Headphone amp design

[budm]

Re: Headphone amp design

So the input is fed to the Inverting input? 1uf and about 3K of input impedance will have poor low frequency, the RC is basically High pass filter, Fc = 1/(2Pi.R.C)
http://www.muzique.com/schem/filter.htm

[lti]

Re: Headphone amp design

I have increased the capacitance to 10uF. It sounds a lot better now. For some reason, higher frequencies were always a little louder with the old Chinese generic 1uF caps.

I am getting unusually high THD and IMD measurements in RightMark. That program isn't very accurate, but it does appear to be right this time. It shows 0.131% THD and 1.382% IMD, which appears to be coming from the inverting amplifier before the MS6308. I used an LM833 for that, which isn't a very good audio op-amp. I think even a 4558 might be better.

[lti]

Re: Headphone amp design

I've been messing around with RightMark Audio Analyzer and these amps.

The APA3541 works well with the filter. The audio source I was using above was a really crappy CD player. It's so crappy that the APA3541 and a TDA2822M are the only two amps that don't clip the input signal. It still has a lot of distortion, but it's better with the filter. I don't know why it would be better, but I got 0.4% THD without the filter, and I could hear the distortion with the headphones sitting on the table. I think I should replace the input coupling caps since they're the same brand and series as the ones that caused the huge distortion and increased output levels at higher frequencies on the MS6308 (Foai CD110 series).

The MS6308 actually worked okay today. I ended up with exactly the same distortion ratings with the MS6308 I put on the prototyping board and the chip in my computer. The IMD was a lot lower today, possibly because I powered it from a good enough 9V battery that the regulator didn't drop out. I need a power supply.

I think I have finished these projects. I've made a decent headphone amp (although it doesn't have much gain) and polished a turd.

Also, RightMark says that the line-level output on my desktop computer hits -3dB at 70Hz. The headphone output with no load has a -3dB point of 30Hz. The headphone output on my Gateway laptop (which now has no display backlight ) hits -3dB at 95Hz and has pretty high distortion, even though it always sounded good to me.

[lti]

Re: Headphone amp design

I added a gain selector switch with two settings (gain of 0.47 or 4.7) and connected the LM833 to the power supply instead of the output of the 78L05 today. I'm getting close to the limits of a 5V-powered op-amp. I made the low-gain setting so I can connect it to a real line-level source without clipping. The high-gain setting is for my laptop.

I haven't found a power supply that doesn't create a ton of background noise. Batteries work, but they drain quickly. I need something with low enough ripple that the LM833 can be connected directly to it.

The amp draws 11mA with no audio playing, and I saw peaks of 18mA with the volume turned all the way up. I have succeeded in creating an amp that can make my headphones buzz.

I also discovered that weaker batteries cause the power to pulse. I can see the power LED flashing, and there is a popping sound on the output. I had to use a nearly new 9V battery with a 470uF cap in parallel.

In the desktop computer, I discovered a few months ago that the line-level output on the back is actually connected to the output of the headphone amp. That would explain why the gain was set to 2. They wanted a line-level output, but the output from the sound chip is already as close to line-level as something powered from 5V can get without some kind of DC-DC boost converter.

[ben7]

Re: Headphone amp design

I added a gain selector switch with two settings (gain of 0.47 or 4.7) and connected the LM833 to the power supply instead of the output of the 78L05 today. I'm getting close to the limits of a 5V-powered op-amp. I made the low-gain setting so I can connect it to a real line-level source without clipping. The high-gain setting is for my laptop.

I haven't found a power supply that doesn't create a ton of background noise. Batteries work, but they drain quickly. I need something with low enough ripple that the LM833 can be connected directly to it.

The amp draws 11mA with no audio playing, and I saw peaks of 18mA with the volume turned all the way up. I have succeeded in creating an amp that can make my headphones buzz.

I also discovered that weaker batteries cause the power to pulse. I can see the power LED flashing, and there is a popping sound on the output. I had to use a nearly new 9V battery with a 470uF cap in parallel.

In the desktop computer, I discovered a few months ago that the line-level output on the back is actually connected to the output of the headphone amp. That would explain why the gain was set to 2. They wanted a line-level output, but the output from the sound chip is already as close to line-level as something powered from 5V can get without some kind of DC-DC boost converter.

Try using a linear regulator, and then try some ferrite beads/chokes (like the SMD ones) to possibly cut down on the high frequency stuff. Also, try a ceramic capacitor of maybe 470nF or 1uF right on the pins of the LM833. You see, capacitors cut out most of the noise, but not all! ... I'm thinking if you get the right ferrite beads/chokes then they will help a lot.

[lti]

Re: Headphone amp design

A linear regulator does help, but I would need a higher voltage power supply. I can use an 8V power supply now.

I was using two unregulated linear power supplies and one 12V switching power supply from a monitor. I think the switching power supply was the best of the three. I have a better chance of getting rid of the higher frequency noise. For a linear power supply, I would need a voltage regulator or a huge filter cap. I guess I could try making a filter using some parts from another monitor power supply.

[ben7]

Re: Headphone amp design

A linear regulator does help, but I would need a higher voltage power supply. I can use an 8V power supply now.

I was using two unregulated linear power supplies and one 12V switching power supply from a monitor. I think the switching power supply was the best of the three. I have a better chance of getting rid of the higher frequency noise. For a linear power supply, I would need a voltage regulator or a huge filter cap. I guess I could try making a filter using some parts from another monitor power supply.

You shouldn't need too big of a capacitor for a linear supply, especially if you are using a voltage regulator. The regulator should cut out all (just about) of the mains ripple.

[momaka]

Re: Headphone amp design

^ Agreed

The amp draws 11mA with no audio playing, and I saw peaks of 18mA with the volume turned all the way up. I have succeeded in creating an amp that can make my headphones buzz.

I have never had any success with building anything good with op-amps, so I just stick to the good old "book-example" class-A BJT amp.

Recently, I built an updated version of this:
https://www.badcaps.net/forum/showthread.php?t=32367
... but I simplified it down alot more (only 2 transistors per channel, and removed many resistors). I also improved the power supply noise rejection and was able to get the output to do a little over 3 Vp-p with an 8V switching adapter.

Maybe that will satisfy your line level needs?

Also, I wonder if you could set-up a BTL configuration with those op-amps of yours. That should (theoretically) get you 2x the output voltage with the same 5V power supply. The only thing is you will need to find a second op-amp IC. But I don't know if that will work. All I know is that I've tried it with BJTs, and it works fine. Also may remove the need for output coupling capacitors.

[lti]

Re: Headphone amp design

The MS6308 and LM833 can output 3.5Vp-p with a 5V power supply. I think a line-level signal should be 2V RMS. The LM833 is currently being used to amplify a line-level signal or the output from my laptop to a level the MS6308 can handle. I'm getting 1.1V RMS into my normal headphone amplifier (a TEA2025-based computer speaker amplifier with 390 ohm resistors in series with the headphone jack) and 1.3V into my cheap Philips SBC-HL150 headphones. I don't know why the voltage would increase as the load impedance decreases. I might be hitting the limits of the MS6308.

Op-amps are easy if you're using a simple amplifier circuit. You can also put other components in the feedback loop to change the output frequency response.

I would like to know how the tone adjustment circuit in the computer speaker amp works. It uses four NPN transistors and is powered by the DC voltage on the output of the TEA2025 before the coupling caps. I guess I'll try to draw a schematic for it some day.