Very infrequent but confusing OVLK issue

[Trydan Am Ddim]

[My first post, been gaining excellent knowledge from this group for many years. However, I had not checked the group since last year and was VERY excited to hear about Dacian new firmware which I plan to try soon. It inspired me to ask the following question which has puzzled me for me several years…]

I’ve used the truly excellent SBMS0 in my camper van, controlling 8 280Ah LiFePO4 in series for a nominal 24V battery, for about 5 years. During that period, I’ve had a small number of occasions when the OverVoltage Lock (OVLK) was triggered, but not the OV. See the attached photos with the Stat value of 24578.

My configuration values for OV, OVLK & OV recovery are slightly different than the defaults; they are set at 3.55, 3.75 & 3.4V resp. Control of my charge source is via a single DSSR20 (diversion part is not used) using EXT4 set to 1. While the OV flag did not seem to be triggered, the DSSR20 did switch off the charging source as seen in the second photo, charging had stopped about 8hrs earlier. I can’t say if this was due to the CFET flag being low or from processing of the ERROR5 variable in the SBMS firmware. Given that charging stopped, I’ve ruled out any issue with the charging controller and its associated wiring. The low, less than 60W, charging occurred during sunrise with partial sun breaking through nearby trees. The cells would have been around 99% full at the time from the previous day’s charging.

The third photo shows that the individual cells have a voltage of 3.416V, or less. Despite being over 8 hrs since the charging was stopped, I’m surprised that if one of the cells had reached the OVLK of 3.75V it would have already been discharged to <= 3.416 given that the only load on the 24v 280Ah cells was the tiny load of the SBMS (+WiFi).

Wondering how the OVLK could happen without an OV. The OV flag should still be set since if there was an OVLK there must have been an OV and the condition for OV recovery has not been reached.

I’ve been scanning the data sheet for the ISL94203 battery pack monitor chip and the SBMSV51d source code to try and figure out why. Does anyone have any thoughts?

 

[Dacian Todea (electrodacus)]

Short answer is that the issue is with one or more of your cell balancing wires.

The OV is trigered when any cell exceeds 3.55V for more than the number of seconds set at Overvoltage delay (typical 6 seconds). This delay needs to be larger than cell balance ON setting since that way false OV is avoided as high cell voltage reading while cell is being balanced is ON for just 3s (default setting).

The OVLK has no time delay it will react instant since in a correct install that will never happen.

So you either have longer than standard cell balancing wires thus higher resistance and higher voltage drop or there is a bad connection (very common for crimped not soldered cell sense wires) or you have a bad power connection between cells and in that case it has nothing to do with cell balancing wires. In that case this will happen when you have high charge current or UVLK for high discharge currents.

The new firmware has tests for both of this conditions one that measures the cell balance resistance and one that measures cell's internal resistance and that includes the buss bars and terminal contact resistance.

[Trydan Am Ddim]

Thanks for the quick reply. All the extra test you have in the new firmware to check for the internal cell resistance, cell balance resistance etc, are somethings I’m eager to try once my ST-Link programmer arrives.

I’m thinking that my inter-cell battery connections are not an issue since the OVLK has appeared during low, approx. 2-3A charging currents. Additionally, when drawing high power for periods of 20+ mins, e.g. near 3,000W loads via a Giandel PS-3000KAR inverter (I’m using a simple external pulse circuit combined with the EXT3 type 2 signal to let the SMBS remotely turn off the inverter if necessary), I have not noticed significant difference between cell voltages. I do periodically check these cell connections with an Infrared Laser Thermometer and recently a thermal camera, while drawing a large load, as the connections have previously loosened on road trips.

While I’m using the ribbon cable length that came with the unit and did solder the ring terminals I use at the ends of the cable, there could be the possibility that there is a connection which might be subject to rare “extra resistance” which could trigger an OVLK condition. From your feedback this seems to be the culprit, but the rarity of these events might make locating it tricky.  The new tests in your updated software will be very helpful.

I did not realize that during cell-balancing periods that an OV condition could potentially be triggered. Due to the apparent “wild” voltage changes that sometimes occur during the balance period, I just wrongly assumed that the ISL94203 knew it was in a balancing period and though voltage values were reported they were not checked against any thresholds. As you mentioned, specifying a long enough value for the Overvoltage delay, relative to the balancing “on” time, makes it simple to mask this issue for OV. However, your comment about the instant reaction of the ISL94203 to a cell exceeding the OVLK threshold makes me realized that this threshold cannot be “masked” by a delay and the voltage changes I'm seeing during the balance period could really be telling me something important about the condition of my balance wire connections. Thanks for this insight.

[Dacian Todea (electrodacus)]

The only way you can do to mitigate the issue until you find and solve the main issue is to increase the OVLK threshold to maybe  3.85V. As it happens just occasionally it means it barely gets to 3.75V peaks.

