According to Pete: Headphone Amp Project/High-Current Output Stages
I design and build a headphone amplifier, and go on a journey of self-discovery (and learn some hard lessons).
Welcome to the latest edition of ATP! This time I design and build a headphone amp specifically designed to operate at the end of a chain of guitar effects boxes, and be able to take stereo input from stereo effects like delay, chorus, etc. But oh, the best laid plans of mice and Petes...
This project tells the tale of many design iterations, bad case cutting and the importance of cool stickers, as well as the value of measurements over blind hypothesis and assumptions. And what starts out as a really simple circuit that doesn’t really work right ends up being, still, a fairly simple circuit but works a whole lot better.
You can’t get all of the components in the project from us, but here’s the list if you want to follow along:
- Aluminum Enclosure
- Barrel Jack
- Proto PCB
- ⅛” Stereo Output Jack
- (2) ¼” Input Jacks
- (2) Black Metal Knobs
- (2) 10k Log Pots
- (3) 100uF Electrolytic Caps
- For power and DC block to headphones
- (4) 0.1uF caps
- Coupling input signal from pots and decoupling reference voltages
- (6) 100k resistors (in Resistor Pack)
- Set up reference voltages and feed to op amp
- (2) 1k Resistors (also in Resistor Pack)
- LM317 Voltage Regulator
- Setup for 7.5Vout with 240 ohm and 1200 ohm resistors - you’ll have to scrounge your stash for these
- (2) 2N3904 Transistors
- (2) 2N3906 Transistors
- Analog Devices OP275
- And a sweet sticker of some kind for when you, like me, inevitably drill errant holes in your case
It’s worth a mention that the ProtoPedal pre-cut case was out of stock when I started this project, otherwise I would have gone with that. But since it was out, I basically free-handed the whole shebang. For better or worse, it’s all about the journey.
Great video! I like when project videos include all the mistakes and rough edges. Too many times I see tutorial videos that are all polished and just show how to build the final design without showing the iterative design process. This can be frustrating for new people when their project doesn't work right on the first try and they think "I'll never be able to magic circuits into existence like Pete does".
The only disappointment I had was at the end when you had your guitar and I thought "Sweet! We're going to get to see Pete play a little guitar!" And then the video ended.
Overall great video! I always get excited when I see a new According to Pete.
Aw, now... I'm a horribly self-conscious player, which is probably another reason for the headphone amp. I did play for a bit when we did the clips of the other guys, but I felt like I played badly and asked for some mighty editing - although I thought some of it would get in, but I guess all my playing sucked that day. Meh. I'll probably get over it one day, but I'm in no great hurry. I get a lot of satisfaction from playing all by myself.
I completely understand the sentiment regarding your playing. I feel that way about a lot of things that I do. As long as it brings you enjoyment, then keep on rocking out with your bad self! Love the vids, keep 'em coming!
I know Pete doesn't need this, but for the rest of us angry pixie wranglers I noticed in my YouTube feed a 3 year-old EEV Blog (#600) where Dave goes over OpAmps. https://www.youtube.com/watch?v=7FYHt5XviKc
I don't know what the problem is with the LM358 but the 80's vintage Carver C-1 preamp uses a simple opamp (RC4136) buffer for its headphone output amp. No external push-pull driver.
Possibly a better op amp to try is the OPA2132. Single supply operation down to 5V and 40mA output current ability. I do think you are going to want a higher supply voltage to get the headroom you need though. Note from the OPA2132 datasheet that for low RL, the output swing becomes asymmetrical with respect to the voltage rails, so the ideal input bias point may be a little higher than Vcc/2. Easy to calculate based on the gain you end up with. I would recommend as a minimum ditching the regulator and just add some capacitors if needed to filter the 9V supply. Check out the CMoy pocket amplifier: https://tangentsoft.net/audio/cmoy/ for other ideas and op amp options..
In your original circuit with the LM358, I assume this was operated on a single supply? Did you directly connect an AC coupling capacitor to the opamp to supply the headphones? If so, you probably could have fixed your original circuit by connecting a 10K resistor directly from the opamp output to ground, then have the series cap to the headphones. The output needs a DC resistive path to ground (or its negative supply) for the output section to operate properly without crossover distortion. This is buried in TI application note AN-116. Bottom line - there's really no reason to need the push-pull transistor output stage.
Also the OP275 is specified for a minimum supply voltage of +/- 4.5V (9V single-supply), so if you have a fairly clean 9V supply, you should eliminate the regulator all together to ensure you are running the OP275 within its minimum supply spec, otherwise, I'm not sure if you are getting optimum performance out of that device.
For other improvements you should look at building an input filter to replicate the relatively narrow bandwidth of a typical guitar amp speaker which rolls off below 100Hz and then pretty sharply above about 3-5 KHz. You can use a properly sized input ac coupling capacitor to simulate the roll-off below 100Hz, then a 2nd order active low-pass filter to simulate the high end roll-off. You will want a filter Q around 1- 2 to simulate the peaking that most guitar speakers have just before they roll-off.
D'oh! I misread the supply spec for the OP275! I thought it was single-ended down to 4.5V. Hmmm... well, it's working well, but it clips lower than I'd expect, maybe that's why. But there's a lot of noise coming off of the 9V wallwart, more than I could remove with just caps (I tried).
I wonder how well using both 7805 and 7905 regulators to get ±5V would work. That would get you 10V to run the OpAmp and still have enough head between the regulator output and the 9V input to provide space for cleaner regulation.
Yes, that's how the LMV358 (and LM) was set up. But when I was doing the testing, I tested with a static load before and after the cap, as well as with the headphones (thought it might have something to do with the reactive nature of the headphones)... at least, I think I did before and after the cap (I'd give about a 90% degree of confidence that that's how it went down). In any case, the signal becomes clean without any load at all, that is, no DC or AC path. So I've got a couple of data points that indicate otherwise... but those are associated with my own degrees of confidence. So... you still may be right. Only further testing will show.
As far as the replicating the narrow BW of a guitar amp, that's something I've hated forever. I wanted to play with a wider BW, and so far I'm not disappointed.