Trouble With A Very Simple MOSFET Amplifier

[/dev/phaeton]

Hello!

I've breadboarded a circuit out of Forrest Mim's _Getting_Started_....
Turn your study guides to page 107, or click the following link:

http://home.mia.net/~phaeton/Anklebiter/mims_mosfet.png

Pretty simple, not a whole lot to mess up there, and we all know that all
the circuits in the Mims book have been verified to work thousands of
times over. Note that the place where the two bus wires cross are NOT
connected, as indicated in the book. If it is uncool to post Forrest's
schematics I'll take it down immediately... I intend to use it to amplify
an electric guitar (~100mV amplitude on the signal). I've built it just
as you see there, but Forrest didn't specify what to use for Q1- the book
just says "Q1 = N-Channel Power MOSFET". I took the liberty of using a
2N7000. Hope that's ok with everyone.

Unfortunately, the one I built doesn't work. I can whang away all I want
on the guitar and there is no output on the speaker. Nary a click, nor
pop, nor gated fizzle..... First thing I took a look at were the
voltages:

power source: 8.61V (sorry, best 9V battery I have)

Drain: 8.61V
Source: 0V
Gate: 0V

I measured these right from the pins. If this were a normal everyday
bipolar transistor I'd know it's a biasing issue and I'd know what to do
to remedy it. But I'm a little fuzzy on biasing FETs. Somehow I recall
the Drain should be half the supply voltage, not equal to it, but I'm not
sure. If someone could throw me a bone here, point me to someplace, etc,
I'd appreciate it. Is the 2n7000 not the right choice for this circuit?
What's the canonical method for determining if you've blown up your FET
with static?

Thanks for any and all!

-Phaeton

 

[Phil Allison]

**Repost from ABSE:


"/dev/phaeton"

I've breadboarded a circuit out of Forrest Mim's _Getting_Started_....
Turn your study guides to page 107, or click the following link:

http://home.mia.net/~phaeton/Anklebiter/mims_mosfet.png

Is the 2n7000 not the right choice for this circuit?


** No - it is not a " power mosfet".

Try something in a TO220 pack.

What's the canonical method for determining if you've blown up your FET
with static?

** It fails to work and the gate draws current.



........ Phil

 

[DJ Delorie]

I measured these right from the pins. If this were a normal everyday
bipolar transistor I'd know it's a biasing issue and I'd know what to do
to remedy it. But I'm a little fuzzy on biasing FETs.

FETs are voltage-driven, with capacitive gates and very little current
flow. So, for enhancement mode MOSFETs, the gate should be a few
volts "away" from the source (same polarity as the drain) to cause
conduction between the source and drain. Find a 2N7000 spec sheet and
look for the Vgs vs Id chart.

To test the biasing, remove the FET from the circuit and see what kind
of voltage range you get on the gate circuit. You want something that
puts the Ids in the right range for the voltage and resistance you
already have - 9v, 30ohms, max 300mA, maybe something in the 200mA
range or whatever your 2N7000 is good for. If you look at the Vgs/Ids
chart, find something in the middle of the flat part of the curve
(where small Vgs changes cause large Ids changes). I downloaded one,
and it looks like anything from 4-8v gives you 400-1600 ma. To keep
it under 300mA, you'd need 3.0 to 3.7 volts. Looks like you're
shooting for 3.5 volts for the gate bias.

What's the canonical method for determining if you've blown up your
FET with static?

Use it as a switch. Put a LED/resistor combo off the drain to Vdd,
source to Vss, and a 1k resistor on the gate. Resistor to Vdd, LED
on. Resistor to Vss, LED off. The spec I downloaded had a simple
test circuit in it.

Also, check for resistance between gate and the other pins. Should be
essentially infinite.

 

[Ben Jackson]

as you see there, but Forrest didn't specify what to use for Q1- the book
just says "Q1 = N-Channel Power MOSFET". I took the liberty of using a
2N7000. Hope that's ok with everyone.

That's not a power mosfet, although it should work for your demo
if you don't let the smoke out. Something like an IRF510 would be
more appropriate and you could get it at Radio Shack. You'll want
a heatsink, which can just be some aluminum bolted to the tab. Beware
that the tab is not tied to the source, it's tied to the drain.

Drain: 8.61V
Source: 0V
Gate: 0V

The gate being 0V is strange. The whole point of the left column of
resistors is to bias the gate so that the small AC voltage at the
input falls in a nice linear range of the Ids vs Vgs curve. You can
get a starting value from a datasheet, but I'm sure 0V is too low.
Try turning it up with no input applied until the voltage at the drain
drops below the battery voltage just a little. Then apply an input
and see if you get a sound.

