High-side N-channel FET Switch (100% duty cycle ?).

[Zero999]

C) You don't need a LDO... A plain old jellybean voltage regulator IC will work since you'll have lots of headroom.

Yes, the plain old LM78L12 will do.

[ArdWar]

78x works, but if you're that concerned with power consumption then you may need to pick something else as its quiescent current is pretty brutal at 2-5mA. Not that LMx3 comparators are particularly low power either.

I'm not sure what's considered jellybean at this performance level, but Diodes AP73xx are pretty popular and the one in SOT-23-5 (SOT-25) package with IN-GND-EN pinout are cross compatible with many other offering from various manufacturers. Do check your pinout however.

[Dannyx]

Thanks for the replies once again. Quiescent current IS important in this application (at least on paper), so the 78XX family was pretty much out. Keeping that in mind, I sifted through some LDOs and picked one with the lowest quiescent current that I could possibly find on the supplier's website, which is a MCP1804. The input voltage is at the very top of the limit (might not survive due to transients) and the output current is fairly low at 150mA, so I don't know if that's going to be an issue in practice, since I don't know how much current the comparator typically draws. "Low" in this case might be plenty.

The AP73XX suggestion appears to be better in just about every aspect, but the supplier doesn't carry any of them  I'm pretty sure I'd have to go with a 12v output though (I mistakenly linked to the 5v version and realised that might be too low for this particular application, so I scrapped it, lest someone read it).

Setting that aside for now, once I DO have an LDO ready to go, there's the matter of deciding if I should power the comparator itself from the LDO, or just use the LDO for the resistive divider that will serve as the reference.

Jellybean or not, I still want to keep it "somewhat professional" (or "industry standard"), but DIY at the same time

[Kim Christensen]

A quick search of Digikey turns up 172 linear regulators in stock that can tolerate more than 36V on their inputs and have less than 100uA of quiescent current draw. That'd be my 1st choice. (Find a better regulator)

Since you have plenty of headroom, you could use a resistor to drop the input voltage to the regulator down a few volts. Size it so that there's enough voltage at max load and low battery voltage, but that it still drops enough voltage at minimum loading and max battery voltage. Combined with the obligatory capacitor on the regulator's input, this should mitigate any short duration voltage spikes from being an issue.
Another option is to use a zener instead of a resistor to drop the required voltage if the current variation is too broad. (in series with the input)

Assuming low quiescent current only really matters when the batteries are depleted, you could make a simple preregulator with a transistor, resistor, and zener. But choose a zener voltage well above your battery's low voltage cutout. That way the zener is only conducting when the battery voltage is high. (Charging)

there's the matter of deciding if I should power the comparator itself from the LDO, or just use the LDO for the resistive divider that will serve as the reference.

If the battery voltage is close to the comparator's limit, then power it from the regulator. You can use the LDO to do both: Be the reference and power the comparator since there's no need for super accuracy here.
One plus for using the LDO for the reference only is that the LDO's current draw will be pretty much fixed. That'll facilitate the resistor voltage dropper method.

[BrianHG]

I wonder if you could use a standard high side driver for the speed with an always-on photovoltaic as your charge pump.

IE: the photovoltaic is your isolated DC-DC supply.

When the mosfet is off, you have the normal diode charging the gate voltage supply.
When the mosfet is on, the photovoltaic across the bootstrap capacitor to hold the gate supply 12v above the source.

[Zero999]

If a tiny amount of voltage ripple is acceptable, you could switch the load off for tens of µs, every second or so and  put a large capacitor across the load to power it, whilst it's off. Another MOSFET will need to be added to discharge the capacitor, when the load is off.

Here's an example of how to do it with the IR2117. The 1000µF capacitor will power the load for a short period of time after the MOSFET is off and the 2N7002 provides a path for the bootstrap capacitor to charge, when it's off. It can't charge though the load, because the big capacitor keeps the load voltage high, so there would be no potential difference to charge it.

Bad circuit attached. Don't use. Corrected version here.

[BrianHG]

@Zero999, wont your circuit potentially send a negative voltage into the Gate if a turn off happens while the 2N7002 is on?

