Copyright 1999 R.G. Keen All rights reserved.
If you're an effects builder or repairer, you know by now that it's a good idea to protect the circuitry in the effect from the consequences of a reversed battery. While many older effects can withstand a momentary reversal, some more modern components can be damaged if you accidentally touch the battery contacts the wrong way during replacement. Sometimes this is subtle damage.
For instance, everyone pretty much knows that CMOS and some opamps can be killed by reversing their power supplies. This happens because the substrate of the entire chip is isolated from the circuit by a reversed bias diode diffusion. If the power supplies are reversed, this diode is forward biased and if any pin is not limited to less than maybe 20-100ma of current, you can get enough current to burn out the bonding wires to that pin - and your fancy opamp becomes a "Darkness Emitting Diode", also known as a "DED". What many people don't know is that although bipolar transistors are not usually killed by being reverse biased, you can permanently degrade the noise performance of a low noise high gain amplifier transistor by reverse-avalanching its base-emitter junction even once!! Can you spell "hiss"? I thought you could.
Ok, so reverse polarity is a Bad Thing. How do we prevent it? The standard wisdom is to use a diode. Not much cleverness needed here, and only two ways to do it - series and parallel. A diode hooked in series with the power supply will allow current to pass only in one direction. If you put the diode in the + lead, the diode takes all the voltage drop if the power leads are reversed. The circuit sees essentially no reverse voltage. The problem is that you pay for that protection in voltage. The diode's forward voltage drop is subtracted from every battery, so a fresh 9.5V battery becomes a middle-of-the-road 8.8V to the circuit that it's powering. And the battery "wears out" 0.6 to 0.7V early as well. Still, better than nothing, and you can get special Schottky diodes with 0.4V drops or maybe germanium diodes for 0.3V drops if you work at it.
A cleverer arrangement is to put the diode in parallel with the effect so it's reverse biased by the normal polarity. When the voltage is reversed, the diode conducts heavily and clamps the reverse voltage to no more than one diode drop. You don't even have to pay the diode forward drop in normal operation. Slick, huh?
Of course, when the voltage is reversed and the diode keeps the reverse voltage clamped to 0.7V, the diode and the battery are engaged in a duel to the death; the battery is determined to bring the voltage up to its internal voltage, the diode is determined to hold it down. The current that flows will often heat a 9V battery so hot you can't hold it in your hand, and may burn out and short the diode - now the pedal is "dead" because the diode died and won't let any voltage reach it the right direction, either!
There is a third way. We'd like to use a modern semiconductor switch to turn the voltage on when it's the right way round and not when it's wrong. Here's a way to do that.
We break the power line and insert a MOSFET transistor. In a typical effect with a positive power supply and a negative ground, we break the + supply line and hook up a P-channel MOSFET with its drain to the battery side and its source to the effect side, and its gate tied to ground with a 1M or so resistor.
"But wait!" I hear you say - "The MOSFET is hooked up backwards. The source should be more positive than the drain for it to work right."
Well, yes and no. If we hooked the MOSFET up the "right" way, source to battery and drain to effect, it would indeed turn on with the normal polarity of supply voltage, and provide excellent forward conduction. However, there is an inherent diode in the MOSFET structure. There is a diode that is an unavoidable part of the way MOSFETs are built that is effectively "connected" with its anode to the drain of a P-channel MOSFET and its cathode to the source of the device. If the MOSFET is put in "correctly", this inherent diode will be forward biased and conduct when the power supply is reversed, and we have no protection at all. If we put the MOSFET in "backwards", the inherent diode is forward biased for the normal polarity, and reverse biased if the power supply is reversed. So the inherent diode works correctly for protection. (The more techno-oriented among you will immediately say - hey! Here's a way around that pesky reverse diode. Yes and no. It only works for big enhancement voltage on the gate to drain and even then only within the maximum gate-source voltage.)
But how about that diode's conduction voltage? And how does the MOSFET work? and?... and? ... and?
The fact is, at voltages less than about one diode drop across the drain/source of a MOSFET, the MOSFET looks like a voltage controlled resistor. If we can turn the MOSFET on hard enough (that is, to a low enough resistance) then the drain/source channel will have a lower voltage drop than the inherent diode, and will conduct all the current! The voltage drop will be less than the diodes - in fact, much less if we get a lot of turn-on voltage on that gate. For a P-channel MOSFET connected as shown in the leftmost schematic, the gate is held MUCH more negative than any point on the drain-source channel, and so the MOSFET is turned on as hard as it can be. For typical MOSFETs, the voltage drop across the drain-source channel will be less than 50mv. In fact, I set up a test circuit for a 50 ma load and a BS250P pulled randomly out of a bag. I could not accurately measure the voltage drop across the MOSFET at 50ma - it was well under 1mv! 50 ma is much more than any typical pedal consumes.
