How to Recondition/Service Your 12V Battery

[generaltso]

Judging by what I just saw, that sounds like a good guess. I checked my battery again, and this time it read 14.7V, so it clearly hit some threshold for the DC-DC converter to kick on. I left it for 30 minutes and it now reads 13.1V. I'm not sure if that's the DC-DC converter ramping down the voltage or if it has turned off and that's just the battery. I'll check it again in another 30 minutes.

After another 30 minutes, the battery is now at 12.8V. So I guess the DC-DC converter is doing what it's supposed to do?

Sponsored

 

[louibluey]

I'm always quick to reach for an instrument, but probably everything needed to watch the LVB and DC/DC converter is in FDRS, where a set of PIDS, some graphed can be monitored by FDRS flight recorder. I posted an example some months back (with tentative permission from a Ford rep to post a FDRS example) of the DC/DC converter reports. FDRS Newbie discovers Flight Recorder!

I will look through the battery management parts and see if I can find the right variables to monitor.

If I can keep FDRS and the laptop both awake long enough (not sure), we can probably watch the system running for some hours after MME goes into rest mode, and possibly starting up DC/DC and/or going to sleep.

 

[Mach-Lee]

I'm always quick to reach for an instrument, but probably everything needed to watch the LVB and DC/DC converter is in FDRS, where a set of PIDS, some graphed can be monitored by FDRS flight recorder. I posted an example some months back (with tentative permission from a Ford rep to post a FDRS example) of the DC/DC converter reports. FDRS Newbie discovers Flight Recorder!

I will look through the battery management parts and see if I can find the right variables to monitor.

If I can keep FDRS and the laptop both awake long enough (not sure), we can probably watch the system running for some hours after MME goes into rest mode, and possibly starting up DC/DC and/or going to sleep.

I've looked at FDRS, not much there other than the SoC and voltage/amperage parameters. When the car decides to run the DC/DC is still somewhat of a mystery. It seems that the BMS likes to keep the battery around 90% when powered on. Perhaps when powered off it allows the battery to drop to ~50% before recharging to 90%? (just a guess) There is also likely some delay before the BMS can properly sense the present SoC and react by turning on the DC/DC. So it's possible that the voltage would sit low for a few minutes before the system detects it and reacts.

Joe, did you happen to run a capacity test on the batteries? It's supposed to be 35 Ah but curious what it really is. You'd need your load to automatically shut off at around 10.5V so you don't overdischarge it.

You could also put a 2A constant load on the car with voltage recording, and see the behavior. 2A is too much for the BMS to sense SoC when the car is off, so that should show voltage triggers.

 

[louibluey]

I've looked at FDRS, not much there other than the SoC and voltage/amperage parameters. When the car decides to run the DC/DC is still somewhat of a mystery. It seems that the BMS likes to keep the battery around 90% when powered on. Perhaps when powered off it allows the battery to drop to ~50% before recharging to 90%? (just a guess) There is also likely some delay before the BMS can properly sense the present SoC and react by turning on the DC/DC. So it's possible that the voltage would sit low for a few minutes before the system detects it and reacts.

Joe, did you happen to run a capacity test on the batteries? It's supposed to be 35 Ah but curious what it really is. You'd need your load to automatically shut off at around 10.5V so you don't overdischarge it.

You could also put a 2A constant load on the car with voltage recording, and see the behavior. 2A is too much for the BMS to sense SoC when the car is off, so that should show voltage triggers.

Not yet, it was on the to do list.

Sure, I'm game. It's the kind of thing I would have done in LabView, but it has been some years. I don't even have it installed on a computer (last updated around 2012 when I dropped my update subscription). Should be easy to predict 10.5V on the bench, in MME with the wild card of the DC/DC converter coming on, probably won't go much below 12 V?

Interesting idea, put the active load on MME. I can do up to 5A easy (or, even 10A with two of the 6632B), any advantage / disadvantage to going higher? I guess we just monitor battery voltage with a DMM? That would show when and if the DC/DC converter comes on. I have a couple of different kinds of DCCT (for the DC/DC converter output cable if it's accessible at the LVB), but that would take a bit more effort to set up, although I used one recently, just have to find the calibration constants.