Also, the pins VB-VS can only be maximum 25v.  If you want a 48v high-side switch, wont you over drive the 25v max?

Also, according to the data sheet, you will still be dragging down the top of the cap with the bottom, so you will still be turning off the top mosfet when the requirement was 100% on time.

I still like my idea.  Copy the example in the IR2117 data sheet, and across the VB-VS cap, place a photo-voltaic optocoupler.  When the IR2117 is turned off, the diode between Vcc and VB provides a high current fast charge for the cap between VB and VS.  When turned on, the photo-voltaic optocoupler keeps a 12v continuous 12v charge on that cap allowing 100% on operation.

In this circuit, the only purpose for the IR2117 is the high speed switching.  Otherwise, just using a photo-voltaic optocoupler all on it's own would be fine.

[Dannyx]

Thanks for the feedback once again. Lots of great ideas in here. One in particular caught my attention:

I wonder if you could use a standard high side driver for the speed with an always-on photovoltaic as your charge pump.

IE: the photovoltaic is your isolated DC-DC supply.

When the mosfet is off, you have the normal diode charging the gate voltage supply.
When the mosfet is on, the photovoltaic across the bootstrap capacitor to hold the gate supply 12v above the source.

Although it increases the complexity quite a lot, I guess it should work, so perhaps someone can provide some pros and cons to this, if it's at all possible.

On the comparator side, adding a resistor in series with the input to the regulator sounds like the simplest possible solution and a decent compromise. It's only going to be using a meager amount of current anyway. I can't imagine the comparator + resistive divider drawing that much current as to cause that input resistor to heat up.

Taking it one step further would be to also use a zener there to clamp to 30v. This was among the suggestions as well, if I understood it correctly. I already have the resistor - why not add a zener to GND as well ? Makes sense. Unless the overvoltage is significant, the zener doesn't even need to be on all the time.

Is an adjustable LDO typically a good idea for this kind of application, or are they considered "less stable" ? It greatly increases the variety, because most of the ones the supplier carries ARE adjustable...

[BrianHG]

I wonder if you could use a standard high side driver for the speed with an always-on photovoltaic as your charge pump.

IE: the photovoltaic is your isolated DC-DC supply.

When the mosfet is off, you have the normal diode charging the gate voltage supply.
When the mosfet is on, the photovoltaic across the bootstrap capacitor to hold the gate supply 12v above the source.

Although it increases the complexity quite a lot, I guess it should work, so perhaps someone can provide some pros and cons to this, if it's at all possible.

No it doesn't.  You follow the bottom end example circuit in the IR2117 data sheet.
Then, you take your photo-voltaic's output side and tie it in parallel with the cap across pins VB and VS.
It will keep the cap charged at a minimum of +12v.

On the LED side, just tie your 5v or 3.3v via series resistor to the LED.
Yes, the photo-voltaic is always on always delivering +12v into the cap.

You drive your IR2117 normally with your high speed digital PWM control signal.

Again, if you do not need the super high speed PWM, ignore everything else and just use the photo-voltaic directly between source and gate on the mosfet.  No other parts needed anywhere.  Just a resistor on the LED side of the photo-voltaic.

Yes, there exist photot-voltaics with 10x faster turn off times than the one listed earlier in this thread, however, the turn on time is identical.

[BrianHG]

High speed voltaic: 8.4v, 53us on, 65us off.
https://www.vishay.com/docs/83469/vom1271.pdf
However, with 8.4v output, this guy need logic level mosfets.

Slower speed, but higher voltage: 13.9v, 35us on, 90us off.  Times 2.  Needs external p-channel j-fet.
https://www.vishay.com/docs/83802/lh1262ca.pdf

[Kim Christensen]

Taking it one step further would be to also use a zener there to clamp to 30v. This was among the suggestions as well, if I understood it correctly. I already have the resistor - why not add a zener to GND as well ? Makes sense. Unless the overvoltage is significant, the zener doesn't even need to be on all the time.

If you need a bit more current capability, you can add a transistor to the zener circuit like the circuit on the left. Then you can get away with a max zener current that is only 5-10% of the barebones zener circuit on the right. (LDO still needs it's own bypass capacitors)
What kind of a battery will you be using? Lead acid?

