CA3 LiFePO4 state of charge

donn

10 kW
Joined
Aug 13, 2018
Messages
875
Location
Seattle
Recently set up, ebay 16 cell LiFePO4 battery and Cycle Analyst 3. "State of Charge" indicator says the battery is full, when it actually should be down to about 40%.

user guide said:
Note: Unlike other chemistries, LiFePo4 has a very flat discharge curve making it particularly difficult to ascertain SoC by voltage. The CA handles LiFePo4 uniquely, disregarding voltage as a SoC metric except at the knees of the curve. This allows it to accurately detect 'full' and 'empty' conditions by voltage while tracking Ah at all other times. This generally gives a reliable charge indication but anomalous LiFePo4 displays are possible if the voltage makes transient low excursions due to sag (more pronounced in cold weather), slow controller capacitor discharge when powering down, etc. In these situations the lower SoC indication due to low voltage can 'stick' as the CA ignores subsequent rise to higher voltages in the middle voltage zone. A fresh charge brings the voltage above the top knee and restores proper indications.

Questions:

1. The way I read this, the computer would need to keep track of the battery's ongoing state for these calculations. Particularly, "... while tracking Ah at all other times." CA displays an Ah statistic on the first screen, but that isn't the one it needs for this - it's just Ah since last time the data was reset, and not tied to battery cycle or even which battery. Can it actually track Ah on the battery and display mid range state of charge? How - how can it reliably reconnect to a battery and know how many Ah it has on it since last full charge? (Extra credit if this hypothetical ability can also be relied on after incomplete charge-up.)

2. This all seems critically dependent on some base voltage. Seller says 16 cells, which CA reckons is 51V; actual reported voltage is 52.2V right now, so that seems fair enough. Sold as 48V. Does CA derive "top knee", "middle voltage zone" etc. from this 51V, or is this another stored state that would have to be tied to battery cycle in some way of which I can't conceive?

(My other "48V" battery is Lithium 2.4Ah cells, but the number of cells needed to get CA to consider it a 48V battery is 13, which would add up to quite a bit more than the advertized 23Ah. If CA needs to know the 52V or something that it actually puts out, I can probably ask the manufacturer what to tell CA.)

Bonus question 3. Just now while checking numbers, I got "Setup changed!", and now I see that it's reporting about one less Ah, and Regen has gone up to an impressive 31%. What's up with that? Before this, I was pleased to see regen at a modest 8% or something. I guess it might be time to reset the data.

The bottom line is of course just that it would be real nice to know where I am on battery charge.
 
donn said:
1. The way I read this, the computer would need to keep track of the battery's ongoing state for these calculations.
The CA stores State of Charge in addition to Ah, etc so the SOC persistence is related to the Trip Reset and not the power cycle. Also, there is only one SOC saved, so switching packs can sort of rejigger the SOC a bit. Non-LiFe chemitries sort this out PDQ but LiFe SOC remains confused because of the open circuit voltage problem.

donn said:
2. This all seems critically dependent on some base voltage. Seller says 16 cells, which CA reckons is 51V; actual reported voltage is 52.2V right now, so that seems fair enough. Sold as 48V. Does CA derive "top knee", "middle voltage zone" etc. from this 51V, or is this another stored state that would have to be tied to battery cycle in some way of which I can't conceive?
It uses the cell voltage to reference a table of discharge curve voltages and so is dependent on a 'typical' cell discharge curve for a given chemistry - not the pack voltage directly.

donn said:
Bonus question 3. Just now while checking numbers, I got "Setup changed!", and now I see that it's reporting about one less Ah, and Regen has gone up to an impressive 31%. What's up with that? Before this, I was pleased to see regen at a modest 8% or something. I guess it might be time to reset the data.
You did not say what firmware you are using. The 'Setup Changed' message appears when the CA sees the Setup change without going into Setup. This happens normally when you use the Setup utility since the CA doesn't really know about it and just sees new settings appear magically. Pretty much any other appearance of this message indicates a program malfunction and a corrupted setting somewhere. This was a problem in 3.1 but was corrected in 3.11 (we believe...). So - if you are running 3.1, upgrade.

