Dell GX280 motherboard capacitor replacement (Rubycon MCZ alternative)

[WhichEndToHoldTheIron?]

I've got an ancient Dell OptiPlex GX280 that runs a CNC machine via its parallel port, but unfortunately it's recently died (no response when you press the power button on the front panel).

A bit of searching indicates this model was notorious for capacitor problems on the motherboard, and the machine's symptoms appear to match those described for this problem.

Some of the caps look to be the Panasonic FL range, for which I can get replacements. However, the caps (for what I understand are the VRMs) use Rubycon MCZ. It looks as though this was an "Ultra Low ESR for PC motherboard" range, and is now discontinued in favour of polymer caps.

I found a datasheet for the Rubycon MCZ which gives the following specs for the parts I need:

6.3V 820uF: 1340mA ripple current, 21mOhm ESR
16V 1000uF: 2350mA ripple current, 12mOhm ESR

Panasonic FM equivalents have similar ripple current, but much higher ESR; and I understand that very low ESR is required for this VRM circuit?

Looking at aluminium polymer capacitors on Mouser I can see models such as the Panasonic 16SEPF1000M (16V 1000uF); which has a much higher ripple current rating of 5400mA, and an ESR of 12mOhm (matching the Rubycon MCZ).

I understand that higher ripple current is "good" (is it good in this scenario?) but that polymer caps have much higher leakage current than electrolytics. Is that a problem for this application? My understanding of ESR, ripple current and leakage current doesn't really extend much futher than their names; so I don't know how to choose an appropriate replacement capacitor.

[amyk]

Those are reservoir caps for the CPU VRM. A little bit more leakage current isn't going to matter; the CPU itself draws dozens of amps.

[wraper]

Do not use polymer capacitors to begin with, it may cause problems due to too low ESR. And if you still decide to do it, use 2 times lower capacitance.

[WhichEndToHoldTheIron?]

> Do not use polymer capacitors to begin with, it may cause problems due to too low ESR

If the rated ESR of the polymer cap is close to the ESR of the Rubycon original, is that OK, or is the use of a polymer cap fundamentally the wrong choice?

>  And if you still decide to do it, use 2 times lower capacitance.

I've seen this advice mentioned elsewhere (in discussions of replacing an electrolytic with a polymer); can you explain (to someone with relatively little electronics knowledge - me) why a lower capacitance part should be used?

[wraper]

I've seen this advice mentioned elsewhere (in discussions of replacing an electrolytic with a polymer); can you explain (to someone with relatively little electronics knowledge - me) why a lower capacitance part should be used?

Because capacitance as such is not important in this case. Total capacitance is selected in a way that capacitors can withstand ripple current, not because there actually is a need to store this much energy. On second thought you can use polymer instead of 16V 1000uF MCZ as it has very low ESR for wet electrolyte, similar to mediocre polymer. On other hand it will be hard to find a direct substitute for 6.3V 820uF. But I personally would not overpay for polymer for no actual reliability benefit and potential problems to have.

[mariush]

I would go for Panasonic FR series , for example go with 1000uF 10v  in 8mm diameter :

1000uF 10v 8x15mm 0.041 ohm ESR, 1240 mA current ripple : https://www.tme.eu/gb/details/eeufr1a102l/tht-electrolytic-capacitors/panasonic/

If you can go 20mm tall, I'd even go with 1500uF 6.3v 8 x 20mm 0.03ohm ESR, 1560mA current ripple : https://www.tme.eu/gb/details/eeufr0j152lb/tht-electrolytic-capacitors/panasonic/

If you can fit 10mm diameter capacitors, those will have better ESR.


If you go polymer, these may be close enough :


560uF 4v 0.016 ohm esr  :  https://www.tme.eu/ro/en/details/a758ek567m0gaae016/tht-polymer-capacitors/kemet/

Yeah, it's rated for 4v, but it's not like you get more than 2v on the CPU, so 4v is more than enough.


