When is it profitable to use OSCON over conventional aluminium capacitors?
An intensive search to answer this question resulted in two examples that encapsulate the issues. Finally conclusions can be drawn.
Example 1.
A reference design defined: www.onsemi.com/pub/Collateral/NCP5332A-D.PDF
The design recommends either eight OSCON capacitors or 10 Rubycon.
Evaluation of this design and calculation reveals the pros and cons of the OSCONS applied in this case.
The Pros.
OSCONS are smaller (often the same diameter), a lower component count.
The Cons.
In summary the choice of eight OSCON caps degrades performance in this example. Apparently not enough to require the adding of one more cap to improve matters.
Example 2.
An Epox 8RDA+ motherboard has six Teapo 3300uF 6.3V making a hefty 20,000uF bulk capacitance at the VRM output. The temperatures measured on the mother board at the VRM input caps is 45C, at the output caps 40C, the ambient temperature being 25C.
The simplest and lowest cost replacement of the Teapo caps is to use like value ultra low ESR caps from Rubicon, Nichicon and Samxon. This will ensure quality and improved performance related to a decrease in ESR and no loss of bulk capacitance.
Endurance .
Endurance is often cited as the reason to install OSCON. I have seen an unconfirmed quote that OSCON caps have an endurance of 200,000hrs at 65C (does this apply to all series?). However is this a realistic benefit?
Taking the temperatures in example 2 above it can be estimated that a quality aluminium cap rated for 2000 hrs at 105C will have an endurance of 128,000hrs at 45C (32 years for 10hrs every day). This is more than sufficient; OSCON is not needed in this case and many other motherboards.
Where is OSCON profitable?
Conclusions
The evidence is quite clear that OSCON caps are not necessary for replacement of bad aluminium caps. Quality aluminium caps have equivalent low ESR, maintain bulk capacitance, cost less, endurance is good and there is minimum change to an existing design. The VRM circuit can produce voltage overshoot (spikes) if radical changes are made, so it is best to stick close to the original design. OSCON caps are best applied where their unique properties can be applied. In the future designers may use more if small size, higher temperature and higher frequency characteristics are required.
An intensive search to answer this question resulted in two examples that encapsulate the issues. Finally conclusions can be drawn.
Example 1.
A reference design defined: www.onsemi.com/pub/Collateral/NCP5332A-D.PDF
The design recommends either eight OSCON capacitors or 10 Rubycon.
Evaluation of this design and calculation reveals the pros and cons of the OSCONS applied in this case.
The Pros.
OSCONS are smaller (often the same diameter), a lower component count.
The Cons.
- The OSCONS cost about 300% more per component.
- Using the formula supplied there is 17% less transient performance.
- The bulk capacitance is down 56% conversely the Rubicon caps are better by 229%.
- The loss of bulk capacitance will further degrade transient performance (this parameter is not covered by the formula, it only considers ESR). To clarify this point, consider ESR without any bulk capacitance. Transient voltage drop and ripple both increase as capacitance decreases. (Note that future VRM circuits operating at Mhz frequencies will not need bulk capacitance).
In summary the choice of eight OSCON caps degrades performance in this example. Apparently not enough to require the adding of one more cap to improve matters.
Example 2.
An Epox 8RDA+ motherboard has six Teapo 3300uF 6.3V making a hefty 20,000uF bulk capacitance at the VRM output. The temperatures measured on the mother board at the VRM input caps is 45C, at the output caps 40C, the ambient temperature being 25C.
The simplest and lowest cost replacement of the Teapo caps is to use like value ultra low ESR caps from Rubicon, Nichicon and Samxon. This will ensure quality and improved performance related to a decrease in ESR and no loss of bulk capacitance.
Endurance .
Endurance is often cited as the reason to install OSCON. I have seen an unconfirmed quote that OSCON caps have an endurance of 200,000hrs at 65C (does this apply to all series?). However is this a realistic benefit?
Taking the temperatures in example 2 above it can be estimated that a quality aluminium cap rated for 2000 hrs at 105C will have an endurance of 128,000hrs at 45C (32 years for 10hrs every day). This is more than sufficient; OSCON is not needed in this case and many other motherboards.
Where is OSCON profitable?
- Where small size is required.
- Where higher frequencies are applied
- At temperature extremes. Perhaps new hotter motherboards or cold conditions.
Conclusions
The evidence is quite clear that OSCON caps are not necessary for replacement of bad aluminium caps. Quality aluminium caps have equivalent low ESR, maintain bulk capacitance, cost less, endurance is good and there is minimum change to an existing design. The VRM circuit can produce voltage overshoot (spikes) if radical changes are made, so it is best to stick close to the original design. OSCON caps are best applied where their unique properties can be applied. In the future designers may use more if small size, higher temperature and higher frequency characteristics are required.
Comment