My First Amp

[ch83575]

I am in an electronics class in college and I am working on a headphone amp for my final project. The class is the first electronics class, so I need to stick to something fairly simple. I need to be able to analyze the entire circuit by hand, so nothing really complicated. I am thinking of a simple voltage follower with no voltage control since it will be used mostly with my computer. How does this look to you guys? Am I doing anything terribly stupid? Any simple changes that would improve things? I know the circuit seems trivial, but I am pretty new at this.

Vdd = 15V
Id = ~300mA

[djk]

May I suggest that you read this thread?

SEWA - Seven Watt Amplifier

[AndrewT]

Hi,
you have shown a CCS loaded, source follower, in single ended mode, powered from a single polarity PSU.

The maximum current available from the follower <bias current.
That bias current flows through your current mirror.
Check your current mirror dissipation.

Check the high pass filter at the input of the amplifier.

PS.
I don't think the SEWA is an example that should be followed.

[ch83575]

I originally used a dual polarity PS thinking that if I found that I needed gain I could add a simple op-amp front end with modest gain. Then, when I modeled the amp without the negative PS the performance was almost identical, so I figured I would get rid of the negative rail. Is there any reason I am missing to stick with a dual supply?

The SEWA is interesting. Don't those biasing zeners constitute a positive feedback loop? In small signal won't they appear as small resistors? Also, what is the purpose of the shunt resistor and cap between the drain and gate?

Thanks for the help!
-Chad

[djk]

"Don't those biasing zeners constitute a positive feedback loop? "

It's a gate clamp to protect the FET from transients above 15V or so.

"In small signal won't they appear as small resistors? "

No. There is a small amount of capacitance.

"Also, what is the purpose of the shunt resistor and cap between the drain and gate?"

I need a link to the schematic you are looking at please.

[ch83575]

I get it... I though they were used to set the DC bias, but I didn't look at the value. They are not conducting with Vin=0 so they wont effect the small signal. The schematic is in the post you (djk) linked to, I don't know how to re-link to a specific pdf. The shunt resistor and cap I was referring to are the 47k and 330uF in the biasing circuit from drain to gate.

Back to my amp: the BJTs in the current mirror dissipate about 900mA, how big of a heatsink should I be looking at? My simulator (Altium) does not give me a power for the mosfet, but it is biased at 300mA, how much heatsink is appropriate for that? Is there a standard method for calculating heatsink size?

Thanks,
Chad

[djk]

There are over 900 numbered posts in that thread, just the post number would help us find the diagram you are referring to.

900mA?

Why?

Headphones are over 100dB/mW, 1W would play over 130dB. Most phones are around 30R, so the bias need be no higher than about 250mA or so.

[godfrey]

Hi Chad

A couple of comments:
a) 1000uF for the input coupling cap (C1) is way to high. Something like 1uF would be better.

b) The current source built around R1, Q2 and Q3 is very inefficient (and not very good). As is, the power dissipation in the amp will be about 4 watts in R1 and about 2 to 2.5 watts each for Q1 and Q2. i.e. About half the power is wasted in R1. A more conventional CCS like those shown below would waste less heat. I'd vote for the 2 transistor one. BTW, the second one uses a LED as a zener. They make nice voltage references - low noise, low dynamic impedance etc, and pretty too.

c) ZTX450 is no good for the CCS, it's power dissipation is too low. Something like a BD139 would be better and can be mounted on a heatsink. See here for a list of other medium power transistors.

d) With single supply and large coupling caps, you need to beware of thumps at switch-on and switch-off. This can be a pain to sort out.

e) Don't forget to check power supply rejection. With mains power and ripple on the supply line, a fair amount of hum will leak into the signal path, mostly through Q1's bias resistor and the resistor feeding the CCS. It's a good idea to add an extra RC filter for these.

f) Heatsinks are rated according to their thermal resistance (in degrees centigrade per watt). e.g. If a transistor dissipating 2 watts is mounted on a heatsink rated at 10 degrees/watt, then the heatsink temperature will be 20 degrees higher than ambient. The transistor's case temperature will be a little higher due to the thermal resistance between case and heatsink, however this is low (maybe 1 degree/watt), so it won't make much difference here in a low power circuit.

Cheers - Godfrey

[godfrey]

Turning the thing "upside down" with a positive-earth power supply can have some advantages:

a) Current drawn from the power supply is constant, so there's no signal current flowing in the power supply wiring. This can help with cross-talk.

b) Any ripple voltage on the power supply will not be applied to the Mosfet, which will help the PSRR, assuming the current source has good rejection and other PSRR issues are sorted out as well.

c) You can probably just short-circuit the output with a switch or relay during turn-on and turn-off to avoid thumps, without damaging anything.

p.s. I'm too lazy to draw all the other changes so pic below only shows the "upside down" configuration.

p.p.s. You need to add a gate-stopper resistor for the Mosfet to avoid oscillation.

[ch83575]

Originally Posted by djk

There are over 900 numbered posts in that thread, just the post number would help us find the diagram you are referring to.

900mA?

Why?

Headphones are over 100dB/mW, 1W would play over 130dB. Most phones are around 30R, so the bias need be no higher than about 250mA or so.

Sorry, that was a typo. I wanted to say 900mW. They each dissipate about a watt at loud listening levels, even more at max output. The mosfet is indeed biased to 300mA.

Godfrey: Thank you very much for the detailed analysis! I will try to incorporate all of your suggestions. The current mirror used here was what was taught in class, and I admit that I didn't do enough research on alternative current sources. The upside down amp is also a really neet idea.

Thanks again,
Chad