 

[phaeton]

is 2n7000 the right device
That's what I had thought. Jameco lists them as a 'power mosfet', but
I immediately noticed the case style. I looked at the datasheet they
supplied (which turns out to be from the wrong manufacturer anyways),
and since it was rated for "0.83W max" I figured maybe it didn't need a
thermal plate built in. (like the LM386 chipamp- no heatsink there for
500mW)....

So actually, this device is more in the 'small signal' neighbourhood
like the J201, BS170, etc, instead of anything you'd associate the word
"power" to?

That silly Jameco. Always pullin pranks... Anyways...


I'll try that tonight, *but* I'm also probably going to swing by Radio
Shack and get an IRF510 (as suggested) to use instead of the 2n7000.

IIRC, I did that, and D to G was infinite, S to G was infinite, D to S
was NOT infinite. I think it was in the 6MOhm ballpark, but I'm not
sure.

really? I'm still a little fuzzy on the concept of 'power' and how
much of it is needed to drive a speaker. What differentiates between a
'power' transistor and a 'signal' transistor, even if some of them run
off the same voltage, and if some signal transistors can operate nearly
rail-to-rail.

I'll hit one on the way home from work. I've been meaning to see if
the local store is about to do the 'Everything 90% off' sale anyways,
like some of them are doing now.

Right.. the 0V part was a red flag for me too...
However, the left vertical column of resistors is NOT connected to the
gate, as per the schematic in the book. It doesn't make sense to me
either, so if it's wrong I wonder if the ink might have worn off that
page there. I've biased BJTs this way so that's why I thought it was
odd as well.

Turning the pot has no effect on the reading I get at the Gate.
Furthermore, Forrest noted the pot as being a gain control, not a bias
control (the opposite of what I would have guessed by looking at it).

Oh well.. let me get the right device in there first, then I'll measure
more stuff.

Thanks for all the help and suggestions!

-phaeton

 

[Ben Jackson]

Right.. the 0V part was a red flag for me too...
However, the left vertical column of resistors is NOT connected to the
gate, as per the schematic in the book.

I don't have your image in front of me, but it *was* connected at the
center pot connection (the wiper). In fact, since the pot is between
two other resistances, you probably can't make 0V at the wiper at all.

Turning the pot has no effect on the reading I get at the Gate.
Bad.

Furthermore, Forrest noted the pot as being a gain control, not a bias

The Vgs to Id relationship is not linear, so the region you select will
have an overall slope that will affect the gain. If you look at that
graph, you'll see that the curve doesn't even start until Vgs(th), which
on an IRF510 is somewhere between 2 and 4V. Below that, it's totally
off. As you turn up the bias, you'll be class B initially (most of your
input signal will, after biasing, fall below the threshold) and you'll
conduct for more and more of your input signal as you turn up the bias.
Eventually you'll have the entire input signal above the threshold voltage
and be class A. If you keep going up (or your input is driven very hard),
the gate voltage will reach the flat spot near saturation and your gain
will go down and distortion will increase. Eventually you'll saturate
the fet and let the smoke out (well, maybe not with a circuit based on
a current limiting resistor). A lot of Class A, B and AB RF circuits
using the IRF510 use a zener + 10R on the gate to absolutely prevent
saturation (around 8V).

 

[DJ Delorie]

I'm still a little fuzzy on the concept of 'power' and how
much of it is needed to drive a speaker. What differentiates between a
'power' transistor and a 'signal' transistor,

In the case of MOSFET, a "power" mosfet just means it has a bunch of
individual transistors all tied in parallel, so you can get bigger
currents. The downside is the higher gate capacitance (i.e. slower
switching times).

That's independent of the max Id current. You can have a small
"power" mosfet, or a big signal mosfet.

As for the speaker, it's 8 ohms. If you want one watt of sound, you
need about 1/R^2 amp (1/64 = 15mA). For reference, my subwoofers are
about 1500 watts each (23 amps equivalent).

Also, just because a transistor is rated for N amps, doesn't mean it
stops at N amps. It means it melts at N amps. So it's up to you to
ensure that the circuit won't try to overload the transistor.

Right.. the 0V part was a red flag for me too...
However, the left vertical column of resistors is NOT connected to the
gate, as per the schematic in the book. It doesn't make sense to me
either, so if it's wrong I wonder if the ink might have worn off that
page there. I've biased BJTs this way so that's why I thought it was
odd as well.

The circuit should have +9v - resistor - potentiometer - resistor - ground.
^
|
gate

That way, like any old resistor divider, it gives you a range of
voltages off the pot. The two fixed resistors can be chosen such that
the pot's whole range covers the useful range of gate bias voltages.

Turning the pot has no effect on the reading I get at the Gate.

Check the resistor wiring.

Furthermore, Forrest noted the pot as being a gain control, not a bias
control (the opposite of what I would have guessed by looking at it).