============Update 8 October 2005 ========
An alert reader found that he got voltage losses across his BS250P p-channel MOSFET of up to 150mv at 20ma. I did some quick garage archeology and found that the actual devices I tested with were labeled "BSP250", not the readily available BS250P. This turns out to make a difference. The BSP250 is available today only in the SO223 SMD package. Its datasheet indicates an Rdson of 0.25 ohm. This leads to a calculated forward drop of about 12mv. Rerunning my original test I got drops of around 1mV with my BSP250 parts. Apparently, I got some in a TO-92 package that's not made any more.
What's an effects builder to do?
Well, you could use the IRFD9024 in the four-pin hexdip package. Its rdson is spec'ed at 0.175 ohms, so it should give great results. In the TO-92 package is the Supertex VP3203N3 with rdson of 0.6 ohms and the Fairchild FQPF11P06, rds =0.175. All of these are between $0.80 and $1.00 at the time I write this.
However: losing anything less than 200mv of power supply is pretty darned good compared to the 600mv of a silicon diode. It may be that a garden variety BS250P or the **cheap** Supertex VP2106N3 at $0.31 would work very nicely for you. Or you could spend $0.50 more and get one of the above super performers.
Note that you probably should use a gate protection zener. For the P-channels, hook anode to gate, cathode to source. For N-channels, hook anode to source, cathode to gate.
=====================================
When the battery leads are reversed, the MOSFET acts like an open circuit, as the now-positive gate cannot enhance the drain-source channel, so there are only leakage currents flowing, and the inherent diode is reverse biased. So - the thing looks like an almost perfect polarity protection switch. Almost zero voltage drop in the forward conduction direction, almost no current flow in the reverse direction.
If you have a Fuzz Face or similar PNP based, positive-battery-grounded pedal, use an N-channel device as shown in the second schematic.
But why not use the N-channel device in the ground lead of the negative ground effect? A couple of second order reasons. This does work just fine from the electrical hookup view. From a practical standpoint, you want the MOSFET in the battery-to-jack lead so the jack can still provide power switching. That makes it awkward to mount the MOSFET, which is after all, a sensitive component in its own right. I chose to switch the non-ground side of the power supply because this can be done with all (hah! both!) parts on the effect circuit board by cutting a trace where the power comes onto the board and splicing in the MOSFET and adding the gate resistor to ground. This makes it easy to retrofit older effects, and the complications of mounting that the battery-to-jack lead presents don't come up.
Retrofitting Old Pedals: First determine if the pedal is (a) positive ground or (b) negative ground, and then whether the pedal has (1) no polarity protection (2) a series diode protector (3) a parallel diode protector. Whether the pedal is positive or negative ground will determine whether it needs a P-channel MOSFET (negative ground pedals) or an N-channel MOSFET (positive ground pedals).
To do this, unclip the battery from the battery clip and turn the pedal power on, perhaps by plugging in a signal cord to one of the jacks, or by flipping a switch. This ensures that the battery leads are properly connected to the circuit. With your meter set to the "diode resistance" scale, measure the DC resistance to ground (outer ring of the input jack) from both of the battery clip terminals. You should find that one of them shows zero resistance to the input jack ring, and the other is either open, shows a diode reading, or briefly goes to zero and then rises in resistance. The one that shows zero ohms at all times is the grounded lead. For the vast majority of pedals, this is the black or negative lead.
By and large, the only pedals that will have the red/positive lead grounded are ones that use PNP germanium transistors, such as vintage or recreation Fuzz Face units or other pedals that probably prominently advertise "germanium".
Suitable MOSFETs are BS250P for P-channel and BS170 or 2N7000 for N-channel. They'll cost you about $0.50 to $0.75 each. Be sure and take static safety precautions with these things, as they are quite sensitive until they're in the circuit.
Just today I've had email from a fellow who wants to use four MOSFETS in a bridge to make effects insensitive to battery polarity. The bridge is arranged to make either battery connection come out right at the effect board, just like a bridge rectifier with diodes, but with almost no voltage loss. This works too, and is one example of the use of MOSFETs as synchronous rectifiers. About the only drawback is that you now have to use four MOSFETs and resistors to get the conduction direction right, and you may not want to pay for this if one MOSFET will get you protection.
So - go protect those pedals!!