Transistor with fast switching at low Idd

[Aggressive_Doughnut]

Howdy all. Have a nano power device that periodically checks battery level.  I turn on this circuit with a MOSFET, but it requires around 10mA to switch on fast enough to hit the ADC sample window, otherwise I have to wait for the timer to come back around to sample it, drawing current the whole time.  Ideally I'd be able to sample within a few microseconds even with only maybe 100 microamps (or less) at the drain. At that current the best I've been able to do is about 50 microseconds which is an order of magnitude off. I have been looking unsuccessfully through digikey since there isn't really a field for switching time test current, and pretty much all of the test currents are 10mA anyway.  Anyone know if there is a specialist transistor (BJT is fine too) with fast(ish) turn on time at low drain current with low gate drive voltage?

[David Hess]

I think you need to find a part with much lower capacitance so that the AC drive requirements are as small as possible.  Small signal MOSFETs exist but are not very common anymore except in RF applications.  A JFET may be suitable if enough gate voltage is available.

[poorchava]

What kind of fet ar ryobi using right now and how much current can you supply for switching?

[Zero999]

What voltage does it require? At 10mA, a couple of MCU outputs or a logic gate might do the trick.

[Marco]

I turn on this circuit with a MOSFET, but it requires around 10mA to switch on fast enough to hit the ADC sample window

Lets say this 10 mA is running into the supply rail capacitance of the circuit ... if that capacitance needs to absorb 10 mA for 5 microseconds before the voltage rises sufficiently to supply the circuit, then that capacitance needs to absorb 10 mA for 5 microseconds before the voltage rises sufficiently to supply the circuit. The MOSFET is irrelevant in this respect.

The MOSFET needs to be fast enough to ramp to 10 mA in a couple of microseconds, but if it does then having a faster MOSFET won't make a difference ... only a smaller supply rail capacitance can make a difference.

[T3sl4co1l]

What is this circuit being turned on?  Why does it need to be on so briefly?  Why does it need to synchronize to something else?

Tim

[Aggressive_Doughnut]

Busy Monday, yeesh.  I think I misled some people by not specifying that the 10mA are at the drain - I have no problem driving the gate with the mcu (I have one unit with a 2n7000 and one with a BS170 - same performance).  If the drain gets less current than that, it switches slowly, increasing the circuit on time to sample. So right now I have a divider with 120 and 180 ohms, which gives me 10mA from my 3V battery (2 D-cells). 

As I said, the device is extremely low power, and I am looking to minimize usage whenever possible - I could just monitor the supply less frequently, or I can just increase the values in the divider and leave the circuit on longer to get a sample, but an ideal solution would be a small signal fet or bjt that turns on quickly (~linearly with current would be great) with small drain/collector current.  Most datasheets list turn on time for 10mA drain/collector current for small signal transistors, but this falls off dramatically within an order of magnitude or two (from ~10ns @ 10 mA to ~100us @ 0.1mA - so 4 orders of magnitude increase in time for a 2 order drop in current).

All told it's not a huge deal - the client won't lose sleep if the device is expected to last 10 years instead of 20 (I know, I know, battery shelf life is less than 10 years, but very low power + supercapacitor + battery monitor means it can be up for a very long time) but I want to make sure I've done my due diligence and there isn't an easy solution I'm overlooking from lack of knowledge.

[Terry Bites]

Forget transistors: ADG801 35nS 0.3R $2 Digikey.

[Zero999]

Is the MCU also running off the 2 D cells? If so, why not just put the potential divider on an output pin?

Failing that, how about a logic gate, with an open drain output, such as the 74LVC1G07?
https://www.ti.com/lit/ds/symlink/sn74lvc1g07.pdf

[Aggressive_Doughnut]

@terry The switch is a good idea, but the ADG801 is more than they will be willing to spend for battery monitoring.  There are other much cheaper switches that I might give a try, but I worry about some of their leakage current specs.  I have seen one, Union Semi UM3156 that seems to have what I want for around 7 cents US, so that's a probable solution.

[Aggressive_Doughnut]

@zero The MCU runns of the D cells, but has logic level 1.8V, no analog/supply out.  Logic gate isn't a bad idea, but the tend to have larger I_q/I_leakage/etc than I would want.  A switch might work if it doesn't leak too much when it's not connected.