The Bulbamp: 1 Watt guitar class A amplifier with a light bulb

[tca]

I've started this project with the idea of designing a solid state 1 Watt guitar amplifier. There are a few projects with similar objectives on the web but I didn't want a IC amp, for instance using the well known LM386. I was looking for some discrete circuit that I could modified to my needs. I also wanted an amp that worked at 9V, so that I could use my pedals power source. After looking or googling about it I didn't find any that would please me.

The basic inspiration came from the amplifiers designed by Nelson Pass (http://www.passdiy.com/) and namely the De Lite amplifier (http://www.firstwatt.com/pdf/art_delite.pdf). And so I ended up with this 3 resistors, 3 capacitors and a mosfet in a class A configuration and, of course, a light bulb.

Here is the circuit.


So there you have it! It couldn't be simpler.

It has about 500K of input impedance (you can change this by increasing the values of R1 and R2) and you can use what ever pre-amp you like, even you favorite booster. The only catch with this design is that it needs a very well regulated power supply. It consumes about 200mA but you should feed it with no less than a 1A power source. The IRF610 should be mounted on a small heat sink. The theoretical output power of this amp is about 0.86 Watts. One of the great things about it is that the light bulb reacts to your picking and playing. Glowing strong if you hit the guitar hard and stays dimly otherwise. This light variation is accompanied with a variation of the resistance of the light bulb and thus introduces a kind of compression on the sound of your guitar.

The next picture shows this amp, the prototype in my prototypes box, in a B box



If you can't find a light bulb that makes this amp work you can substitute the bulb with a 47Ohm 10W resistor. Full details are given here: http://www.diale.org/bulbamp.html

I've also recorded some sound samples (http://www.diale.org/mp3/Bulbamp/), and please don't pay any attention to my playing These were recorded with a portable mp3 recorder at 1 meter from the speakers. I've also test the Bulbamp with the speakers of an old VOX AC-30, with a Chandler tube driver as a pre-amp. It sounded amazing.

Comments and sugestions are welcome.

Cheers.

[brett]

 
Great work.
Hint: power supplies for inkjet printers are robust and suitable (usually a couple of amps at 10 to 30V).

[John Lyons]

Very nice!

[PRR]

The bias on this plan is VERY uncertain. It "expects" that 4.5V at Gate will flow 200mA. That may be "typical for IRF610", but there is at least a 3:1 spread from one to the next. A low-side '610 won't flow enough current to be audible, a high-side '610 will bottom itself and clip half the signal.

If you omit R2 and return R1 to Drain, it will self-bias with Drain at 4V-6V for most IRF610. However input resistance falls, maybe below 100K, and non-linear when clipped. You may raise R1 to 10Meg, 50Meg, quite high, so the input impedance stays high.

> The theoretical output power of this amp is about 0.86 Watts.

No, <0.16 Watts clean.

The 3W 12V lamp is nominal 0.25A.

You are not giving it full voltage, so current will be less, though probably over 0.1A. I accept your 200mA figure.

For a maximum positive swing, force the MOSFET "off", all lamp current is diverted to speaker. Assume it stays 200mA or 0.2A.

0.2A peak on an almost-clipped Sine is 0.14A rms. 0.14A in 8 ohm load is 1.13V rms. 0.14A*1.13V= 0.158 Watts almost-clipped sine.

A more refined computation (allowing for lamp current drop on peaks) gives 0.087 Watts almost-clipped Sine. (Maybe a zero was lost in the innernet?)

Power efficiency for 0.087 Watts out from 9V*0.2A input is then 4.83%. This looks bad, but is actually nearly as good as it gets (8.58%).

[tca]

No, <0.16 Watts clean.

By theoretical I mean, really theoretical!!!!

P=I*V/2

50% of the power of the power source

[tca]

Hint: power supplies for inkjet printers are robust and suitable (usually a couple of amps at 10 to 30V).

I dind't know that, thanks.

[Gus]

You need to also think about the input capacitance of the MOSFET.  A MOSFET circuit like that is sensitive to the output drive of the stage before it.  The output resistance of the stage before it and the input capacitance form a low pass filter.

If this is going to be used as a guitar amp why is C2 4700uf?  Guitar amps don't need low bass.

[tca]

If this is going to be used as a guitar amp why is C2 4700uf?  Guitar amps don't need low bass.

I'm also using this amp to listen to some music (just pluging in the output of my sound card).

[PRR]

> 50% of the power of the power source

This is correct for transformer-coupled SE amps, sine signal, no DC in the load.

