Li-ion batteries in series or parallel?

Hi to all,

I am asking again this common question, because I want to make things clear in my mind. In the attached file you can see the two different configuration that I have in mind for my project. I want to deliver regulated 3.3V to the electronics from battery voltage. One solution is to use two batteries in parallel (actually one battery could do the same job, two are for increasing capacity) and use a step down/step up DC/DC regulator like TPS63001. The second solution is to put two batteries in series and then use a step-down DC/DC regulator. like the TPS62056. My question is which of the two configurations give bigger battery capacity and thus longer autonomy for my circuit.

Assuming that both DC/DC regulator have an effiency of η and that the capacity of all batteries is 1.2Ah. For both configurations, the available energy has the same value: J = η * [(3.6V * 1.2Ah * 2]

Can anybody help me on how to compute the autonomy of a theoretical circuit that has an average current consumption of (let's say) 200uA per hour? Which is the best to achieve bigger lifetime and how can it be justified?

Can anybody help me with this?

Thanks,
Nikos

Series will give you more voltage, = 3.6 + 3.6 = 7V, parallel will give you 3.6V. If you need less voltage, then, the rest will be wasted

john blue said:

Series will give you more voltage, = 3.6 + 3.6 = 7V, parallel will give you 3.6V. If you need less voltage, then, the rest will be wasted

Click to expand...

Yeah, the regulator will convert the input voltage to 3.3V with an efficiency of η. And the rest power is lost. But can this justify that the series will give more lifetime? You recommend to use the series circuit? I have the feeling that the parallel configuration gives more total capacitance ( => lifetime), but I cannot justify it.

Parallel. Series you got 7V to 3.3. It's not efficient at all

I suggest you use parallel. From 3.6V to 3.3V, you can use a 3.3V LDO, the effieciency is about 91%.

If series, the DC/DC converter from 7.2V to 3.3V has effieciency maybe less or about 91%, but the circuit is more complex, and would have some spurs.

tony_lth said:

I suggest you use parallel. From 3.6V to 3.3V, you can use a 3.3V LDO, the effieciency is about 91%.

If series, the DC/DC converter from 7.2V to 3.3V has effieciency maybe less or about 91%, but the circuit is more complex, and would have some spurs.

Click to expand...

Hi tony_lth and thank you for your reply.

You are totally correct in what you are saying and I agree that if this was exactly the case, it would be better to just use a LDO. However, what happens if the temperature falls or the battery starts loosing its energy? Then, the voltage of the battery will fall below 3.6V and even below 3.3V. And in that case, the LDO would not be able to regulate the voltage. On the other hand, if I use the DC/DC regulator, even if the battery voltage falls I will be able to keep the 3.3V rail stable. Thus, the DC/DC regulator needs to be used.

In fact, I think that the efficiency of both series and parallel configurations is the same, what the regulator concerns. The problem is, how to compute the lifetime of the electronics in both cases and if there is any difference between them. I'm sorry for my noob questions, but I'm a little bit troubled.

The power consumption is 3.3V*2e-4A=6.6e-4W. Assume the baterry aging discount is 20%, and DC/DC efficiency is 90%, so the lifetime is (3.6V*2*1.2Ah)*90%*(1-20%)/6.6e-4W=9425hours. Consider the DC/DC voltage should be greater than 3.3V, so the real lifetime should be less than 9425hours. You can check the battery datasheet to find the degrade plot.

tony_lth said:

The power consumption is 3.3V*2e-4A=6.6e-4W. Assume the baterry aging discount is 20%, and DC/DC efficiency is 90%, so the lifetime is (3.6V*2*1.2Ah)*90%*(1-20%)/6.6e-4W=9425hours. Consider the DC/DC voltage should be greater than 3.3V, so the real lifetime should be less than 9425hours. You can check the battery datasheet to find the degrade plot.

Click to expand...

So, actually, there is no big difference between the series and the parallel connection in this case, right? The available Wh is (3.6V*2*1.2Ah)*η_p*80% for the parallel and (7.2V*1.2Ah)*η_s*80% for the series case. And if the efficiencies (η_p, η_s) of the two different regulators is around the same, then I can choose either of the two connections without loosing lifetime. Thanks tony_lth for your help!

Parallel has a advantage that is more robust. In series mode, if one battery fail, the whole battery system fail. But parallel not.

I searched in the web, found Li-ion battery voltage has some component like regulator to make voltage stable, so pls confirm this point from the vendor of the battery.

tony_lth said:

Parallel has a advantage that is more robust. In series mode, if one battery fail, the whole battery system fail. But parallel not.

I searched in the web, found Li-ion battery voltage has some component like regulator to make voltage stable, so pls confirm this point from the vendor of the battery.

Click to expand...

Yes, it's true that the parallel is more robust, but on the other hand requires three wires for the connection with the PCB (whereas the series requires two wires) and moreover I haven't found any battery holder for two batteries in parallel (whereas a hundread of holders exist for two or more batteries in series). Thus, I have to use a custom battery pack. From this point of view, the series is more easy to use.

I didn't completely understand your last point. You mean that you found that some Li-ion batteries have an internal mechanism to stabilize battery voltage?

nickagian said:

). I didn't completely understand your last point. You mean that you found that some Li-ion batteries have an internal mechanism to stabilize battery voltage?