FET curves are non-linear. Depending on where on the curve you are,
the slope varies. Since in an AC coupled amp, slope == gain, the gate
bias determines the gain of the transistor.

 

[default]

Is the 2n7000 not the right choice for this circuit?


** No - it is not a " power mosfet".

Try something in a TO220 pack.

This is a little milliwatt amp intended for a 9 V battery a 2N7000 is
probably good. A single ended class A with the speaker in the circuit
is not going to want a high DC offset on the speaker . . . for that
one needs a transformer, or resistor (or constant current source)
replacing the transformer and a large electrolytic to couple to the
speaker.

 

[default]

Hello!

I've breadboarded a circuit out of Forrest Mim's _Getting_Started_....
Turn your study guides to page 107, or click the following link:

http://home.mia.net/~phaeton/Anklebiter/mims_mosfet.png

Pretty simple, not a whole lot to mess up there, and we all know that all
the circuits in the Mims book have been verified to work thousands of
times over. Note that the place where the two bus wires cross are NOT
connected, as indicated in the book. If it is uncool to post Forrest's
schematics I'll take it down immediately... I intend to use it to amplify
an electric guitar (~100mV amplitude on the signal). I've built it just
as you see there, but Forrest didn't specify what to use for Q1- the book
just says "Q1 = N-Channel Power MOSFET". I took the liberty of using a
2N7000. Hope that's ok with everyone.

Unfortunately, the one I built doesn't work. I can whang away all I want
on the guitar and there is no output on the speaker. Nary a click, nor
pop, nor gated fizzle..... First thing I took a look at were the
voltages:

power source: 8.61V (sorry, best 9V battery I have)

Drain: 8.61V
Source: 0V
Gate: 0V

I measured these right from the pins. If this were a normal everyday
bipolar transistor I'd know it's a biasing issue and I'd know what to do
to remedy it. But I'm a little fuzzy on biasing FETs. Somehow I recall
the Drain should be half the supply voltage, not equal to it, but I'm not
sure. If someone could throw me a bone here, point me to someplace, etc,
I'd appreciate it. Is the 2n7000 not the right choice for this circuit?
What's the canonical method for determining if you've blown up your FET
with static?

Thanks for any and all!

-Phaeton

The gate should have about one to four volts with respect to the
source. Disconnect the gate and see what the gate divider is reading
adjust for ~1+ volts. Reconnect to the gate and it should start
pulling some current down through the speaker (AND YOU SHOULD STILL
SHOW THE ONE VOLT ON THE GATE- or the device is fried) - one volts
isn't carved in stone, but should put the 2N7000 in the linear region
with a 9 volt supply.

If you download the data sheet for the 7000 you'll see the operating
bias curves.

www.fairchildsemi.com/ds/2N/2N7000.pdf


That should work with reasonable volume - increasing the current
capability of the mosfet might not be a good idea. That design is
called single ended Class A. Increasing the power supply and
transistor current might put an unhealthy amount of DC on the speaker.

One can make some pretty good sounding simple single ended mosfet amps
with a current limiter in the source (even a 3 terminal regulator will
work) If you want more power check out Nelson Pass' site

"This is the second installment of a trilogy of construction projects
centered around the performance obtainable from absolutely minimalist
circuitry. Part one described the Zen amplifier, a 10 watt
single-ended class A power amplifier using a single MOSFET gain
stage."

The Zen amplifier looks like the Forest Mims circuit on steroids

http://www.passdiy.com/legacy.htm

It is fun to read about even if you don't build anything - if you do
build ; it won't break the bank, will sound great (rivaling some
"audiophile" stuff) and is simplicity itself.

 

[Phil Allison]

"default"
"Phil Allison"

This is a little milliwatt amp intended for a 9 V battery a 2N7000 is
probably good.

** Nope.

A single ended class A with the speaker in the circuit
is not going to want a high DC offset on the speaker . .

** A fragile 2N7000 may easily wind up damaged in that circuit.

Looks like the OP's already is.

A "power mosfet" is specified and wisely so.





.......... Phil

 

[default]

"default"
"Phil Allison"


** Nope.




** A fragile 2N7000 may easily wind up damaged in that circuit.

Looks like the OP's already is.

A "power mosfet" is specified and wisely so.





......... Phil

2N7000 is good for 200 ma and 400 mw dissipation, should be plenty for
50 + milliwatts out with a suitable high impedance speaker and enough
voltage drive - this ain't no stereo system.

 

[Phil Allison]

"DJ Delorie"

As for the speaker, it's 8 ohms. If you want one watt of sound, you
need about 1/R^2 amp (1/64 = 15mA).

** Huh ????????????

What sort of twisted thinking is this ?

Power = I squared R.

So I = sq rt P / R

1 / sq rt 8 = 0.35 amps.


