Ground-controlled Transistor Switch?!?! Any ideas?!

[quietchannel]

Hello all!
I desperately need some help finding a solution to this small conundrum. I'm building a useful audio tool which runs off of a 9V battery, but has no power switch.

There is only one normalled terminal which could be used to turn the power on and off... but its behavior is opposite of the Normally Open switch on a typical guitar effects pedal.

This Normally Closed switch terminal is CONNECTED TO GROUND when the circuit should be OFF. Terminal is NOT CONNECTED (ie:floating) when the circuit should be ON.

Is there some configuration where I can use a transistor or two as a switch?
...with a very low current draw when the switching circuit is connected at some point to ground by the normalled terminal (OFF)...
...but the switching circuit passes the positive rail to the main circuit when the ground is disconnected from the switching circuit (ON)?

Thank you for any feedback! It is very much appreciated.

Robert

[MrStab]

I get the feeling this is too simple an answer, that you've either already considered or isn't the right solution, but do you mean something like this?



EDIT:

This Normally Closed switch terminal is CONNECTED TO GROUND when the circuit should be OFF.

...suggests that what i just said totally wasn't what you meant lol. sorry bout that

[MrStab]

maybe a PNP feeding another PNP, so the ground control switches the first one ON, which in turn would send current to the 2nd PNP's Base, turning that one OFF? not at all elegant, i just felt guilty for posting a redundant answer. lol

[~arph]

Or just replace the pnp in your drawing with a npn?

[PRR]

[R.G.]

Geofex.com, 2012:
http://geofex.com/FX_images/PNP_power_switching.pdf
and earlier in the power switching circuits for switching bipolar power with one contact.

A PNP works fine. A P-channel MOSFET may or may not get you to lower voltage drop across the switching device depending on current level.

[quietchannel]

Thanks all!

RG-- I think your drawing still works opposite of my situation, right?
I need power to the circuit when the switching contact floats, and no power when grounded.

PRR-- Is there a reason to use a hex inverter? would a dual inverter work?

Does anyone know if a combination of MrStab and ARPH's techniques will work?

Thanks again.

[PRR]

> Is there a reason to use a hex inverter?

Note that the 1-to-5 connection allows hex inverter OR hex NON-inverter (buffer); whichever you got.

Why hex? Cheaper than dirt. Low loss due to paralleling.

> would a dual inverter work?

I do not know of a DIP CMOS dual inverter; if it exists it may be no cheaper than a hex.

I know the SMD line sometimes has less stuff per pack; if SMD is your thing AND you can confirm enough current with useful voltage, that can work.

What you really want is One Buffer with CMOS-like input, low-low idle current, and ample output.

[PRR]

> too simple an answer
> PNP_power_switching.pdf

These appear to be "grounded = ON".

OP seems to be asking for "CONNECTED TO GROUND when the circuit should be OFF".

Or am I mis-reading something?

[SISKO]

A PNP works fine. A P-channel MOSFET may or may not get you to lower voltage drop across the switching device depending on current level.

RG: Supossing were using the device in the saturation region, dosent a BJT have lower Vce than a mosfet? At least, that is the way i understood it and it was because the bjt being  a bipolar device rather than a unipolar device (a is a mosfet when its conducting). I dont know if that aply to small signal devices too.

[quietchannel]

PRR-- nope, you're not missing anything. You've got the right idea!
Thanks for the suggestions. I only asked about dual inverter DIP because I have limited PCB real estate.

Is it not true that two inverting transistor amp stages could be used? Maybe their idle draw would be too high... ?

[R.G.]

RG: Supossing were using the device in the saturation region, dosent a BJT have lower Vce than a mosfet? At least, that is the way i understood it and it was because the bjt being  a bipolar device rather than a unipolar device (a is a mosfet when its conducting). I dont know if that aply to small signal devices too.

A bipolar saturates to a low voltage if the current is small and if the effective gain is quite low - perhaps 5 or 10.  Bipolars can get down under 100mV if you bang the base HARD at low currents. When a bipolar is truly saturated, the collector voltage is below the base voltage and you're getting direct transfer from base to collector. It's possible in some circumstances to have the base current actually make the collector voltage rise, although these are degenerate cases.
MOSFETs "saturate" to a resistance. You still have to put a large enhancement voltage on the gate, but for that you get to a low resistance. It's less power to hold a MOSFET to a low resistance than to hold a bipolar to a low saturation voltage in many cases.