[BennoG]

A completely different approach.
Integrated power rail switchers.

VN7050ASTR      https://nl.mouser.com/datasheet/2/389/vn7050aj-1852754.pdf

BTS5010     https://nl.mouser.com/datasheet/2/196/Infineon_BTS5010_1EKB_DS_v02_00_EN-1511519.pdf

Benno

[Dannyx]

Thanks for the input again. Some more info:

Again, if you do not need the super high speed PWM, ignore everything else and just use the photo-voltaic directly between source and gate on the mosfet.  No other parts needed anywhere.  Just a resistor on the LED side of the photo-voltaic.

I saw someone mentioned PWM. I won't be doing any PWM, nor involve a micro in this project - just a basic on/off control straight from a comparator output when the battery drops below a certain value.....at least that's the plan. The idea is to prevent erratic behavior on the downstream devices when the voltage drops below a certain threshold and also prevent too deep a discharge.

Then:

What kind of a battery will you be using? Lead acid?

Yes, I'm planning on using 2 12v/20Ah SLAs in series, since I already have them on hand. I know SLAs are fairly resilient and can take some abuse with little to no risks, esp. when compared to LiIon cells, but I wouldn't want to abuse them and go ham on them either.

The LM9061 also seems a viable option for driving the cut-off FET, all things considered.

As far as photovoltaics are concerned (whether on their own or in combination with a separate driver), during my searches I came across a FDA217. I haven't done any deep dives into the datasheets to work out how it performs vs. the suggested VOM121, but what the FDA has going for it is definitely the package: a DIP is MUCH more suited for my particular case, since I will be hand-assembling onto perfboard. I CAN technically botch-solder SMDs there as well somehow, but......y'know......

EDIT: I may be wrong after all.....the speed difference between the FDA and the VOM is significant and I suppose it's directly proportional to the output voltage each produces. 8.4v is a tad on the low side, but it's faster....makes sense.

[Kim Christensen]

a DIP is MUCH more suited for my particular case, since I will be hand-assembling onto perfboard. I CAN technically botch-solder SMDs there as well somehow, but......y'know......

Don't forget that SOIC to DIP adapter boards are pretty cheap to buy.

[Zero999]

@Zero999, wont your circuit potentially send a negative voltage into the Gate if a turn off happens while the 2N7002 is on?

Also, the pins VB-VS can only be maximum 25v.  If you want a 48v high-side switch, wont you over drive the 25v max?

Also, according to the data sheet, you will still be dragging down the top of the cap with the bottom, so you will still be turning off the top mosfet when the requirement was 100% on time.

I still like my idea.  Copy the example in the IR2117 data sheet, and across the VB-VS cap, place a photo-voltaic optocoupler.  When the IR2117 is turned off, the diode between Vcc and VB provides a high current fast charge for the cap between VB and VS.  When turned on, the photo-voltaic optocoupler keeps a 12v continuous 12v charge on that cap allowing 100% on operation.

In this circuit, the only purpose for the IR2117 is the high speed switching.  Otherwise, just using a photo-voltaic optocoupler all on it's own would be fine.

I messed the circuit up because I missed a transistor off. Here's the corrected version.

a DIP is MUCH more suited for my particular case, since I will be hand-assembling onto perfboard. I CAN technically botch-solder SMDs there as well somehow, but......y'know......

Don't forget that SOIC to DIP adapter boards are pretty cheap to buy.

I wouldn't bother with those adaptor boards for ⅟₂₀" pitch 8 pin parts. Bend two of the middle pins opposite one another (2 or 3 and 6 or 7) up, then you can solder the other pins to the tracks and the middle pins can be connected to the necessary parts of the circuit via thin enamelled wire.

[langwadt]

a DIP is MUCH more suited for my particular case, since I will be hand-assembling onto perfboard. I CAN technically botch-solder SMDs there as well somehow, but......y'know......

