diy solar

diy solar

BMS manufacturing way behind current technology?

RedTechNeck

Tech Assistance for Escaping Grid Slaves
Joined
Mar 6, 2022
Messages
16
I've been noticing that BMS units currently manufactured and advertised seem to employ large banks of MOSFETS, and brag about being less than .02 ohm resistance. Did any of you know that the present state of the art in silicon MOSFET technology is less than .001 ohm "on resistance" in one transistor! I've seen some, depending on voltage rating at 600 micro-ohms or less. If you bank say 4 of these together you would get 150 micro-ohms. Amazingly the current limit for these modern mosfets individually is usually only limited by the current capacity (melting) of the lead frame and wire bond to the die. BMS units typically need 2 mosfet banks back to back to control charging and discharging.

Using more up to date components a 400 amp BMS is easily obtainable with only four mosfets per bank and only about 150 micro-ohms resistance. This would save a lot of power and extend battery discharge time. Better for cloudy days.

Conclusion is that old generation dirt cheap mosfets are being used in most "current" BMS units. But......... you can buy them real cheap too! I've been designing with mosfets for several decades and just couldn't help noticing this issue. Guess I'll have to eventually build my own BMS units after I buy some cheap ones to get me by for the a little while.

Also, it seems like these BMS units being sold now days might need additional heat sinks and cooling if you run them at max ratings for any length of time.

Paralleling existing BMS units on one battery stack to get more amps is not the best idea either as small imbalances will exist that will cause headackes and trouble.
 
Interesting, and sounds like a good idea to me. From what I've seen, the runtime difference probably isn't huge. I think it was @upnorthandpersonal who looked at one and found a group of mosfets, each rated for full amperage, with a total full-current loss of 12.8W. Even if you figure 1/4 of that average, you're looking at <400Wh over five days. I'm doubtful you could have a bank big enough to come close to even the latter loss figure and run for five days, so you're almost certainly looking at a multi-string mSnP bank with n BMSs (no two BMSs on the same cell, so no fighting each other) with even more mosfets and net resistance lower still.

Where I expect this would be much more beneficial is fail rate, especially as mosfets readily fail short circuit, which is usually undetectable unless there's a problem, and that's when you need them to work. With fewer mosfets a .x%/year fail means the risk of your BMS failing to protect your system goes down significantly.

This would also have an advantage for cells >200Ah, as it would make it easier to get a BMS that can handle the allowable cell discharge rate without breaking the bank on a high end relay based BMS.
 
depending on voltage rating at 600 micro-ohms or less

If you go to Mouser and search all MOSFETs with a Rds_on <= 600uR, you only have them with a Vds <= 40V. The first usable MOSFET for a 48V LiFePO4 battery (let's take 60V - no margin) has 680uR Rds_on. The MOSFETs on the JK BMS are rated at 100V Vds, and they're all > 1mR on Mouser.

What MOSFET with a Vds rated at 100V or over do you have in mind that has a Rds_on < 600uR?
 
When I get some time, I'll look at the types I have in my lists, which are not entirely up to date with 2022.

There are lots of 75 and 80 volt mosfets with .002 ohm or less. I use some of these. When properly paralleled the total Rds On drops accordingly. For a BMS up to at least 36 volts these should be fine. For 48 volts maybe, but will require more careful design. It's hard to depend on Mousers listings. They don't stock all manufactured versions. International Rectifier is one of the pioneers in this technology, but they have been reacquired by some foreign interests I believe. Seems like most all semiconductor manufacturers have changed ownership in the last decade. The best way to find these is to use the manufacturers listings, then find who sells that device. Future, Arrow, Digikey, and many more.

As far as voltage ratings go, it is always a good idea to use transient protection external of the BMS. Most spikes actually don't come from lightning, but are from motors starting and stopping, and ordinary switches being switched on and off. Mosfets will avalanche if voltage is exceeded. This does not always harm them like it does with bipolar transistors. Gate voltage should never be able to go over rated limits in any competent design. If it does, you have a crappie board level design with a lot of corner cutting and omitted protection components. In any case, for a BMS this is only an issue when they are cut off! This only happens when charge voltage is maxed out or minimum discharge voltage has been reached.

Be aware that most BMS units use a proprietary chip to drive or indirectly drive the mosfets. This chip handles the measuring of the cells voltages, temp, etc. from sensors and various inputs. Not sure if it is all analog or a dedicated computer and AD converter based chip. Also not sure how accurate the voltage measurement sections of these chips are. Hopefully using band gap references internally. This can be a big deal for battery cycle life as battery oxidation/reduction chemistry is exact (same as most other chemical reactions) and will not tolerate loose voltage measurement errors.

Cell balancing can be done using several variations of passive or active device designs on same board.

It all boils down to battery life and getting the most out of what energy you can get and store. Generator fuel costs money, sun electricity is free but doesn't happen at night. As for wind power, in summer the wind never blows much at night to run that AC unit. In winter it blows a lot but now you don't need it because your burning wood or propane to heat. (Mother nature has a warped sense of humor) Modern TV's and computers don't use much juice and it's easy to store enough for that.
 
Since writing the above, and doing some looking at which BMS to get, largely debating the JBD relay to minimize MOSFET failure only noticable in time of need, or the JK 200A, I've done some number crunching based on an expected real world usage.

With a 13kWh/day usage on a 48V system, and using the nominal 48V that's 271Ah/day. JK MOSFET rating is .32mOhm net MOSFET resistance. Counting the entirety of this as coming from the batteries, and putting 170Ah as peak load consumed across one hour, with the remainder spread through the other 23hrs, I come up with a total consumption of 5.9Wh/day lost to the BMS MOSFETs. Add in that any system consuming this much power is likely to have two or more strings of batteries, each with its own MOSFET BMS, and a lower still resistive loss, and I don't see any real-world measurable loss of concern.

That said, I still like this idea because fewer MOSFETs seems like less to go wrong and a much lower probability that, should something be overcharging the batteries, one or more MOSFETs will have failed short and allow the battery to be damaged or destroyed. That possible failure is one of two primary reasons I'm considering the JBD relay BMSs with ~60Wh/day parasitic load as my system will be configured.
 
Back
Top