Meanwhile - I can tell you that 3.2b1 has a fix for LiFePo4 SOC that works nicely. I've been riding with it for a while and it gives good SOC after power cycles regardless of how slowly the power goes down or comes up. Not sure when 3.2 beta will be released (Sept, I think) - JLE is doing the marriage/honeymoon thing and so we haven't yet discussed the near term beta roadmap.
 
teklektik said:
donn said:
1. The way I read this, the computer would need to keep track of the battery's ongoing state for these calculations.
The CA stores State of Charge in addition to Ah, etc so the SOC persistence is related to the Trip Reset and not the power cycle. Also, there is only one SOC saved, so switching packs can sort of rejigger the SOC a bit. Non-LiFe chemitries sort this out PDQ but LiFe SOC remains confused because of the open circuit voltage problem.

So ... if I clear the trip data and I'm back to 0 Ah, the computer starts a fresh voltage-read SOC and counts trip Ah relative to the new SOC, as long as the trip lasts? And since the voltage-read SOC is not much use with LiFePO4, it's ideally better to keep trip Ah going.

teklektik said:
It uses the cell voltage to reference a table of discharge curve voltages and so is dependent on a 'typical' cell discharge curve for a given chemistry - not the pack voltage directly.

The chemistry alone - I don't have to worry about whether I got the number cells right, etc.?
 
donn said:
The chemistry alone - I don't have to worry about whether I got the number cells right, etc.?
It uses the number of cells to figure out the per-cell voltage from the pack voltage. It just has different curves for different chemistries...
 
donn said:
So ... if I clear the trip data and I'm back to 0 Ah, the computer starts a fresh voltage-read SOC and counts trip Ah relative to the new SOC, as long as the trip lasts? And since the voltage-read SOC is not much use with LiFePO4, it's ideally better to keep trip Ah going.
Yes and yes.
Except that there is that business of the CA paying attention to voltage near the knees of the curve, so the SOC may drop a bit after the reset if it is near full charge, but won't change at all if it's at half charge since it doesn't trust the lower voltage as much as it trusts the knee voltage. Odd, but works fine for most use cases.
 
teklektik said:
donn said:
The chemistry alone - I don't have to worry about whether I got the number cells right, etc.?
It uses the number of cells to figure out the per-cell voltage from the pack voltage. It just has different curves for different chemistries...

OK, though from my perspective just running the setup, I don't enter a pack voltage, it seems to be computed as a function of chemistry and number of cells. Anyway, if I say there are 16 cells because that's what they told me, and there are really 15, then apparently it will think I'm running low voltage per cell. (I just changed it from 16 to 15, and battery is now 48V as advertised. No apparent change in SOC, but then this is LiFePO4 anyway.)

[ edit ] Oh wait - I get it - all LiFePO4 cells are the same voltage!? So the chemistry alone is sufficient. I am not knowledgeable about batteries, so it takes a little while for the illumination to penetrate.
 
donn said:
Oh wait - I get it - all LiFePO4 cells are the same voltage!? So the chemistry alone is sufficient. I am not knowledgeable about batteries, so it takes a little while for the illumination to penetrate.

CA Setup Utility ToolTip and Help File said:
[ Batt->Volts Disp ]
Chooser to select whether the main CA screen displays just the battery pack voltage or alternates between pack voltage and average cell voltage. The average cell voltage is not measured but is computed as (pack voltage)/(number of cells).
 
So it seemed to me a fairly safe bet to derive the number of cells from the actual voltage. The LiFePO4 battery is a little over half discharged and reads 52.2; 52.2/16 (the published cell count) matches the CA user guide Battery SOC Curve value close enough at 3.26. The fully charged AllCell Lithium battery was 54V, so at a 4.2V cell per the curve, it must be 13 cells.
 
If it's helpful, where the "voltage" is the nominal rating, not the full charge voltage:

A LiFePO4 "48v" battery is typically 16s.

A Li-Ion/LiCo/LiMn/NMC/etc "48v" battery is typically 13s

A Li-Ion/LiCo/LiMn/NMC/etc "52v" battery is typically 14s

Sometimes confusing, when actual series number of cells isn't specified, and/or full pack voltage is not given.


The LiFePO4 16s and the Li-Ion 14s are generally compatible for full charge voltages, but the LiFePO4 has a very "flat" discharge curve: Almost all of it's capacity is right around 52v, so it is difficult by voltage alone to tell what it's SoC is. When it starts dropping in voltage it's generally close to empty, though there's a bit of usable capacity there.