Don't forget that SOIC to DIP adapter boards are pretty cheap to buy.

something like this: https://www.lcsc.com/product-detail/Power-Distribution-Switches_Infineon-Technologies-BTS6133DAUMA1_C2680389.html

wouldn't be too hard to fit on a perf board, just lift the two center legs

[BrianHG]

Thanks for the input again. Some more info:

Again, if you do not need the super high speed PWM, ignore everything else and just use the photo-voltaic directly between source and gate on the mosfet.  No other parts needed anywhere.  Just a resistor on the LED side of the photo-voltaic.

I saw someone mentioned PWM. I won't be doing any PWM, nor involve a micro in this project - just a basic on/off control straight from a comparator output when the battery drops below a certain value.....at least that's the plan. The idea is to prevent erratic behavior on the downstream devices when the voltage drops below a certain threshold and also prevent too deep a discharge.

Ok, just use a single https://www.vishay.com/docs/83802/lh1262ca.pdf and be done with it.
No high side switch.
No J-Fet needed as you do not care about it being ultra fast.

Just 1 photovoltaic optocoupler and a series resistor for the LED side.

As for the low voltage comparator, just use a CMOS input micro-power one and wire it for some hysteresis with a voltage reference on the inputs.  The output should drive the LED on the photovoltaic.

When in undervoltage shutdown, your circuit should draw a few microamps of current, IE, power just the comparator.
When turned on, your circuit should draw the 10ma it takes to power the led on the photovoltaic.

If you decide to use a P-Channel Mosfet instead (You can have 2 or more in parallel for huge currents, low on resistance), IE: No gate driver needed, the output of the comparator goes directly to the mosfet's gate with a series resistor.  With this circuit, whether on or off, the quiescent current will be the same few microamps.  For a P-Channel mosfet in a 12v circuit, make sure the chosen one has a gate which can survive at least 15v, maybe 24v.

Example 16v comparator with a quiescent current at around ~2ua.
https://www.digikey.com/en/products/detail/texas-instruments/TLV3702IDR/1669452


Note that a 40v version of a ~1ua quiescent comparators exist in the 2.25$ each range like this one:
https://www.digikey.com/en/products/detail/texas-instruments/TLV1861DBVR/17748398
(This one needs a pull-up resistor when used with a P-Channel mosfet...)

These guys would be useful for a battery system operating in the 24v-30v range.
For the P-Channel mosfets, you might need a series resistor to the gate with a ~15v protection zener diode.

[Dannyx]

The last post made me reconsider my options. I started off with a basic On/Off control which I was going to accomplish the easiest way possible: a LOW side N-channel driven by the comparator. I then kicked it up a notch with a HIGH side P-channel FET (still relatively easy to accomplish, but unsatisfactory in terms of resistance). I then kicked it up once more with a HIGH side N-channel FET AND a driver. This increased the part count and complexity.

I now realize I missed one in-between option, which was mentioned just now: 2 (or more !) HIGH side P-channel FETs. Would this be feasible ? Can a comparator drive such a gate capacitance reliably ?

The TLV family is unobtanium unfortunately....What I WAS able to source based on suggestions so far are:

-IR2117
-FDA217
-VOM1271
-MCP1804 as the LDO I mentioned
-LP2901 as a "ultra low power comparator". Not sure if this last one is relevant/good/useful in any way. It's also a VERY old part....

[BrianHG]

The last post made me reconsider my options. I started off with a basic On/Off control which I was going to accomplish the easiest way possible: a LOW side N-channel driven by the comparator. I then kicked it up a notch with a HIGH side P-channel FET (still relatively easy to accomplish, but unsatisfactory in terms of resistance). I then kicked it up once more with a HIGH side N-channel FET AND a driver. This increased the part count and complexity.

I now realize I missed one in-between option, which was mentioned just now: 2 (or more !) HIGH side P-channel FETs. Would this be feasible ? Can a comparator drive such a gate capacitance reliably ?

The TLV family is unobtanium unfortunately....What I WAS able to source based on suggestions so far are:

-IR2117
-FDA217
-VOM1271
-MCP1804 as the LDO I mentioned
-LP2901 as a "ultra low power comparator". Not sure if this last one is relevant/good/useful in any way. It's also a VERY old part....

