Stabilizing drift? I need some help...

[zachomega]

I've been working on a pedal that uses 3n201 Mosfets. 

The thing sounds great but after about 10 - 30 minutes of play the sound flubs out.  The mosfets begin to draw more current as the thing goes on and on.  I have the second gate and the source tied together and grounded through an unbypassed 100 ohm resistor.  As it stands now, the thing draw quite a bit of current at idle (1k drain resistor for approximately 1/2 of the 9 volt supply). 

Anyway, the bottom line is, I need ideas on how to stabilize the mosfets to keep them from drifting and drawing more and more current.  Is a larger source resistor enough?  Any ideas would be greatly appreciated.

Thanks,
-Zach Omega

[Paul Perry (Frostwave)]

I can't say much wihtout knowing more about the circuit, but 1K seems pretty low.
Personally, I'd throw in 10K for a start, just to see if 1. it sounds OK still and 2. if it stabilises. Nothing to lose.

[R.G.]

The 3nxxx MOSFETs are from the very first generation of MOSFET devices. They were use a lot in very high-end radio receiver front ends and IF stages. The second gate was used for gain control. They are fragile as all get out.

I've never seen one in real life. This info is just regurgitated from a friend who worked a RCA back when they did broadcast work and RCA existed as something more than just a brand name.

I never got close enough to them to know how to stablize them.

[zachomega]

RG, you want some to play with? 

-Zach Omega

The 3nxxx MOSFETs are from the very first generation of MOSFET devices. They were use a lot in very high-end radio receiver front ends and IF stages. The second gate was used for gain control. They are fragile as all get out.

I've never seen one in real life. This info is just regurgitated from a friend who worked a RCA back when they did broadcast work and RCA existed as something more than just a brand name.

I never got close enough to them to know how to stablize them.

[zachomega]

The 1k drain load is appropriate for the transistor.  Trust me, I wish it could be higher, but that'd require a larger power supply and I like to run from batteries.  Needless to say, with batteries, the current draw is quite high. 

These particular mosfets have a great sound.  They just aren't stable for much more than 15 minutes. 

-Zach Omega

[johngreene]

I've been working on a pedal that uses 3n201 Mosfets. 

The thing sounds great but after about 10 - 30 minutes of play the sound flubs out.  The mosfets begin to draw more current as the thing goes on and on.  I have the second gate and the source tied together and grounded through an unbypassed 100 ohm resistor.  As it stands now, the thing draw quite a bit of current at idle (1k drain resistor for approximately 1/2 of the 9 volt supply). 

Anyway, the bottom line is, I need ideas on how to stabilize the mosfets to keep them from drifting and drawing more and more current.  Is a larger source resistor enough?  Any ideas would be greatly appreciated.

Thanks,
-Zach Omega

Typically you use the 2nd Gate to bias the device (or for gain control). I've never seen it used with it tied to the source. Seems a waste. You need feedback to hold the bias steady over temperature. The nice thing about a dual gate MOSFET is you can use drain feedback biasing and not effect the input impedance of the device.

--john

[zachomega]

John,

I'm not exactly following what you are saying...I suppose I should add that these are very unusual mosfets in that you don't need to bias the gate positively in order for conduction between the drain and source.  That is, you can have no source resistor, a resistor from gate 1 to source for a voltage reference, gate 2 tied to the source/ground, and a drain resistor and get amplification until the runaway begins. 

-Zach Omega

Typically you use the 2nd Gate to bias the device (or for gain control). I've never seen it used with it tied to the source. Seems a waste. You need feedback to hold the bias steady over temperature. The nice thing about a dual gate MOSFET is you can use drain feedback biasing and not effect the input impedance of the device.

--john

[johngreene]

... get amplification until the runaway begins. 

exactly.

[zachomega]

haha...now I'm more confused than ever.

-Zach Omega

... get amplification until the runaway begins. 

exactly.

[johngreene]

haha...now I'm more confused than ever.

-Zach Omega

... get amplification until the runaway begins. 

exactly.

Just because it will work, doesn't mean it will be stable. The operating point is changing significantly with temperature so you are going to need to stabilize it with feedback. The cool thing about a dual gate MOSFET is that you can set up the bias pretty much independently of the parts of the circuit that set the gain. I've worked with these devices a lot in the respect I am tasked with designing them -out- of the circuit, so when it comes to the details of the design, I'm not going to be of much help. They were used frequently in older radio designs as VGAs and mixers because of their low current requirements. However they are not very linear (bad for an amplifer) and they have zero LO rejection (bad for a mixer).

--john

[zachomega]

So how are you suggesting I attempt to bias this thing? 

-Zach Omega

haha...now I'm more confused than ever.

-Zach Omega

... get amplification until the runaway begins. 

exactly.

Just because it will work, doesn't mean it will be stable. The operating point is changing significantly with temperature so you are going to need to stabilize it with feedback. The cool thing about a dual gate MOSFET is that you can set up the bias pretty much independently of the parts of the circuit that set the gain. I've worked with these devices a lot in the respect I am tasked with designing them -out- of the circuit, so when it comes to the details of the design, I'm not going to be of much help. They were used frequently in older radio designs as VGAs and mixers because of their low current requirements. However they are not very linear (bad for an amplifer) and they have zero LO rejection (bad for a mixer).

--john

[zachomega]

Or rather, where do I get the feedback from?

-Zach Omega

[db]

Use an op-amp 

[Paul Perry (Frostwave)]

http://www.btinternet.com/~dsergeant/rx.htm

the second circuit on this page shows the bias arrangement using the second gate.
It's a RF application, but the principle is the same.

