Using an opamp as an headphones amplifier

[ratatax]

Hi,

In my search for designing a very simple amp that can drive headphones, I found an opamp with quite high current capabilities, the RC4580. The datasheet says clearly it can be used as an headphones amplifier : http://www.ti.com/lit/ds/symlink/rc4580.pdf

If I connect it to the output of my DAC, would a simple schematic like this work ?


(the bottom tracks are the inputs)

Some circuits have a capacitor in series at the outputs, I guess it's to kill any DC offset but why there would be one ?

Also I have no idea how to calculate the gain for a useful headphones amp especially if the headphones are high impedance. I know my DAC output is 0.88V RMS, and I power the opamp with +15/-15V.
I currently have a gain of two... Is it high enough ? (well of course I could just test it and tune the values, but since I don't have the parts nor the pcb currently, I'd like to avoid as much trial-and-error as possible...)

[oPossum]

A capacitor on the output would typically only be used if the op-amp was powered by a single supply instead of a dual supply.

[Audioguru again]

High impedance headphones are piezo that sound awful. What is the impedance of your headphones?
The datasheet for your headphones will list the required input level for a certain loudness.

Why are your resistor values so low? The datasheet shows an output voltage loss of 5V p-p when the feedback resistor value is as low as your 470 ohms and the voltage loss is higher when loaded.

[John B]

Yes, what are the impedances of the headphone(s) that you want to use? I'm assuming you will use 32 ohm consumer headphones. If you have +/- 15 to play with, you can increase the output impedance of the op amps to around 120 ohm. This will change the power transfer characteristics to provide more consistency, and safety, if you use any headphones between 32-250 ohms.

If you have a low impedance output, and you set the gain for use of 32 ohm headphones, they will be too quiet when used with higher Z headphones, or if you set the gain for hi-z headphones, then plug in 32 ohm phones you may blow your ear drums and/or the phones.

Due to bias currents and offsets, you will probably end up with a DC bias on the output. So if you do not wish to capacitively couple the output, you'll probably have to manually trim the output.

I've made amps for 250 ohm headphones, and have found that an output swing of +/- 3V to be sufficient, although if you do have +/- 15V rails, distortion free output to +/- 5V would be easily achievable. These are only desired to reproduce transients without clipping or distortion.

[magic]

Max output voltage of a few V RMS should be "enough for everyone™", unless you use crazy stuff like AKG K1000 or Hiifman HE6.

1.7V RMS may be enough for many uses too. Check if your headphones are listed on this site and check "power output for 90dB SPL" for reference.
https://www.innerfidelity.com/headphone-measurements

BTW, 3x more voltage is only 10dB more SPL.

RC4580 seems to be a clone of NJM4580, those are often used on outputs of audio gear, should be OK.

The 10Ω resistor will help against oscillation but it may affect frequency response because it sets output impedance to 10Ω. You may consider lowering it for specmanship or changing it to get whatever "sound" you want (if the effect is noticeable with the HPs you use) but check for oscillation by connecting capacitors like 10pF/100pF/1nF/10nF or pieces of unterminated cable before you plug in anything valuable.

DC offset may be produced by the source and some also exists in the opamp. The sum of these two is amplified by the gain (2x as of now). A few mV isn't going to damage anything. You are basically building a "CMOY" amplifier and AFAIK people typically build them without capacitors.

However, make sure that your power supply can never supply only one rail. When one rail is shut down and the input is shorted to ground, the output may be driven hard to the remaining rail.

15V is probably higher than you need and may be a problem with low impedance loads which draw a lot of current but little voltage, meaning that almost all power from those 15V is dissipated in the opamp.

I see no problem using 470Ω for feedback. That's 1kΩ load in addition to the headphone, makes little difference.

Inverting configuration will have somewhat less distortion for the price of 3dB more noise (at higher gains the noise difference is even smaller). Those are quiet opamps and feedback resistance is low, noise is unlikely to be audible in headphones so going inverting may be worthwhile. My soundcard uses inverting NJM4580 on the outputs.

[ratatax]

High impedance headphones are piezo that sound awful. What is the impedance of your headphones?
The datasheet for your headphones will list the required input level for a certain loudness.

Why are your resistor values so low? The datasheet shows an output voltage loss of 5V p-p when the feedback resistor value is as low as your 470 ohms and the voltage loss is higher when loaded.

Where do you find this info on the datasheet ? After some searches it seems my values are indeed quite low, that may waste power. 10K seems closer to what is commonly used, and > 1M isn't very recommended. 10k would be a better bet for my application?

Yes, what are the impedances of the headphone(s) that you want to use? I'm assuming you will use 32 ohm consumer headphones. If you have +/- 15 to play with, you can increase the output impedance of the op amps to around 120 ohm. This will change the power transfer characteristics to provide more consistency, and safety, if you use any headphones between 32-250 ohms.

I have headphones with impedance around 200 ohms ("studio" type).

So to  support those I need to increase the output voltage (gain) but also increase the output resistor to still safely support low impedance headphones without burning them ?


BTW thanks for the other advices

[Zero999]

Keeping the feedback resistor values low is good because it reduces the noise. I couldn't find the current noise specification for the RC4580, but it's quite likely to be fairly high, as it's a BJT input device.

The RC4580 doesn't have any short circuit protection, so if there's a fault in the headphones or the gain is set too high and the impedance is low, it will smoke.

Adding a current limiting resistor is a good idea, but it should be within the feedback loop, so the output impedance is low, to ensure a good damping factor.

[John B]

Yes, what are the impedances of the headphone(s) that you want to use? I'm assuming you will use 32 ohm consumer headphones. If you have +/- 15 to play with, you can increase the output impedance of the op amps to around 120 ohm. This will change the power transfer characteristics to provide more consistency, and safety, if you use any headphones between 32-250 ohms.

I have headphones with impedance around 200 ohms ("studio" type).

So to  support those I need to increase the output voltage (gain) but also increase the output resistor to still safely support low impedance headphones without burning them ?


BTW thanks for the other advices

Yes that's right. The output resistor should cause a voltage drop, which is fine if you have 15V rails. If you do the calculations of power dissipated in the load with a given source impedance, something around 120 ohms is a compromise which allows the use of 32-250 ohm phones. Otherwise if the output impedance is low, the power in the load can roughly be calculated as V^2/R. If you accidentally swap from 250 to 32 ohm phones, you'll suddenly be pushing 7-8x the power. You'll throw the headphones pretty quickly off your head I can guarantee!

[magic]

BTW, you may consider studying the "Objective2" design. It's thoroughly tested, used in the wild and people have measured the hell out of it. It's based on a similar NJM4556 opamp and appears to withstand short circuits fine when it's run from a reasonable supply. You could probably get away with ±9V given your output voltage needs.

As for input noise current, I noticed that at least for NE5534/5532, it is specified exactly in accordance with Schottky equation, which is simply the unavoidable shot noise present in the bias current. If this generalizes to other opamps without bias cancellation, NJM4580/4556 should perform on par with NJM2068 since their Ib is similar. And the latter is known to be dead silent in the O2 with 1k5/1k feedback network, which is even higher resistance than 470/470. So nothing to worry about.

Gain is not a big deal, you can always increase it by swapping resistors if it turns out insufficient.

BTW, you don't need much current for 200Ω at low voltage. NE5532 might be enough and it's guaranteed short-circuit proof because it has an active current limiter at 40mA. Many others too, like the OPA2134 which people often use in CMoys.