Use the TLV1861DBVR.  This one's supply current is 440na per channel, not the ~100x 60ua of your bipolar one.  (x2 and x4 versions exist, in dip 8 and dip 14 packages.)
You get 40v so that charging a 24v cell close to your 30v would break your chosen comparator.
Also, it has CMOS inputs, so, when dividing the battery voltage to test against your voltage reference or regulator, you may use higher resistance resistors keeping you batteries load to a minimal and your resistor divider cool.  (IE: less input current, measured in the picoamp range compared to your bipolar comparator nanoamp range.)

I now realize I missed one in-between option, which was mentioned just now: 2 (or more !) HIGH side P-channel FETs. Would this be feasible ? Can a comparator drive such a gate capacitance reliably ?

The more capacitance just means a slower turn on/off.  So long as you meet 0v for off and a good -12v to -15v for on, you will be fine.  It will probably operate at the speed of the photovoltaic if you use a big series gate resistor like over 1k ohm.

[BrianHG]

Try this comparator:
https://www.digikey.com/en/products/detail/texas-instruments/TLV1812QDRQ1/18716516

With push-pull output, all you need is a series resistor from the comparator's output to the P-Mosfet gates.
It has +/- 30ma output drive.
Use 1 resistor per P-Mosfet.
If you choose a P-Mosfet with a built in gate zener diode protection, that is all you will need, otherwise, you will need 24v zener diodes.  (Choosing a P-mosfet which can handle 24v at the gate means 0 current with a 24v battery supply.  Or, very little when using a 100k or 10k series resistor.)

[BrianHG]

Find a low quiescent negative voltage regulator of -12v or -15v and all you need is the comparator, a few resistors  and caps with the P-Mosfets.  No zener gate protections.  You can also drive the gates with the full 30ma drive.

Basically, the comparator's VCC is tied to the bat + once again, but the comparators GND is tied to the output of the regulator giving it a false ground 12v down from the batteries 24v.  The comparators push-pull output will swing between the batteries V+ and down to 12v below, perfect for the gate.

You voltage sense will be a resistor divider going all the way down to the true 0v, 24v below, divided so you will see around +15v at it's - input.  The + input will be a fixed voltage divided reference between the -12v regulators output and the bat V+.

I know this is a little up-side-down, but when the battery voltage drops, that +15v also drops and once it goes below the + input reference, the comparator's output will go high turning off the P-Mosfet.

[Postal2]

A completely different approach.
Integrated power rail switchers.

VN7050ASTR      https://nl.mouser.com/datasheet/2/389/vn7050aj-1852754.pdf

BTS5010     https://nl.mouser.com/datasheet/2/196/Infineon_BTS5010_1EKB_DS_v02_00_EN-1511519.pdf

Benno

These (above) are very unreliable chips, do not use them. Use something like BTS432E2.
https://www.farnell.com/datasheets/1836336.pdf

[Kim Christensen]

Basically, the comparator's VCC is tied to the bat + once again, but the comparators GND is tied to the output of the regulator giving it a false ground 12v down from the batteries 24v.  The comparators push-pull output will swing between the batteries V+ and down to 12v below, perfect for the gate.

Yup... That's how I'd do it with a P-FET. Simple circuit.

[Dannyx]

Thanks for the info.

Like I said, I cannot source a TLVxx comparator anywhere unfortunately. Shipping is way too expensive.

With that in mind, what should I look out for in a comparator ?

-maximum operating voltage, that's probably at the top of the list as #1
-maximum sink current (or source current if it's a push-pull), that's #2
-whether it's a push-pull or open-collector config, that's #3
-other more intricate stuff like bias or offset ?

If a comparator appears to be suitable in all other aspects but fails the voltage range, it's probably not necessarily a deal-breaker, since I could use the LDO to act as both the supply AND the reference, as previously discussed.

Actually, let's try this: HERE'S a link to the website where I get my sh!t from. I linked the comparator section. That way, you can see what I'm seeing. The results may differ because of your location, but it should be good enough to play around with the filters...

[BrianHG]

If you use my negative LDO regulator trick, then your comparator only need to operate up to 12 to 15v.
Does Aliexpress ship to you in a cheap manner?

Also, if you can do microcontrollers,  thing like cheap 8bit PIC guys have built in voltage reference and ADCs if you want to software manage your battery, but, that's a whole other can of worms.