[Sir H C]

If I understand the device correctly, it is a self-cascoding MOSFET.  The second gate sets the cascode voltage, and the first gate sets the biasing.  One of the curses with MOSFET devices is since they are lower gain compared to bipolar devices, you need more degeneration on the source to keep it stable.  The other technique would require two of these devices, matched, and set up so that you use one for biasing (diode connected), and then with a large resistor tie it to the other gate.  Curse is using 2 devices to bias one.

[johngreene]

I've been working on a pedal that uses 3n201 Mosfets. 

The thing sounds great but after about 10 - 30 minutes of play the sound flubs out.  The mosfets begin to draw more current as the thing goes on and on.  I have the second gate and the source tied together and grounded through an unbypassed 100 ohm resistor.  As it stands now, the thing draw quite a bit of current at idle (1k drain resistor for approximately 1/2 of the 9 volt supply). 

Anyway, the bottom line is, I need ideas on how to stabilize the mosfets to keep them from drifting and drawing more and more current.  Is a larger source resistor enough?  Any ideas would be greatly appreciated.

Thanks,
-Zach Omega

Good night's sleep, hot cup of coffee, fresh brain.....

hmmmm, I wonder since you do not have anything driving the second gate if it isn't oscillating at RF.....

This is an RF transistor so when you design a circuit with it, you need to design it using RF techniques. Your intent might be to use it at audio frequencies but without careful attention to what the circuit looks like at VHF frequencies, you could be designing an oscillator up there that will be dominate. Do you have a way of checking if it is oscillating up to the hundreds of MHz? (oscilloscope).

According to the datasheet, if you have 0V VG2-source voltage and 0V G1 to source voltage this thing is going to bias at around 4.5mA which is pretty much what you are seeing. The device should be able to dissipate this much heat on its own so it looks like there is something else going on. You don't mention if the transistor is getting hot to the touch?

--john

[zachomega]

haha...What fun would that be? 

-Zach Omega

Use an op-amp 

[zachomega]

John,

I did some more testing last night...

Firstly, the transistor does not get hot at all.  Additionally, if I hold my finger on the transistor, there is absolutely no drift.  I'm beginning to think it is not a temperature thing. 

I usually use battery power...What I noticed is the voltage drop on the batteries is pretty substantial due to the current draw of the device (and more so in the design I have which uses 2 of them - for a total draw somewhere around 10 mA. 

What I think is happening is that the battery cannot constantly supply this current...so it drops lower and lower until the voltages are too low.  I hooked up a battery measuring 8.75 volts and when in circuit it was putting out less than 2.5 volts with just one 3n201. 

Next I hooked it up to a power supply which was putting out around 11 volts.  I left the thing on for nearly two hours and the voltages head steady the entire time. 

So what I think is that the voltage drift appeared to be the transistor shifting when in reality the power supply is shifting.  Is this possible? 

The circuit I was using was a 100 ohm resistor from source and gate 2 to ground, 1 meg resistor from gate 1 to ground, and a 1k resistor from drain to B+.  The 100 ohm resistor measured around .38 volts and the drain measured around 7.1 volts with 11 volts on the supply. 

Does any of this make sense?  Or am I having some other problems that are being masked in my test condition? 

I was thinking if this is the problem, the solution to make it "battery friendly" would be to put 2 9 volts in series and then put in a voltage regulator.  Hopefully the two 9 volts would be able to deliver enough current.  I'd really hate to have to use a power supply for one pedal. 

-Zach Omega

Good night's sleep, hot cup of coffee, fresh brain.....

hmmmm, I wonder since you do not have anything driving the second gate if it isn't oscillating at RF.....

This is an RF transistor so when you design a circuit with it, you need to design it using RF techniques. Your intent might be to use it at audio frequencies but without careful attention to what the circuit looks like at VHF frequencies, you could be designing an oscillator up there that will be dominate. Do you have a way of checking if it is oscillating up to the hundreds of MHz? (oscilloscope).

According to the datasheet, if you have 0V VG2-source voltage and 0V G1 to source voltage this thing is going to bias at around 4.5mA which is pretty much what you are seeing. The device should be able to dissipate this much heat on its own so it looks like there is something else going on. You don't mention if the transistor is getting hot to the touch?

--john

[johngreene]

The data sheet doesn't show any characteristics for a DS voltage less than 15 Volts. All the parameters are defined with a supply voltage of 18V. So having a supply of less than 9 V may very well be your problem.

Holding your finger on the transistor can be doing several things, two of which are: 1) providing temperature stability, or 2) adding capacitance to the case. The case is attached to the source. I still wouldn't rule out oscillation. Your finger might be adding enough capacitance to the circuit to kill the oscillation.

--john

[zachomega]

I noticed that as well...and 90% of the literature also has 4 volts on grid 2...If I end up going the power supply route, I may end up playing with that as well since it increases the gain considerably.  I was just trying to maximize battery life, but batteries just might not be an option. 

As for touching the transistor, my house is pretty chilly so my body temperature would definitely increase the temperature of the transistor compared to the room temperature.  However, as you mention, the capacitance from touching it could also stop any problems...But as I said, I didn't notice any shift touching or not touching when run on the power supply.  I'm starting to look towards batteries as the problem...Although with my luck, I'm sure I'll run the unit I built on a power supply and be surprised when it misbehaves again. 

-Zach Omega

The data sheet doesn't show any characteristics for a DS voltage less than 15 Volts. All the parameters are defined with a supply voltage of 18V. So having a supply of less than 9 V may very well be your problem.

Holding your finger on the transistor can be doing several things, two of which are: 1) providing temperature stability, or 2) adding capacitance to the case. The case is attached to the source. I still wouldn't rule out oscillation. Your finger might be adding enough capacitance to the circuit to kill the oscillation.

--john