I think post-acquisition G-Luxon is better and pre-acquisition G-Luxon is equally bad. Not deeming G-Luxon good or even average, though... same goes for OST. Another poor decision that came to my notice with the motherboard itself was the use of freestanding FETs in the VRM - as though the board doesn't already run hot enough. It screams of "planned obsolescence" IMO.

I could hear a quiet ticking noise. I removed the rubber seal from the Adda fan and it was very dry. I put two drops of oil in there. I like that Adda used a very tight rubber seal. I had to use a pocket knife to get it out, where as I normally use my longest fingernail.

Delta, ADDA, Foxconn, Yate Loon, and AVC use very tight rubber caps for their sleeve bearing fans. Sunon not so much, but in more recent years they've sealed their fans with a metal cap that can't be removed.

Okay so I was SO close to being done removing the glue...and then literally on the last component that was smothered in glue...this zener diode cracked in half and I was so careful too. Add solder to one leg, wait a minute....remove solder, wait a minute....repeat on the other leg. Then the stupid thing just breaks in half. It's covered in glue so I can't read anything on it. Is there anything I can do? I have lots of spares, but I've never had to replace a zener before so I don't really know what to do here. I don't want to guess and have something go wrong.

I'll appreciate any advice you can give me!

You can't just replace a zener diode with any diode. You'd have to match the voltage of the original zener as well. Zeners can conduct in the opposite direction, unlike regular diodes. My guess is the zener ultimately fell apart from ~52k hours of thermal duress, so just the heat from your soldering iron was enough to finally crack it (it's easy to crack zeners anyway). You'd need a voltmeter or something to test the zener with to get an idea of its voltage rating, but you can't test it if the zener is broken in half. Do you know what trace (what output) the zener diode was connected to at the anode and cathode legs?

[goodpsusearch]

Re: Very Worn Out HP DC5100 SFF Computer

Try to take out the glue and with a very good led flashlight maybe you could read part of the marking

[Pentium4]

Re: Very Worn Out HP DC5100 SFF Computer

Thanks for the replies. I'm going to try and see if I can use hot water to try and get the glue off the zener. Even with young eyes the thing is pretty hard to see lol.

So far I've done some tracing of the zener. The cathode is connected directly to the pin 1 anode of the NEC 2561 opto which goes through the pin 2 cathode and traces all the way to the 5VSB pin on the secondary monitoring IC (SITI PS224)
And it looks like the anode of the zener traces to the 3.3V output pin on the PS224. I'm going to upload a picture in a little bit.

Another poor decision that came to my notice with the motherboard itself was the use of freestanding FETs in the VRM - as though the board doesn't already run hot enough. It screams of "planned obsolescence" IMO.

Yeah I usually don't like free standing FETs either. But there is a shroud that goes from the intake fan to the heatsink, so at least half of them would be getting direct airflow. Plus, a 4 phase VRM with 16 FETs seems pretty adequate even for a Prescott P4

[Wester547]

Re: Very Worn Out HP DC5100 SFF Computer

Yeah I usually don't like free standing FETs either. But there is a shroud that goes from the intake fan to the heatsink, so at least half of them would be getting direct airflow. Plus, a 4 phase VRM with 16 FETs seems pretty adequate even for a Prescott P4

That shroud should help concentrate airflow over the FETs and the heatsink and depending on how thick the copper planes are, those 16 FETs might be sufficient even upstanding. But SFF machines are generally very limited in terms of airflow and ventilation. And that's a Lite-on PSU so I think it may be worth the effort.

[Pentium4]

Re: Very Worn Out HP DC5100 SFF Computer

The zener in question fits in those holes above 511 resistor. The cathode lead goes on the left. Is it possible to just get an "overspecced" zener for this? As in, equivalence to installing a 10V capacitor on the 3.3V rail?

[ratdude747]

Re: Very Worn Out HP DC5100 SFF Computer

The zener in question fits in those holes above 511 resistor. The cathode lead goes on the left. Is it possible to just get an "overspecced" zener for this? As in, equivalence to installing a 10V capacitor on the 3.3V rail?

Nope.

First, it may not even be a zener diode. Some regular diodes come in glass cases. The 1N4148 is a common example.

Second, assuming it is a zener, they have two common uses. One is used as a means of overvotage protection on input; if the voltage across the diode exceeds the nominal voltage of the zener diode, the diode conducts and clamps the voltage down. This property is used for the second major use, as part of a voltage regulation circuit, which will usually use 1-3 transistors; depending on the circuit, you can get short-circuit protection and/or variable output. In either case you don't want to change the value of the zener diode, as that will screw something up.

[ChaosLegionnaire]

Re: Very Worn Out HP DC5100 SFF Computer

^what he said^ diodes much less zener diodes function and work in a much different way than caps. what works for caps wont hold true for a diode. a diode does not store charge like a cap does!

dont be blowing up any good ol' stuff now! through carelessness or whatnot. i get upset when good ol' stuff gets trashed due to "bad accidents".

[Pentium4]

Re: Very Worn Out HP DC5100 SFF Computer

Nope.

First, it may not even be a zener diode. Some regular diodes come in glass cases. The 1N4148 is a common example.

Second, assuming it is a zener, they have two common uses. One is used as a means of overvotage protection on input; if the voltage across the diode exceeds the nominal voltage of the zener diode, the diode conducts and clamps the voltage down. This property is used for the second major use, as part of a voltage regulation circuit, which will usually use 1-3 transistors; depending on the circuit, you can get short-circuit protection and/or variable output. In either case you don't want to change the value of the zener diode, as that will screw something up.

Thanks for the info. Do you think my next best course of action is to try and remove the glue on the diode to see if I can read the marking?

^what he said^ diodes much less zener diodes function and work in a much different way than caps. what works for caps wont hold true for a diode. a diode does not store charge like a cap does!

dont be blowing up any good ol' stuff now! through carelessness or whatnot. i get upset when good ol' stuff gets trashed due to "bad accidents".

I wasn't trying to compare the functionality of a zener to a capacitor at all. I just don't know much about zener diodes and was making a reference to "overspeccing" as in reference to installing a 10V capable capacitor in a 3.3V circuit. Just like for a regular diode in a circuit that needs 35 reverse voltage, you can install a diode that is capable of 60 reverse voltage. That's the only reason I brought up the capacitor reference. I too, don't want to break this due to a dumb mistake, which I've already started. I doubt that glue on the zener would have ever caused a problem in the first place

Pentium4 - It turns out I did the full recap on a DC510, not the 5100.......(I have slept once or twice since then!) The 510 was my scanner radio uploader. I have recapped a DC5100, replacing everything except the ones around the VRM.

In a worst case scenario, do you think you could check the value in yours if it isn't a big hassle for you?