[Trydan Am Ddim]

Within just a few minutes of using the new 6.1c firmware I’m noticing things that I had not seen before. Towards the end of charging, during the cell balancing period I’m noticing peak cell voltage of over 3.6V and colour coded in the histogram!  It is much easier to see these voltage peak values compared to the previous firmware and I now see how close these can get to the OVLO threshold.  This new insight was just from the basic cell monitoring page. I haven’t yet had time to experiment with all your other new features.  

I had previously examined the v51d source code and had seriously consider rewriting a small part of the code adding a user option to select not to show the changing voltages during the cell balancing periods.  The lack of an ST-Link programmer immediately on hand and the slight possibility of bricking(!) my SBMS0, dissuaded me from the task. I now have an ST-Link programmer but I now also have more insight that the temporary voltage readings whilst balance can be illuminating.   

[Dacian Todea (electrodacus)]

The behavior is the same is just that in the new version the cell voltage bar graph is color coded so it is much easier to spot when cells get above thresholds.

It is normal at the end of charge that cells due to voltage drop on cell sense wires will exceed 3.55V (default) limit but they should not get anywhere close to 3.75V.

Worse case with included cable you get 50mV drop per wire so around 100mV is worse case for a cell that is between two balancing cells that will peak at 3.65V still 100mV away from 3.75V but if one or more of the cell sense wires has high resistance due to bad contact or longer than standard wires then it can exceed 3.75V OVLK limit.

I also considered hiding the voltage drops due to cell balancing but that will provide no feedback to user that cell balancing is performed and no diagnostic data. So I decided not to mask or hide it.

[Dave McCampbell]

Trydan.  I don't know which cells or terminals you have, but when I changed out my old cells with female screw terminals, several that I had damaged by over torquing by hand tightening, I ordered stud terminals on the new EVE cells.  At first I just wiped the terminals and connectors clean and hand tightened them to what I thought was sufficient.  

Then after reading about proper cleaning and torquing of cell terminals, I carefully cleaned all the terminals and connectors with alcohol and torqued them to EVE's recommended torque.  My new connectors also had one small threaded female terminal mid connector where I could connect the SBMS0 sense wires without interfering with the cell terminal connections.  I have not soldered the sense wire ring terminals, yet.  Now I can get the balance among all 4 cell sets to within 1mv.  This seems sufficient.

Maybe Dacian can comment on his recommendations for how he does his cleaning and torquing of the cell connectors and other connections.  Hopefully this will help you with your OVLK problem.
Dave

[Dacian Todea (electrodacus)]

I installed my EV cells sometime in 2022 at the spec torque and never checked the connections other than using the new firmware as a test. The connections are excellent as they where when initially installed.

But my battery is stationary installed in my house so no vibrations or thermal cycling with fairly stable ambient temperature in the 18C to 27C range.

My cells had M6 tapped holes as you can see in photo and I got bolts of exact length so they almost get to the bottom but not quite. Depending on connection I needed to add an extra washer and I used aluminum lugs where I connected flexible silicone wires for the series connection between cells and paralleled with the included flexible buss bars.

Is easy to check connections even without the new SBMS0 firmware by just using a multi meter set a constant high load  or during the day while charging in full sun just measure the voltage drop across the connections. If the are equal then there is no connection problem.

I don't remember exactly but I may have used a bit of Vaseline (petroleum jelly) on the top of the terminals before making the connection to prevent air from oxidizing the surface. Torquing to spec was maybe fairly close to the limit of this connectors but it ensures the connection will remain perfect for many years.

[Trydan Am Ddim]

Dave,

Your comments about the cell terminal likely point to the cause of my very infrequent problem and matches Dacian’s feedback regarding the cell balancing wires. My EVE cells originally had female screws terminals like yours. I had two problems in attaching the connection between cells and the balancing wires – i) not to overtighten the male screws that hold the connectors and ii) having the correct amount of thread depth on the screws. At the time I had a torque wrench but not a finer setting torque screwdriver and was very concerned about going too far when tightening the screws. However, I was pretty sure that my inter-cell connections were good.

My Fluke 189 multimeter has excellent resolution, down to 0.001mV in the 50mV range so in theory I can measure really tiny voltages and initially I did a constant high load test similar to what Dacian mentioned, checking the voltage drops across my flexible buss bars. This test coupled with checking temperatures across the terminals with a 100A load running for 20 mins made me think that inter-cell connections were fine. I did not pay as much attention to the cell balancing wires – I wrongly thought that e.g. if the ring terminals were slightly tarnished having been kicking around in a spart parts draw for a couple of decades it would not matter to the voltage measurements. I realized that during balancing that I might not be getting a "full" balance current but that was no big deal. What I did not realize is that the voltages measured during the balancing period, when mA currents are flowing in the balancing wires, are used by the ISL94203 and compared against the various programmed threshold to determine any fault conditions. I wrongly assumed that as these measurements were not the “real” cell voltages, they were not important. Dacian corrected my false thinking earlier.