If you don't mind waiting for packages to come from LCSC, Aishi PZ series is fairly high esr for a polymer capacitor :

AISHI(Aihua Group) SPZ0JM471E07O00R  470uF 6.3v  6.3mm  x 7 mm  0.020 Ohm 3900mA current ripple : https://www.lcsc.com/product-detail/AISHI-Aihua-Group-SPZ0JM471E07O00R_C112524.html

[wraper]

Looking more into it, yeah, there is really no direct substitute seem to be available for MCZ, Different manufacturer series with similar specs seem to be discontinued as well. Although you could could use Rubycon ZLG, 1200uF instead of 1000uF 16V and 1000uF instead of 820uF 6.3V. So either do something like this or use polymer and hope there will be no issues (most likely will be fine).

[WhichEndToHoldTheIron?]

Because capacitance as such is not important in this case. Total capacitance is selected in a way that capacitors can withstand ripple current, not because there actually is a need to store this much energy.

Interesting. I have a (basic) understanding of why you'd choose a capacitor with a particular capacitance, but this scenario (choosing a cap for the required ripple current) is new to me. Is there a particular name for this technique, or something I could read to understand it? I know that's not critical for fixing this motherboard, but it would be good to learn.

On second thought you can use polymer instead of 16V 1000uF MCZ as it has very low ESR for wet electrolyte, similar to mediocre polymer. On other hand it will be hard to find a direct substitute for 6.3V 820uF. But I personally would not overpay for polymer for no actual reliability benefit and potential problems to have.

My assumption is that the MCZ must have been chosen specifically for their ultra low ESR (for a wet electrolyte); hence why a polymer would be a suitable replacement?

I had spotted the Rubycon ZLG (mentioned in your later post). The 820uF has an ESR of 28mOhm (vs the 21mOhm of the MCZ). Quite a lot higher in percentage terms, but I don't know if that's critical.

I would go for Panasonic FR series , for example go with 1000uF 10v  in 8mm diameter :

1000uF 10v 8x15mm 0.041 ohm ESR, 1240 mA current ripple : https://www.tme.eu/gb/details/eeufr1a102l/tht-electrolytic-capacitors/panasonic/

If you can go 20mm tall, I'd even go with 1500uF 6.3v 8 x 20mm 0.03ohm ESR, 1560mA current ripple : https://www.tme.eu/gb/details/eeufr0j152lb/tht-electrolytic-capacitors/panasonic/

If you can fit 10mm diameter capacitors, those will have better ESR.

If you go polymer, these may be close enough :

560uF 4v 0.016 ohm esr  :  https://www.tme.eu/ro/en/details/a758ek567m0gaae016/tht-polymer-capacitors/kemet/

Yeah, it's rated for 4v, but it's not like you get more than 2v on the CPU, so 4v is more than enough.

If you don't mind waiting for packages to come from LCSC, Aishi PZ series is fairly high esr for a polymer capacitor :

AISHI(Aihua Group) SPZ0JM471E07O00R  470uF 6.3v  6.3mm  x 7 mm  0.020 Ohm 3900mA current ripple : https://www.lcsc.com/product-detail/AISHI-Aihua-Group-SPZ0JM471E07O00R_C112524.html

Those Panasonic FR caps have much higher ESR than the MCZ; so I assumed that would be a problem; but am I mistaken?

[mariush]

Those Panasonic FR caps have much higher ESR than the MCZ; so I assumed that would be a problem; but am I mistaken?

The way I see it is that those ultra low esr capacitors have water based electrolytes which degrade over time ... by the time those capacitors were leaking and cracking at the top, they were already degraded and their ESR was worse than the value listed in datasheet for quite some time yet the system still worked.  The system probably ran for months or years with degraded capacitors, so there's still a reasonable tolerance, slightly bigger ESR won't be a big deal.

FR series if i remember correctly is made with some semi-aqueous electrolyte that can handle high temperatures better than the electrolyte in ultra low esr so they should maintain their performance for long time.

[WhichEndToHoldTheIron?]

I've ordered replacements for the Panasonic FL caps, so I'll make a start on the board when they arrive.

I've got a PEAK Atlas ESR70, but it reports "in circuit/leaky" for most of the caps on the board, so I'll then start removing the MCZ caps and test them to see what state they're in (and then make a decision on what to buy to replace them with).

