Seeking a low voltage shut off with just a MOSfet and a zener (maybe a resistor too)
LoopyByteloose
Posts: 12,537
Okay, I am back into ponder Li-ion and NiMH cells need for a low voltage cut-off to protect from immediate and fatal damage due to going below the specified voltages.
For a long-time I have figured I would have to use a comparator and a zener diode with a few resistors to get an output that simple was high or low depending on the battery status.
From there, the pulse could drive a power MOSfet on or off depending of the batteries voltage when compared to the zener's voltage (which would set the low voltage limit.
============
But I still dislike the part count and am wondering if just a zener, a MOSfet and a few resistors might have some way of doing the same thing.
Does anybody know?
For a long-time I have figured I would have to use a comparator and a zener diode with a few resistors to get an output that simple was high or low depending on the battery status.
From there, the pulse could drive a power MOSfet on or off depending of the batteries voltage when compared to the zener's voltage (which would set the low voltage limit.
============
But I still dislike the part count and am wondering if just a zener, a MOSfet and a few resistors might have some way of doing the same thing.
Does anybody know?
Comments
http://datasheets.maximintegrated.com/en/ds/MAX8211-MAX8212.pdf
Thanks I will keep that in mind.
I have discovered two other options. Trying to sort out what I like.
So far, for Li-ion, it seems easiest to go with an existing protection board for each and ever cell.
And for NiMH, a DIY solution may still be required. NiMH in series are prone to reverse polarity damage when the voltage on any one cell drops below 1.0VDC.
My gut feeling is that it is best to just forget NiMH and use Li-ion solutions with pre-fabricated protection.
Comparators and Op-Amps will allow one to create hysteresis in actual operation. So I suspect the Zener white noise might not be an issue.
A. An op amp that includes an internal voltage reference -- the LM10
B. A MN1380 series Voltage trigger used with a voltage divider to create a voltage reference (search for a Panasonic MN1811-r)
And for Li-ion, there seems to be boards that attach to 14500 and 18650 cells.
http://www.all-battery.com/1S_Li-Ion.aspx
– Monitor and trigger threshold for overcurrent protection during high discharge/charge
– Overcharge Detection (OVP) current operation or battery overcharge conditions.
– Depleted battery monitoring and protection.
http://www.mouser.com/ProductDetail/Texas-Instruments/BQ29703DSET/?qs=sGAEpiMZZMsfD%252bbMpEGFJeXatD6U5GsFkL2%2fMOeGv%252bQ%3d
Needs a dual-fet added for the actual switching
Most are listed under dual common drain:
http://www.mouser.com/Semiconductors/Discrete-Semiconductors/Transistors/MOSFET/_/N-ax1sfZscv7?P=1z0xrwj&Ns=Pricing|0
So unless you already have a high side switch that you want to utilize then DIY is not the way to go.
But I think with a resistor divider you could make the MCU itself monitor it and then put a external-circuit-LDO enable low and then go to deep sleep to consume uA.
Could use a dedicated uA IQ 100mA LDO just for the MCU
Or some type of power on reset circuit when a fresh battery is inserted, you hold a button down to enable the 3V LDO to override its internal pull-down on Enable, after 1second the MCU takes over (need to boot up) and sets the pin high.
And when MCU sees low voltage (3.1V or lower Vbat) it put the pin as input and shuts the whole circuit down, even itself.
with LiPo I recommend to use 3V LDO as the whole system voltage and not 3.3V as to get the most out of the battery considering ldo voltage drop
http://datasheet.sii-ic.com/en/battery_protection/S8261_E.pdf
It is worth a glance at the data sheet to see what you get in one of these chips.
The following is how I implemented that chip in a prototype, with the dual mosfet...
emesystems.com/images/S-8261_circuit.gif
That is typical of the protection circuit that is built into battery packs. Multiple cells are more difficult due to the need for charge balancing.
I came up with the following circuit for use with small (1.4Ah) LiPo, and it is performing as expected. This uses a micro-power voltage detector and a hi-side pmosfet. I'd never consider using a separate ref +comparator+hysteresis for this, because these chips designed for microprocessor reset have it all built in.
http://emesystems.com/images/ISL88003_cutoff.gif
When the battery voltage falls below 3 volts, the IC cuts off power to Vbb and to the rest of the circuit, and the residual battery current is about 2µA. I have an SD card detect switch set up as user controlled power on-off.
