Macintosh polymer recap & ESR considerations

[JDW.]

Originally Posted by stj

i meant the YXA,B,F series

you obviously speak japanese - why not email Rubycon and ask them if they have old datasheets or compatability tables?

Here's the YXA datasheet and this page says it was last sold in 2012.

[stj]

damn, well i think YXF is still sold.

[JDW.]

Originally Posted by stj

damn, well i think YXF is still sold.

Well, it's certainly not sold on Mouser. And since I usually share my recapping knowledge with others in the form of a replacement capacitor list, a readily available replacement capacitor that works well is critical.

Using my DE-5000 meter, I made the following measurements today spanning several minutes for each test to ensure the most accurate measurement, all at 120Hz because higher frequencies produced no results on my meter...

Stock Rubycon 2200uf 10V capacitor (SH-series, 105°C) (inside Sony PSU in Apple HD20SC):
• θ = -75°
• Q = 3.7
• R = 0.1Ω (only the tenths place was shown, so it could be 100, 150, etc.)
• D = 0.270

Nichicon 2200uf 16V capacitor (UHE1C222MHD1TO, 105°C):
• θ = -88.6°
• Q = 45
• R = 0.0Ω (only the tenths place was shown, so it could be 20mΩ, 60mΩ, etc.)
• D = 0.022

LCC 2200uf 25V capacitor (new but 16.5 years old, 85°C):
• θ = -87°
• Q = 20
• R = 0.0Ω (only the tenths place was shown, so it could be 20mΩ, 60mΩ, etc.)
• D = 0.049

Note that D is Dissipation Factor, aka tan δ, which is almost always shown in capacitor datasheets even when ESR is not.

After recapping, when I use the stock Rubycon cap, all is well and there is not beep/squeal shortly after I switch on power, and the peak-to-peak ripple on the scope shows as it did prior to recap -- about 50mVp-p. The ESL spikes are 100mVp-p max.

When I use the Nichicon cap, the ripple is about 70mWp-p and ESL spikes exceed 100mVp-p, and there is a beep/squeal sound shortly after I switch on power. I don't see a buzzer on this PSU, so I think the sound is coming from one of the two transformers, as per what my ear is telling me.

I then thought it best to try a new but very old cap I got when I bought my JAMECO electrolytic radial capacitor kit back in April 2003 (more than 16 years ago). That cap too is 2200uF but 25V and rating at only 85°C. It's marked LCC and I cannot find any data sheet for it. I doubt JAMECO had these kits in stock for more than 1 year, and I doubt the cap was made more than a year before the kits were assembled, so I would guess the age of the cap could be as old as 18 years. But again, it was new, never used, and always stored in a location where people are (i.e., no long term temperature extremes or excessive humidity). With this old LCC cap, there is no beep sound, and the scope measurements are exactly what I see with the stock Rubycon cap. This is interesting because if you look at my measurements above, the LCC still has much better test results than the Rubycon. The LCC's Dissipation Factor is more than twice the Nichicon's though, so perhaps that is the key here? Even so, it's still much lower than the 0.270 of the stock Rubycon.

In any case, the problem is easily solved for me by using either the stock cap or the LCC, but if I share a capacitor replacement list with others, the issue of that one cap is a sticking point. Sure, I could just advice using the stock cap, but what about cases where that cap is bad? And so I am trying to find a suitable replacement 2200uF 10V (or higher) cap that will physically fit (13mm x 26mm) and produce a stable output voltage. I think the beep after the switch-on is telling me of some instability, and the higher ripple voltage of the Nichicon replacement seems to bear that out.

Any further thoughts on this would be appreciated.

Thank you.

[stj]

well 13mm is wrong, the official size is 12.5mm,
if your trying to size-match there is nothing wrong with increasing the voltage to 16 or even 25v

have you seen this chart?

[JDW.]

Originally Posted by stj

well 13mm is wrong, the official size is 12.5mm,
if your trying to size-match there is nothing wrong with increasing the voltage to 16 or even 25v
have you seen this chart?

