Auto-on circuit. An old one, or this one?

[brett]

Hi
has anyone designed a circuit to auto-detect a signal at the input and turn the power on for a pedal?

In the absence of finding anything about, I have had this idea.
A JFET common source stage could detect and amplify the signal to 3V or more.  The high high input impedance (e.g. 2.2M) would prevent any effect on the signal path.  Next, if that signal were rectified (e.g. by a schottky diode) and used to positively charge the gate of an N-channel MOSFET, the MOSFET could connect the power.  (ie. if the MOSFET drain was connected to V+ and the drain was to the circuit). 

For low power demand (50mA) a BS170 or 2N7000 would do.  For high (50mA to 1A) an MTP3055 would be needed.  Either way, the total cost would be about US$3 for the components, and the space required should be less than 1.5 x 1.5 cm (3/4 inch x 3/4 inch).

The length of time that the circuit would stay turned on with no signal present should be several minutes with a 0.1 to 1uF cap and a low-leakage diode (e.g. 1N4004) used as a "bleed resistor" lleaking a few nanoamps (as per the Millenium II circuit of RG). (Or does an electrolytic cap leak that much anyway?)

Mybe I'm missing something, and it won't work?  Thanks for any help.
I'll draw up a schematic tomorrow (at work.  Shhhh!)

[Sir H C]

Might well work.  Other things to think about:

1.  Zener to protect the gate from too much voltage if you hit the device really hard.
2.  Electrolytics often have some big leakage, can get 1uF non-electrolytic that should leak much less.
3.  The jFET could be a MOSFET instead and then you get the same high input impedance, run it with the gate below threshold then you could keep that whole path low current, as the FET would be off when not seeing signal.

[brett]

Damn! I forgot that the JFET or MOSFET detector would need power all the time.  That might shorten the battery life somewhat.
I'lll draw something soon.

[Paul Perry (Frostwave)]

Think microamps for the MOSFET. It is certainly doable. But one problem if you have very very high input impedance, is that stray static fields might turn it on.

[brett]

Thanks Paul
Yeah, I've decided that about 4.7M is plenty, and to aim for a decent strum to turn it on.
I worked out that the "always on" issue can be solved by using a large drain resistor (e.g. 100k) in the detector and biasing the JFET/MOSFET to "almost off", thereby providing another few hundred kohms resististance.  So a battery might last 6 months or more while waiting for a signal (?).

[brett]

Here's the schematic, using a JFET sensor and a MOSFET switch.
I'm not sure about the JFET biasing.  I suppose the gate is at 0V and the source is biased a bit like the cathode in a cathode-biased tube.  ie current x ohms = V of bias. Current from the load-line on the datasheet??

All suggestions are more than welcome.

[Paul Perry (Frostwave)]

Is that mosfet going to act as a switch? It's bound to cause unhappiness if it finds itself operating in a linear way at any time.
Maybe there is some kind of ultra low quiescent current switch....

[brett]

The idea is that the MOSFET gets a solid voltage (>3V) on the gate and turns on, and the leakage from the gate via the diode is so little (nanoamps) that it will hold "on" for several minutes without any signal coming through.  It is true that after a few minutes it will make a slow transition to "off". 

Oh yeah, I see what you mean.  The voltage drop across the MOSFET will be a few volts and the current might be significant, so the wattage could be an issue.  Maybe an MTP3055 (10A, 2W, US$1) will be needed.  Thanks for the pointer.

[Paul Perry (Frostwave)]

For all I know, ther emight be a "solid state relay" that gives an almost out-of-the-box solution. At least the SSR solution would be guaranteed OK on the output side.
Anyone know if ther eis a "standard" solid state relay? Hate to design around stuff that disappears overnight..

[zachomega]

Here is a practical question...I am not too knowledgeable about this sort of thing...but will a J201 be able to gain sufficient swing to get the 3 volts?  You'll need to know exactly where the thing is biased, but you'll need >3.6 volts to get 3 volts on the gate of the mosfet.  That is a pretty big gain considering the average guitar pickup puts out something like 100-200 mV.  I don't really know what I'm talking about here...But Just a thought I had anyway.

-Zach Omega

[brett]

Hi
a JFET source-follower often has a gain of 15 to 20.  So a chord = 200mV x 15... yeah, it might not be enough.

The CD4066 has several switches in it.  
I figured that a large, complicated divise like that was excessive when I just wanted one switch.  However, I now realise that it uses cascading MOSFETs to make sure that there's almost no linear region, which would be a real advantage.
The quiescent current is only 0.01uA, so that's cool, but the "on" resistance is typically 180 ohms, so I'd need to parallel several of the switches.  But that's no problem.  It might be suitable to use as a detector, too.  (ie switch the power through to several others in parallel that turn the circuit on).
Alternatively, a good detector would be a CD4050B (buffered).  Only .02uA quiescent current, and presumably it would have a gain of thousands (?).  Maybe that would give an almost instant transition between on and off.
Hmmm... Back to the drawing board....

[Processaurus]

For how much pedals (especially boutique pedals) cost they have no excuse to not have something like this idea, everybody leaves their stuff plugged in occasionally.  Nice idea.  First time you leave it on overnight it'll pay for the MOSFETs.

[R.G.]

... um...

CMOS opamp and CMOS 555 timer. The opamp gets you high input impedance, triggers the 555, which turns on power for X seconds by pulling down on the base (gate) of a PNP (P-channel) transistor (MOSFET).

Standby current should be well under battery self-discharge if you do it right.

I can do a schemo if that's not enough description.

[Paul Perry (Frostwave)]

I was hoping you would come to the party, RG! Nice one!
And I wouldn't mind seeing the right hand half of the circuit, at least..

[brett]

Hi
thanks RG
I'm totally in agreement with that.  An op-amp is a whole lot less pins and size than a hex buffer (8 vs 16 pins).  The OPA336 only draws 20uA.  It works off a single supply, too.  I must admit I had only ever heard of CMOS input op-amps, like the CA3130, and didn't even know that 100% CMOS op-amps existed.  Maybe a reader will be aware of a more common CMOS op-amp.  One that I can get in the Oz bush, maybe?

The TLC555 solves the linearity problem.  If I switch the output up when on (rather than down), I guess it can use an N-channel MOSFETs, like the cheap and available BS170 and 2N7000.  The main issue is that the datasheet says it draws up to 300uA, making for only 80 days of stand-by battery life.  I was hoping for 3 or 4 times that much.  But the TLC will work with a lot less than 9V, and only needs to output 3V to turn the MOSFET on.  Maybe a resistor in its power supply line would work.  The chip itself appears as a 30,000 ohm impedance when "off", and a little more when "on", so a 33k resistor might halve its power consumption and run it at about 4.5V.  Or not ??
thanks for any ideas
I'll draw up something tomorrow night.

[Sir H C]

I don't know if Micrel still makes the 1555 and 1557 (#s are from memory) which are SOT35 versions of the cmos 555 timer, simplified with less current usage.  Still if you want minimal current usage, you will either have to roll your own or look at some of the insanely low current parts (Maxim has a bunch of cool ones).  In the world of micropower, 1uA total current is a current hog, and Maxim has some really cool parts that use well under that including some microphone pre-amps.  One of those might suffice into the diodes.

[gez]

http://www.rapidonline.com/netalogue/specs/82-0782.pdf

http://www.rapidonline.com/netalogue/specs/83-0426.pdf