Parallel Supercapacitors with LFP bank

[Hedges]

I will be AC charging this system quite often, so the supercap will be useful also for removing the AC charge ripple.

I'm not sure how effective supercaps are at 60 Hz. Do you have data on frequency response or inductance?
Could be electrolytics work better. Although, even if supercaps only present 10% or just 1% of their capacitance at that frequency (impedance is 10x or 100x higher than pure capacitance suggests), could still be beneficial.

I found some curves, but not ESL in a data sheet.
Curve page 6 shows linear decrease in impedance with frequency, expected of a capacitor, only to about 5 Hz:

https://www.cap-xx.com/wp-content/uploads/datasheets/CAP-XX-GW103-GW203-Datasheet.pdf

The capacitor will draw AC current and heat up as it tried to smooth the 60Hz ripple.
The data sheet I found has an amps rms spec. Looks to me like 0.6% of voltage spec would be max ripple, so your system needs enough capacitor to spread out current draw and limit ripple to that amount.

Supercaps are meant to supply current for seconds. In electronics design we use a hierarchy of caps with different value and frequency response. The electrolytic caps in your inverter probably take care of most 60 Hz ripple.

 

[GXMnow]

With my XW-Pro, I was rather shocked when I saw the amount of ripple current both charging and discharging the battery bank. Since it is a true PFC corrected system, the current is a 120 hz full wave rectified sine wave. If you look at the current on a scope, you have the first half of a sine wave, and then the second half of the sine wave is positive again, and this just keeps repeating. So when it is charging at 30 amps, it is actually going from almost 0 to about 42 amps. This is filtered a little by the capacitor bank, but not near as much as I expected. I am going to look into adding an LC or Pi filter to try and smooth it a bit at my battery bank. I used to have a few of the power filters we used on Xenon lap supplies, the ripple from even a 3 phase bridge would chew up the bulbs if the filter caps failed. The series inductor was designed around the 3 phase 360 Hz ripple frequency though, so I would need about triple the inductance. The caps also had to be sized for the 150 volt open circuit before lamp ignition, so they were only about 12,000 uF and still 12 inches tall and 3.5 inches around. I do have 2 of them here. I tried just paralleling one of them at the battery bank, but with just #12 wire to the cap, it made no difference to the ripple at the batteries. Seeing this ripple current, I totally understand why the current reading in the BMS jumps around a bit, it is just taking a sample off the shunt without enough of a filter.

 

[Hedges]

The series inductor was designed around the 3 phase 360 Hz ripple frequency though, so I would need about triple the inductance.

Single or split phase delivers power that ripples with the sine wave, but 3-phase ideally has no ripple at all.
If fed through diodes to charge caps, of course power factor would be bad.
A good PFC should draw current in proportion to voltage from each of the 3 phases, and deliver steady DC current that is the sum of the power drawn from three phases. Just the reverse function of a grid-tied inverter.

A PV inverter feeding single phase AC has to draw ripple current from the caps. A 3-phase inverter should impose no 60 Hz ripple current at all on the caps, only the ripple from its higher frequency switching. Inductors would carry the piecewise linear or stairstep current synthesizing each sine wave, but the three phases together sum to no sine wave at 60 Hz, just steady-state current.

"current reading in the BMS jumps around a bit, it is just taking a sample off the shunt without enough of a filter."
I would filter the sense lines; that is low current so doesn't need large filter components.

 

[GXMnow]

3phase does have far less ripple, but there is still some. Without the caps, the xenon bulbs would only live about 1/10 life and the electrodes looked like they were beaten with a mace. Just 5% ripple was enough to shorten their life. And the ripple is up at 360 hz. Like I said, the inductor was only about 50 turns of #3 solid copper on an E I core, so it is not enough to filter the 120 hz ripple I am seeing.

I am not in a big hurry to open up the BMS and try to find the shunt sense line to add a filter, but yes, that would not need large components like it will take to filter the battery current line.

 

[Dhopewell]

I have a naive question. If you need supercapacitor properties could a lithium titanate battery pack do? They cycle more than lifepo so certainly could deal with surge and ripple, have a pretty high C rate / low internal resistance and about the same voltage as super caps with higher energy density. Are there caps that discharge that much more quickly, and do you really need that much speed for a load?

I’m not aware of cheap super caps - could someone post a link?

 

[RCinFLA]

I think the price of Maxwell Ultracaps has skyrocketed since Musk/Tesla bought them out. It might be due to their getting new business direction and not making caps for general market so all there is left is what is in the existing inventory pipeline.

I have tried a few of the Chinese supercaps and found them to be pretty poor. Very high internal resistance compared to Maxwell Ultracaps.

 

[toms]

LTO cells would be great, in ten years time when i’m looking for my next battery they will be on my list if they haven’t shown significant failures before then.

In the meantime they aren’t cost effective.

Problem with batteries is that to cost them properly you need to wait until the end of their life - most simulations rapidly age them.

I’m still waiting for the first pack i have been involved with to get below 80% capacity - and that one is just over 10 years old.

All i can do is look at packs that have degraded, and see what is being done differently to the ones that are being used without degradation.

I can show you the scope traces of reduced AC ripple, and lowered surge currents- but you will likely have to wait more than 10 years for a result on increased cell longevity.

 

[Dhopewell]

That is the point I see here - for normal use, say below 5C - it’s possible to find cells that do just fine and life expectancy isn’t an issue unless the cells are hot. But it would only take a small pack of LTO to serve as a super cap. It’s too late to do the math but I might figure out what it takes to deal with ripple or gigantic inrush, or slamming the brakes on a scooter, etc