Has anyone else tried out the "Headbanger" headphone amp, based on the LM386 opamp?
Complete schematics and description is at "www.minidisc.org/headbanger"
I did it just a view hours ago. It sounds really crispy but has ahellofalot noise in it.
Volume seems to be ok, although it distorts a bit when the 100k pot is fully open.
I used the headphone-out of a small Mackie console, because I need to make a headphone distribution amp with 1 in and 6 outs. Power is supplied by a 15VDC 1amp wallwart.
One thing I noted: It is somehow not clearly stated how to connect the separate grounds but on the bottom view of the amp board they are clearly connected together. A corresponding image is available at the link mentioned above.
Any idea of how to control the noise? Maybe by using lower input voltage, perhaps 9 - 12VDC?
Phantombox
Complete schematics and description is at "www.minidisc.org/headbanger"
I did it just a view hours ago. It sounds really crispy but has ahellofalot noise in it.
Volume seems to be ok, although it distorts a bit when the 100k pot is fully open.
I used the headphone-out of a small Mackie console, because I need to make a headphone distribution amp with 1 in and 6 outs. Power is supplied by a 15VDC 1amp wallwart.
One thing I noted: It is somehow not clearly stated how to connect the separate grounds but on the bottom view of the amp board they are clearly connected together. A corresponding image is available at the link mentioned above.
Any idea of how to control the noise? Maybe by using lower input voltage, perhaps 9 - 12VDC?
Phantombox
I built this amp 10 years ago now and it has worked great ever since! I power mine from a 9V battery and put a .1uf cap across the battery and that reduced the noise a lot. I built mine on perf board from Radio Shack parts in a hotel room so the construction isn't great, the layout isn't great and still it works great! it gets extremely loud but is a bit harsh sounding.
But for me, my goal was to be able to plug it into my laptop and watch movies on a plane over the loud engine noise. so a little bit of hiss and harshness is drowned out by the jet engines LOL!
Zc
But for me, my goal was to be able to plug it into my laptop and watch movies on a plane over the loud engine noise. so a little bit of hiss and harshness is drowned out by the jet engines LOL!
Zc
LMH6672 could be a bit of a leap for most building LM386 amps - soic, power pad, 80 MHz CFA - not exactly a socket and perfboard project amp choice - and the OP's 15 Vdc supply will toast the chip
the 1st question for a headphone amp is always - what headphones? - they vary in Z and sensitivity by orders of magnitude - no "standard" like 8 Ohms for loudspeaker
IEM need step down Xfmr to get good S/N, for some 200 mVrms will drive them to 120 dB SPL
"mainstream" circumaural headphones range from 16-600 Ohms
the 1st question for a headphone amp is always - what headphones? - they vary in Z and sensitivity by orders of magnitude - no "standard" like 8 Ohms for loudspeaker
IEM need step down Xfmr to get good S/N, for some 200 mVrms will drive them to 120 dB SPL
"mainstream" circumaural headphones range from 16-600 Ohms
LM386s are, in fact, very noisy. Effective input voltage noise is about at 4556 level, but 40 dB of voltage gain is just excessive for headphones. These chips were designed as small speaker drivers and are not exactly hi-fi. The Headbanger amp itself seems to be about as good as LM386 amps get, though I would have used a smaller-value volume pot.
If you're looking to build a very simple headphone amp, I'd recommend a classic Cmoy with a NJM4556 opamp, which AFAIK is pretty much what's inside a Grado RA-1. This will be lightyears ahead of any LM386 amp. Gain should be chosen depending on headphone sensitivity, it'll usually be somewhere between 2x (6 dB) and 10x (20 dB).
If you're looking to build a very simple headphone amp, I'd recommend a classic Cmoy with a NJM4556 opamp, which AFAIK is pretty much what's inside a Grado RA-1. This will be lightyears ahead of any LM386 amp. Gain should be chosen depending on headphone sensitivity, it'll usually be somewhere between 2x (6 dB) and 10x (20 dB).
