Submitted by asgard on Thu, 2013-01-31 14:52

This is J.R. Stoner's blog post.

This is a picture of my muse, Mr. Humphrey Lyttleton Katz or Chairman Humph, for short.

He is saying here, "Go away. You're so boring.  Get over yourself."

Actually Humph is probably the neediest cat I ever had, in his own weird way.

Enough about Humph.  Here is a listing of some of the projects I have been working on, of late:

This one is called the Powah.  It is just a little on/off toggle with an indicator LED.  It is designed to mount on one end of my breadboard, so as to reduce the wear and tear caused by excessive insertions of power bus leads.  The board is approx 1 cm^2.

Pretty nifty, eh?  I searched around and the closest thing to this is a power switch from Pololu, which I did not find that interesting (too large and did not have the 100-mil form factor. Here is a schematic:

It is very simple, as you can see.  The main limitation is that the power for the IC is tied to the input, so the DC level is limited to 6V.  Turned on, the current loading from the Powah is 3.06mA. Turned off, the loading is unmeasurable with my crusty old B&K multimeter.  If I can sell enough of these, and perhaps some other ones I am working on I can spring for a nice Fluke one.

Here is an alternative design I have been noodling right now:

The main advantages here are:

The MOSFETS are complementary and on the same die, so they are matched.  And they are much beefier so the current passage has more safety (I would be comfortable passing at least 2A through this device).

The Zener limits the gate power supply to 5.1V, so the input switching can go all the way to the breakdown limit of the P-channel MOSFET, which is 20V.

Turning on the switch requires only a short button press, while turning off the switch requires a longer button press, as a fail-safe.

The inherent hysteresis of the Schmitt front-ends on the gates means that the reset signal on the second pair requires a time interval before the latch signal is active.  This is a free way to debounce the circuit.

The main disadvantages are the increased complexity and part-count requires the board real-estate to also increase.  I have managed to fit the whole works on a 0.8 x 0.5 inch board.  There are, to my knowledge, anything like a 74LVC2G7002 device out there.  I wish there were, as I think I could fit 2 SO8's better than the SO14.  Here is a 3D rendering of what that board is likely to look like:

Anyway, enough of my blather.  For now.


Submitted by Camo on Thu, 2013-02-07 12:39


Can you attach an Eagle or gerber file and a BOM for lazy folks like myself?

I myself would probably use a switch but w/e. I like it!

Submitted by Grand Flibbergibiter on Thu, 2013-02-07 22:34


Sorry to sound critical, but toggle pushbutton controls are probably the worst user interface strategy ever conceived.  It is more than confusing to have a single button do something, and also do the inverse of that thing.  I think of the many Prius drivers who have died while trying to get the button marked "Power" to turn the power off.  Of course, the Prius "Power" button has an led to show you the current power status.  Too bad that it's completely unreadable in sunlight. 

The unambiguous push button strategy is to have a button for ON and one for OFF.  Perhaps ON being a smaller green button and OFF a larger red button, perhaps octagonal.  The component count would drop from 20 in the above design, to 3 for the 2 button design.  Each of the two buttons would feed the inputs of a simple R-S flip flop.  Contact bounce is forever absolutely zero.  The flip flops come 4 to a chip, which costs about a dime and is available in any logic family.

Falling in with looking cool is not without some costs.  My Prius just went out of warranty and it will soon have a real OFF switch.  I think we have all have had the experience of a computer not doing that which we ask.  When the computer in my car can most definitely kill me, I think it's time to have a real switch for me.  The problem isn't really the floor mats, it's the driver being unable to turn off the engine when there is a problem with the floor mats.  As an engineer, I am certain that Toyota could have invented an OFF button which is not a request to a computer.  But seems like they gotta be modern!  Clearly, unambigious user interface design still has a legitimate place twixt man and machine.

Submitted by Camo on Fri, 2013-02-08 15:19


I think that rant is a bit much considering the actual use of this design.

I do hope that asgard is going to post this to a version control system (github, bitbucket) so we can fork it ;)

Submitted by feurig (not verified) on Sat, 2013-02-09 01:43


Agreed. As one of the moderators here I almost deleted this account since even G.F's own description of himself screamed troll. And given that he went to all the trouble in order to rant this rant, I am not sure my my instincts were wrong. 

Asgard. Nice post! Its odd when I follow a link from Dangerous Prototypes (which I read like some people read slashdot or hack-a-day) and it comes back here.