Why do switching FETs just stop working?

[Mark Hammer]

In responding to a thread on MEF, the topic came up of the venerable 2SK30A JFET, used for switching purposes.  Often, someone will post "My Boss/et al pedal doesn't switch anymore.  How come?", and the final remedy will be to replace the switching FET with a new one.

But at the same time, just try and imagine how many K30As are out there in 4 and 6-stage phasers, humbly and flawlessly serving their masters for years on end.  Why do they seem to give up the ghost when used for switching, but apparently not for phasing?

Is there something about a switching JFET that doesn't survive well or risk-free in the switching environments we provide? 
Is there a limited lifespan, and XX thousand turn-on pulses later it's toast? 
Are there switching FETs which are safer bets for longer life?  If so, do they have tradeoffs?
Is there anything one can do or at least avoid, with respect to FET-switching to extend their life or reduce likelihood of failure?

I know that DOD has turned to the J113 for switching in a great many of their products, and I tend to see very few posts from people saying "My Grunge/Milkbox/Classic Tube pedal just stopped switching, why?". 

Perhaps those who repair such things for a living might have something more informed to say about failure rates.

[O]

Ha! 

I was actually just thinking about the exact same thing since I'm staring at 3 "dead" Blues Driver pedals. I think the switching FETs are more sensitive at either reverse voltages or static (or in this case; MODS!!!). In most cases of reverse voltage death I've seen, the diode goes and takes the fet switches with it. As mentioned, these 3 pedals came in with incomplete mods so I'm thinking that at some point, so stray voltage fried them and they stopped switching.

With the amount of failed Boss switching FETs, I think it would be beneficial to have a quick and dirty replacement cross reference chart. It would be perfect for everyone out there that has fried their pedal to get them up and running again.

[Mark Hammer]

While there is the difference in JFET used, I might note that the circuit used for switching FETs in DOD boxes is typically different than the discrete flip-flop used in Boss pedals.  I wonder if that has any bearing on the possible sources of FET-failure you allude to?

[O]

You'd think that the conversation would be flip flopped (bad pun) as DOD is generally, lower quality than say, Boss. It definitely would be interesting to compare/contrast both circuits to see what makes DOD's implementation a bit more sturdy.

[Mark Hammer]

Well at this point, I don't know if it IS more sturdy, since I'm only going by what I can remember of people posting things about repairing "stuck" DOD pedals over the years.  For all I know, we may not see many such posts because owners are more likely to sell or trash such pedals.

Having said that, I can't recall any cases in memory where someone was advised to just get another J113 to replace the one that was fried.

[Fender3D]

While there is the difference in JFET used, I might note that the circuit used for switching FETs in DOD boxes is typically different than the discrete flip-flop used in Boss pedals.  I wonder if that has any bearing on the possible sources of FET-failure you allude to?

I repaired lotsa Peavey Autographs, they have 4 J117 (or 113?) muting the output while switching presets (then uP served) shorting signal to ground when failing...
IIRC a mixer series with similar FET switching had same failures too...

[JDoyle]

If a JFET is failing in a 9V pedal, I would hazard a guess that it has something to do with forward biasing the gate and exceeding the Ig limit - which is almost always 10mA.

Regards,

Jay Doyle

[Earthscum]

Probably not much insight, but I was pondering this, kinda, when I was putting a J201 in a DOD Overdrive to switch the diodes in and out. I was wondering which way I should put it in, and considered using an MPF102 instead because of the symmetrical channels (and lower "ON" resistance).

I wonder if they don't get forced into a reverse-biased situation somehow? My buddy blew his Zen clone (RC4558P) and all that happened was that they shut down stage main power before he had his gear unplugged. It seemed like it caused some kind of fly-back pulse that went through the cords or something, like he had his cable maybe coiled over the mains AC extension cord when they powered down.

[R.G.]

I'm with Jay on this one. JFET gates are not particularly sensitive to reverse breaking, but too much current in either forward-diode or reverse breakover kills them. The channel is generally good unless you exceed the power rating with too much current.

Pulses, leakages, and other stuff can cause this. There is a whole set of circuitry intended to make inputs and outputs safe for the world; these are much like the protection circuits on CMOS logic, with series resistance and diodes inside the resistances to clamp voltages to the power supplies. Done properly, you never hear these parts. However, it can make a pedal be higher-voltage-mod-unfriendly. 

[Mark Hammer]

So is this an argument for placing some sort of series resistance ahead of the gate?  Alternatively, are there some JFETs that are more likely to be "immune" to outlier/anomalous conditions?

