I am trying to make a circuit that slowly discharges a multi cell lithium battery to storage voltage. I already have a comparator circuit that can turn on or off discharging. I want to support 2s to 6s batteries so we are talking about voltages ranging anywhere from 7.6v to 25.2v. I am shooting for 500mw of power dissipation so I can achieve my form factor requirements for the PCB (11mm x 15mm). Ideally, the same circuit can discharge any cell voltage with constant power. So far I have tried/considered:

  • PTC Thermistor load. This will essentially run the PTC perpetually in its "tripped" state. I have an NTC thermistor in series to handle the inrush current until the PTC heats up and protects the circuit. This best captures what I want. However, I contacted a support engineer and they say the use case is not specified and could negatively impact the life of the device. https://www.murata.com/en-us/products/thermistor/ptc/prg

  • BJT with an NTC thermistor switching a resistive load on or off. I would preferentially avoid this due to extra components necessary.

  • I would like to stay away from MCU controlled constant heaters for simplicity sake.

  • Buck or Boost converter is not practical for form factor and cost.

Any help will be appreciated.

More context:

  • I am creating a battery discharge circuit small enough that can be integrated into a battery connector (like an XT60 adaptor).
  • Users may inadvertently connect a higher cell count than selected or thermally insulate the device so I want overtemperature and overcurrent protection.
  • Users can choose cell count with a DIP switch so I want a constant power load.
  • I also want to decrease the BOM cost of course, so one device that does it all would be wonderful :). So far a PTC thermistor fits the bill but I am wary of leaving it in a tripped state for extended periods.

Edit: One last requirement that I overlooked (I apologize). The device will be "plug in and forget" so it needs to have a very low quiescent current so it can be left on the battery for months or even years without over discharging the battery (which will damage it). The solution I have right now has a leakage current in the microamp range.

  • Current Schematic: Schematic
  • \$\begingroup\$ Sounds like you are in the market for a depletion mode MOSFET. \$\endgroup\$
    – jonk
    Apr 11, 2018 at 18:01
  • 1
    \$\begingroup\$ I know you said you don't want a DC-DC. But you actually do. Use a 5V DC-DC to drive a 56 Ohm 1 Watt resistor. PTC will draw large current prior to trip and will have dissipation that is very sensitive to ambient temperature. \$\endgroup\$
    – user57037
    Apr 11, 2018 at 18:07
  • \$\begingroup\$ I agree with the DC-DC being the easiest way to go. Any solution with decent repeatability will require feedback, and DC-DC controllers are cheap and easy and smaller than anything that needs discrete op-amps for feedback. \$\endgroup\$
    – Selvek
    Apr 11, 2018 at 18:26
  • \$\begingroup\$ @jonk can you explain how a depletion mode MOSFET can be used in a constant power load? I can only find use cases for a constant current source. \$\endgroup\$ Apr 11, 2018 at 18:31
  • \$\begingroup\$ Thanks guys! I actually do want heat dissipation to be sensitive to ambient temperature. The device I am making will fit on the end of an XT60 plug so users may stuff a thermally isolated bag full of batteries with the device on. By experiment I find 500mw to be a safe conservative maximum to be safe to touch but it is desired to increase power in a cold environment or lower in a hot environment. As for cost, I am using a TL431 as both the comparator and the shunt. I do not think it can get any simpler / cheaper than that. \$\endgroup\$ Apr 11, 2018 at 18:38

2 Answers 2


The type of PTC you need is a ceramic one, I don't think the polymer "polyfuse" type are suitable. PTC ceramics were commonly used as fairly constant temperature heaters, and probably still are in hair-tongs etc, where they use a PTC thick film resistor printed on a ceramic plate. However these are commonl;y used today for pretty high temperatures, which might be more than you would want inside a soft plastic connector. If you use overcurrent protectors, you will also find that these are high temperature devices - they will have to idle at (say) 120C. If you use an SMD device. this means the soldered joints and pcb have to take that, which they might not like long term. Against that, this is not really a very high value or very long term device so it might not matter. Leaded disc types transfer much less power to the pcb and solder, so would be a more reliable choice. There are PTCs for sensing with s lower slope, and which have a much more constant slope rising from lower temperatures.

Motor protection PTC's have much more that range you are looking for. When you look at the lower sense temps (50-90C e.g. B59100M1070A070) you will see that the don't have the nasty NTC part at the start of the curve.

FYI, here is a constant power, constant current circuit. This also meets your objective using very cheap components, and without requiring a very specific thermistor.

The constant current emitter follower Q2, has its reference voltage subtracted by the current mirror Q1Q3 in proportion to the supply voltage. The result is a linear dropping current i.e. constant power when the slopes are balanced. R3 is the primary slope adjuster.


simulate this circuit – Schematic created using CircuitLab

We can simplify this using Q4 to replace the zener (Q3 cancels Q2 VBE, so Q4 provides the constant reference voltage). This gives an NTC, but this is what we want. 2 dual-npn and 3 resistors is about as cheap as you can get.


simulate this circuit

Flipping the circuit around we can gate it with a TL431 to control stop voltage, and give near zero off current


simulate this circuit

  • \$\begingroup\$ Indeed the PTC device I am looking at is the ceramic BaTiO3 based one. This one specifically is perfect murata.com/en-us/products/… \$\endgroup\$ Apr 11, 2018 at 22:05
  • \$\begingroup\$ PTC fuse type thermistors are just not that great for this job, in particular they have an NTC region. See additional notes above \$\endgroup\$
    – Henry Crun
    Apr 11, 2018 at 23:37
  • \$\begingroup\$ @HarrySvensson I liked the smaller circuits, but then they can't be edited/simulated anymore. Any solution? \$\endgroup\$
    – Henry Crun
    Apr 11, 2018 at 23:56
  • \$\begingroup\$ @HenryCrun I've given up on CircuitLab because it constantly bugs for me. It's either bugged out because I touched them, or bugged out because that's how CircuitLab is, or bugged out because of the letter "m". - Okay, I went into your edit, removed the "m" from the top schematic, then scrolled down and clicked "edit this circuit", that worked. (I did not apply this edit) - Perhaps you want to revert your edit to the moment before I changed all of them. - You can revert by pressing "edit" and then at the top it says "Rev", scroll to the one you want. - So it bugged out because of the letter "m" \$\endgroup\$ Apr 12, 2018 at 0:38
  • \$\begingroup\$ Thanks for the reply. I understand they have an NTC region, but as soon as the PTC part kicks in it will protect itself and hover somewhere around it's curie point. I do not see why the NTC region is that problematic. \$\endgroup\$ Apr 12, 2018 at 1:26

I'd use a LM239 or any other cheap dual comparator with open collector outputs.

Comparator #1 with a cheap voltage reference/LDO and a resistor divider to set the discharge voltage.

Comparator #2 with a thermistor as temperature sensor makes a thermostat.

Open collector outputs are joined together, creating a logic AND to only draw current when temperature is below a certain value, and voltage is above the threshold.

You'll need a transistor as a switch, and a SMD resistor to dissipate the heat. Make sure to place the temperature sensor next to the resistor and that their ground pads are connected with lots of copper. If you want to use the PCB as a heat sink you'll need a copper pour anyway.

  • \$\begingroup\$ Thank you for your suggestion. Unfortunately, I think the quiescent current will be too high for my Low Iq requirement. I apologize I left that requirement out in my original description. \$\endgroup\$ Apr 11, 2018 at 22:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.