PS: OK, so upon re-reading the OP wants a headphone distribution amp. I'd go with the suggested bunch of 4556s then. One per channel is fine for general purpose applications, just make sure you have some decently high supplies.
A 15 VDC wallwart isn't exactly ideal, that would mean either many or big coupling capacitors. I'd copy the O2 power supply which only requires an AC wallwart (transformer).
A 15 VDC wallwart isn't exactly ideal, that would mean either many or big coupling capacitors. I'd copy the O2 power supply which only requires an AC wallwart (transformer).
The venerable lm386 chip can be made to play ball. That is what the Headbanger amp is all about. I could argue that a gain of 12 is higher than most people would require. But it's all a trade off. It has the power for basically any headphones known to man.
The very first amp I built was a headbanger and in many ways, it is better than the numerous C-moys that I have since built. No Cmoy amp made can power my 32ohm phones properly.
The headbanger could power two sets in parallel with ease. The THD of the 386 based headbanger amp is well within acceptable limits.
However, the problem of the one amp not having enough power for more than one set of phones is only a minor problem here. That is why it is suggested that you use one output buffer for each of the outputs respectively. That solves a host of other problems that will arise if you try to do a one chip output to several sets of headphones at a time.
In the case of a distribution amp, the LM386 will just increase noise and THD where the NJM4556 could do the same job with better results. You will still require one buffer for every output. Why not use a dual supply chip with its advantages? In this case, many.
The very first amp I built was a headbanger and in many ways, it is better than the numerous C-moys that I have since built. No Cmoy amp made can power my 32ohm phones properly.
The headbanger could power two sets in parallel with ease. The THD of the 386 based headbanger amp is well within acceptable limits.
However, the problem of the one amp not having enough power for more than one set of phones is only a minor problem here. That is why it is suggested that you use one output buffer for each of the outputs respectively. That solves a host of other problems that will arise if you try to do a one chip output to several sets of headphones at a time.
In the case of a distribution amp, the LM386 will just increase noise and THD where the NJM4556 could do the same job with better results. You will still require one buffer for every output. Why not use a dual supply chip with its advantages? In this case, many.
The Headbanger is named appropriately. It could probably power all 6 sets of 32 ohm headphones. The LM386 isn't a first-rate performer, but it has its advantages when used as a headphone amp. These days, a headphone amp is a necessity IMO, and the Headbanger is a great way to dip your toe in the diy waters.
Mine is about deck-of-cards size and will play for endless hours. And it'll make my 16-ohm ear buds scream. I love its portability.
The headbanger is a practical DIY project. The LM386 is often used in minimal-parts configuration where there is high-gain and high distortion. But the headbanger designer has adjusted the circuit for minimal distortion and pleasing sound. If it does not sound good, there is something amiss in the circuit or the input source. All grounds should converge to a single point.
...Use an output series resistor?
I did a bit of simulation using the LM386 model you find in the LTspice group. I don't think this is particularly accurate, but then again, real-world '386s seem to vary considerably, too! The model rather seems to represent the junky ones. Even that gives reasonably acceptable distortion into 32 ohms in the Headbanger config (gain = 20 dB). A good version would seem to be performing reasonably well, though still at fairly high noise levels.
I don't think LM386s are terribly attractive for headphone amps any more. Gain-hacked TDA2822Ms or equivalent make better single-supply cheapies, and NJM4556D equipped Cmoys don't exactly break the bank either.
I did a bit of simulation using the LM386 model you find in the LTspice group. I don't think this is particularly accurate, but then again, real-world '386s seem to vary considerably, too! The model rather seems to represent the junky ones. Even that gives reasonably acceptable distortion into 32 ohms in the Headbanger config (gain = 20 dB). A good version would seem to be performing reasonably well, though still at fairly high noise levels.
I don't think LM386s are terribly attractive for headphone amps any more. Gain-hacked TDA2822Ms or equivalent make better single-supply cheapies, and NJM4556D equipped Cmoys don't exactly break the bank either.
Now look at that, I just managed to improve the simulated LM386's distortion performance quite dramatically by cranking up its gain to maximum and using it like an ordinary opamp, with an ~10x non-inverting gain using feedback Rs of 100R and 1k.