Different subject, this is my original LVB still powered at 13.65V on the bench ... check this out (25+ hours):

 

[louibluey]

... did you happen to run a capacity test on the batteries? It's supposed to be 35 Ah but curious what it really is. You'd need your load to automatically shut off at around 10.5V so you don't overdischarge it.

You could also put a 2A constant load on the car with voltage recording, and see the behavior. 2A is too much for the BMS to sense SoC when the car is off, so that should show voltage triggers.

LVB load test (MME resting, frunk light out) at 4A (active load, steady 12V battery drain at 4A) underway. It is cool, 45F, I am hoping for less than 35 Ah, so this does not take too many hours.

Two goals to this first test, check LVB measured Ah (at cool temps), and to note at what voltage the DC/DC converter comes on (if it does) in the MME idle/resting state (power off, no charge cord). As best as I understand, stopping at 10.5V both defines measured Ah, and should still be a high enough LVB voltage to just be full discharged (0% LVB soc), without damaging my LVB.

Set PS voltage of the active load to 10.2V with the voltage drop across the current leads, that should stop the test around 10.5V if I forget or miss the 10.5V point. When the DC/DC converter comes on, the hp will just continue to sink the extra 4 amps, MME should not care (the Voltage at the monitored jump points will go up to somewhere between about 13.4V and 15.0V. I think there is an error for unexpected continuous drain, so MME and/or FordPass might complain at some point (TBD). Or, if the DC/DC converter does come on before 10.5V, the Ah test is over (because at that point, MME will be charging the LVB from the traction battery).

hmm, just occurred to me, by using the jump points, I am bypassing the battery current sensor on the negative terminal of the battery. So probably MME will not notice the extra 4A, not sure if that will affect whether the DC/DC converter comes on or not?

Okay first result is in, did not take long at all! The LVB voltage started at 12.322V at 12:43 pm (about 12.13V at the bottom of that first curve). By 2:05:36 pm, 11.885V, the DC/DC converter came on to charge LVB and save the day! So for this first test, MME resting, no charge plug, MME off, frunk light off, about 45 degrees F, about 11.9V is the magic number for the converter to come on. The frunk light did come on, when the DC/DC converter started. See graphs below.

 

[louibluey]

... you may want to test the float current with your new battery vs. your current one to see if there are differences there.

Final report on the float test for the one year old LVB removed from MME. After some 35 hours on the bench powered at 13.65V, the current actually dropped to the same value as the new one, about 32 mA.

 

[Mach-Lee]

LVB load test (MME resting, frunk light out) at 4A (active load, steady 12V battery drain at 4A) underway. It is cool, 45F, I am hoping for less than 35 Ah, so this does not take too many hours.

Two goals to this first test, check LVB measured Ah (at cool temps), and to note at what voltage the DC/DC converter comes on (if it does) in the MME idle/resting state (power off, no charge cord). As best as I understand, stopping at 10.5V both defines measured Ah, and should still be a high enough LVB voltage to just be full discharged (0% LVB soc), without damaging my LVB.

Set PS voltage of the active load to 10.2V with the voltage drop across the current leads, that should stop the test around 10.5V if I forget or miss the 10.5V point. When the DC/DC converter comes on, the hp will just continue to sink the extra 4 amps, MME should not care (the Voltage at the monitored jump points will go up to somewhere between about 13.4V and 15.0V. I think there is an error for unexpected continuous drain, so MME and/or FordPass might complain at some point (TBD). Or, if the DC/DC converter does come on before 10.5V, the Ah test is over (because at that point, MME will be charging the LVB from the traction battery).

hmm, just occurred to me, by using the jump points, I am bypassing the battery current sensor on the negative terminal of the battery. So probably MME will not notice the extra 4A, not sure if that will affect whether the DC/DC converter comes on or not?