I should have emphasized how rare this OVLK event is. It has happened about 6 times between 2021 and 2023. But has not happened since 2023. What happened in 2023…

I added M6 x 20mm hex socket grub screws to the 16 terminals to my cells. This made attaching the inter cell connections much easier as I now used a washer and nut rather than as screw and I felt I could better torque the connection and the screw depth issue was fixed. In the process of doing this upgrade, I removed and reattached my sense cables. Don’t remember specifically cleaning them but the reattachment might have removed some extra resistance that could have been on there.

Since that time, I have not had the OVLK issue but was constantly thinking that it was a lurking problem as when it happens all loads are turned off.  Very annoying to return to your camper van to find both your 12v refrigerators and freezers turned off for hours despite your battery being fully charged. In the original source code for the firmware, I saw where the loads where shut off when an OVLK occurred and had contemplated doing a small “fix” to the firmware; possible to do this fix as this logic is in the firmware and not within the ISL94203. I decided against messing with it in case I bricked my SMS0. With Dacian’s new firmware I knew I would be messing with the SBMS0, so many must have features!! so I raised the question of how OVLK can happen but no OV. I had not considered the possibility of previously poor balancing wires causing this. If I had raised this question back in 2022 I would have been enlightened by Dacian much earlier.

[Dave McCampbell]

Trydan.  With my previous RJ cells I did the same by adding a grub stud in order to make it easier to torque the connecting bars without terminal damage.  I did previously damage a couple of them but was able to repair them by using a 1/4" grub screw in the 6mm holes.  This is a fix I found on another group and it works well.  The new EVE cells, like Dacian's, are much better quality and come with welded studs so that I can easily torque nuts on them.  Dacian's cell connector scheme is a third option that allows use of the flexible connectors.  There are many small electronic torque wrenches available on the internet that work well for this.

I have also read that, unlike Dacian's home setup, it is important to recheck torqued nuts at least once a year on boats and campers due to vibration issues.  I will be doing that in a few months, but so far no indication of a problem with my LFP battery.

Dacian.  I'm curious as to why you used the silicone wire instead of flat Cu connectors to connect your cell groups?  I really like your compression scheme.

Dave

[Dacian Todea (electrodacus)]

Dave,

Cells expand and contract with the state of charge so connection between cells needs to be flexible in order to allow for this small amount of expansion/contraction else the stress will be on connection points as something will need to move.

The silicone wires I had around and are less expensive than flexible buss bars.

[Trydan Am Ddim]

Dave, I’ve learnt from experience that I need to check my cell connections after long trips in the camper van, especially after travelling on washboard patterned gravel roads. One way I do this is by applying a large, >50A, load, and check the cell connectors’ temperatures via an IR temperature gun.  A quick check after about 1-2 minutes of load, and if all is well check again about 15-20 mins later. Usually, I do this if I need to run my small 120VAC water heater in the campervan.  I recently brought a thermal camera on the pretext that I “needed” it to check these connectors, but the cheap temperature gun works well.

When I got my EVE cells, I was amazed about the connection mechanism via the not particularly long 6mm diameter female screw threads into the soft battery terminals. It seemed inadequate considering the current that could be drawn; I went for a 24v system to halve the currents. As you mentioned, the newer cells have much better connection mechanisms.

I ended up making my own flexible connectors between cells, partially because I had a large quantity of Belden ground braid on hand. I took sections of this and bent it over itself twice, to get a length with quadruple the current capacity, and placed on each end a section of ½” copper pipe. I then used a large vice to “crimp” each end until it was flat. They are similar to the ones in Dacian photo, but have a distinctly more “DIY” look to them. Because of the DIY nature of these connectors, I initially spent quite some time checking for voltage drops around them, replacing a few of the earlier ones I made as I refined my technique.

Spent all the time on these but completely ignored the quality of my cell balancing connections!!   

[Dacian Todea (electrodacus)]

Trydan,

You can not use copper on aluminium. The cell terminal are aluminium and you need plated copper to make the connection. The copper pipe is for sure making you problems as there is a reaction between copper and aluminium.  

The lugs connectors I used are also aluminium but they have a thick tin plating so aluminium is not in direct contact with copper. 

[Trydan Am Ddim]

Thanks for pointing that (Cu on Al) out. I’ve been lucky over the past 5 years in that significant corrosion has not set in. Perhaps I’ve been helped by the fact that I live in a semi-arid climate but obviously I can’t completely defy the rules of physical chemistry. I’m likely to do some battery maintenance later this year and replace my cell connectors with tinned copper lugs.