Minimalist Mosfet amp

[Spiny]

Having been a reader for a fair while the forum has encouraged me to try building an amp again - the last one was a valve amp in the 60's this is based on various ideas on the forum and Nelson Pass's thoughts on Mosfets.

Having read the pass lab and solid state posts a Mosfet amp was in order from the parts bin (junk box) This is my attempt at a minimalist class A amplifier, 10 components per channel on veroboard. I'm surprised at its quality given the simplicity.

Some notes, I run this at 9 volts though 8 to 25+ seem to work with variations in R2, Testing seem to show the mosfet matters little, various IRF and some odd switching mosfets out of bust PSU board work.
Set the Bias current with VR2, at 9 volts on V+ 1 amp produces a good sound. set with an ammetre or measure accross R2. As its Class A a decent heat sink is needed, and an old CPU one worked well for me.

Thanks to all the posters for reviving my interests in audio electronics, its been fun building this and I'd like to share the results.

Alan

 

[Nelson Pass]

Have you tried it without the diodes (and c2)?

 

[Spiny]

Nelson

I did try without diodes, Clipping distortion without the voltage lift from the diode junctions occured, I originally designed this with a resistor in place of the diodes but the power disappation was to much for the junk box so a pair of diodes went in....

why 9v (9.8 actually) is due to a scrap 8volt 4 amp toroid I had in the junk box.

I read as many of your insights and thoughts on audio as I find, finding them inspirational, you are one of the reasons I have come back to real audo after a long break,
Alan

 

[yaron]

Hi Nelson

The diodes give 1.2 v to the source of the FET. He needs the gate to be in lower voltage from the source, so he put the diodes (and the capoacitor filter ac changes).

 

[GRollins]

Hi Nelson

The diodes give 1.2 v to the source of the FET. He needs the gate to be in lower voltage from the source, so he put the diodes (and the capoacitor filter ac changes).

The IRF830 appears to be an enhancement mode device, not depletion. The Gate will need to be more positive than the Source.

The load resistor (R2) is 2.5 Ohms. At 1A bias, that's going to give you a 2.5V drop. At that point, the amp will clip asymmetrically. You could increase the bias, but that would likely blow the MOSFET in short order, as it's only rated for 4.5A, or increase the value of the load resistor to the point where the Drain of the MOSFET settles somewhere close to half whatever rail voltage you choose. That will (within the limits of the MOSFET) give you better clipping performance. Assuming a 9V rail and 1A bias, a 4.7 Ohm resistor will come pretty close. Use at least a 10W resistor and ventilate it well. 20W would be better--you'd be less likely to scorch your fingers.

Grey

 

[syn08]

Use at least a 10W resistor and ventilate it well. 20W would be better--you'd be less likely to scorch your fingers.

So you think a 20W resistor will stay significantly cooler than a 10W resistor?

 

[Spiny]

thank you for all the observations, The parts used were what was to hand, designed the old fashioned way - scope and meter I don't have a simulator .
Grey, I will acquire a suitable pair of power resistors and play, maybe I can drop the diodes that way, without them I got asymetric distortion far to easily.
The fun in building this and your comments have encouraged me to continue to see what it will do with a better set of components now on order.
Currently its playing music from a FM tuner and has been doing the last few hours through a pair of fairly sensitive home brew speakers - even the better half agrees it sounds good.

 

[GRollins]

So you think a 20W resistor will stay significantly cooler than a 10W resistor?

One of the first demonstrations they gave back when I took electronics in school was to run three identical value resistors in parallel across the output of a power supply. The only difference between the three was the wattage: 1/2W, 1W, and 2W.
Since all three resistors had the same value and were in parallel, it was obvious that each was dissipating the same amount of heat. The difference in wattage rating, however, made it instantly clear why higher wattage resistors are bigger--the 1/2W was too hot to touch, the 1W was touchable, while the 2W was merely comfortably warm. Why? Surface area. The amount of heat radiated per unit of surface area decreased as the physical size of the body of the resistors increased.
So, yeah, a 20W resistor will be cooler to the touch than an equivalent 10W resistor, assuming the same heat dissipation.

Grey

 

[syn08]

One of the first demonstrations they gave back when I took electronics in school was to run three identical value resistors in parallel across the output of a power supply. The only difference between the three was the wattage: 1/2W, 1W, and 2W.
Since all three resistors had the same value and were in parallel, it was obvious that each was dissipating the same amount of heat. The difference in wattage rating, however, made it instantly clear why higher wattage resistors are bigger--the 1/2W was too hot to touch, the 1W was touchable, while the 2W was merely comfortably warm. Why? Surface area. The amount of heat radiated per unit of surface area decreased as the physical size of the body of the resistors increased.
So, yeah, a 20W resistor will be cooler to the touch than an equivalent 10W resistor, assuming the same heat dissipation.

Grey

Well Grey, I was expecting from you a little bit more than reminiscing a high school physics experiment

Let's focus on your example, so it is about a 4.7 ohm resistor under a 1A DC current. This would make a DC power dissipation of 4.7W, say roughly 5W.