I do have some 3.3VDC 1 AMP linear voltage regulators with an enable. But I hate to insert them to waste the additional power of a 3.7V 18650 Li-ion. The penalty of wasted power is too high. I'd like to drive the Propeller over a full range from 4.2VDC to 3.0VDC and risk shorter useful life. And I'd like to not bother with the Propeller monitoring the shut-down. This is not about wasting a pin for the input. It is about adding the extra code to monitor that extra input for a shut-down.
In any event, it looks like each individual Li-ion or NiMH cell requires its own protective device to really get longevity out of these cells. That appears to be what Tesla is doing with the 18650 cells. NiMH may have a significant economic disadvantage as the cells output a nominal 1.2VDC versus the Li-ion 3.7VDC. That means more than double the equivalent amount of protection devices.
But I am finding that it is harder and harder to locate AA NiCd. They seem to be slowly being pushed out by NiMH. The NiCd certainly provide less energy, but are less sensitive to abuses.
I think I am finally ready to admit that I should go all Li-ion and buy 10 or more protection boards at a time for a bulk discount. Li-ion seems to charge faster than the NiMH as well.
That's old out of date talk. The newish (been in the shops for at least 8 years now) generation NiMH known as "low self-discharge" are very good. They may not have the very low impedance of NiCds but really do handle a lot of abuse otherwise. No need to worry about pre-discharging for example. Cooking them on high power doesn't appear to hurt at all and of course dead flat is no problem. Capacity is mid-range, about the same as Lithium-Phosphate, and will happily charge on a NiCd charger, although most new chargers are rapid charging down to a 1 hour charge.
Their big selling point was always how long they can hold a charge for when not being used or only very tiny power like a wall clock. They have totally replaced alkalines for me. I put them into everything that uses AA and AAA.
What might need to be changed in your circuit for bigger 3V7 LiPo - say 8000mAh?
It seems that the same link I mentioned about provides higher amp rate options. But my impression is that the amp rate is linked to the charge and discharge rate and less to the actual cell size. Of course, using a lower amp rate on a bigger cell capacity implies a more limited discharge and a slower charge.
In other words, much depends on what the project demands rather than a close relationship to the AH capacity of the cell.
Take a look at the 7 amp board at the link below. Is that what you desire??
http://www.all-battery.com/1S_Li-Ion.aspx
Larger current comes down mainly to using a mosfet capable of higher current. The FDN304 as written is rated at 2.4A, 60mΩ Rds.
The circuit I posted covers only the low-voltage-cutoff function, between the battery and the circuit load. The mosfet is not in the charging path. It is only in the discharge path and In my case the circuit being supplied draws only a 10 mA average and 350mA in short bursts. The charging IC, not shown, supplies up to 450mA from a USB port or 700mA from a USB charger, and it handles the current limit and overcharge protection.
The battery protection ICs that Loopy mentioned are a bargain. You can see a couple of soic8 mosfets on the board in addition to the actual protection IC in a 6-pin soic. For a 6-pin protection IC like the Seiko S8261, both of the mosfets are in series with one leg of the battery, and they have to carry both the charge and the discharge current.
A primary goal with a low-voltage-cutoff is low battery drain, ideally zero to stretch out the remaining life of the battery above the point of no return. The ISL88003 draws less than 200 nA, similar to the S8261.
My main goal is to achieve battery direct powering of the Propeller without a voltage regulator wasting a significant fraction of the battery's power.
The extremely small current demands of a low-voltage cutoff scheme allows to do so without damaging rechargeable cells. But there is an equally valid alternative of using non-rechargeable cells in these micropower schemes as they generally pack more power into a given amount of space.
In one desires to use non-rechargeable batteries, there may be a valid alternative of just creating a brown-out cutoff before the micro-controller and related chips begin to generate trash data.
The first few days the voltage is 4.2V, way above the safety region of most 3.3V IC devices.
