An improvement on the soft-latching power switch.

[asgard]

Here is something I consider an improvement on the soft-latching power switch circuit Dave posted in #262.
The difference here is turning on the power requires only a short press on the switch, while turning off the power requires a full 3-second press of the switch.  This avoids the bad news when a device is computing away and a momentary glitch on the mechanical switch turns the whole thing off.  Bummer.

I have also enclosed a photo of the breadboarded test of this circuit.  I hope you guys like this hack.

J.R. Stoner
BDG

[asgard]

For funsies I have this video of the test circuit functioning.

[Mcfly]

Love It! thanks

[johnnyfp]

What's  the standby current on this now?

[asgard]

I'm measuring approx. 0.2 uA.

[johnnyfp]

Nice! Which cap does the off delay?

[onlooker]

Here is something I consider an improvement on the soft-latching power switch circuit Dave posted in #262

This idea is not new. I have posted it twice in eevblog forum.

1). https://www.eevblog.com/forum/blog/eev-blog-262/60/
and
2). an earlier specific  version for UT61E mod.
https://www.eevblog.com/forum/chat/ut61e-auto-power-off-modification/

Have you happened to read them before your posting?

In fact, your R3, C2 and C3 are redundant. 

The key to my circuit is a). the diode,b) the location of C1, c). the switch location (touch ground) and d). on/off is related to how long the switch is pressed. For these, I would say it is a different design than an improvement.

To make it easier, I quoted my earlier post below.

This would be my solution if switch-on-ground and load-independent are required. This is actually a simplified and extended version of one I used for UT61E mod. In that sense, it works as a real circuit.

It works like this:short press of switch (<0.5s) does turn-on; long press of switch (>0.5s) does turn-off. The time constant can be changed by changing C1.

For resistive loads, M3Vto-2 is not needed and loads should be parallel to R4. M3Vto-2 is purely for separating power supply path from switch ctrl path.

[asgard]

No I did not read every single one of the postings in the forum before posting this circuit.  R3 is not redundant, since it establishes the decay slope through the 4.7uF cap so that the off-delay is set to 3 seconds.  This is clearly a simple circuit, so of course it is not going to be "new".  I just established that nobody has marketed anything like it in a form-factor suitable for directly incorporating in a breadboard.  My production board has four holes lined up so only four single-pin headers are needed to make it instantly usable in a breadboard.

[asgard]

What's  the standby current on this now?

...In fact, not having a uCurrent handy (yet) my meter doesn't accurately measure currents this low in the first place.  My SPICE model says the quiescent current through the pass MOSFET is below 4nA.

[onlooker]

No I did not read every single one of the postings in the forum before posting this circuit. R3 is not redundant, since it establishes the decay slope through the 4.7uF cap so that the off-delay is set to 3 seconds.

You did not really answer my question. I would say the transition from your earlier circuit in,

https://www.eevblog.com/forum/projects/two-mosfet-soft-latching-startup-switch/

to the current one is quite striking (I also do not want to try to understand why there is the need to start a new thread on the same topic).

Anyway, if you can't realize R3,C3 and C2 are redundant, you did not really understand how this circuit work.

In this case, C2 is not just redundant, it can be actually harmful. As for R3 and C3, the circuit looks more like a bad copy of the
UT61E mod I did (https://www.eevblog.com/forum/chat/ut61e-auto-power-off-modification/
).

I have there two RC timing circuits (one of them is like your R3 C3) for good reason since I have two power off ctrl signals to handle: one is from manual button press the other is from APO.

The APO signal should do power off ASAP, hence it uses the short time RC. On the other hand, The manual power off needs to have a (3sec) delay since it shares an existing button that short presses are already assigned for the meter use, hence it uses the longer time RC (your R3C3). Also due to the shared nature of the button, one leg of the switch must be on ground.

Now in your case, the longer time RC becomes redundant since you deals only one power off signal. Adjusting C1 is enough to change delay time.   

This is clearly a simple circuit, so of course it is not going to be "new".

Well, this is not entirely true. The self-latching part is a common knowledge. But, as for this particular power on/off mechanism, I designed it from fresh (for a particular use). If someone else did it already, I would like to be enlightened. And I can still claim I independently designed it.

[asgard]

*SIGH* The 100nF cap at C1 is necessary for the purpose of ensuring the initial state of the latch is turned off.  I also want to prevent as much as possible the initial inrush current from being passed to the load.

[onlooker]

What made you think it will do what you said? Did you do any analysis, simulation or real experiment to verify it? If you did, please show it.

Assuming you have a 10V power source.  When the switch is pressed, C1 and C2 are instantly fully charged (assuming perfect caps)  and form a voltage divider with a voltage ratio=C1/C2. That is, about 9V is instantly applied to S2~G2, this in turn turns on mosfet#2 instantly.  For this turn-on behavior, with or without C2 does not make any difference, the difference is just instantly turn#2 on by 9V or by 10V.

Your point about "initial state" was wrong too.

By the way, your choices of the values for resistors are not following some basic and obvious rules (really bad). I am not talking about the RC timings. Do you know why they are bad and how to make proper choices here? I leave this for you to figure out if you can.

Anyway, if you can't do a rudimentary analysis I guess it is pointless to explain more.  I have already spent too much time on this.