At ~5vpp into 32 ohms (~100 mW), 1 kHz harmonic distortion for the standard Headbanger circuit simulated as dominant 3rd (closely followed by 4th), with 3rd @-66 dB.
With the circuit above, it turned into dominant 2nd @-76 dB, with 3rd and 4th at ~-85 dB.
Hmm, AC analysis showed some high-frequency gain peaking. A 100p (ceramic NP0) in parallel to 1k feedback resistor R5 took care of that for now. Distortion is virtually unchanged.
Bandwidth now is really wide, so you may want to include some input RC filtering to keep RF out.
The circuit didn't blow up in simulation, so it may not do in real life either. Sure seems worth trying.
At ~5vpp into 32 ohms (~100 mW), 1 kHz harmonic distortion for the standard Headbanger circuit simulated as dominant 3rd (closely followed by 4th), with 3rd @-66 dB.
With the circuit above, it turned into dominant 2nd @-76 dB, with 3rd and 4th at ~-85 dB.
Hmm, AC analysis showed some high-frequency gain peaking. A 100p (ceramic NP0) in parallel to 1k feedback resistor R5 took care of that for now. Distortion is virtually unchanged.
Bandwidth now is really wide, so you may want to include some input RC filtering to keep RF out.
The circuit didn't blow up in simulation, so it may not do in real life either.
Sure seems worth trying.
As with any "magic tweak", there had to be a downside, and that's stability with capacitive loads. The original Headbanger circuit is utterly unimpressed by capacitive loading. This one will do 22 nF max, which is about 10 times the amount I'd ever expect from a headphone cable.
HOWEVER, a measly 1 ohm resistor in series with the load - which is plenty low for cans, and it's not like you'd want to use the kind of super-critical IEMs on a noisy LM386 output anyway - will tame even the critical 47n..100n region.
Distortion at 5 Vpp into higher-impedance loads is a bit higher numerically but very well-behaved - 2nd @ -65 dB, 3rd @ -75 dB, 4th @ -85 dB and quickly dropping further. (That's about 10 dB better than the Headbanger circuit.) I also tested with 10 kHz, and results are almost the same. Distortion does rise noticeably for 32 ohm loads - 2nd @ -66, 3rd @ -77, 4th @-79 dB. This trend continues at 20 kHz.
HOWEVER, a measly 1 ohm resistor in series with the load - which is plenty low for cans, and it's not like you'd want to use the kind of super-critical IEMs on a noisy LM386 output anyway - will tame even the critical 47n..100n region.
Distortion at 5 Vpp into higher-impedance loads is a bit higher numerically but very well-behaved - 2nd @ -65 dB, 3rd @ -75 dB, 4th @ -85 dB and quickly dropping further. (That's about 10 dB better than the Headbanger circuit.) I also tested with 10 kHz, and results are almost the same. Distortion does rise noticeably for 32 ohm loads - 2nd @ -66, 3rd @ -77, 4th @-79 dB. This trend continues at 20 kHz.
Last edited:
As with any "magic tweak", there had to be a downside, and that's stability with capacitive loads. The original Headbanger circuit is utterly unimpressed by capacitive loading. This one will do 22 nF max, which is about 10 times the amount I'd ever expect from a headphone cable.
HOWEVER, a measly 1 ohm resistor in series with the load - which is plenty low for cans, and it's not like you'd want to use the kind of super-critical IEMs on a noisy LM386 output anyway - will tame even the critical 47n..100n region.
Distortion at 5 Vpp into higher-impedance loads is a bit higher numerically but very well-behaved - 2nd @ -65 dB, 3rd @ -75 dB, 4th @ -85 dB and quickly dropping further. (That's about 10 dB better than the Headbanger circuit.) I also tested with 10 kHz, and results are almost the same. Distortion does rise noticeably for 32 ohm loads - 2nd @ -66, 3rd @ -77, 4th @-79 dB. This trend continues at 20 kHz.
a newbie question, will a small 8 Ohm speaker, instead of high impedence headphones, make the above any different?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.