Yes, I had seen that chart when I visited the Rubycon website, but it does not mention the SH series from the 1980's, which isn't too big a surprise since I cannot find a mention on their website of any cap made prior to 1999. I also believe the USH series mentioned on page 3 of that PDF is not the same as the SH-series of the 1980's because USH is rated for 85°C whereas my stock cap in this power supply is rated 105°C. The presence or absence of the "U" is yet another differentiator that shows the two series are different.

13mm is not wrong within the context of what I wrote, but I see now I need to clarify. A capacitor with a Diameter of 13mm will physically fit into the space provided for that capacitor on the SONY PSU inside an Apple HD20SC power supply. That implies that a 12.5mm, 12.0mm and smaller capacitor diameter will also fit.

I fully understand there is ordinarily nothing wrong with choosing a higher voltage cap as a replacement, but in this case there is, insofar as the higher voltage caps tend to have lower ESR and Dissipation Factor, and on some switching power supplies, an ESR that is too low can cause phase shifts, oscillations and other undesirable behavior that negatively impact the stability of the said power supply. This info is often written into the data sheets of switching power supply controllers, many of which will specify very clearly how much ESR is required. In those cases, at least for modern switchers, it's common to see an ultra low ESR ceramic cap used in series with an SMD resistance of several hundred milliohms.

So in summary, I'm looking at capacitors that no only physically fit, but have sufficient ESR and/or Dissipation Factor to function in a stable way in this particular power supply.

[JDW.]

I purchased 3 capacitors from Mouser, and measured data using my DE-5000:

Illinois Capacitor 228CKS010M 2000hr@85°C:

ESR
@100Hz = 60mΩ
@120Hz = 50mΩ
@1000Hz = 43mΩ
@10,00Hz = 39mΩ
@100,00Hz = 42mΩ

DF@120Hz = 0.90
Capacitance = 2.11mF
Q@120Hz = 11.04
θ@120Hz = -84.7°

Panasonic EEU-EB1A222 10,000hr@105°C:

ESR
@100Hz = 90mΩ
@120Hz = 90mΩ
@1000Hz = 80mΩ
@10,00Hz = 74mΩ
@100,00Hz = 65mΩ

DF@120Hz = 0.138
Capacitance = 1.95mF
Q@120Hz = 7.2
θ@120Hz = -82.0°

Cornell Dubilier SEK222M010ST 2000hr@105°C:

ESR
@100Hz = 40mΩ
@120Hz = 40mΩ
@1000Hz = 31mΩ
@10,00Hz = 29mΩ
@100,00Hz = 27mΩ

DF@120Hz = 0.070
Capacitance = 2.06mF
Q@120Hz = 14.13
θ@120Hz = -85.8°

I soldered each of the above into my power supply and tested the resulting ripple and ESL spikes on my scope. None of these 3 caps caused the audible 1kHz beep sound at power-on, which was nice. Surprisingly, the highest ESL spikes came from the Panasonic, and the ripple was higher than any of the other capacitors (due to its ESR). It's rated for 10,000hrs but the ESL spikes were much higher than the other 2, so I won't be using it. Ripple and ESL spikes were almost the same for the Illinois and Cornell caps. I intend to use the Cornell due to its 2000hr@105°C rating versus 2000hr@85°C for the Illinois, and for its lower ESR and higher Q.

So the moral of this story is that its difficult to know which cap is best without testing. I had thought the Panasonic would win, but due to the higher ESL spikes it produced (which still somewhat baffles me), it was put out of the running.

Thanks to all who contributed to this thread.

[stj]

the dubilier cap had the lowest esr - that's why it had the best performance.

[JDW.]

Originally Posted by stj

the dubilier cap had the lowest esr - that's why it had the best performance.

Consider well what I discussed before. The Nichicon capacitor I used originally had even lower ESR but the performance was worse, resulting in higher amplitude ESL spikes and a 1 kHz audible beep sound at power on. That needs to be taken into serious consideration. “Lowest ESR” does not guarantee the capacitor will be the best performer.

[stj]

yes, but there is a limit to how low the esr can go, usually related to the frequency of the circuit.

[JDW.]

Yes, that's what I have been saying. It's simply a lack of a full schematic that necessitated experimentation with various capacitors on my part to find the right one -- one that is still commonly found and I could recommend to others doing the same recap job.