Using a "constant current" load, 25%.

The efficiency of a resistor loaded stage is often neglected, and sometimes computed wrongly. One generally wise author extrapolates "12.5%". This is in fact correct if idle DC current flows in load; we generally do not want that much DC in our speakers.

It really doesn't get better than 8.58%.

You could "improve" your amp, efficiency-wise, by changing the DC load (lamp) to 1.4 times the signal load and biasing the active device to about 30% of supply voltage. For 9V 8 ohms: 12 ohm lamp 2.7V across transistor. The current increases to 0.525A, more power input. Efficiency should be above 7%, output could be a whopping 0.3 Watts rms. The clipping action will be quite different, and if you like the way it is now, you may not like the tone of the "improved efficiency" amplifier.

> the input capacitance of the MOSFET

It's not just a low-pass. Look at Pass' articles. The capacitance is non-linear. This adds even-order distortion at the top of the audio band. That may even be good for guitar.

[Gus]

Something to read
http://sound.westhost.com/project36.htm

[tca]

Something to read
http://sound.westhost.com/project36.htm

I know about that page but I do think that Nelson Pass approach to class A amps, using mosfets, is much honest and simple and by far the most ingenious way of thinking about audio amplification.

[Gus]

There is nothing wrong with a BJT design.

Read what PRR posted about the capacitance,  I only posted the simple part about input capacitance.  Look up how the gate charges and a MOSFET turns on  

Look up the Hood amp.

[Nasse]

http://www.elektor.com/magazines/2003/november/fet-amp-with-valve-sound.55417.lynkx

[goulashnakov]

I'm intrigued by this project, never thought to use a bulb as a load resistor.  Incandescent light bulbs would have some inductive properties too, right? (depending on bulb design).  But I don't know how significant those inductive properties would be for an audio device.  Either way, I am interested in throwing this on the breadboard to hear it for myself.

Just a thought, after reading through the comments on efficiency.  Would a SRPP approach do anything to improve efficiency on this?  And how might that affect other sound characteristics, like the mild compression effect tca mentioned in his OP?

[tca]

I'm intrigued by this project, never thought to use a bulb as a load resistor.  Incandescent light bulbs would have some inductive properties too, right? (depending on bulb design).  But I don't know how significant those inductive properties would be for an audio device.

I don't know either but I suspect that the inductance of the bulb does not matter in this case.

Just a thought, after reading through the comments on efficiency.  Would a SRPP approach do anything to improve efficiency on this?  And how might that affect other sound characteristics, like the mild compression effect tca mentioned in his OP?

I've thought of that, but replacing the drain resistor with a  constant current source would mean, almost, doubling the number parts of the circuit, although it would increase gain and power of the amplifier. If you add a CCS you end up with a circuit similar to the amplifier part of Nasse's link (http://www.elektor.com/magazines/2003/november/fet-amp-with-valve-sound.55417.lynkx (if you can't read the PDF just google the title of the article)).

Of course adding a CCS will alter the harmonic content of the output signal and this depends how you bias the upper fet.

Either way, I am interested in throwing this on the breadboard to hear it for myself.

I would like that very much!

A few mods that you can try:

1. you can use a 47 Ohm 10W resistor instead of the bulb;
2. decrease the value of the input capacitor if the sound looks to muddy, this happens if you try to overdrive it;
3. decrease the value of the output capacitor if you plan to use the amplifier as a guitar amplifier (as suggested by Gus).

[tca]

Look up the Hood amp.

I do like this version better: The Classic JHL-Pass Style http://www.passdiy.com/pdf/PLH_amplifier.pdf

[rogerray]

I bet this circuit will generate big distortion..But doesn't matter for guitar amp. Infact that's what a guitar amp want, right?

[tca]

I bet this circuit will generate big distortion..But doesn't matter for guitar amp. Infact that's what a guitar amp want, right?

Not so big. For a medium input voltage, my stract, is actually quite clean. But check the sound samples to get a clear view. In that case I've used a SHO as a pre-amp.

[tca]

The bias on this plan is VERY uncertain. It "expects" that 4.5V at Gate will flow 200mA. That may be "typical for IRF610", but there is at least a 3:1 spread from one to the next. A low-side '610 won't flow enough current to be audible, a high-side '610 will bottom itself and clip half the signal.

Another thing that you could try is to put a pot instead of the voltage divider and adjust the bias voltage accordingly to the mosfet that you use.


This would solve the bias problem that as PRR pointed out. I've tried this one and it works great.