Circuit/components to "help" battery through spikes in current draw

[schaplan]

We're designing a portable instrument with an embedded computer. It's a computer that gets very hungry for power sometimes, and needs to be powered by batteries at times. We're stuck using off-the-shelf UN38.3 certified batteries at the moment. 18.5V ("5s" or 5-in-series for li-ion cells) is the preferred voltage as it can efficiently be regulated to 16v and happily power the three sub-systems, one of which is the computer. The prototype system should demonstrate the ability to hot-swap batteries, or use two batteries together to extend battery life, as well as be able to charge the batteries in the unit when supplied with external power.

Normally, the entire system will draw about 15W, or close to 1A. We've identified several 50 to 70 Watt-hour battery candidates, which would provide the minimum 3-hour run-time required.

The problem we run into, is the computer can draw significant current in short bursts (less than half a second), and when the battery can not provide that, the computer crashes. We're willing to reduce the computer power in the bios, limiting it to say 20W, but early testing on these settings still shows that it will draw over 6A for very short periods.

The batteries we'd like to use, for the most part, are limited to 4A discharge current.

My background is not electrical, but I have heard of capacitors. Is there a common approach to using capacitors or other components, potentially a cap and an IC, to assist during short periods of high current draw?

We have a set volume for the batteries which somewhat limits us to the 50-70 Wh and 4A discharge rate batteries described above. However, in the unit we have a custom PCB with the battery charging circuit and voltage regulator where we could potentially include additional components to assist the batteries during these times.

Another solution we looked at are employing batteries with higher discharge rates. We'd identified some available graphene li-po batteries that meet the spec, but do not have on-board battery management system and our battery charging circuit does not support the cell balancing cable included with them.

Thank you for any direction you can give on this need. I apologize if this has been posted before; I attempted to search but each situation that came up in results seemed quite different than this.

 

[3DDave]

Supercapacitors might be used, but they are low voltage devices and aren't nearly as power dense as electro-chemical cells. You would need to look up the specifics; every time I have considered them the low voltage and cost was prohibitive and that was enough to end my investigation.

You are looking to store 60W-seconds, not including losses in the storage.

I didn't see a specification for the minimum voltage it works with.

I too have been hit with EEs saying "here's the power we require" and failing to mention the power spikes that can be 10X higher, though the worst was "this component generates less heat than the original" and omitting "and it can only take a 10C rise when the old one could take a 100C rise. Have any idea how much larger a heat sink has to be to dump heat with less than 10C to work with? The guy who spec'd it left the company before his landmine blew up in my face.

 

[lucky555]

I would do a bit more homework to see if something is making this worse than need be.
Is the battery's BMS cutting out, i.e. seeing the overcurrent and disconnecting the pack causing the reboot glitch?
What is the undervoltage lockout on the computer's power supply? It might be cutting out when the input voltage sags if it's say below 15V.
So these cutting out prematurely can aggravate things.

Supporting 6A peaks in the msec range is done with large value electrolytic capacitors in the 1,000's of uF. You could try some as it also filters out the brief voltage dips, which can prevent false triggering of BMS or UVLO or the computer's reset line.
It seems odd the computer's power requirement is so high - look at what a cellphone can achieve.
If you need 0.5 second hold time, it's going to get big and expensive and normally a larger battery would get used.

 

[schaplan]

Thanks for the replies! I'm glad to hear using capacitors isn't a crazy idea at least in theory. Based on what I've seen so far employing capacitors in the 1000s of uF up to 100,000 would be beneficial from a physical volume standpoint. However, it doesn't look like I'm going to get the EE support on this one so we are looking at a bigger battery and another regulator. Previously, both were limited to 4A which would explain the failures.

It looks like available options for UN38.3 rated batteries that fit in our envelope is not going to leave us much headroom, and I still like the idea of using capacitors to handle less than 1% of operation where these short current spikes occur

Say we used something like this 11000 uF cap

https://www.digikey.com/en/products/detail/united-chemi-con/ELBK250ELL113AL40S/5824547

Where would it go in the circuit? What sort of support components would be recommended with them? Perhaps a resistor so they don't draw too much current themselves when charging?

I'm thinking they'd go in series on the power after the regulator before the computer, so after they charge, they sit at the regulated voltage. If so, could we use multiple in that fashion? Would they be in parallel or series?

As far as the computer drawing a lot of power... that's what they can do! It's incredible to see how quickly that processor can just demand power when it wants it. It goes from a trickle to opening up the floodgates in almost in a millisecond, and generates the heat to boot...!

 

[IRstuff]

Not a practicing EE anymore, but someone like your missing EE needs to do some analysis to see how the battery and BMS can be designed/configured to handle instantaneous high loads. A typical alkaline AA battery can discharge 10A when called upon, and that's mainly because of its low internal resistance. Some RC racing fanatics will "condition" their racing batteries by intentionally loading their batteries, presumably to force the internal resistance even lower.

However, a REALLY good BMS should be able to manage to maintain the output voltage, even in the face of a sudden current surge.

So, it may just be that you have, at worst, a combination of poor internal resistance in your batteries, coupled with an insufficient BMS design/configuration

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[3DDave]

UN38.3 appears to be only about safety, not performance.

Probably an NTC thermistor that limits inrush current would be good for charging the capacitor.

See:

https://product.tdk.com/en/techlibrary/applicationnote/howto_ntc-limiter.html

For discharge look for a low-resistance transistor that is set to detect when the cap voltage is greater than the battery voltage. I would worry that that spike in draw would be too short to let the NTC thermistor to drop in resistance, leading to too much voltage drop for the CPU to withstand. MOSFETs for example.

 

[Comcokid]

The internal impedance of Lithium batteries is low. I'm surprised you have too much voltage drop during transient current events. You mentioned you are using 5S (5-series) as "the preferred voltage as it can efficiently be regulated to 16v".

Are you sure it is not your regulator that is your voltage drop issue? Are you using a linear voltage regulator? Linear regulators typically have a minimum voltage drop of < 0.4V (for a very good low-dropout type) to 1.2 V (for the older type of voltage regulators). Use a regulator with a lower input-to-output dropout if this is the issue.

OR, can you use a 6-S pack instead? If you're using a linear regulator this will increase it's heat. You could then change the regulator circuit to a buck-type. Efficiency is much higher and heating is lower.

 

[sunnysky]

The max current of 4A is quite low for high-quality batteries. This limits the I^2R=P power dissipate and R is the sum of each cell ESR or Rs which rises rapidly below 10% charge (SoC).

The power capacity W=VI and voltage drop at max load depend on the above parameters.

Battery cells are like massive ultracaps where the ESR*C=T time constant varies with component type and quality except C values are in the 100k Farad range with ESR ranging from 20 mohms to 200 mohms for 18650 cells. Your results may vary. measure and average step pulse loads and compute ESR=ΔV/ΔI. Then if you need to reduce ΔV use ΔIc = C ΔV/Δt which says the changer in current Ic in the capacitor, C in Farads will drop in voltage ΔV in time Δt then minimize the instant drop in voltage due to ΔVd= ESR * ΔI.