Take a look at this power resistor spec http://www.yageo.com/pdf/yageo/Leaded-R_SQP_NSP_2008.pdf they are fairly standard ceramic body wirewound resistors, of 1W to 25W power. A 10W power resistor under a 5W load (50% nominal) will heat up in free air convection up to 125 centigrades. A 20W power resistor under 5W load (25% nominal) will heat up to 95 centigrades. 95 centigrades is certainly lower than 125 centigrades (btw, it is the convection that keeps the higher power device cooler, not radiation!) although I doubt you'll be able to keep your fingers on it. However, cased amps certainly cannot be considered as a "free air convection" environments. In a closed case (even if well ventilated) the temperature difference between the two resistor will certainly be much smaller. How small? Can't tell precisely, because it all depends on the case construction, but I wouldn't set hopes for more than 15-20 centigrades... In the extreme case of a complete sealed case (which is certainly not true for a power amplifier) the temperatures will be almost identical.

Then the legitimate question arises, why and when should we use high(er) power resistors? The answer is reability. All devices (passive and actives as well) share the same type of failure rate vs. temperature dependency, and that's generally an exponential shaped curve. We certainly shouldn't use high(er) power resistors only to avoid finger burnings, we are anyway not supposed to touch those but for avoiding reability issues, and here comes the power derating curve. According to the same datasheet, a 10W resistor will handle 5W (50% nominal) up to 150 centigrades ambient, assuming an otherwise perfect heat transfer to ambient. 125 centigrades is certainly less than 150 centigrades, and even if you consider the non ideal case convection, the resistor temperature won't exceed 150 centigrades, which pretty much makes using a 20W device pointless. My rule of thumb for wirewound power resistors is: load them up to 50% of the nominal power, or otherwise put, choose the nominal power as double the load. For film resistors, I chose 25-30% loads only, because of the worse convection properties and higher failure rates vs. temperature.

Interesting enough, the closer you get to the nominal power, the larger the temperature gap becomes. I would for example have no reserve in recommending for this application a 10W resistor over a 5W resistor, but certainly I could barely justify the cost and size of a 20W device over a 10W device.

 

[Spiny]

The IRF830 appears to be an enhancement mode device, not depletion. The Gate will need to be more positive than the Source.


Performance. Assuming a 9V rail and 1A bias, a 4.7 Ohm resistor will come pretty close. Use at least a 10W resistor and ventilate it well. 20W would be better--you'd be less likely to scorch your fingers.

Grey

Having acquired a couple of 4.5 ohm 20 watt resistors - I too don't like scorched fingers - and clamped them to a small finned heatsink, dumped the diodes and set the bias to @ 5 volts. Actually best trimmed with a scope and a signwave the result is a bit better - quality the same but more headroom.

If anyone builds this you need a clean power supply - eithor lots of capacitance or a regulator

 

[Spiny]

Forget to add, Thanks Grey for your comments on the original posting - this little amp will do me for a while while I consider my next move.

 

[brianuk]

This interests me. What amount of distortion does something like this have? I run active so low power doesnt bother me and might be just the ticket for the treble?

Thanks.

 

[Spiny]

I've nothing to measure distortion Need to build a clean
oscillator, on my list of things to do.
However it appears to be fairly clean to my ears. and without feedback should have a wide bandwidth.

 

[gootee]

Well Grey, I was expecting from you a little bit more than reminiscing a high school physics experiment

Let's focus on your example, so it is about a 4.7 ohm resistor under a 1A DC current. This would make a DC power dissipation of 4.7W, say roughly 5W.

Take a look at this power resistor spec http://www.yageo.com/pdf/yageo/Leaded-R_SQP_NSP_2008.pdf they are fairly standard ceramic body wirewound resistors, of 1W to 25W power. A 10W power resistor under a 5W load (50% nominal) will heat up in free air convection up to 125 centigrades. A 20W power resistor under 5W load (25% nominal) will heat up to 95 centigrades. 95 centigrades is certainly lower than 125 centigrades (btw, it is the convection that keeps the higher power device cooler, not radiation!) although I doubt you'll be able to keep your fingers on it. However, cased amps certainly cannot be considered as a "free air convection" environments. In a closed case (even if well ventilated) the temperature difference between the two resistor will certainly be much smaller. How small? Can't tell precisely, because it all depends on the case construction, but I wouldn't set hopes for more than 15-20 centigrades... In the extreme case of a complete sealed case (which is certainly not true for a power amplifier) the temperatures will be almost identical.

Then the legitimate question arises, why and when should we use high(er) power resistors? The answer is reability. All devices (passive and actives as well) share the same type of failure rate vs. temperature dependency, and that's generally an exponential shaped curve. We certainly shouldn't use high(er) power resistors only to avoid finger burnings, we are anyway not supposed to touch those but for avoiding reability issues, and here comes the power derating curve. According to the same datasheet, a 10W resistor will handle 5W (50% nominal) up to 150 centigrades ambient, assuming an otherwise perfect heat transfer to ambient. 125 centigrades is certainly less than 150 centigrades, and even if you consider the non ideal case convection, the resistor temperature won't exceed 150 centigrades, which pretty much makes using a 20W device pointless. My rule of thumb for wirewound power resistors is: load them up to 50% of the nominal power, or otherwise put, choose the nominal power as double the load. For film resistors, I chose 25-30% loads only, because of the worse convection properties and higher failure rates vs. temperature.

Interesting enough, the closer you get to the nominal power, the larger the temperature gap becomes. I would for example have no reserve in recommending for this application a 10W resistor over a 5W resistor, but certainly I could barely justify the cost and size of a 20W device over a 10W device.

Besides reliability considerations, I think that sometimes there is also another possible valid reason to choose a resistor with a larger power rating. i.e. In some locations in some circuits, distortion can be reduced by using a resistor with a higher power rating, since it would lower the change in resistance due to resistor self-heating during signal amplitude excursions, which would otherwise cause distortion.