Active balancer, make it smart!?

[Vi s]

You mentioned earlier that the switch on voltage for the balancer was measured to be 3.43 Volts.
Assumption:
The balancer checks only the voltage of cell 1. In your test cell 1 may never have exceeded 3.43 V, which explains, that the balancer remains off.

Your assumption was spot on!! ?
After the cell 1 hit 3.44V the balancer started. After that i stopped charging the first cell to see when it stops again. Right now cell 1 is at 3.40V and the balancer still works. I guess it stops once it has reached the end of balancing. Let's see.

 

[Vi s]

Yes you are right.
But the problem with all those flying capacitor balancers is, they can only move charge and balance between two adjacent cells.

If for example there is a string of 16 cells, and cell 1 is low, cells 2 to 15 are perfect and cell 16 is very high.
Cell 16 cannot transfer charge into cell 1 directly.

Cell 16 starts to increase cell 15 and cell 1 draws charge away from cell 2.
Gradually 15 starts to bleed charge into 14, and at the other end of the stack a similar thing happens with cell 3.

It can take a very long time before everything settles down.
It would be nice if a central "brain" could see that cell 16 is high and cell 1 is low and shift some charge directly from 16 directly into 1.
But that is not how any of the cheap flying capacitor or ringing inductor balancers work.

The inductor balancers might work like that (adjacent cells) but not the flying capacitor ones.

 

[Warpspeed]

What you say is quite true, as long as your battery only has eight cells.
Any more than that and you have to overlap two or more balancer boards across a central common cell, and the simplicity then breaks down.

 

[Vi s]

What you say is quite true, as long as your battery only has eight cells.
Any more than that and you have to overlap two or more balancer boards across a central common cell, and the simplicity then breaks down.

I don't know about the 16s ones (i don't understand why they would have to overlap two or more balancer boards across a central common cell), you might be right about that one. Maybe @BradCagle could chip in some thoughts about that.

I noticed one thing, cell 16 (B16) was the highest (minimum 20% more capacity than the lowest cell) after letting the active balancer stay connected 24/7 over a couple of months. In Andy's video it was the same, cell 16 was the runner cell. Not sure how that phenomenon is created.

 

[Vi s]

Updates on testing HANKZOR's customized active balancer:

The kind customer agent i was always referring to was actually the very well known Nami many know here already. She does really a great job, they must be proud to have her!

So this is the information Nami supplied me with so far (she said after the weekend she can reply more accurately):
Only the voltage of B1 (cell 1) is measured. These are the specs of the chip which is doing that.

She wrote:
"Sorry, the engineer checked the resistance of the sample and found that the resistance used by this sample is 3.3V. As long as B1 is lower than 3.3V, it will stop. The voltage detection chip only detects the voltage of B1. When B1 is lower than 3.3V, it will stop, and when B1 is higher than 3.465v, it will restart. 3.3*1.05=3.465 startup"

As well:
"It can be made more accurately and needs to increase the cost"

I checked and according to my new JBD BMS and pretty accurate RC3563 voltmeter which seconds the BMS measurement to the mV, my customized active balancer starts if B1 is at 3.456V and stops again if B1 reaches 3.323V.

 

[BradCagle]

I don't know about the 16s ones (i don't understand why they would have to overlap two or more balancer boards across a central common cell), you might be right about that one. Maybe @BradCagle could chip in some thoughts about that.

I noticed one thing, cell 16 (B16) was the highest (minimum 20% more capacity than the lowest cell) after letting the active balancer stay connected 24/7 over a couple of months. In Andy's video it was the same, cell 16 was the runner cell. Not sure how that phenomenon is created.

I haven't examined the 16s, only 8s, and 4s. But I couldn't image they would do something weird like overlapping. I mean all they have to do is share a common parallel buss, and the switching signal to be in sync across both sides. Looking at the pictures of the 16s the two halves are back to back right where the parallel bus is. Stands to reason they would be connected there. Of course take that with a grain of salt, just speculation.

 

[BradCagle]

Actually may be possible to link two 8s (or 4s) balancers together at the parallel buses. Now the switching will be out of sync, but might could add an extra (larger) cap to compensate.

 

[Warpspeed]

That should certainly be possible, but the pictures I see of these balancers does not seem to offer that option.

One other issue of directly linking together (and synchronizing the switching) would be the voltage rating of the mosfets.

 

[Samcat]

Updates on testing HANKZOR's customized active balancer:

The kind customer agent i was always referring to was actually the very well known Nami many know here already. She does really a great job, they must be proud to have her!

So this is the information Nami supplied me with so far (she said after the weekend she can reply more accurately):
Only the voltage of B1 (cell 1) is measured. These are the specs of the chip which is doing that.
View attachment 99038
She wrote:
"Sorry, the engineer checked the resistance of the sample and found that the resistance used by this sample is 3.3V. As long as B1 is lower than 3.3V, it will stop. The voltage detection chip only detects the voltage of B1. When B1 is lower than 3.3V, it will stop, and when B1 is higher than 3.465v, it will restart. 3.3*1.05=3.465 startup"

As well:
"It can be made more accurately and needs to increase the cost"

I checked and according to my new JBD BMS and pretty accurate RC3563 voltmeter which seconds the BMS measurement to the mV, my customized active balancer starts if B1 is at 3.456V and stops again if B1 reaches 3.323V.

0.165 V is not bad......if you shooting for 3.4 V for cells max V that will bring you in worst case scenario up to 3.565 . Definitely interested in more details. More tight ( accurate) deltaV trigger will be better choice.

 

[BradCagle]

That should certainly be possible, but the pictures I see of these balancers does not seem to offer that option.

One other issue of directly linking together (and synchronizing the switching) would be the voltage rating of the mosfets.

The parallel buss on these things never see voltage higher than the average of all cells so max 3.65 (or 4.2 for li-ion) Therefore the mosfets never see a voltage higher than that, even if they were linked. The parallel bus is on at the rear of the balancer (or middle for the 16s). You can scratch the enamel down to the traces, and solder wires there.

Again just speculating about what can be done through modding. I have attached to the parallel bus on these and can read the average voltage there.

 

[Hans Kroeger]

Yes you are right.
But the problem with all those flying capacitor balancers is, they can only move charge and balance between two adjacent cells.

If for example there is a string of 16 cells, and cell 1 is low, cells 2 to 15 are perfect and cell 16 is very high.
Cell 16 cannot transfer charge into cell 1 directly.

This is true for many "flying capacitors" Balancer, but does not apply to the concerned Hangzu / Heltec Balancer.
See the above video which explains how these Balancers work. I have tested in detail quite a number of these Balancers for 12 V and 24 Volts.
My conclusion: the best Balancer available for the price,.......but only, if you turn it off below some 3. 35 V, and turn it on above 3.37 V.

 

[Hans Kroeger]

Updates on testing HANKZOR's customized active balancer:

The kind customer agent i was always referring to was actually the very well known Nami many know here already. She does really a great job, they must be proud to have her!

So this is the information Nami supplied me with so far (she said after the weekend she can reply more accurately):
Only the voltage of B1 (cell 1) is measured. These are the specs of the chip which is doing that.
View attachment 99038
She wrote:
"Sorry, the engineer checked the resistance of the sample and found that the resistance used by this sample is 3.3V. As long as B1 is lower than 3.3V, it will stop. The voltage detection chip only detects the voltage of B1. When B1 is lower than 3.3V, it will stop, and when B1 is higher than 3.465v, it will restart. 3.3*1.05=3.465 startup"

As well:
"It can be made more accurately and needs to increase the cost"

I checked and according to my new JBD BMS and pretty accurate RC3563 voltmeter which seconds the BMS measurement to the mV, my customized active balancer starts if B1 is at 3.456V and stops again if B1 reaches 3.323V.

Good job! Thank you for sharing this information.
I will think about which voltage level would be the best compromise, taking into account the a.m. hysteresis.
Nevertheless one drawback is the fact that it is only cell 1 being monitored. This is no problem, once the cells are balanced. But one objective of a Balancer is to accept unmatched / unbalanced cells and bring them to a balanced status.

 

[Vi s]

Good job! Thank you for sharing this information.
I will think about which voltage level would be the best compromise, taking into account the a.m. hysteresis.
Nevertheless one drawback is the fact that it is only cell 1 being monitored. This is no problem, once the cells are balanced. But one objective of a Balancer is to accept unmatched / unbalanced cells and bring them to a balanced status.

Welcome, my pleasure.

I also asked about why only B1 is measured and she answered:
"Measuring the voltage requires self consumption of power, which not only needs to consider the power consumption, but also the cost. The balancer is cheap, so no more changes will be made"

Yes for an unmatched pack one will have to take the Arduino road or buy more expensive equiqement. These relay boards with an integrated esp32 module cost just 4 usd. Hardware is cheap and no issue, there is just no code available for that yet. Should be rather easy to do for someone who can code.

 

[Warpspeed]

The parallel buss on these things never see voltage higher than the average of all cells so max 3.65 (or 4.2 for li-ion) Therefore the mosfets never see a voltage higher than that,

The parallel bus certainly never sees more than one cell voltage ACROSS THE BUS. And neither do the capacitors.

But what if you have a 100v battery string of thirty cells?
There will then be a 100v difference between the first lowest cell, and the thirtieth highest cell.
The mosfets that are off might then see up to 100v across them.

Its not a big deal for 12v to 48v batteries, but there will be a definite limit to how many of these you may be able to stack and link up together.

 

[Vi s]

After some more testing i can say i am very pleased with it. It balances as good and fast as my stock one and on top of that it can't mess up my top balance anymore.


It goes to ? latest at 3.311V (3.323V to 3.311V). There is only 4-5% of capacity at 0.26C charge and discharge between 3.456V and 3.311V (tested multiple cycles). So as soon as it starts balancing (3.345V to 3.346V) the balancer has a minimum of 4% of capacity to do its job before going into ?mode again.

In my case most of the time it will by in sleep mode (which i like) anyway since i don't go over 3.45V often. I will only charge higher if i see that my battery needs a new top balance. Then the balancer will be ready help with that in a convenient way. So that fits my battery usage very well.

If i forgot something you are interested in let me know.

 

[Hans Kroeger]

After some more testing i can say i am very pleased with it. It balances as good and fast as my stock one and on top of that it can't mess up my top balance anymore.
View attachment 99056

It goes to ? latest at 3.311V (3.323V to 3.311V). There is only 4-5% of capacity at 0.26C charge and discharge between 3.456V and 3.311V (tested multiple cycles). So as soon as it starts balancing (3.345V to 3.346V)the balancer has a minimum of 4% of capacity to do its job before going into ?mode again.

In my case most of the time it will by in sleep mode (which i like) anyway since i don't go over 3.45V often. I will only charge higher if i see that my battery needs a new top balance. Then the balancer will be ready help with that in a convenient way. So that fits my battery usage very well.

If i forgot something you are interested in let me know.

It fits your needs, that is good news.
For my usage the switch on voltage level is a bit too high. The switch off voltage would be OK.
Just a note concerning your statement:
"the balancer has a minimum of 4% of capacity to do its job before going into ?mode again."
The "job of the Balancer" has nothing to do with the 4% of capacity. The Balancer just moves charge between cells until cell voltages are equal. The current which moves charge is proportional to the difference between the individual cell voltage and the average value of all cells. The higher the capacity of the cell is, the more time the Balancer will need to do its job. But this is all independent of the amount of charge (4%) to be moved between the switching points of the Balancer. The Balancer doesn't know anything about the charging or discharging process. It simply reads voltages.....

 

[BradCagle]

The parallel bus certainly never sees more than one cell voltage ACROSS THE BUS. And neither do the capacitors.

But what if you have a 100v battery string of thirty cells?
There will then be a 100v difference between the first lowest cell, and the thirtieth highest cell.\

None of the mosfets see more than a single cell voltage no matter how many cells are strung up.

The mosfets that are off might then see up to 100v across them.

I assume you're talking about the series mosfets. Mosfets in series the voltage rating is additive as long as they are all switched at the same time.
If you have 30 10v mosfets in series, and they are switch the same time you have 300v of maximum drain-source blocking capability.

 

[Hans Kroeger]

The parallel bus certainly never sees more than one cell voltage ACROSS THE BUS. And neither do the capacitors.

But what if you have a 100v battery string of thirty cells?
There will then be a 100v difference between the first lowest cell, and the thirtieth highest cell.
The mosfets that are off might then see up to 100v across them.

Its not a big deal for 12v to 48v batteries, but there will be a definite limit to how many of these you may be able to stack and link up together.

Mosfets can easily handle drain to source voltages of 100 Volts and more.....

 

[Warpspeed]

Not if they have a lower voltage rating they won't.

 

[Vi s]

It fits your needs, that is good news.
For my usage the switch on voltage level is a bit too high. The switch off voltage would be OK.
Just a note concerning your statement:
"the balancer has a minimum of 4% of capacity to do its job before going into ?mode again."
The "job of the Balancer" has nothing to do with the 4% of capacity. The Balancer just moves charge between cells until cell voltages are equal. The current which moves charge is proportional to the difference between the individual cell voltage and the average value of all cells. The higher the capacity of the cell is, the more time the Balancer will need to do its job. But this is all independent of the amount of charge (4%) to be moved between the switching points of the Balancer. The Balancer doesn't know anything about the charging or discharging process. It simply reads voltages.....

Thanks, yes i know the job of a balancer has nothing to do with capacity but concerns only voltage.

Why i measured and mentioned it has following reason.
As you know already for especially lifepo4s voltage and capacity doesn't behave linear. A difference of 56mV (3.4V to 3.456V = 56mV) has a different significance depending where you are on the charging curve (it is 50%capacity in the flat curve area according to my testing). So i wanted to put the deviated (from what i ordered and was confirmed before ordering) switch ON and OFF voltage points in perspective. Is the difference from 3.4V (or 3.43V since that is the number they tested in their lab and shared with me before ordering) to 3.456V significant or not was my question. Will it be an issue since the deviation is higher than the 50mV they promised before? 3.43V (what they said they tested in the lab) would have been within the promised 50mV tolerance. 3.456V (56mV) was already outside but let's say that is still fine because my voltmeter has an 0.5% error at a scale of 0.001V. 3.311V (89mV = 78% higher than 50mV) was way out of tolerance. So after testing of how much difference it makes in relation to capacity i thought, well never mind, i can live with a +-2%capacity (-89mV 3.4V +56mV) tolerance (it will be of course more or less if i would reduce or increase the charge or discharge rating), not worth the trouble of complaining. Overall, although not perfect, this customized balancer is already way better, more convenient for me than the stock version. Of course i would also have preferred if it would have switched on at 3.4v. 3.456V is indeed borderlining but still usable since i know it and can adapt to it.
So I just shared the numbers of whatever i tested, the 4-5%capacity (in between 3.311V and 3.456V) was just an information for those who also have questions like me above.

Sure everyone has a different system, different charging end points, different demands etc so i understand judging from your final charge voltage of 13.9V, 3.45V is too high since there could be instances where the balancer wouldn't even turn on which would render it a useless accessory.

After all one has to keep in mind, this customized version is their first try, a prototype, maybe they will improve it and then bring it eventually on the market. If they want to commercialize it, switch ON should be at 3.4V (or little bit lower), switch off at 3.3V (or little bit higher). that would cater most demands and charging profiles. I already gave them a feedback and recommended those numbers for a commercial special version for Lifepo4 batteries in solar systems.