Okay first result is in, did not take long at all! The LVB voltage started at 12.322V at 12:43 pm (about 12.13V at the bottom of that first curve). By 2:05:36 pm, 11.885V, the DC/DC converter came on to charge LVB and save the day! So for this first test, MME resting, no charge plug, MME off, frunk light off, about 45 degrees F, about 11.9V is the magic number for the converter to come on. The frunk light did come on, when the DC/DC converter started. See graphs below.

Alright, so 1.37 hrs x 4A = only 5.5 Ah? Your discharge rate is approximately 0.1C, so I would estimate the battery was drained down to about 30% SoC before it turned on (assuming voltage reading was accurate). That would mean the total battery capacity is only about 8 Ah? Seems way too low.

It looks like you're using dual leads and measuring everything right at the jump points? Important that the voltage sense is directly on the terminal and not clipped onto another alligator carrying current? Double check that your voltage meter reading under load is very similar to a separate multimeter reading directly on the battery terminals.

The jump terminals are not bypassing the BMS, the current will still flow through the negative battery cable with the sensor.

Also interesting that it detects a severely discharged battery and goes all the way up to 15.1V. Does the graph continue so we can see charge times and termination behavior?

Maybe these batteries have an actual capacity that's much less than their rating, and that's causing some of the issues? I'd suggest a bench capacity test.

Final report on the float test for the one year old LVB removed from MME. After some 35 hours on the bench powered at 13.65V, the current actually dropped to the same value as the new one, about 32 mA.

That's consistent with what I've seen in terms of time it takes. This suggests that the used battery may be returned to a similar condition as new with the long term charging.

From those graphs it looks like your total circuit resistance is 0.35 Ω, which is very high for battery connections (that would cause 1.8V drop at 5A current). Personally I'd get rid of those gold plated terminals because it seems like they have too much of a voltage drop affecting results, I'd get standard lead ones without any shiny coatings, or use copper terminal clips directly on the posts.

 

[louibluey]

Alright, so 1.37 hrs x 4A = only 5.5 Ah? Your discharge rate is approximately 0.1C, so I would estimate the battery was drained down to about 30% SoC before it turned on (assuming voltage reading was accurate). That would mean the total battery capacity is only about 8 Ah? Seems way too low.

It looks like you're using dual leads and measuring everything right at the jump points? Important that the voltage sense is directly on the terminal and not clipped onto another alligator carrying current? Double check that your voltage meter reading under load is very similar to a separate multimeter reading directly on the battery terminals.

The jump terminals are not bypassing the BMS, the current will still flow through the negative battery cable with the sensor.

Also interesting that it detects a severely discharged battery and goes all the way up to 15.1V. Does the graph continue so we can see charge times and termination behavior?

Maybe these batteries have an actual capacity that's much less than their rating, and that's causing some of the issues? I'd suggest a bench capacity test.



That's consistent with what I've seen in terms of time it takes. This suggests that the used battery may be returned to a similar condition as new with the long term charging.

From those graphs it looks like your total circuit resistance is 0.35 Ω, which is very high for battery connections (that would cause 1.8V drop at 5A current). Personally I'd get rid of those gold plated terminals because it seems like they have too much of a voltage drop affecting results, I'd get standard lead ones without any shiny coatings, or use copper terminal clips directly on the posts.

Agree, it looks like a bench test for a good capacity test.

I will study your notes, some of high R might be that I am using regular test leads and Pomona banana cords for the current path (150 milliohms or so both ways), but the clips are on the test points in different locations on the lugs (not clip on clip). It should be a good 4 wire Kelvin measurement at the battery (or jump points, but with relatively large drops between the battery and power supply), but maybe I missed something.

I see 15V often from the DC/DC converter since the weather turned cool/cold. For example, when programming I set the PL6100 to 13.4V, and often see the converter come on to before 14.4V, but more recently 15.0V or 15.1V. Could be battery soc, but I'm starting to think they may factor in ambient temps too. Since this discussion, I am setting the PL6100 to 13.7V, I guess every little of bit of time helps.

