Guitar Headphone Amplifier

[tech_builder]

Hi,

I am designing a simple guitar headphone amplifier for my son, and I would like to get some feedback on the design I have come up with. I've spent some time looking around at various incarnations of headphone amplifier and decided that I wanted to use jellybean parts and not specialty parts like an LM386 just to make it more interesting. I’m not interested in pristine audio quality as it is a for guitar, but I don’t want noticeable distortion when playing clean tones either.

It seems to me that one of the departures guitar headphone amplifiers have from a typical headphone amp design is the requirement to manage the input attenuation of the device as guitar pickups vary a lot in their output. I decided to gang opamps in series with the first one attenuating and the second one acting more like a volume once the guitar level was set. This method does not let me turn the follow all the way to zero though. Any suggestions on making this better would be welcome.

Once past the input stage the output is a conventional class AB amplifier with a simple Tl082 driving the transistors. I’ve seen some designs with an output cap and some without, I would like to use one to deal with possible DC offset, however I’m not too sure what size to use or what type. The ones I’ve seen seem large enough that they would need to be an electrolytic cap, but I’ve heard that electrolytic are not very good for audio applications (at least in the audio path). 220uF was one value I saw, but 47uF was used in the MXR headphone amp. I saw another one with 1000uF too.

The power stage is a buffered virtual ground design I scrounged on the internet. I’m not sure if it is overkill, or if a simple resistor and capacitor divider would work just as well. I thought it would be fun to try

I’ve built the circuit up on a breadboard and for the most part it sounds ok. Getting it to not distort at a decent listening volume can be challenging, which is something I’m hoping to get some advice on before committing to a circuit layout. I’m also happy for any pointers to existing designs that I can look at as well.

Thanks!

[DrGeoff]

You might want to set the input impedance of the input stage to around 47K, as this is what most guitar pickups expect to see.

[Keicar]

My understanding was that an input impedance around 1M? was more typical.

[Zero999]

The input has no biasing resistor, meaning the output will quite likely saturate at either supply rail, when the input is disconnected. Add a resistor from the input to 0V: 1M does seem reasonable.

Running the TL082 off a total supply voltage of 9V is marginal. It's designed for higher voltages (>10V) than that. Most of the datasheet doesn't specify it below 10V (+/-5V) and the lowest supply voltage in any of the graphs (figures 6 & 12) is 7.5V (+/-3.75V) which won't give you much battery life. It also depends on how much voltage swing you require. I'd consider using an op-amp which is fully specified to work down to 6V.

http://www.ti.com/lit/ds/symlink/tl082.pdf

[Buriedcode]

Most guitar amps have an input impedance of ~1M.  Lower and it can roll off the highs (not that electric guitars have much in the way of high frequency content).

I'm assuming by 'headphone amplifier' you mean a box you plug your guitar into, and headphones into, so he can hear the guitar - rather than just a headphone amp one would plug into the 'headphone out' of a guitar practice amp which isn't loud enough ?

In this instance, whilst starting off just 'making an amp' would be good, I think you should think about the possibility of adding some sort of overdrive and/or EQ.

If he's learning, then although it completely takes the fun out of DIY, there are some very nice cheap portable multieffects units about, which are pretty much designed for headphone practice, and have all manner of 'tones'.  I found when learning a touch of reverb completely changed how I played, and also.. the 'nicer' tone made me play more.

As for the opamp choice, I've seen guitar outputs hit +/- 4V for the higher output humbuckers, so if running off 9V you might consider something with rail-to-rail inputs.  But as you said 'jellybean' parts that makes it difficult.  A TL082 is just fine for guitar (many guitar pedals use the TL072, but a TL082/4 is in quite a few, electric guitars dont' really need much bandwidth). But as you noticed, clipping might be an issue (input or output).  You can always just use two 9V batteries, but I hate those things, expensive, and won't last long.

[tech_builder]

Thanks for all the replies!

DrGeoff: I’ve never heard that guitars expect to see 47k impedance. Where does that spec come from? Does it change from pickup to pickup?

Hero999: Thanks for the heads up on the input biasing resistor, it was definitely saturating! I decided to change the op-amp to something rail to rail. I found an interesting website that looks at a number of modern op-amps for application to audio:

http://www.cycfi.com/projects/six-pack/op-amp-shootout/

I went through his list and found a few that I thought could be suitable candidates: OPA1664 looks like it could be good, and it’s relatively inexpensive. I also found the OPA4171 as well.

Buriedcode: He has an amp already with headphone output, but he wants something that he can move around with and won't be a big distraction for his baby sister. He's not too picky, and it is a win-win situation for us both . He gets a guitar headphone amplifier, and I get to make something cool. I will probably iterate on it in the future sometime, but I have some other projects I would like to get to first. As I mentioned above, I decided to go with different op-amps that are designed to operate at lower voltages and are rail to rail.

I’ve attached an updated revision on the guitar headphone amplifier. I included the input biasing resister mentioned by Hero999, as well as the new op-amps (in this case the OPA1662, which was modeled in TINA TI). I also changed the volume portion of the input to be an inverting amplifier so that I could get a gain lower than one. I am having trouble getting TINA TI to run with these op-amps in the circuit though. It keeps telling me I have convergence problems, and I’m not too sure what that means. The circuit simulates fine with TL082s.

[BrianHG]

Don't you have the + and - inputs of opamps U3&U4 wired backwards?
And wont the way you wired the volume in U2 only go from full volume loud to super loud? (Unless you intend to distort the sound)
Also, having a volume control knob wired into the feed-back circuit of an op-amp.  This isn't usually done in practice for interference and noise & performance/signal gain issues, but it can function.
C5 is also shorted out and of no use in this new schematic, it does not match the first schematic.
There were fewer mistakes in your first posted schematic.
Also, the output drive has enough drive for both ears of your headphones.  I recommend getting rid of 1/2 of your output stage if you want to save parts and a bit on battery life.  Just use a second 220uf cap from the transistor emitters to the second earphone channel to maintain low frequency response.

[MagicSmoker]

...
http://www.cycfi.com/projects/six-pack/op-amp-shootout/

I went through his list and found a few that I thought could be suitable candidates: OPA1664 looks like it could be good, and it’s relatively inexpensive. I also found the OPA4171 as well.
....

Op-amps for driving headphones need a bit more grunt that the usual "audio" op-amp like the venerable NE5532 or LM833. Personally, I really like the sound of STMicro's TS922AIN, and it has 80mA max output, which is extraordinarily high for an op-amp in a standard SO-8 package. It is a bit of a power hog, though, requiring 2mA supply current minimum.

The OPA1664 looks good, too. In fact, there are dozens of op-amps that will do an excellent job, and trying them out could be a worthwhile pursuit in itself. Use a socket for the op-amp to make it easy to swap it out.

[Audioguru]

The 10nF capacitors that are parallel with all negative feedback resistors cut all the mid and high audio frequencies.
For example when P1 is set to halfway, its 25k ohms and the 10nF capacitor cuts frequencies above only 640Hz.

[BrianHG]

The 10nF capacitors that are parallel with all negative feedback resistors cut all the mid and high audio frequencies.
For example when P1 is set to halfway, its 25k ohms and the 10nF capacitor cuts frequencies above only 640Hz.

I said C5 in his second updated schematic.  That one isn't wired correctly.  It's wired correctly in the first schematic.
All the other feedback 10nf caps are functional and fine in both schematics.

[tech_builder]

That's embarrassing...its what you get when you only have 20mins to work on something...it looks like when I did a swap for the U3 and U4 opamps they weren't oriented correctly, and the shorted cap

I was wondering whether one of the output transistors could power both headphones, so I guess it's worth trying out. I'll look through all of this again when I get another chance and hopefully produce something better

I have ordered the op-amps I mentioned. Can't wait to see how those work out!

[BrianHG]

That's embarrassing...its what you get when you only have 20mins to work on something...it looks like when I did a swap for the U3 and U4 opamps they weren't oriented correctly, and the shorted cap

It's alright.  You should see how many times I end up editing my posts after I do the post.  Dooohhh... 
As long as you catch 99% before you may a PCB....

[Audioguru]

Here is a simulation of the frequency response with the 10nF capacitor negative feedback capacitor cutting mid and high audio frequencies:

[danmcb]

what you have is basically OK, but as others have said, you want a resistor of somewhere between 470k and 1M to deck at the + input of the input stage to set input impedance. It is also common to place a -3dB point for the LF response with a blocking cap  before that. You can set the LF point at about 40Hz, as the E string of a guitar is round about 80Hz. You also want to set an HF 3dB point - around 10 or 15kHz is fine for a guitar pre. This is better done on the power amp stage. You might want some HF rolloff on the preamp for stability but it will vary with the feedback resistance, so push it up out of band, about 50-100kHz  so that the tone does not alter with gain.

You are making a bit of a meal of this though. On a thing like this there is not much point having separate gain and volume - unless you have a SIG LED or something to help set the gain. They are just going to do pretty much the same thing from the point of view of the user.

And why two output amps? for left and right of a headset? just drive them in parallel.

So you can do the whole thing with a dual op amp like a TL072 (FET type is better than 5532 for this to get the input Z high). One preamp stage with a GAIN pot, and a mini power amp (opamp with push pull o/p to give current drive,  as you have shown).

[MagicSmoker]

I'm embarrassed to admit I just now looked at the schematic... I agree with the other comments, but also want to add that you can greatly simplify this circuit by using one of the op-amps I recommended to directly drive the headphones. That is, no output buffer will be needed.

I'm not sure how much voltage gain you need in a guitar amp, but I suspect it is not so much you need to break the amplification into 2 stages (plus a unity gain buffer), which is what you have now. It is also best to put the volume pot in series with the input, and not in the feedback path. The capacitor across the feedback resistor is to prevent oscillation and roll off the gain at high frequencies; the voltage gain will be cut in half (ie - down 3dB) when the reactance of the capacitor equals that of the feedback resistor. If the feedback resistor is 10k then an appropriate choice of shunt capacitor is 470pF. Note, however, that this capacitance is seen by the output of the amplifier and can contribute to instability itself if it gets too large in value (and 10nF is almost certainly too large in value, both because of this reason and for rolling off the output way too early, as mentioned above).

Finally, when you use an active rail splitter to create a bipolar supply you don't need the big DC blocking capacitor in series with the speakers. You do, however, need to remember that the signal ground is actually at 1/2 the supply voltage and so might not get along well with other devices sharing a common chassis ground. In this case there shouldn't be any problems as a guitar is a floating signal source and headphones are a floating load.

[Zero999]

Here is a simulation of the frequency response with the 10nF capacitor negative feedback capacitor cutting mid and high audio frequencies:

But that's not what the schematic says. C5 is bypass with a piece of wire, so does nothing.

[Audioguru]

But that's not what the schematic says. C5 is bypass with a piece of wire, so does nothing.

The circuit has two parts that cut high audio frequencies. Before I showed the worst one that was accidently shorted by a wire and now I am showing the other one that is not shorted. Both filters added together destroy the audio.

[tech_builder]

Once again, thank you for all the feedback. I finally found some time this weekend to breadboard a new design I came up with after reading through what has been discussed and doing some more research.
The first thing I decided to change was the opamp responsible for volume change. I just swapped it out for a simple pot. I had a 100k on my desk and it worked well.

I still wanted to have a gain and volume, so I put together a simple clip circuit with a comparator to tell me when the output from the first opamp was over about 2.8V, which was where I figured the distortion would begin. I figured it might be difficult to see the clipping LED so I put a simple sample and hold type circuit on it to give it a little bit of persistence. Seems to work good.

The OPA1664 gave a much much better performance than the TL082 opamp did so I will move forward with this opamp. The OPA1664 is a 4 opamp package so I only need to power all of them once. I’m not too sure what to do with the unused opamp, any ideas?

The other thing I did was switch to a simpler power supply strategy by removing the transistors, which seemed to work fine in testing.

I changed the resistor values in the output (biasing?) stage from 4.7k to 1k in a bid to increase the voltage swing I could get out of the headphones before distortion occurred. Unfortunately this increases the current consumption...So far with my limited testing the whole circuit pulls about 8mA idle and about 12-16mA while strumming the guitar at a reasonable volume. A 9V battery with about 500mAh it should last 30-40 hours. Maybe I'll look into getting some small lithium cells and stringing them in series to get above 9V, but then I have to figure out how to charge them...

Hopefully this effort is better than my last one  I appreciate any feedback you may have on this iteration. Thanks!

[Zero999]

The unused op-amp can be connected as a unity gain buffer with its input connected to 0V.

Bear in mind that the clipping voltage depends on the supply voltage and it will reduce, as the battery discharges. The comparator's reference could be changed so it indicates clipping when the output voltage exceeds the supply voltage minus a three diode drops.

[Audioguru]

On your new headphones amplifier circuit I calculated that its maximum output level will be low if the transistors are not replaced for higher current gain.
If the current gain is low and the base-emitter on-voltages are high for the 2N3904 and 2N3906 output transistors then the peak output voltage will be far less than 1.6V.

100mW is loud in a headphone. For 100mW the RMS voltage must be 1.25V then the peak voltage must be 1.77V.

Your high value for the emitter resistors is throwing away much output voltage swing and is causing crossover distortion.
C1 is preventing the opamp from reducing output distortion.

I use my hearing as a clipping detector because I do not listen to RAP, instead I listen to music.

[Zero999]

On your new headphones amplifier circuit I calculated that its maximum output level will be low if the transistors are not replaced for higher current gain.
If the current gain is low and the base-emitter on-voltages are high for the 2N3904 and 2N3906 output transistors then the peak output voltage will be far less than 1.6V.

100mW is loud in a headphone. For 100mW the RMS voltage must be 1.25V then the peak voltage must be 1.77V.

Your high value for the emitter resistors is throwing away much output voltage swing and is causing crossover distortion.
C1 is preventing the opamp from reducing output distortion.

I use my hearing as a clipping detector because I do not listen to RAP, instead I listen to music.

Those are good points. I don't think it will be that bad, since the transistors will be much better than the minimum values, especially after they've warmed up a bit, as the current gain has a positive temperature coefficient.

I think the biasing resistors could all be increased to save battery life and the whole biasing network AC bypassed with two large electrolytic capacitors to couple the signal from the op-amp to the transistors.

[MagicSmoker]

Or, you know, the entire "booster" amplifier stage could be dispensed with by choosing an op-amp with a higher output current rating like the TS922AIN I suggested before. That op-amp's max output of 80mA can deliver over 200mW into the typical pair of 32 Ohm headphones all by itself.

[Zero999]

Here's what I was talking about earlier on. 1000µF may sound large but the capacitors only have around 700mV on them, so can be a low voltage rating such as 6.3V.

The 220R biasing resistors may need to be adjusted to give low power draw and low crossover distortion.

[Audioguru]

Hero has a good idea of using the capacitors so that the opamp can turn ON the output transistors with plenty of current instead of turning OFF the transistors.

[calexanian]

add say a 470 ohm resistor directly from the output of the op amp to the load for stabilization purposes. That circuit works fine for non audio but without a resistor there you may have noise or junk in the signal. I used to build that circuit all the time when a LM386 would not cut it and I had a  dual power supply. Oh, and you don't really even need those capacitors. If you want boot strapping of that to above the swing of the op amp there are other circuits to use.