Headphone amp design

[lti]

I want to improve some headphone amplifiers that I have. One is based on an MS6308 (a TDA1308 clone), and the other uses an APA3541.

I have two MS6308 chips. One is used as the headphone amp in my computer, and the other is on a motherboard that I have been taking parts from. In the computer, I just need to reduce the gain to match the output of the speaker amp. Replacing the 20K ohm feedback resistors with 15K ohm resistors should work.

For the second chip, I might try to remove it and use it. I would like to know why all of the TDA1308 circuits I see use the inverting input and place resistors in series with the input. Is there a fault with the chip that makes the non-inverting input not work properly? In the computer, the inverting input was used, but the audio signal it receives is inverted by a TL064.

The APA3541 sounds horrible. It is the worst excuse for an amplifier I have ever heard. I also have two of these chips, and they both sound the same. I am currently using the circuit board from a CD-ROM drive that had the chip already installed. I modified the circuit to match the application circuit, but with a 20K ohm potentiometer on the input and a ferrite bead on the output (it was already on the board). The high frequencies are extremely loud and distorted, but the rest of the audio range sounds normal. I think I need to add a low-pass filter. It sounds okay with a 0.22uF cap from the input to ground or a 47uF cap from the output to ground, but those aren't ideal filters.

I would like to make the APA3541 usable (if it is possible) and use the non-inverting input of the TDA1308.

[budm]

Re: Headphone amp design

Do you have te diagram of the circuit you are using? What do you use on the power amp feed back, any cap for limiting high frequency band width? Any resistor at the output for buffering the Hadphone (resistor in series with the head phone), any RLC network at the output of the power amp?

[lti]

Re: Headphone amp design

Do you have te diagram of the circuit you are using? What do you use on the power amp feed back, any cap for limiting high frequency band width? Any resistor at the output for buffering the Hadphone (resistor in series with the head phone), any RLC network at the output of the power amp?

If you're asking about the APA3541, I have no way to change the feedback. I have attached the datasheet for this chip and a picture of the board I am using. I have changed the component values to match the application circuit. There are no resistors in series with the output. The inductors at L701 and L702 on the board (marked "Bead") are in series with the output. I have also left the 4.7 ohm resistor in series with the 5V supply. Bypassing this resistor doesn't affect the sound.

I have also uploaded the MS6308 datasheet. It is similar to the TDA1308 datasheet and has identical specs, but it has more application circuits. The manufacturer's website redirects to some kind of cartoon porn site. I don't know why anybody would want to look at crappy drawings of people ing.

[lti]

Re: Headphone amp design

I made an acceptable low-pass filter for the APA3541. I am using a 4.7K ohm resistor in series with the input and a 0.01uF cap connected to ground. There is still some distortion (I have no idea where Anpec got 0.03% THD). Shorting the inductors might have had a slight effect, but I'm probably imagining it.

[lti]

Re: Headphone amp design

Is the TDA1308 really inverting the signal? The voltage divider that places half of the supply voltage on the non-inverting input is confusing me.

Also, what kind of caps should I use for output coupling? The computer currently uses 100uF tantalum caps for output coupling. Should I replace these with electrolytic caps?

[momaka]

Re: Headphone amp design

I made an acceptable low-pass filter for the APA3541. I am using a 4.7K ohm resistor in series with the input and a 0.01uF cap connected to ground. There is still some distortion (I have no idea where Anpec got 0.03% THD). Shorting the inductors might have had a slight effect, but I'm probably imagining it.

Maybe try playing with the input, output, and bias capacitors (swap for higher values).
For decent headphone bass output (IMO), you'll want at least 1000uF output capacitors. For input, 1 uF should be okay since the amplifier's input resistance is quite high (180 kOhms). Just keep in mind that if you change the input coupling caps, you might also have to change the bias capacitor to satisfy that equation in the data sheet.

As for the series inductors on the output - if anything, those should be doing some high frequency blocking too. It's also possible that this amp is just not powerful enough to drive that much bass/low frequencies through headphones.

[redwire]

Re: Headphone amp design

Chip amps have a few quirks to getting them to sound reasonable. I use LM386, TDA820, TDA2822 for 8ohm loads. First you need RF filter on input as AM radio gets in and things sound distorted. Second, since the output stage is quasi-complementary it can oscillate and a RC snubber 10ohms+0.1uF from output to ground helps with this.
The chip data sheets look fudged- that low distortion spec is awesome but they don't say at what load if any.
Check the chip's bias - the output pin should idle at 1/2Vcc.

