Effect of ripple load on Li-ion battery

[Allen3]

I'm investigating the design of a bi-directional switching DC-DC converter that will be connected to a Li-ion battery. The goal is to make the switching frequency relatively high, in the range of 50-100 KHz. Due to the switching, the converter will place a load on the Li-ion battery that contains a DC component and a ripple component at the switching frequency. Similarly, when returning power to the battery, the converter will put a ripple current back into the battery.

I'm trying to figure out how Li-ion batteries handle this high frequency ripple. Is there recommended limits for the ripple current and frequency when charging and discharging? How does the performance of the battery change as the amount of high frequency ripple changes?

Can anyone share some information on this or point me to a good reference? I've done a web search, but it didn't get very far. Thank you.

 

[IRstuff]

Unless the ripple exceeds the capability of the ESR, why do you think there's a problem?

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[Allen3]

Yes, DC+ripple will have higher I2R losses than just DC. Is that the only thing I need to worry about? Maybe the high frequency ripple affects (adversely or beneficially) the chemical reactions inside the battery. I don't know.

 

[itsmoked]

The ripple will heat the battery according to the ESR. If the ripple is low enough or the ESR is low enough to keep the temperature rise low enough for the battery's expected environment you shouldn't have a problem. Consider that most Li-Ion feed switching supplies that result in ripple.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Allen3]

OK, thanks. I expected the battery would tolerate some amount of ripple, but I didn't know how much and what the tradeoffs were.

 

[bacon4life]

The Importance of Monitoring AC ripple Voltage and Current explains how ripple affects batteries.

 

[Allen3]

Thanks for the tip. That paper does say it applies to VLRA (lead–acid) batteries, not Li-Ion batteries, with 50/60 Hz ripple, not high frequency ripple.

For the sake of completeness in this thread, I have learned that the skin effect can be important, i.e., at high enough frequencies, the current paths inside the battery will be narrowed by the skin effect and this will increase their effective resistance and cause additional heating inside the battery. In addition, current ripple will induce a changing magnetic field inside the battery which will cause the same type of losses that are found in magnetic cores used for transformers and inductors, i.e., magnetization hysteresis and eddy current losses. These will also heat the battery.

 

[Handl0r]

Hey Allen,

Im also trying to do some research in this field (ripple effects on l-ion). May I ask which kind of literature you used?

greets
Handl0r

 

[Allen3]

Just papers that are publicly available and indexed by Google, which unfortunately isn't a lot. You might be able to find more in a good pay-walled research index. FYI, in addition to the effects described above, in some chemistries, local polarization of the electrolytes can occur. This reduces the reaction rate, reducing power output, and may increase effective resistance and power losses. Also, with a Li-ion battery, I believe you always need to be concerned about minimizing how often you change the polarity of the current, i.e., minimize the number of transitions from a charging state to a discharging state, because just about all Li-ion batteries suffer from capacity fade as they are charged and discharged. As long as the ripple is smaller than the DC, then you will be either charging or discharging and not cycling between the two, so this would not be a problem.

 

[Handl0r]

Many thanks for your hints in this case. I also have access to the IEEE library, but i can not found informations about high frequency effects. Modeling is done only for low frequencies or pure dc current. I would appreciate if you can send me some links to some paper stuff.

 

[Allen3]

The only page I bookmarked is this one:

http://www.electrochem.org/dl/ma/200/pdfs/0135.pdf

. I remember seeing some other papers that looked at battery impedance in a lot more detail and contained similar graphs. I believe that above 5 KHz or so, the only additional effects you need to worry about are skin depth and magnetization effects (eddy currents and magnetic hysteresis) that I mentioned in a post above. These a highly dependent on the construction of the battery (materials and layout). I think to play it safe, you should keep those frequencies out of the battery by using a series inductor. That will move the magnetic core losses out of the battery and into the inductor where you'll have better control over them and they'll heat the inductor instead of the battery.

 

[Handl0r]

Again thanks for the link and your help. I will do some more research in skin effect and magnetization effects.

 

[IRstuff]

Upon further thought, I think you guys are barking up the wrong tree. No system in the commercial world uses solely the battery for high frequency current draw; that task is handled by capacitors, both large and small. The reason has to do with the frequency response of the battery, compared to a capacitor. Even with active power supplies, the frequency response of the outputs might only be in the low kHz, compared MHz demands. Capacitors smooth all that out.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[Allen3]

There are always tradeoffs such as volume, weight, cost, operating conditions (primarily temperature), efficiency, reliability and lifespan (high voltage, high power capacitors are prone to early failure). A series inductor of course has the same effect as a parallel capacitor, or you can use both. And no matter what you do, some amount of high frequency ripple is going to make its way into the battery, so its a good idea to understand the effects.

 

[DHambley]

Allen, I've worked with battery manufacturers who had limited or no test data on high frequency effects. They had ESR curves at a few different temperatures which were produced using low frequency stimulation but no info on the heating effects you mentioned. In any case like this, go with the recommendations of this thread and add L-C filtering so that the C takes most of the high-frequency ripple. If just C and no L, you have to analyze how much of the ripple is shared by the C vs the battery. I don't know the power levels you are designing to but mind your loops with high di/dt. Once I worked with a 100kW system for which the original designers used no inductor, just a capacitor at the PWM connected via huge cables a foot apart; The cables open-air inductance of a few hundred nano-H's was the only L in the system and so, the voltage on this 'inductor' blasted EMI out of the system far, far higher than what is allowed by the FCC.