13s LiIon is a lower overall voltage, maxing out at around where the other two have their median voltage.
 
That would be good info for the user guide. The setup menus don't particularly care about that nominal battery voltage, so when you configure the right chemistry and it says OK, that's a 41V battery, it can be a puzzlement. Then given the right cell count, it changes that to a 51V battery. That list of typical battery configurations could shine a lot of light on this stuff for people like me who don't make batteries.
 
The CA3 wants pack Ah input during initial setup. Isn't this data also used in computing and displaying SOC? My fuel gauge/battery icon is always very close to actual state of charge based on CA monitored and displayed AH usage as compared to my total pack Ah input number during setup. I always use trip reset just before I re-charge.
 
BVH said:
The CA3 wants pack Ah input during initial setup. Isn't this data also used in computing and displaying SOC?
Yes. The problem is that it uses both Ah and V, but voltage is not a good indicator for LiFePo4. All the other chemistries are actually calculated a little differently from LiFePo4 because of this. This led to some odd situations where a LiFePo4 pack could be power cycled at mid-charge and come up showing no charge at all - whereas the CA has high confidence in Li-ion voltage and so those have always power-cycled okay in mid charge because the CA 'believed' the voltage was a good indicator of SOC. Anyhow - fixed in the next beta....
 
Yes, like that. The user guide version could probably go into a little more depth for LiFe - if I have a "48V" battery, now what - is it 15, or 16? I can tell by actual voltage, so maybe some key values or a pointer to the charge curve chart. That the cell count is important, that's useful information too.
 
donn said:
Yes, like that.
The point was that tool tip information already exists and people don't bother to use it. It's even invoked directly on the parameter that needs to be entered and still it's not used as your remarks and request here indicate.

This does not build a compelling argument for additional effort - particularly for LiFe - an older and unpopular chemistry for the ebike demographic...

And, of course, there is no User Guide for the 3.1 firmware family.... There is the Help File, but focus is on videos for the people who don't read manuals or ToolTips.
 
teklektik said:
The point was that tool tip information already exists and people don't bother to use it. It's even invoked directly on the parameter that needs to be entered and still it's not used as your remarks and request here indicate.

I was not aware of the existence of that Setup application until you mentioned it yesterday. When I went out looking for docs on Cycle Analyst 3, I got the user guide. That's the maximum I would expect for anyone in my situation who has acquired a configuration with a Cycle Analyst, until they have some reason to need the Setup application, which I guess would be to upgrade the version.

This does not build a compelling argument for additional effort - particularly for LiFe - an older and unpopular chemistry for the ebike demographic...

I guess I will let the demographic speak for itself on that. I found them quite readily available, I will say that.

And, of course, there is no User Guide for the 3.1 firmware family.... There is the Help File, but focus is on videos for the people who don't read manuals or ToolTips.

Argh. Up to this point, my hope was only to make a helpful suggestion, and certainly not to pick a fight over how this Cycle Analyst product is documented. But videos ... I guess here my annoyance is directed at those people, who evidently can consume information only from videos, to the extent that videos are now getting to be the sole documentation for a product. What a god-awful waste. If you bought something and you need the install procedure for example, which is better: an ordered list of the actions you need to take, with diagrams etc. as required, that you can print and set beside you while you go through those steps, maybe scribbling notes if you're that organized? Or a video with some guy waving his hands, occasionally giving something the wrong name, picking it up, putting it down, yakking about stuff that doesn't really matter, and somewhere in there mentioning at 3:12 that the plug with the blue, brown and green wires goes to the plug with the blue, yellow and red wires?

Whatever. Here's one person who looked up a User Guide, for Cycle Analyst 3 if not 3.1.
 
donn said:
I was not aware of the existence of that Setup application until you mentioned it yesterday. When I went out looking for docs on Cycle Analyst 3, I got the user guide.
...
What a god-awful waste.
...
Whatever. Here's one person who looked up a User Guide, for Cycle Analyst 3 if not 3.1.


If you go to the manufacturer site you will find the CA3 page (indexed off the nav bar) with a variety of information including the exact Help files on-line with the ToolTip content I snapped above, a link to the Unofficial Guide, and downloads for firmware, the Setup Utility, and drivers.

No searching required.
 
Back
Top