[kwijibo]

Well at this point, I don't know if it IS more sturdy, since I'm only going by what I can remember of people posting things about repairing "stuck" DOD pedals over the years.  For all I know, we may not see many such posts because owners are more likely to sell or trash such pedals.

Having said that, I can't recall any cases in memory where someone was advised to just get another J113 to replace the one that was fried.

I don't think DOD pedals can really be compared in this case. Up until the late-90s Digitech series DOD were using a much cheaper mechanical actuator than BOSS, and the pedals tended to fail from either a footswitch actuator that would not return-to-zero or a broken tactile switch. I've replaced mounting hardware and tactile switches in about 10 DOD pedals now. It doesn't seem like the mechanical components last long enough for the electrical components to reach age of failure.

[R.G.]

So is this an argument for placing some sort of series resistance ahead of the gate?  Alternatively, are there some JFETs that are more likely to be "immune" to outlier/anomalous conditions?

Using a resistor by itself can help, but there is a value of voltage which will reach 10ma (or whatever the death current is) for any resistor. The standard input protector is a series resistor followed by some variation of diode or zener clamp, possibly followed by another series resistor. The diodes/zeners clamp the voltage to a safe value, the first series resistor holds down the current that flows so it does not kill the clamp diodes for any reasonable voltage. Clamp diodes can be 100ma (1N4148) or 1A (1N400x or 1N5819 Shottky) pretty easily and cheaply, and the bigger ones will have 100A peak current ratings, so even for hundreds of volts (stray AC) or thousands of volts (static discharge with limited energy storage) the diodes live through it and clamp the voltage to safe values. That gets  you CMOS protection, where there is no diode action on the gates.

JFETs can be forward biased by a diode drop. So the second resistor can hold down the current in the forward bias direction to less than the death current; it's sometimes impractical to make a JFET gate not forward conduct, as a JFET distortion circuit may depend on it.

Kwijibo brings up an interesting point. Switching circuit JFETs may not go to a jack, but they *do* go to off-board components in the footswitch, which is another path in for transients. Good CMOS practice is to never take a CMOS input pin off-board without some kind of protection/limiting on the PCB between the wire and the pin.

[amptramp]

Check a typical JFET switching design like this one:

http://www.generalguitargadgets.com/boss/BossBD-2.gif

Notice that the gates of all JFET's are pulled down to shut off, but the gate is driven through a diode.  When the JFET is supposed to turn on, the pulldown is removed, but there is nothing other than diode leakage and the diode is not driven high - leakage would take it to about the same voltage as the JFET channel.  But since it is not driven by anything other than leakage (and a reverse-biased diode does not provide much that can be called drive), there is no telling what the voltage is sitting at on the gate.  Maybe the JFET picks up fatal gate transients when it is not driven or maybe it just doesn't pull up and remains low, looking like it is failed but actually just responding to the gate voltage.

The National Linear book shows a JFET switch where the gate was pulled down by a diode as in the example above but pulled up positively to the source voltage by an op amp cannected as a unity-gain buffer driven from the JFET source and driving a resistor to the gate.  When the diode pulled the gate down, it was down.  When the diode was allowed to float up, the gate is returned to the source voltage by the resistor, but since the resistor is driven by a buffer, it doesn't upset the voltage at the JFET source because the current comes from the buffer, not the signal line.

[Paul Marossy]

Kwijibo brings up an interesting point. Switching circuit JFETs may not go to a jack, but they *do* go to off-board components in the footswitch, which is another path in for transients. Good CMOS practice is to never take a CMOS input pin off-board without some kind of protection/limiting on the PCB between the wire and the pin.

That my first thought, the footswitch. I fixed a foot controller for one of my old FX units where the CMOS chip had no protection from the footswitches and the CMOS chips eventually crapped out and the only fix was to replace those chips. Worked great after I replaced those chips. But before I did the fix, less and less of the buttons did what they were supposed to do, so it was failing progressively. But with an FET, there's only one input, so when it goes it goes.

[diagrammatiks]

The DOD pedals actually do it a bit differently.

It seems to me that maybe the boss style is all discrete?

the dod's use an ic.

[Earthscum]

The DOD pedals actually do it a bit differently.

It seems to me that maybe the boss style is all discrete

the dod's use an ic.

DOD uses cmos switching. Essentially it actually is doing the same thing as Boss, using mosfets instead of bjt's.

[Paul Marossy]

The DOD pedals actually do it a bit differently.

It seems to me that maybe the boss style is all discrete

the dod's use an ic.

DOD uses cmos switching. Essentially it actually is doing the same thing as Boss, using mosfets instead of bjt's.

Yep.