[WhichEndToHoldTheIron?]

if it is the VRM ?
i would think it needs low ESR for the Cout ?
maybe other PSU experts like to chime in

That's the gist I get, both from replies here and things I've seen on other forums. The capacitance isn't critical but the high switching speed requires extremely low ESR. Why exactly that is I don't know.

[wraper]

if it is the VRM ?
i would think it needs low ESR for the Cout ?
maybe other PSU experts like to chime in

That's the gist I get, both from replies here and things I've seen on other forums. The capacitance isn't critical but the high switching speed requires extremely low ESR. Why exactly that is I don't know.

Because resistance results in voltage drop and heat. For example adding 5 milliohms in series with 100A current means 0.5V voltage drop and 50W power dissipation. It's not about switching speed as such, it's about very high current at very low voltage. If you reduce converter switching speed, ESR still would need to be extremely low, however the more you reduce it, the more bulk capacitance will play a role.

[StrayElectrons]

When I recapped my Dell 8300 Dimension back in the day, I used ultra low ESR capacitors and computer lived on daily use at work for about 5 years and was pulled still working.  And in same era of P4 boards, there were mixture of polymer and electrolytic capacitors as well.    Many people did all polymer mod for the VRM section and rest of board, with no issues as well due to lack of good ultra low ESR capacitors recently.

Cheers,

[WhichEndToHoldTheIron?]

Because resistance results in voltage drop and heat. For example adding 5 milliohms in series with 100A current means 0.5V voltage drop and 50W power dissipation. It's not about switching speed as such, it's about very high current at very low voltage. If you reduce converter switching speed, ESR still would need to be extremely low, however the more you reduce it, the more bulk capacitance will play a role.

I've been reading up on ESR and ripple current and I think I understand that now (the combination of ESR with ripple current resulting in power dissipation).

The one thing I've not understood yet is the scenarios where too little ESR can be a problem (apparently stability in some types of circuit).

[WhichEndToHoldTheIron?]

When I recapped my Dell 8300 Dimension back in the day, I used ultra low ESR capacitors and computer lived on daily use at work for about 5 years and was pulled still working.  And in same era of P4 boards, there were mixture of polymer and electrolytic capacitors as well.    Many people did all polymer mod for the VRM section and rest of board, with no issues as well due to lack of good ultra low ESR capacitors recently.

Cheers,

Thanks - that agrees with comments I've seen on other forums about the subject.

In combination with other comments in this thread I feel reasonably confident that polymer caps (with appropriate ESR and ripple current values) should be safe for the VRM sections on this board.

[WhichEndToHoldTheIron?]

I've made some progress on the motherboard, but with somewhat unexpected results.

All of the Panasonic FL caps (6.3v 1800uF and 6.3v 1500uF) I've removed from the board actually measure OK with the PEAK ESR70. I can't find a datasheet for this series, but all the caps were within 10% of the rated capacitance and between 0.02 and 0.03 Ohm ESR. I have replaced them with new ones though.

I removed the ultra-low ESR Rubycon MCZ caps (the ones that I was planning on replacing with poly caps), and they don't measure too bad either. All capacitances were 13% of rated (or better), with the ESR values being a bit higher than rated:

6.3V 820uF: 21mOhm ESR rated, 30mOhm ESR measured
16V 1000uF: 12mOhm ESR rated, 20 to 30mOhm ESR measured

Would the above values for the MCZ caps be enough to prevent the board from running? Feels to me like it may not be the caps that are the cause of the failure as they don't seem that bad 

[wraper]

Would the above values for the MCZ caps be enough to prevent the board from running?

Very unlikely.

[WhichEndToHoldTheIron?]

Thanks. That was my (admittedly uneducated) thinking. Probably something else that's gone then.

Apparently the PSU is common fault area with this model; but that died years ago and I replaced it with another brand. Perhaps that one's gone too - but it'd be the first PC PSU I've ever killed 

[nctnico]

Be prepared to replace the capacitors again after a couple of years. This model pushes the capacitors very hard. IIRC I have one of these around as well and I recall having to replace the capacitors for a second time.