The bias current needed for 1 watt of audio into 8 ohms with a simple class
A stage is actually 0.5 amps.

Signal current swings from 1 amp to zero giving an 8 volt swing.





........ Phil

 

[Phil Allison]

"default"

2N7000 is good for 200 ma and 400 mw dissipation,

** You are simply not following the point.

THAT particular circuit will allows a 2N7000 he mosfet to exceed it ratings
!!

should be plenty for
50 + milliwatts out with a suitable high impedance speaker and enough
voltage drive - this ain't no stereo system.

** You are an asinine, MORON.

The OP's posted circuit is THE ONE at issue here !!!

Is has a 8 ohm speaker with a 9 volt supply, so a current of over 1 amp is
possible.

Dissipation can be as high as 4.5 volts times 0.56 amps = 2.5 watts.





......... Phil

 

[DJ Delorie]

Power = I squared R.

Sigh, I had I R squared. No wonder the results seemed incorrect. I'm
blaming my arthritis. Yeah, arthritis, that's it.

 

[DJ Delorie]

Looks like the OP's already is.

Based on the diagnostics he's sent in so far, I think he just wired up
the bias resistors wrong and the circuit just hasn't done anything
yet.

 

[Don Bowey]

"DJ Delorie"



** Huh ????????????

What sort of twisted thinking is this ?

Power = I squared R.

So I = sq rt P / R

1 / sq rt 8 = 0.35 amps.


The bias current needed for 1 watt of audio into 8 ohms with a simple class
A stage is actually 0.5 amps.

Signal current swings from 1 amp to zero giving an 8 volt swing.





....... Phil

He only want a few mW......

Don

 

[default]

Is has a 8 ohm speaker with a 9 volt supply, so a current of over 1 amp is
possible.

Dissipation can be as high as 4.5 volts times 0.56 amps = 2.5 watts.

And you may be missing the point - someone needs to go back and find
out what speaker the original plans call for. I see the speaker as a
problem - half amp DC continuous through the speaker is likely as not
going to cause distortion (not that someone with an electric guitar
might object to that). The speaker should be up around a 30-100 ohms.

 

[Phil Allison]

"default"


** You are an asinine, fucking MORON.

Drop Dead.





......... Phil

 

[phaeton]

Sorry that this caused such a controversy, everyone :-(

I guess I failed to note on my diagram that the speaker is 8 ohm.

Last night I stuck an IRF510 into the circuit instead of the 2n7000
(observing different pinout, of course). Now the speaker at least
crackles a little bit when I touch parts of the circuit and/or plug in
the guitar, so it's a little more alive. I still didn't get any guitar
sound out of it, but the voltage measurements were much closer to what
has been discussed here. It seems that turning the pot changes the
drain and gate voltages and their relationship to each other. I moved
it all around and got no sound out of it, still.

I did find that the cap was leaking, so I changed it. I measured the
resistance of the speaker to test for continuity (7.5 ohms was what i
read). I poked around the circuit with an audio probe* and got guitar
signal on all 3 pins, with the drain being the loudest. In fact, i got
it everywhere, including where the speaker connections are, just no
sound out of the speaker. Ran out of time before I could change
speakers, but if the speaker had continuity and was roughly in the
expected range, why wouldn't it do something?

I'll have to piddle with it s'more. Thanks for all the suggestions
thusfar.

-Phaeton



* Audio Probe: essentially a piece of wire plugged into a known,
working amplifier, isolated with a 2.2uf capacitor in series.

 

[Michael A. Terrell]

Sorry that this caused such a controversy, everyone :-(

I guess I failed to note on my diagram that the speaker is 8 ohm.

Last night I stuck an IRF510 into the circuit instead of the 2n7000
(observing different pinout, of course). Now the speaker at least
crackles a little bit when I touch parts of the circuit and/or plug in
the guitar, so it's a little more alive. I still didn't get any guitar
sound out of it, but the voltage measurements were much closer to what
has been discussed here. It seems that turning the pot changes the
drain and gate voltages and their relationship to each other. I moved
it all around and got no sound out of it, still.

I did find that the cap was leaking, so I changed it. I measured the
resistance of the speaker to test for continuity (7.5 ohms was what i
read). I poked around the circuit with an audio probe* and got guitar
signal on all 3 pins, with the drain being the loudest. In fact, i got
it everywhere, including where the speaker connections are, just no
sound out of the speaker. Ran out of time before I could change
speakers, but if the speaker had continuity and was roughly in the
expected range, why wouldn't it do something?

I'll have to piddle with it s'more. Thanks for all the suggestions
thusfar.

-Phaeton

* Audio Probe: essentially a piece of wire plugged into a known,
working amplifier, isolated with a 2.2uf capacitor in series.

Don't pay any attention to Phyllis. She's on the rag, as usual.


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida