Overcoming Zener Knee for BMS

Sunder said:
Oh really? I thought the ratio of the transformers windings fixes the voltage.
Only in unregulated linear supplies.

In nonisolated switching supplies, there is no ratio; there's just one coil. Pulse width determines the output voltage.

In isolated switching supplies, there is a ratio, and the ratio is set so that you get maximum power (at minimum input) at close to maximum pulse width. But you can still vary the output voltage via adjustment of pulse width.
 
Yes for AC, but then it is rectified into DC and then the buck regulator switches the DC downward in voltage (one way to do it). With a transformer they more likely extend the switching regulation circuit across the transformer and switch on the input side while feeding back from the output side, effectively feeding the transformer with variable voltage as needed to suit the output voltage and current needs.

The biggest problem may be finding an isolated adjustable low output voltage DC-DC converter that has well behaved current limiting on the output, they often don't have that feature, or they drop into hiccup mode rather than deliver the limiting value of current.

Here's an engineering note showing some topologies:

http://www.st.com/content/ccc/resource/technical/document/application_note/2a/8c/7f/9e/70/6b/43/62/CD00003910.pdf/files/CD00003910.pdf/jcr:content/translations/en.CD00003910.pdf
 
you can buy 30 small transformers with a unregulated voltage of around 12V@1A or so (no load is around 17V). those are very cheap and if you put a full bridge recitifier after that (and a massive cap) you can put those cheap buck converters behind that and make your own parralel charger that charges each cell individually. can be made prettly cheap. at least a LOT cheaper then any of the alternatives.
say you grab a dual core 12V transformer like this one:
https://www.aliexpress.com/store/product/YHDC-PE5424K-M-Power-20VA-Input-230V-Output-2-12V-Encapsulated-transformer-PCB-Welding-isolation-transformer/834088_32757809922.html?spm=2114.12010612.0.0.43246c09QuYazN

you need 15 of them for a 30S battery. each line can supply 0.8A at 12V. at 2.5V with losses that gives about 3.5A of charging current per cell. i would recommend adding a diode so the cell does not discharge into the buck converters and that adds another 0.6V depending on the diode you get. that increase in voltage makes it also easyer for the buck regulators to have a better output. usually they tank around 2.5V. adding the diode gets you above 3V making it more stable.

with those buck converters you can make a big charger for less then 120 bucks. if you spend more you can get more current.
one big advantage is that you can use simple plugs to charge each block of cells to keep the connector size manageable and is basically scalable to infinity and beyond.
 
Just for comparison, the 35s version of a chinese LTO BMS goes for US $142.60. The 20s version is $127.88. Together, you could do 55s.
https://www.aliexpress.com/item/100...fab2-47c0-b8d4-8a6517bb617b&priceBeautifyAB=0

This is about the cheapest one I could find that would possibly handle 55s worth of LTO cells.

Individual CC CV buck converters with separate supplies is another approach. The little buck converters that have actual current limiting are about $2 each. 5V, 1A wall warts go for about $1 each. The buck converters are rated for 3A, but can't do that continuously. 1.5A is more like it. I've used these with good results and they have LED indicators for CC and CV modes, which is nice.

CC CV Buck Converter.JPG
 
I got 20x of UltraCap boards yesterday. They were USD12.50 x 10pcs and free shipping as I ordered them, apparently they raised the price now. Looks like easier way to get there.
https://www.ebay.com/itm/New-10pcs-lot-2-7V-500F-Super-Capacitor-Protection-Board-Balancing-Plate-Module/252010862770?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649
 
The ultracap boards are nice. The picture shows a pair of 100 ohm bleed resistors, so at 2.5v, would be 50ma. If the charge current is over 50ma, then it may go over the set point.

Just for fun (and because I have some LTO cells around), I tried placing 4 1N4001 diodes in series and tested. It starts to conduct around 2.5v, but at .5A, it goes way up to over 3.5v, so this does not seem like a good approach. A 3W white LED was pretty similar. The knee is too soft.


Alan B said:
Nice little buck converters.

Just to be crystal clear, these are not isolated so each one must be powered by a separate isolated supply.

Right. It would end up looking like this, only with 55 wall warts. You might run into issues trying to plug in that many at the same time. Big inrush current might blow the house breaker.

Chargers.JPG
 
parabellum said:
You mean a pair of 10ohm resistors and 500ma current, right? :wink:

I guess that must be it. If they were 100 ohm, they should be marked 101. Description says Balanced Current: ≤1A. Both 50ma and 500ma are ≤1A.

500ma would be pretty decent.
 
fechter said:
Right. It would end up looking like this, only with 55 wall warts. You might run into issues trying to plug in that many at the same time. Big inrush current might blow the house breaker.

Chargers.JPG

I think a bunch of 12v dual output pcb tranformers are a nicer solution. 12v input for the buck converters allows more power to flow then you can get with the 5v wall warts. With a tiny heatsink and a fan you can get 5 amps if you really push it. But if you blow one it costs less then 2 bucks to replace so trial and error investigation should be done. Inrush might be an issue with 25 little transformers, but that can be mitigated with a simple precharge resistor with a delay relay bypass to limit the inrush.
Slightly more elegant then stringing 20 extention cords together and have 4 dozen shitty usb chargers running at their peak output.
Bonus: you can charge any pack at any voltage this way. Even ones extremely out of balance.
 
fechter said:
Just for comparison, the 35s version of a chinese LTO BMS goes for US $142.60. The 20s version is $127.88. Together, you could do 55s.

This is about the cheapest one I could find that would possibly handle 55s worth of LTO cells.

Option 3... (or maybe about my 9th or 10th option... I've lost count.)

The problem with two separate BMSes, is that it's harder than it looks to balance between sub-packs. So I with my 3 x 12S, If two of the packs were at the balance voltage, the third one would find it almost impossible to balance and be brought up with the other two. I'm not sure if that's because of some odd change of resistance issue, which causes most of the current to go off to the shunt resistors, rather than the cells acting as resistors.

However, I have seen whole battery balancers. So I do wonder if I made it say, 56S, and got two 30S BMSes then (running as 28S), then tried to balance at a pack level as well as a cell level, whether this would make the above problem a bit easier.

I can't find one for high voltage packs at the moment, but they claimed that any time there was ever more than a 100mv difference between the batteries, it would shunt power from the higher to the lower, completely isolated, so the battery could even be in use while this occurred.

All that said, my favoured solution is currently the Supercapacitor balancers. I'd seen them in 6 strings before, and assumed they were like a 6S BMS, but seeing them in single capacitor formats, makes me think that they are effectively just a cheap but better version of what I am trying to do with the Zeners - a blind and dumb circuit to waste off energy as soon as the trigger voltage is hit.
 
Yes, you are correct that having two BMSs on the same pack will be harder to deal with. Once approach is to use two chargers, one for each BMS so they operate independently during charge.

The ultracap balancers seem pretty good. The only thing is you won't have any way to detect when one cell goes over or under voltage. As long as they cells are well balanced in the beginning, you could probably get away with periodically checking the balance. Also don't try to use the full capacity of the pack. Pack level LVC should help avoid problems.
 
fechter said:
The only thing is you won't have any way to detect when one cell goes over or under voltage.
Theoretically overcharging a little bit will visually show that all cells are at same level, those boards have LEDs, they get on when shunt activates.
 
parabellum said:
Theoretically overcharging a little bit will visually show that all cells are at same level, those boards have LEDs, they get on when shunt activates.

I missed the LEDs. That's great. At least you can easily check it manually. The charger could also have a mode where the current limit is lower than the shunt so it can stay on for long periods if needed for balancing.

For my application (standby backup power) I can live with a pretty slow charge rate.
 
fechter said:
I missed the LEDs. That's great. At least you can easily check it manually. The charger could also have a mode where the current limit is lower than the shunt so it can stay on for long periods if needed for balancing.
Push button overriding charge voltage resistor or diode inline. Cool, right? :)
 
Those modules start balancing at 2.7 and stop at 2.68V.
Unfortunately, I found 2 out of 20 modules dead and I do not think it is static related.
It finally looks like this in 11S2P configuration:
 

Attachments

  • 2018-05-15 23.14.12.resized.jpg
    2018-05-15 23.14.12.resized.jpg
    120.2 KB · Views: 1,402
fechter said:
Order extras.
Fortunately I need 11.
fechter said:
How hot do they get?
After 24h the cells were at ~2.6V/cell
Procedure:
Ambient temp: 30C.
Open as in picture, without forced circulation.
I started charging at 2A, the modules started draining at 100-200mAh in, at 15 min, 500mAh at about ~2.75V/cell, still charging.
Modules draining >10min
Max registered temperatures (they vary between boards):
At resistor (opposite to bus screw) 96C
Bard (just beside former resistor) 72C
Control circuit 46C
 
Wow. No wonder they die if that's the heat they reach.

I have a fair bit of space, so I wonder if a small fan that only turned on during charging would help. Just get get a 240v fan, and join it with the charger cable.

I'm now getting impatient for my cells. They were supposed to be shipped Monday, but no tracking number yet. :(
 
Good test results. Thanks for posting them.
Those temps sound within the rating for the parts. Barely. Even a small fan would make a big difference.

Seems like you could set the charger for 2.6v/cell and they would stay off most of the time.
 
Sunder said:
Wow. No wonder they die if that's the heat they reach.
Dead 2 came this way.
Resistors can handle up to 175C, the 2x 3 leg components are sanded to hide markings but I suppose 90C would be tolerated.
Board is well conductive, the screw side of board is ~ 20C cooler, because of heat sunk to the terminal and in my case cells absorb enough to use as is.
In worst case scenario of 3V/cell overcharge, each board will produce ~1.8W of heat.

Maybe stick on alu sheet and let it open to convection?

2018-05-16 23.57.58.resized.jpg
 
I would replace the resistors with bigger ones and get them off the board. Much better cooling that way. Slo the pcb doesnt get as hot so the other stuff should last way longer
 
parabellum said:
Maybe stick on alu sheet and let it open to convection?

That would work. I've done something similar in the past. I put Kapton tape over a piece of aluminum heat sink, then glued it over the resistors with silicone. The tape is just to prevent the aluminum from shorting anything. About 2"sq surface area would dissipate 2W nicely.
 
flippy said:
Slo the pcb doesnt get as hot so the other stuff should last way longer
Yes, that will work for sure, or just removing 1 of the resistors and halving the heat generation. Sure, they will take longer for balancing and allow cells go higher V, but hey, aren’t those batteries super tolerant to over/under volting and never need balancing? :D
 
parabellum said:
2018-05-16 23.57.58.resized.jpg

Hey Parabellum(or anyone else reading this thread that bought the above balancers). Could I ask a small favour ? I've ordered a whole bag of these, but the only option they had was economy shipping, so they won't arrive for another 4 weeks. In the mean time, I want to 3D print a retainer that will hold both the cells and these balancer. I want to raise them 5-10mm, so they can get airflow under them.

I need to know what size bolt I could put through the holes in the centre (that the legs of the capacitor would normally go through) to retain it. If you could let me know, I can get the retainer printed before the balancers arrive, saving me a fair bit of time.

Thanks in advance.
 
Back
Top