So you add a schottky diode to drop it 0.3V, but this drop is also there when Vbat is 3.4V and lower
Diodes are not much cheaper than a extreme low IQ LDO that maybe have as low as 0.12V drop
Devices that can handle Vbat you don't have to power through the LDO, just take precaution when 3V system is interfacing with this higher voltage sub-block (like using 2K series resistors)
And don't have system where the protection diodes are on a lot as that could drain 1mA continues.
As you can see with the black line on chart below, usable Vbat voltage is around 3.2v (the final drop just takes days after that)
So it's recommended to base the system on a 3.0V LDO
300mA with 20uA current draw
http://www.digikey.com/product-detail/en/MIC5307-3.0YD5%20TR/576-2855-6-ND/1858607
It have Enable so you could implement ~200K+200K resistors going to a prop pin sensing Vbat voltage that will use another prop pin to shut down the ldo, and with a manual push switch you bootstrap a boot up.
But only needed of you are using plain battery-cell as most commercial battery pack have built-in under voltage cutoff, so you will not experience a stay in the brownout region as it goes from 3v/2.8v to 0v instant.
I was aware of the S-curve of battery charge and discharge, but not that the curve flatten out with lower rates.
Many microcontrollers will operate at 5VDC and down to zero, unlike the Propeller that seems to have been targeted for 3.3V due to the higher frequency operation running cooler via PPL.
People have tried 5VDC operation of the Propeller and gotten away with it (for awhile). Since 4.2VDC would be significantly less voltage that 5.0VDC, I presume it would be proportionately less heat (and less potentially damaging). This is a gray area of use, similar to over-clocking the Propeller. Without actual statistical research into the durability, nobody really knows.
In sum, I can't be as bad as 5VDC.
There is an alternative.
One could place a 1 watt 3.6 volt zener with a current limiting resistor in parallel to the battery output. That would accelerate the batteries discharge down to 3.6 volts (a safer region). Just because Li-ion charge fully to 4.2VDC, doesn't mean that one has to use all of the capacity.
Of course, an in series diode to safely drop the voltage is no good - basically it will result effectively the a diode drop on the high end, but creates problem with an early low voltage cut-off on the low end (unless one attempts to run the Propeller down to 2.7VDC).
What is really causing the problem?
Battery manufacturers provide batteries nominal voltage that is the middle or mean of the S-curve discharge. The S-curve chart is the real model of performance.
AND>>> Yes, you are convincing me that a 3VDC LDO voltage regulator might be useful in some instances if it has an additional Enable pin to cut out discharges below 3.0 VDC.
But I think you can just create a voltage divider on the Enable pin that would trigger a shut out without having an microcontroller involvement. Using a small 10 turn pot as your voltage divider might be optimal to get a precise cutout point. Then 10 turn pot might be 100K or higher (seems only 1uA or less to control the Enable) and wired across the battery V+ and V-
Without the cut-off, the LDO is just likely to drag a Li-ion in the damage region. It will work fine with primary cells that will never be recharged.
> Just because Li-ion charge fully to 4.2VDC, doesn't mean that one has to use all of the capacity.
Contradictory statements, you want to quick-drain the top 15% power out of the battery so it stays in safe region.
Using a Prop pin (as long you have plenty available) comes with free schmitt-trigger and software hysteresis.
Trust me, using the recommended circuit I mention and you will get longest battery life.
Most MCU are not 5v (it's not 1980's), the low power leader MSP430 can only go up to 3.6V.
With proper programming techniques the msp430 will only use 1uA at 1Mhz as it can wake up its own DCO in 1-2uS (one or two ticks) and it auto turn it on/off if timers and serial ports use it intermittently.
>LDO is just likely to drag a Li-ion in the damage region
As manufactures don't want to handle warranty claims from irresponsible users, most batteries have under voltage protection and also short circuit protection (the later could be the law as to import to USA)
I just do NOT see any advantage to having the Enable managed by two pins on the Propeller unless you desire some sore of 'orderly shut down' procedure within the Propeller to retain data or to prevent a mechanical hazard.
You device has a 1 Volt On and a 0.2 Volt off on the Enable. And the manufacturer's PDF show the Enable is already attached to a Schmitt Trigger.
