# Range for low ESR capacitor

I'm new at the circuit design biz. I'm designing a battery powered device around the TI TPS61025 boost converter, and I'm rather confused about the proper range for the low-ESR output capacitor. The data sheet says says

With the calculated minimum value of 24 μF and load transient considerations the recommended output capacitance value is in a 47 to 100 μF range. For economical reasons, this is usually a tantalum capacitor. Therefore, the control loop has been optimized for using output capacitors with an ESR of above 30 mΩ.
So I understand I want a cap in the 47-100 μF range with an ESR of at least 30 mΩ. But what's a reasonable upper limit for the ESR? Is there an upper limit?

This is hard to answer without knowing the what exactly is going on inside the control loop, which is probably not described with much detail in the datasheet. Saying it is "optimized" for 30 mΩ doesn't say what it really needs or what happens when it's not exactly 30 mΩ. Since they make a issue of the output cap ESR, there should be a min/max ESR spec elsewhere in the datasheet.

Hopefully 0 ESR is OK, else it gets inconvenient. Lower ESR is generally better, and certainly closer to ideal, for a capacitor. Capacitor spec sheets therefore often only spec maximum ESR. In that case you either have to get the guaranteed minimum from the capacitor manufacturer, or add deliberate resistance. A better answer is to stay away from chips that specify a minimum output cap ESR to be stable.

That chip is probably old, as ceramic capacitors have come a long way in the last few years. These have much lower ESR than tantalum, and can now approach such capacitances at reasonable cost. Newer chips now make use of the lower available ESR to get better performance instead of requiring the capacitor to have a high resistance. This is a all around better strategy.

• The transfer function of the error amplifier is given in the paragraph after the one quoted in the question.
– Mark
Commented Aug 9, 2011 at 20:21
• I've seen several controllers which required a minimum ESR on the output capacitor to maintain loop stability. It's a nifty cheat, IMHO. Another issue that often gets overlooked is that large ceramic capacitors can emit acoustic noise due to the piezoelectric effect. In my line of work (sonar systems), this is a big deal, so even if a controller works fine with a ceramic output cap, we try to use a small ceramic + large tantalum combination. This results in the ceramic cap's ripple current being in the higher frequencies that don't bother us. Commented Aug 9, 2011 at 20:53
• I agree with the above. Most likely the internal compensation is type-2-ish, and cannot deal with the steep roll-off slope caused by the low ESR. Since the internal compensation is hidden, trying to stabilize for low-ESR with external Rs and Cs could be an exercise in futility. Commented Aug 10, 2011 at 14:09
• Whats hidden about the internal compensation? The transfer function is given and they even provide a recommendation on the external components needed to provide stability with a very low ESR output cap. I really don't see how they could have made it any easier to sort out.
– Mark
Commented Aug 10, 2011 at 19:35

The upper limit is defined by your required maximum ripple on the output voltage.

Higher ESR = more ripple.

This is discussed in the paragraph of the datasheet directly before the one you quoted in this question.

• I'm sorry, but I don't understand. That following paragraph, "Small Signal Stability", starts out "When using output capacitors with lower ESR...", but the quoted paragraph says it's optimized for ESR above 30 mΩ. So don't I want higher, not lower, ESR? Commented Aug 9, 2011 at 21:38
• What you want depends on what you need. That paragraph discusses what needs to be considered if you choose to use a capacitor with ESR < 30mOhm. Why would you do that? Maybe you need very low ripple, maybe you can't afford the cost of a tantalum output capacitor, etc.
– Mark
Commented Aug 9, 2011 at 21:48