[lti]

Re: Headphone amp design

Maybe try playing with the input, output, and bias capacitors (swap for higher values).
For decent headphone bass output (IMO), you'll want at least 1000uF output capacitors. For input, 1 uF should be okay since the amplifier's input resistance is quite high (180 kOhms). Just keep in mind that if you change the input coupling caps, you might also have to change the bias capacitor to satisfy that equation in the data sheet.

As for the series inductors on the output - if anything, those should be doing some high frequency blocking too. It's also possible that this amp is just not powerful enough to drive that much bass/low frequencies through headphones.

The bias cap doesn't seem to have any effect on the sound quality. The circuit originally used 0.1uF caps on the input and a 47uF bias cap.

I don't think the output power is the problem. The low frequency response is actually better than the TDA1308. The filter I added has a -3dB point above 3KHz.

Chip amps have a few quirks to getting them to sound reasonable. I use LM386, TDA820, TDA2822 for 8ohm loads. First you need RF filter on input as AM radio gets in and things sound distorted. Second, since the output stage is quasi-complementary it can oscillate and a RC snubber 10ohms+0.1uF from output to ground helps with this.
The chip data sheets look fudged- that low distortion spec is awesome but they don't say at what load if any.
Check the chip's bias - the output pin should idle at 1/2Vcc.

The APA3541 has no background noise. I never tried adding a snubber or checking the bias. The MS6308 inside my computer picks up some background noise (including some subsonic noise), but that was reduced significantly when I reduced the gain. I ended up using 10K ohm resistors, and the headphone output is still a little louder than the speakers. Unfortunately, the next lower value of surface mount resistors I have is 1K ohm.

The LM386 makes a surprisingly good headphone amp. The LM386 chip I have sounds horrible with a load of less than 16 ohms. With headphones (32 ohm load), the sound quality is the same as a TDA2822M (the high frequencies are a little too loud), but there is very little background noise.

[ben7]

Re: Headphone amp design

Maybe there is a noisy resistor or two in there?

Edit: Oh, and what about the power supply you are using? Brand of headphones?

[redwire]

Re: Headphone amp design

I kinda hate the LM386. It's noisy, oscillates like crazy and has high distortion with loudspeaker loads (<16ohms). It also makes an awesome AM radio receiver...

The problem with all chip-amps, even the big ones i.e. LM3886 is that putting a PNP power transistor on an IC takes up a lot of the die and is expensive. You need that PNP for the output stage. Instead, they use a NPN so you get a lop-sided output stage and more distortion and less stability.
So I've given up on a low-distortion chip-amp for headphone use.
Lately, I tried a Chinese clone of the Lehmann amp from eBay. It is an op-amp driving a discrete transistor output stage, biased heavily into class A so it runs hot. If you're a real headphone kinda guy, it might be worth looking at.
Silicon Chip magazine did a couple good designs too.

[ben7]

Re: Headphone amp design

I've seen some articles of small MOSFET and op-amp based head-amps online. From what the constructors say, they give quite decent sound quality.

[lti]

Re: Headphone amp design

Maybe there is a noisy resistor or two in there?

Edit: Oh, and what about the power supply you are using? Brand of headphones?

Which amp would have noisy resistors?

For the MS6308, I am powering it with a 78L05 regulator fed from the computer power supply's +12V rail. The APA3541 has been powered by many sources. I am using crappy $10 Philips headphones.

I kinda hate the LM386. It's noisy, oscillates like crazy and has high distortion with loudspeaker loads (<16ohms). It also makes an awesome AM radio receiver...

I never picked up AM radio with one. I did get it to oscillate by connecting it to a portable CD player. It definitely isn't a hi-fi amp, but it works good enough for me. I have even seen "hi-fi" headphone amps based on the LM386. I think an LM386 is better than the typical inverting amplifier based on an op-amp that is commonly used for "hi-fi" headphone amps.

So I've given up on a low-distortion chip-amp for headphone use.
Lately, I tried a Chinese clone of the Lehmann amp from eBay. It is an op-amp driving a discrete transistor output stage, biased heavily into class A so it runs hot. If you're a real headphone kinda guy, it might be worth looking at.

The headphone amp I usually use is the amplifier from some old computer speakers. It uses a TEA2025B and a tone control circuit. It is picky about what coupling caps are used between the tone control circuit and the final amplifier, but it works very well. I can just use cheap Philips headphones and it sounds as good as larger, more expensive headphones. If you look at the TEA2025 datasheet, you wouldn't think an amp based on that chip could ever sound good. Of course, it isn't very portable. I would need to find an enclosure for it and bring a ton of batteries, and I want to find replacement drivers for the speakers (one blew after the caps failed in the amp).

[ben7]

Re: Headphone amp design

For the MS6308, I am powering it with a 78L05 regulator fed from the computer power supply's +12V rail. The APA3541 has been powered by many sources. I am using crappy $10 Philips headphones.

Do you have power supply decoupling caps a plenty? 78xx series vregs will get unstable if you don't have a small low esr cap on the output (like a 100n ceramic). It would be nice to have a 10uF on the output too, and a 100n placed directly across the amplifiers' power pins.

Edit: The datasheet for the APA3541 actually says that it does need good supply bypassing in order for the lowest THD to be reached.

[lti]

Re: Headphone amp design

I would have to check that. The regulator and MS6308 are on the computer's motherboard. The APA3541 sounds the same on all power supplies. I have run it from batteries most of the time. I currently have a 100uF filter cap, which used to be one of the output coupling caps.

[momaka]

Re: Headphone amp design

The circuit originally used 0.1uF caps on the input

That's kind of low. Even with the amp's 180 KOhm input resistance, the cut off frequency is around 9 Hz. Try at least 1 uF. And you don't want the output caps to be the limiting factor either, so go with 1000 uF.

[lti]

Re: Headphone amp design

I think you missed the part where I replaced the caps with the values used in the application circuit in the datasheet. I am using 1uF input caps, 220uF output caps, a 10uF bias cap, and a 100uF filter cap. I could install larger output caps, but I don't think I would be able to tell the difference with these headphones.

I think I have made that amp sound as good as it ever will by adding the low-pass filter. I don't have any problems with power supply noise or bass response. The only problem I had with it was that the high frequencies were too loud and distorted (it was probably amplifying high frequencies so much that it was clipping). I want to move on to the MS6308. That amp has poor bass response in its standard circuit design, even with large output coupling caps (I used 470uF).

In the computer, I stacked two more tantalum caps on top of the original output caps. The capacitance is now 247uF, and there is some weird low-frequency noise on the output that I couldn't hear before. The new feedback resistors aren't soldered on very well because I didn't want to remove the motherboard. The second MS6308 (if I can remove it from the board) will be a standalone headphone amp for some future project.

[momaka]

Re: Headphone amp design

I think you missed the part where I replaced the caps with the values used in the application circuit in the datasheet.

No, I saw it but I forgot

In the computer, I stacked two more tantalum caps on top of the original output caps.

Tantalum? In a signal path? I don't know if that's such a good idea. (or am I mistaken here?)

[lti]

Re: Headphone amp design

I don't know if it is bad, but I didn't have room to solder through hole electrolytic caps in place of the original 100uF tantalum caps and the largest surface mount caps I have are only 100uF. Compaq never really knew what they were doing, so it could be bad to use tantalum caps. I think I asked about this earlier.

The stack of caps would look nice in the ghetto mod thread.

[Th3_uN1Qu3]

Re: Headphone amp design

Tantalums are good for use in low power audio amps - they do it in laptops all the time. Better than lytics actually.

Ceramics are the one you want to avoid, because of their tendency to "ring" and pick up external noise (physically) - using ceramic caps in a signal path (especially on the input) is like having a microphone across your input signal.

[lti]

Re: Headphone amp design

Ceramics are the one you want to avoid, because of their tendency to "ring" and pick up external noise (physically) - using ceramic caps in a signal path (especially on the input) is like having a microphone across your input signal.

Due to the design of the circuit feeding the amp, there are three ceramic caps on each input of the MS6308 in the computer. I can't tell what the circuit is doing since it is on a multilayer board.

Of course, the APA3541 datasheet recommends ceramic caps.