Thank you for taking the time to look over the data, and for the pointers. I am terribly rusty, and never really knew much about batteries.

 

[Peaceowl]

Thank you for your time and info.

 

[louibluey]

Just a brief look at the terminals. I am no big defender of them, I find the clear coat over gold plate offensive, not to mention the thin layer of copper between the thin layer of gold like coating and whatever silver colored mystery alloy makes the bulk of them. That said, I used a dremel tool to lightly grind all of the conducting surfaces, and a thin coat of dow corning d4 dielectric grease to prevent corrosion. Rough measurements with a milli ohmmeter from one interior side to a front connection was less than 1 milli ohm.

After I finished the 13.7V run on the original battery on the bench, after some hours of resting, I measured it with the KW710 directly on the battery terminals. It occurred to me while we have no idea what the KW710 absolute accuracy is, it could be a quick check for the terminals, so I bolted in some short sections of freshly cleaned 1/4" copper pipe and remeasured at the copper terminals. The first result directly clamped to the battery terminals was 5.57 milliohms, just now with the terminals installed and placing the clamps on the copper pipe, I get 5.64 milliohms.

So as above, I am no huge fan of these terminals, but I think the resistance they add for this testing is fine, probably well under 1 milliohm.

They are convenient, because of the variety of hole sizes and set screws. Plus whatever that mystery metal is, there is a lot of it, they are substantial devices with some bulk to them.

I do think the relatively light duty clips I used on the jump points today were less than ideal, but, I had 4.00A indicated the whole time and no glitches in the voltage measurements, so I think the data is good. I have not yet reviewed the short Ah data, that might make less sense.

Next step is probably an uninterrupted ah test on the bench, that might tell us more, just wanted to rule out a problem with the terminals before proceeding, probably tomorrow with the Ah test at possibly 2A, 3.5A, maybe higher. The old hp makes some pretty good noise at 4A, so the lab/office will be noisy for some hours if the actual Ah is anywhere near 35 Ah.

I am a little unclear on the unloaded voltage to soc curve and measuring voltages during the discharge test, but we can probably sort that out later. Hopefully, after 10.5V is reached in circuit, I can remember to get an open circuit LVB reading too.

 

[louibluey]

MME 12V LVB ampere hour (Ah) test underway at 2A. Should know something by tomorrow midday, will update this post. The Pomona 2948 series patch cords are 2 ft #18 wire (used to do precision measurements work years back, they have beryllium copper springs, brass body, all gold plated, pretty nice). I doubled up the current legs, but ampacity should be a non issue to 5A (max sink for this active load, although I could double the load up to to 10A). Moved this testing to the basement, so I don't have listen to the active load fan for hours.

I have an inverter with #2 cables and Andrews fork lift connectors from inverter work I was doing with Chevy Volt (similar to what @dtbaker61 has been working on for MME), so later could possibly test the LVB at some relatively high discharge currents too.

Update: The 2A test was setup for 20 hours, naively thinking the 35 Ah battery would run to 0% soc at 2A, in about 17 hours. This battery had a 13.1V open circuit Voltage before the test. The battery terminal voltage curve has been a virtual straight line since about 12.652V (loaded) at about 7:13pm Sunday evening, to this Monday morning, 11.686V at 9:57am. That is a change of -.966V over 15.2 hours. At this discharge slope, the test will end well above 10.5V in 20 hours early afternoon. About 70F in the basement. hmm, maybe a recharge, rest, and a run at 10A next. Note that the battery terminal voltage is considered, the power supply just makes whatever voltage it needs to provide, to keep 2.001A flowing through the battery.

Update: 2A test complete, one year old AGM taken out of MME performed fine. After 20 hours, my original one year old LVB was still making 2A. The voltage was down to 11.16V. I know we need 10.5V to the get the official reading, but based on the above, at least at 2A, 70F, the LVB is okay (40 Ah already produced). No issues with LVB capacity at 2A. Monitoring open circuit recovery a bit, then going to recharge, and run again at 10A.