Click to expand...

Yes, some web file saidl so.

tony_lth said:

Yes, some web file saidl so.

Click to expand...

Ok, thanks a lot for your help!

nickagian said:

And if the efficiencies (η_p, η_s) of the two different regulators is around the same, then I can choose either of the two connections without loosing lifetime.

Click to expand...

I should recommened you that when battery voltage varies, the battery efficiency will decrease.

PS:
Because your current is very low, you can't neglect the GND current of the LDO. Such as ADP1711, the GND I=40uA @ load current =100uA.
If you use DC/DC converter, the efficiency is also very low at low load current and also can't neglect the GND current. Check the datasheet.
You can check TPS54620 efficiency vs. load current.

Let's check some facts,

First of all, explain a little concepts:

- Parallel batteries would give you more capacity if needed. If you need 2A output and have two batteries of 1.5Ah, certainly you cannot connect them in series to provide you the current. Naturally, they will give you more application lifetime and voltage stability. If your application requires let's say 150mA out of these batteries, you will have 20 hours of running time since each one will be providing 75mA and they're rated at 1.5A (in this example).
- Series is for voltage boost. Simple as that, 3.7+3.7 (at nominal voltage) = 7.4. Running time won't increase since each battery will be sourcing 150mA rather than 75mA, so, the running time will be reduced to 10 hours. (the same as one battery). What was your gain then ? Well, the increased voltage. But, in your case, you don't need it.

If you're DC/DC converter is a buck/boost configuration, there's no problem of using a single battery and letting it work all the battery voltage range, and, since you said something about 200uA comsumption.. it looks like for the user it won't make any difference putting them in parallel, it would be everlasting anyway. Moreover.. if you want more lifetime... why don't you buy a battery with double the C-rate ?

tony_lth said:

I should recommened you that when battery voltage varies, the battery efficiency will decrease.

PS:
Because your current is very low, you can't neglect the GND current of the LDO. Such as ADP1711, the GND I=40uA @ load current =100uA.
If you use DC/DC converter, the efficiency is also very low at low load current and also can't neglect the GND current. Check the datasheet.
You can check TPS54620 efficiency vs. load current.

Click to expand...

You have right regarding the GND current of the LDO, but I have already taken it into account. Actually, the number that I've written is the mean current consumption. But, anyway, I use a DC/DC converter. And indeed, the efficiency is much lower at low load current, but most of them have some special mode which increases the efficiency at low load current.

arghpok said:

Let's check some facts,

First of all, explain a little concepts:

- Parallel batteries would give you more capacity if needed. If you need 2A output and have two batteries of 1.5Ah, certainly you cannot connect them in series to provide you the current. Naturally, they will give you more application lifetime and voltage stability. If your application requires let's say 150mA out of these batteries, you will have 20 hours of running time since each one will be providing 75mA and they're rated at 1.5A (in this example).
- Series is for voltage boost. Simple as that, 3.7+3.7 (at nominal voltage) = 7.4. Running time won't increase since each battery will be sourcing 150mA rather than 75mA, so, the running time will be reduced to 10 hours. (the same as one battery). What was your gain then ? Well, the increased voltage. But, in your case, you don't need it.

If you're DC/DC converter is a buck/boost configuration, there's no problem of using a single battery and letting it work all the battery voltage range, and, since you said something about 200uA comsumption.. it looks like for the user it won't make any difference putting them in parallel, it would be everlasting anyway. Moreover.. if you want more lifetime... why don't you buy a battery with double the C-rate ?

Click to expand...

First of all, thank you very much for your reply. It is really very helpful!! I understand now that the series configuration is for doubling the output voltage.

Secondly, I cannot use a battery with double the C-rate (as you suggest) because in that case I should buy a C or D size battery, the size of which is unacceptable for my box where the platform is integrated. Thus, I have to go only with AA batteries. That's the reason why I want to use two AA batteries in parallel. And indeed, my DC/DC converter is a buck/boost configuration and thus, I could use a single battery. But I need to use as much capacity as possible.

Basically, what I said is true, but I forgot some details. I assumed that the regulator device was a linear regulator in the explanation; thus, the current required from the batteries would be the same in each topology and the regulator would dissipate as much power as needed to obtain your vout.

Well, this isn't the case cause we're talking about DC/DC converters, which operate as dc transformers, it is, theoretically they have zero dissipation. According to these, with 100% efficiency, it wouldn't be any difference between series or parallel. Here's why:

- Let's say Vbat = Vout = 3.5V. If you series de batteries you'll have 7V. That would make the converter to operate as buck. The equations for a buck converter are:

Vout = D*Vin (3.5V = 0,5*7V)
Iout = Iin/D (150mA = 75mA/0.5)

That is, you manage to amplify your current when you reduce your voltage. So, the current from the battery would be half of the output current. According to this, you would get your runtime doubled, but...

There are some other considerations. Efficiency on DC converters is better when the conversion factor is low (close to 1, bucking or boosting). So, using a voltage as close as possible to your output will save you some energy. However, since you say that you cannot find holders and it's complicating the layout of the project itself, you may want to trade-off some energy for real-state area and simplicity.

In conclusion, series would give you nearly the same runtime as parallel (depending on the efficiency of the DC converter, the quality of the passive elements..., etc) and simplify the PCB.

Hope this helps,