So I tried to design/build a guitar headphone amp..

[jonny]

So I tried to design/build a guitar headphone amp and thought it'd be simple right...

So here it is below. But basically it's a buffer into a tone stack then a little bit of gain and op amps to drive the headphones. It simulates okay in spice but when I went to make it I have no luck.

My problem is noise. It's sounds quite good, but there's just so much noise. The noise is about as loud/ or louder than the guitar. I'm using a LT1054 to make a bipolar supply, I'm thinking it could be an issue somehow. Or just bad design haha? Any ideas?

[Jdansti]

Can you post a schematic of the LT1054 supply?

[jonny]

It's the inverter taken directly from the datasheet, I've used it before in a few other designs no problem. But I have noticed if you don't keep the coupling close to the op amps it can be noisy. I'm gonna try moving it a bit closer but i'm not sure it'll help.

[Jdansti]

That could be. Also, if you have an adjustable soldering iron, make sure it's off. They put out a lot of RF noise.

[R.G.]

My problem is noise. It's sounds quite good, but there's just so much noise. The noise is about as loud/ or louder than the guitar. I'm using a LT1054 to make a bipolar supply, I'm thinking it could be an issue somehow. Or just bad design haha?

Probably bad layout and wiring, and lack of decoupling.

Noise as loud as guitar in a circuit like that probably means that it has bad RF oscillation problems. As National Semiconductor used to note in their (...1970s...)  data books, although power supply bypassing and proper grounding are not shown in schematics, they're always assumed. They used to recomment one 0.01 or 0.1uF ceramic cap from both power supplies to ground as close to the power pin(s) as it was possible to get them on *every* IC. That's usually overkill, but it's necessary for some high gain/high impedance circuits. RF oscillation has an "angry" quality to the hiss, not the steam-escaping of thermal noise.

It could be the LT1054 leaking through, but in my experience that tends not to give the hissing noise of RF oscillation.

[PRR]

You need to bring 20mV guitar up over a Volt at the headphones, gain of 50.

The Fenderish tone-stack has loss near 10:1, so you need total gain more like 500.

You have a sorta-weak signal (guitar) go through the lossy tone-stack without gain in front. Don't take loss before you make gain. That just puts the signal down deeper in the universal hisss. "U4" should be a gain stage, not just a buffer.

Headphones only need about a Volt rms, 3V peak-to-peak, so an ideal power stage should only need 3V supply. Indeed all iPods and such run about 3V supply. But those are somewhat specialized chips. A single 9V supply may be easier to design around.

The gain you can take from guitar with 9V supply is limited. Guitar can put out over 0.5V; a 9V amp won't do over 3V clean. So max front-end gain is maybe 5 or 6 (or adjustable...).

I would _really_ plagiarize (or research) any of the many op-amp based guitar amps, especially ones with a "gain" pot in the first stage.

TL072 is not good about driving 164 ohms to headphone level, and applying 18V isn't making it any happier.

LM386 is really a very good headphone driver, and happy with single 9V power. Yes, this means a Vref for your opamp stuff.

First-stage gain say 5. Tonestack loss 10:1. LM386 gain of 200. Total gain about 100. You could rig a switch on the first stage for gain of 5 or 25, to cover all cases.

[Jdansti]

+1

Here's a classic that's easy to build. I used two output jacks in parallel. One to go to a speaker cab, and the other for head phones with a 100 ohm, 1W resistor in series with the jack to avoid blowing my ear drums. It's a great sounding amp with nice overdrive-sounding gain.

http://www.beavisaudio.com/projects/NoisyCricket/Noisy_Cricket_Schematic.gif



The only other mod I made was to place a SPDT switch in series with the gain pot because I couldn't get a completely clean signal by just turning down the pot.  Do a search on YouTube for Noisy Cricket Amp and you'll find some good examples.

[jonny]

Thanks for the help everyone. Still no luck with it for the moment, I made this design one night in 30min when I was sending off a few other pcbs to a board house. Yeah, I did think I should have added gain to the buffer stage, but even as is it should work. I mean the noise is louder than the guitar.. it's just not right.

Here's a pic of my board..



I tried increasing the coupling/ adding it closer but it didn't help. I noticed that when I attached my ground of the scope to ground the hum/noise was drastically reduced?? - (Although still not perfect). With the scope I can see a whole lot of noise, I think it's only there after the last op-amps.. but I might need to double check that. hmm.

I did think about using the 386 but my logic was that a cmoy works and I have op amps around so why not just go with that.

Here's my eagle files if anyone's interested. The layout is pretty much just auto routed and placement was pretty rushed. It was really just a project I thought I'd do while I was sending other PCBs off. But probably need to put a bit more thought into it now.. I really want to get it working. Or at least my flatmates as well haha.

Eagle files:
http://www.mediafire.com/?7pcdimuudgdoy1m
http://www.mediafire.com/?vqk2jq1kcywmeqq

[Jdansti]

This has been hashed out here with excellent comments by PRR in addition to what he's provided above:

http://www.diystompboxes.com/smfforum/index.php?topic=88880.msg751834;topicseen

Based on PRR's info and what I've found on the web, I think you may be fighting the limitations that the TL072 has for this application. It can be done, but you'll have to give the TL072 some additional help.

BTW, as you probably are aware, simulators do some amazing things, but we sometimes see (hear) different results than what the simulator estimated.  (For you simulators lovers, not trying or gonna get into a simulator argument with you here ).