Open circuit recovery after disconnected from 2A active load at 11.16V (all data 1 second / point)

hmm, if the recovered voltage from 11.2V is 11.4V, and this graph says 11.55 open circuit is 0% soc, maybe 10.5V is too far for AGM testing? Maybe 11.35V under discharge load?

KW710 on a discharged LVB

Setting up for the 10A test, and recharging at 20A (went from 10A to 20A after pic)

 

[louibluey]

The 10A load test is complete. My one year old LVB, despite all of the minor problems MME has had properly maintaining it from time to time (including the 12V LVB Sync4 and FordPass warnings to get 12V service), is fine. Note that I did Recondition/Service my 12V Battery before all of this testing as per @Mach-Lee 's instructions, which probably helped the one year old LVB as well.

The one year old LVB delivered 10A for 3.5 hours without problem (35 Ah, as rated). The small glitch early on is where I decided to double check 10A total by adding the ammeter DMM.
(Set acquisition to 5 hours, stopped at 1:30 remaining after this screen shot)

Open circuit recovery after load test

 

[Mach-Lee]

The 10A load test is complete. My one year old LVB, despite all of the minor problems MME has had properly maintaining it from time to time (including the 12V LVB Sync4 and FordPass warnings to get 12V service), is fine. Note that I did Recondition/Service my 12V Battery before all of this testing as per @Mach-Lee 's instructions, which probably helped the one year old LVB as well.

The one year old LVB delivered 10A for 3.5 hours without problem (35 Ah, as rated). The small glitch early on is where I decided to double check 10A total by adding the ammeter DMM.

Open circuit recovery after load test

Wow, it did better that rated! The industry spec is to give amp hours at the 20-hour rate. Based on your testing the battery could be rated at 40 Ah! Testing at a higher amperage will reduce the observed capacity due to Peukert's Law. Based on your two tests I can estimate the Peukert constant is 1.08, which is really good for an AGM (an ideal battery would have a constant of 1.00/unity).

These results would seem to suggest recharging the battery for 36+ hours successfully rejuvenated it. I wonder how many batteries replaced under warranty were actually "bad" and how many just needed a weekend charge? Any remaining issues are likely due to poor charge management by the Mach-E BMS, which may have been resolved with the latest software updates.

The 10.5V vs. 11.0V discharge termination is sort of a judgement call. If discharging at high amps I usually use 10.5 due to the Peukert effects, but at slow discharge (5+ hours) 11.0V is the correct termination for AGMs. It somewhat depends on the shape of the curve, you want to stop it after it starts turning down but before it "falls off the cliff".

 

[generaltso]

After another 30 minutes, the battery is now at 12.8V. So I guess the DC-DC converter is doing what it's supposed to do?

Well, I got the 12V Battery Fault message again today. I think I've done all I can do at this point. I guess I'll ask my dealer if they're ready for the glass recall yet so I can have them do both at once. Of course, it's probably unlikely that they'll order a new battery for me without looking at the car first.

 

[louibluey]

... Does the graph continue so we can see charge times and termination behavior? ...

Ran another discharge test at 5A today (click on graph for larger version), and let it run for two charging cycles. Power off, MME at rest, active load connected to the 12V jump points (as you noted 12V gnd sensor in circuit), used an OBD "power cord" for 12V measurement (5A is not flowing through any of the voltage sense path, the short red/black fused wires to a dual banana plug to the DMM driver's floor, LXI bus, ethernet bridge back to the computer in the lab). Horizontal axis is hours. MME was not plugged in during this test.

The negative 0.3V spikes are not glitches. When MME wakes to start the DC/DC converter, apparently a number of loads come on.

Not sure if 11.9V is optimal for the low end voltage, however what is clear is that MME is reliably identifying the need to charge, then charging the LVB. Unclear why some are still have 12V LVB low voltage problems. I am running the latest MME (FDRS) software as of November. Turned off the 5A active load at the very end right side.

Sponsored