Both at low currents. At high currents and for high voltage standoffs, the bipolar starts to win again because you have to dope the MOSFET for higher voltages, and that runs the saturated channel resistance up.

Is it not true that two inverting transistor amp stages could be used? Maybe their idle draw would be too high... ?

Two inverting stages work. Use an NPN for the first stage, with a resistor feeding it base current, and the collector resistance connected to the base of the PNP on the positive supply. Use the switch to shunt the base current on the NPN to ground for off. When the switch opens, the NPN conducts, forcing the PNP on. You get a composite gain through both NPN and PNP that's quite high, so the base current "wasted" in the shunt can be small.

If you replace the NPN with a 2N7000 and 1M or so to V+, and the PNP with a BS84P or similar, you can cut the current for the N-channel gate down by using a 100K to 1M for it, and you may well not need a resistor to the P-channel gate, although it would be smart to use a gate-source resistor to pull up the p-channel gate when it's supposed to be off.

[PRR]

I am not seeing two ordinary BJTs giving low-low-low "off" current. Three (plus several R) could be done, but that gets near the bulk/complexity of a 14-DIP.

Darlingtons tend to show 1.2V drops, and the allowable drop has not been specified(!).

P-devices, BJT or MOSFET, can easily get drops under 0.2V, but need significant drive (BJT needs current, MOSFET needs voltage).

Neither has the Load Current been specced-out. I ass-umed that >>10mA was unlikely. For BJT, that suggests the driver idle current flowing while "off" has to be a good fraction of an mA. If load is really 0.1mA, then the drive might maybe be down in the uA range.

Two P-MOSFETs and two 5Meg resistors is the simplest I see (without clearer requirements). Layout better be just-so, or that much raw gain in a small space is liable to oscillate in the transition. The CMOS DIP is known to be manageable. However GHz bursts on switch-bounce may not be a big deal.

[MrStab]

given that all the replies seem to point to this being a fairly-uncommon and complicated way of going about things, is ground-control definitely the only solution to switching this thing off? i don't wanna probe at your design ofc, but maybe the limited board space should tip the scales away from the ideal switching you first planned. i guess 2 transistors and a few resistors/diodes isn't much, but then there's the caveats.

[SISKO]

A bipolar saturates to a low voltage if the current is small and if the effective gain is quite low - perhaps 5 or 10.  Bipolars can get down under 100mV if you bang the base HARD at low currents. When a bipolar is truly saturated, the collector voltage is below the base voltage and you're getting direct transfer from base to collector. It's possible in some circumstances to have the base current actually make the collector voltage rise, although these are degenerate cases.
MOSFETs "saturate" to a resistance. You still have to put a large enhancement voltage on the gate, but for that you get to a low resistance. It's less power to hold a MOSFET to a low resistance than to hold a bipolar to a low saturation voltage in many cases.

Both at low currents. At high currents and for high voltage standoffs, the bipolar starts to win again because you have to dope the MOSFET for higher voltages, and that runs the saturated channel resistance up.

Thanks R.G. Your knowledge and experience is invaluable!

[quietchannel]

Thanks a ton for all of the replies, especially your in-depth consideration RG.

Here's the basic situation--
This needs to work like a guitar effect pedal-- an input (or output) plug insertion turns the device on. An plug removal turns it off.

The tricky thing is to be more universal it uses XLR/combo jacks.
And it accepts balanced or unbalanced 1/4" inputs or XLR inputs, so the standard guitar effect pedal trick can't work.

There is a version of the neutrik combo jack which offers normalled connections for every single terminal, including an extra one for the shell of the XLR plug, but they are all normalled closed, and open when the plug is inserted.
http://www.neutrik.us/en-us/plugs-jacks/jacks/combo-i-series/ncj10fi-h

No one seems to make an XLR or combo jack with the opposite normalling scheme.


With this in mind, can anyone see some other possible solution aside from my original (opposite from usual) ground switch?

[MrStab]

a parallel to this dilemma seems to be stereo effect switching, or even balanced TRS instead of XLR. maybe that could help with trying to find a tried & tested approach.

an alternative to avoid taking up extra space on the top (though more vertical space) could be a switched volume pot, or some fancy arrangement with a voltage on half a dual pot to activate a transistor over a threshold