Output capacitor of a DC/DC converter

[onemilimeter]

Hi,

Let's say we have two types of capacitor, i.e. 35V 10uF tantalum capacitor (ESR@100kHz=0.1ohm) and 35V 10uF ceramic capacitor. Which one of them will you choose to be the output capacitor of a DC/DC converter for best performance?

Cheers.

[Psi]

i'd use the ceramic.
Depending on brand/spec the ceramics ESR should be better than that tantalums, or atleast no worse than it
Probably in the range of 0.1R - 0.01R.

There maybe other issues than ESR to consider though, multilater ceramics have the peizeoelectric effect.



 

[onemilimeter]

There maybe other issues than ESR to consider though, multilater ceramics have the peizeoelectric effect.

If we look the unit price quoted in Farnell:

[ A ] VISHAY SPRAGUE - TR3D106K035C0125 - CAPACITOR, TANTALUM, 35V, 10uF, ± 10%, 0.125ohm, -55°C to +125°C
£1.05

[ B ] MURATA - GRM31CR6YA106KA12L - CAPACITOR, 1206, 35V, X5R, 35V, 10uF, -55°C to +85°C
£0.22

Apart from the operating temperature range and consider that ceramic has lower ESR (a feature required by DC/DC converter output capacitor) than tantalum cap, the ceramic capacitor (item B) above appear to be the right choice. However, I agree with Psi that there may be other issues than ESR. Please share if you know any of them that make the tantalum capacitor more preferable.

Cheers.

[Conrad Hoffman]

IMO, tantalums don't have any special high frequency properties compared to electrolytics designed for HF operation. Tantalums tend to fail shorted. Aluminum electroyltics tend to fail open, or they get high losses and burst. Life of tantalums tends to be longer, unless they're in severe service with high current pulses. A good multilayer ceramic will beat them both for HF performance, and last near to forever. They can still fail, usually shorted, but the technology today is darn good if you're willing to pay the price, which isn't usually too much. The only thing to watch for is that Rs (esr) can be low enough to cause ringing problems against circuit inductance. Stick an extra fractional ohm in there if you need it or just keep stray circuit inductance low.

[onemilimeter]

IMO, tantalums don't have any special high frequency properties compared to electrolytics designed for HF operation. Tantalums tend to fail shorted. Aluminum electroyltics tend to fail open, or they get high losses and burst. Life of tantalums tends to be longer, unless they're in severe service with high current pulses. A good multilayer ceramic will beat them both for HF performance, and last near to forever. They can still fail, usually shorted, but the technology today is darn good if you're willing to pay the price, which isn't usually too much. The only thing to watch for is that Rs (esr) can be low enough to cause ringing problems against circuit inductance. Stick an extra fractional ohm in there if you need it or just keep stray circuit inductance low.

Very good information... Thank you very much

[PeteH]

I usually use a mixture of tantalum and ceramic output capacitors. Depending on your DC bias voltage, the ceramic output capacitors will degrade in capacitance by a significant factor. Most people do not consider DC bias when designing output stages. When your calculations result in a particular value, depending on the application, you should consider DC bias, temperature, age, etc.

Using a larger package  (1210) can reduce this effect. When you jump to 2220 you're looking at higher costs and stress sensitivity (flex and expansion). In many industries, electrolytic (non-solid) are not allowed to be used. A balance between HF bypassing and LF decoupling depends on your application. Take a look at the DC bias effect on that 10uF capacitor and compare how many ceramics you would need to get 10uF of effective capacitance.

I would throw down a tantalum further away from the DC converter (closer to the load) and ceramics to take the switching ripple currents. Tantalum capacitors can be reliable if used properly. If you don't use them in high ripple current, high temperature applications, they're great to work with. There are several organic polymer tantalums (more expensive) that eliminate the typical tantalum failure mode. These caps are the T520 series (Kemet); featuring low ESR and high current capability.

Of course if you're costs aren't a priority and you have space (what are the odds) you can throw down the tantalum and some smaller valued ceramics to take the switch transition currents.

Feel free to add extra requirements or details about your project/design. I would love to provide more insight.

[onemilimeter]

I usually use a mixture of tantalum and ceramic output capacitors. Depending on your DC bias voltage, the ceramic output capacitors will degrade in capacitance by a significant factor. Most people do not consider DC bias when designing output stages. When your calculations result in a particular value, depending on the application, you should consider DC bias, temperature, age, etc.

Using a larger package  (1210) can reduce this effect. When you jump to 2220 you're looking at higher costs and stress sensitivity (flex and expansion). In many industries, electrolytic (non-solid) are not allowed to be used. A balance between HF bypassing and LF decoupling depends on your application. Take a look at the DC bias effect on that 10uF capacitor and compare how many ceramics you would need to get 10uF of effective capacitance.

I would throw down a tantalum further away from the DC converter (closer to the load) and ceramics to take the switching ripple currents. Tantalum capacitors can be reliable if used properly. If you don't use them in high ripple current, high temperature applications, they're great to work with. There are several organic polymer tantalums (more expensive) that eliminate the typical tantalum failure mode. These caps are the T520 series (Kemet); featuring low ESR and high current capability.

Of course if you're costs aren't a priority and you have space (what are the odds) you can throw down the tantalum and some smaller valued ceramics to take the switch transition currents.

Feel free to add extra requirements or details about your project/design. I would love to provide more insight.

Thank you for your reply. I learn many new knowledge today. For example, the effect of DC bias voltage on ceramic capacitor.