lm10 as comparator

[bigtrout]

Hi...i have a question with a need for some advanced help. Basically i am building an ultra low power solar controller using two seperate comparators. One comparator turns a mosfet off if battery voltage rises above 14 volts( using resistor divider against a reference). The other comparator senses current direction to turn another mosfet on and off. These mosfets back to back on the negative line between the battery and pv panel and act as an ultra low power diode/solar panel disconnect. The circuit works with an lm393 comparator and a tl431 as a voltage reference. It draws 69 milliwatts when off...and 86 milliwatts when running.

I broke my leg and have time to focus on making this draw even less power as a pet project. I can use a lt1004 as reference voltage and lt1017 as comparator and get it down to 5.5 milliwatts off and 25.5 on. I would have to order these parts...but i recently was given a supply of lm10c chips and have been intrigued with the design.

Can i use the lm10c as my reference and two comparators? Datasheets and app notes show there is a reference voltage....a reference amp...and a precision op amp on the same chip...my question....can i use the reference 200millivolts and compare that to battery voltage and run that thru the reference buffer...using the output to turn on one of my mosfets? Then the other op amp for my current sensing circuit? This would let me realize my power savings goal and simplify my circuit without buying new parts. I know its an unconventional approach but it seems this versatile chip is used in alot of unconventional circuits.

 

[(*steve*)]

I would take a look at something like the LMC7221. You'll probably get a couple of orders of magnitude reduction in power usage right there.

The next challenge is to find a voltage reference that will work with a couple of uA or less.

You will have to be very careful in the selection of pull-up resistors, although if you can ensure that pull-up resistors are only carrying current in the ON state for your device (when presumably there is excess power available) you may be OK.

The other trick is to power your circuit from the solar panel rather than from the battery. As long as it behaves sensibly at low input voltages, there is probably no need for the charger to do anything when the solar array is not generating power. Also presumably the solar array will generate sufficient voltage to operate the charger prior to it reaching a voltage high enough to charge the batteries.

If it's powered from the solar panels then the power used by the charger is less important as long as it is small compared with the power available.

Hopefully some of this will be useful

 

[bigtrout]

Thanks...nice comparator there, even less current than the lt1017 at 40uA but im not sure that the large input offset will work for my current sense compare. I will remember it for future projects. And as for the voltage ref...the lt1004 only draws 8uA. Your advice about powering from solar may work to eliminate a mosfet...and the need for current direction sensing so i could save power knock another 10mW or so off the on state power draw.

I still would like to know if my unconventional plan for the lm10 will work because:
I wont have to buy anything to build and just for my own knowledge. The chip is unique and may give me a few new tricks for other builds.

 

[(*steve*)]

Try it. If you can get it powered from the solar panels only, the quiescent current draw will not be as much of an issue (I presume you want to keep the current drained from the batteries at night as low as possible)

 

[bigtrout]

I am going to prototype and see...looking at the app notes and datasheet...the regular op amp can handle 20ma....the reference amp 3ma so i will adjust my mosfet gate resistor accordingly....it doesnt need to be a fast circuit...so it just may work....

Any tricks for powering up from solar when the both the battery and solar 12V are tied together and im doing my switching on the 0V rail? I can power the whole circuit on the solar panel side but still have to read battery volts.

 

[bigtrout]

Well...i remade my solar charger circuit and modeled it in LTspice. There was no lm10 model but the lt1635 is a very close relative to the lm10 so i used it. The reference amp works with a few cautions from the lm10 app notes. It only sources 3mA and sinks 10uA and will get only go to .8 volts of the positive rail. However...in my circuit its driving mosfets to replace the backfeed diode and limit battery charge to 13.8 volts for a low power solar so it doesnt need to switch fast. The model worked perfect. So I breadboarded the circuit....and it appears that it works great. One LM10...five resistors...two shottky diodes..two low rds n channel mosfets and i have a solar controller that draws 5mW...my diode losses using the mosfets are less than 20mW on an 11 watt panel...im happy tomorrow the soldered version gets built.
So anyone wondering about the lm10...yes u have a 200mv reference voltage and in reality two op amps...what a chip!