I must admit that my last posting rambled a bit. Is that making you argumentative? I am just trying to collaborate.
PICs have certainly been around far after the 1980s with 5V use. What do you think Parallax used to get started with the BasicStamps. AVR chips are 5VDC to 0. The TI device, the msp430 is a late comer that was specifically designed for extreme low power operation.
A lot of MPUs require 5VDC to achieve peak clock rate and operate at 3.3VDC with a significantly lower clock rate.
I could the see the problem when LDO shuts off the Vbat goes up a little as always happen when under no load.
So now it re-enables the Enable so now you will have days of Self-Oscillation with 1000's of resets/reboots of the mcu.
Add flip-flop etc, it's starting to look like a home-made tiny mcu.
Using the MCU to monitor it will also give you an opportunity to shut-down with planning, like saving data to eeprom.
While it 's imperative to be familiar with battery data sheets, to be really serious about it you have to have a specific application in mind and characterize the battery running in situ. Here for example is a charge-discharge graph of a Prop device that averages 30mA from a 1.4Ah LiPo, at room temperature.
You see, a scheme to drain off the battery down to 3.6V would waste almost all of the capacity. That is because this battery operating at 0.02C delivers almost all of its capacity above the 3.6V mark. It is quite a different picture if the device being powered is operating at higher current to power a motor or something. Even so, it the load is variable, the battery terminal voltage will be variable too.
Thanks. Still one could use a Zener scheme to prevent voltage from going over the 4.2 Volt mark if there is such a hazard present. Though it may just be an unnecessary addition.
Regarding comments about bounce in an Enable's voltage divider, I suspect a capacitor of appropriate size and placement would eliminate that.
While the 3.0VDC LDO voltage regulator is appealing for a board build. It does cost $1.50 for the chip alone. There are still many projects that may desire to avoid a 'full design and construction' and just want to deploy a 40-pin DIP Propeller. (I do like the idea as these small micropower regulators are generally ignored by Propeller users that think they must use a 1 amp regulator... 300ma is likely enough with good management of loading the i/o.)
As Tracy Allen stated, one really has to have a project in mind as the goal in order to finalize which choice is best.
It's more than a bounce, it's a rise of 0.1V so if the Enable does not have some type of hysteresis every time it shuts off it re-enable itself over and over......
A really large cap (costly) would slow it down a little, but as there is not load it will only take seconds before it reach Vbat level as caps are never used to create hourly delays.
> It does cost $1.50 for the chip alone.
Here is a 50cent 250mA , IQ 12 uA at no/minimal load = would take 3.8years on a 400mAh battery if that was the only thing on, battery itself and its protection ic probably have another 10uA self-drain
http://www.mouser.com/ProductDetail/Texas-Instruments/LP5907MFX-30-NOPB/?qs=sGAEpiMZZMsGz1a6aV8DcHd1yzolGFo2R37WJCJ2GlM%3d
250mA is more than enough as large current devices should use battery directly
hint if you use p-mosfet for high side switching turning it off with a 3v mcu pin (equal -1v Vgs) is not enough many times so apply a 100K resistor to Vbat-level on its gate and another 100K in series to MCU pin in case
protection diodes kicks in to get much smaller drain. (if its ratings of -1.5 vgs is fully on) and use input/output-low (open drain) to control it.
An actual zener diode is not a good option in any case, following on the subject line of your thread. The zener graph of voltage-current has a very soft knee, especially at low voltage ratings. You can get chips usually called parallel regulators or references that have very sharp points above which they draw lots of current and below which they draw practically none. Combine that with a comparator, hysteresis and a mosfet, and you have your circuit. But why bother? Microprocessor reset chips have the necessary reference and the comparator built into one tiny micropower chip, and cost about the same as the discrete but inadequate zener.
Another design decision is how you want to charge your battery. If you want to leave it in place for charging, it will be subject to the full charge voltage, and the charger itself will reliably limit the voltage to 4.2. Here for example is a graph of a Prop device drawing ~30mA, 700mAh LiPo, with a Linear IC charger chip (LT4054). You can see the cycles as the device is left connected to USB and then, when disconnected, a long fall to a cutoff point.