Hi, I have limited experience in DIY amps and have some extra LM3886 chips left over from another project. I want to make stereo a headphone amp/monitor for a bass guitar player to use instead of his big rig. Could I set the gain to about 20 and be done or would other changes be needed for a typical 8 to 32 ohm headphone range? I have a left over +15/-15 regulated supply that could be used to power it. Thanks!
It's going to sound a bit loud, consequently not particularly low noise. The solution to those issues would be to fit an output transformer - typically for low impedance headphones you'll only need a buffer or at most up to 6dB of gain. Thus a 10:1 step down transformer would be a workable solution.
It will cost more for the transformer than buying a headphone driver chip. The lm386 has a minimum preset gain of 20 (+26dB) and may be a bit too loud. But worth trying.
Last edited:
Almost certainly too loud and can't run on +/-15V - would run on just +15V quite happily though if its the N-4 variant. Why buy an LM386 though when it provides no advantage (other than quiescent current) over LM3886 which the OP already has? He can determine if loudness (and hiss) are a problem with the existing LM3886 amp chip before shelling out any money.
surely any reasonably designed low power amplifier will have less noise than a 3886.
There must be some low cost headphone driver chips of good audio quality out there that have lower gains. Nice if they were +0dB to +12dB. That's probably all we need for the vast majority of headphones.
One can set the 3886 to lower gain than the minimum of +20dB , but the compensation to suit has to be determined.
The noise gain is high but that noise can be pushed to well above the audio band.
I did that with the 3886 set to 10times for active computer monitors. I think Tomchr has done something similar but he keeps his secrets.
Even close up I can't hear any extra noise compared to normal at 2.5m listening distance. The 3886 is not as quiet as discrete, but is is good enough, even in active implementation.
There must be some low cost headphone driver chips of good audio quality out there that have lower gains. Nice if they were +0dB to +12dB. That's probably all we need for the vast majority of headphones.
One can set the 3886 to lower gain than the minimum of +20dB , but the compensation to suit has to be determined.
The noise gain is high but that noise can be pushed to well above the audio band.
I did that with the 3886 set to 10times for active computer monitors. I think Tomchr has done something similar but he keeps his secrets.
Even close up I can't hear any extra noise compared to normal at 2.5m listening distance. The 3886 is not as quiet as discrete, but is is good enough, even in active implementation.
Use the LM3886 as a roughly 15W amp with that supply, but add 100 ohms 1W resistors in series with headphone jack, what every standard headphone out does.
I'd also walk an extra step and make a little combo, think a 3" or 4" woofer in a book (not bookshelf) sized box, since you have to pack amp and supply somewhere anyway.
Bass through headphones is horrible and will damage his hearing , a small woofer will let him hear better at 3 AM bedroom levels .
Minimum gain is around 30X (22k/680r NFB, check the datasheet) , not stable under that.
Also add a speaker jack, if/when there's a speaker, *any* speaker available, he can use it.
I'd also walk an extra step and make a little combo, think a 3" or 4" woofer in a book (not bookshelf) sized box, since you have to pack amp and supply somewhere anyway.
Bass through headphones is horrible and will damage his hearing , a small woofer will let him hear better at 3 AM bedroom levels .
Minimum gain is around 30X (22k/680r NFB, check the datasheet) , not stable under that.
Also add a speaker jack, if/when there's a speaker, *any* speaker available, he can use it.
The 3886 i stable down to +20dB, 10times. National tell us that.
But when you look at the gain plot you will see the phase margin is at 65° and that will give significant overshoot and ringing.
It will not oscillate into resistive loading, it is stable !
For good stability margins, to suit good audio sound quality, one needs to set gain to around +29dB
But when you look at the gain plot you will see the phase margin is at 65° and that will give significant overshoot and ringing.
It will not oscillate into resistive loading, it is stable !
For good stability margins, to suit good audio sound quality, one needs to set gain to around +29dB
Not quite correct, in its plain vanilla form you can set it up for much lower gain without compensation.
For 8dB of gain I would suggest Rf/Ccomp 2.5K//220pF, Rg 1.0K.
You can download the reference design from TI and run it in the freeware version of TINA.
is Rf/Ccomp a series pair across the -IN to +IN chipamp pins?
+8dB = 2.51 times, that would indicate a 1k5:1k0 NFB pair.
Is that what you meant by Rg=1.0K, or 1k0
or
did you mean 1k0:680r for the NFB pair?
Is there a way to predict the stability margins for that +8dB gain?
+8dB = 2.51 times, that would indicate a 1k5:1k0 NFB pair.
Is that what you meant by Rg=1.0K, or 1k0
or
did you mean 1k0:680r for the NFB pair?
Is there a way to predict the stability margins for that +8dB gain?
Last edited:
The LM3886 is not the best candidate for a headphone amp. First off, you don't need 40+ W unless you plan to catch your headphones on fire. Secondly, while the LM3886 is pretty low noise for a power amp, you'll likely find it too noisy for a headphone amp. Basically, you'll end up listening through a blanket of hiss.
Buffer chips such as the BUF634 and/or LME49600 would be better suited for a headphone amp. Many op-amps can drive headphones directly as well.
Tom
Buffer chips such as the BUF634 and/or LME49600 would be better suited for a headphone amp. Many op-amps can drive headphones directly as well.
Tom
Someone mentioned the LM386. This is a popular choice for a quick and dirty headphone amp for electric guitar and bass. It is not hi-fi and it is not low noise, but it runs just fine off a 9 volt battery and works with a minimum of components. It will provide far more power than you would ever need.
Asking if the 3886 would make a good headphone amp is like asking if you should use a sledgehammer to drive a tack. Forget about it.
Asking if the 3886 would make a good headphone amp is like asking if you should use a sledgehammer to drive a tack. Forget about it.
100 Ohm series output is not standard with today's 32 Ohm headphones - especially not the multidriver IEM
with 130 dB SPL re 1 Vrms sensitivities these simply aren't suited to LM3886, if you have a good enough quality audio output step down transformer most of these wouldn't need any amp with most line outputs once the Z transformation is considered
LM3886 could be somewhat reasonably used with low sensitivity Orthodynamic Planar Magnetic like Highfiman or Audeze where up to 6 W could be desired for "audiophile" dynamic headroom into 50 Ohms
with 130 dB SPL re 1 Vrms sensitivities these simply aren't suited to LM3886, if you have a good enough quality audio output step down transformer most of these wouldn't need any amp with most line outputs once the Z transformation is considered
LM3886 could be somewhat reasonably used with low sensitivity Orthodynamic Planar Magnetic like Highfiman or Audeze where up to 6 W could be desired for "audiophile" dynamic headroom into 50 Ohms
Last edited:
No problem using LM3886 for headphones at all.
I tried it (using my chipamp with gain = 30 with a Neutrik 6.35mm socket).
It is quiet as any headamp on my AK701.
Just did not sound particularly nice, for which I have no explanation.
Also tried LM1876.
Patrick
I tried it (using my chipamp with gain = 30 with a Neutrik 6.35mm socket).
It is quiet as any headamp on my AK701.
Just did not sound particularly nice, for which I have no explanation.
Also tried LM1876.
Patrick
That should be the +N part though. These chips would appear to be running a decent amount of bias current and with headphones should be running in Class A a lot of the time. (20 mA is good for 1.7 Vrms into 62 ohms, which is about 105 dB SPL with K701s.) High THD at low levels would mainly appear to be a problem of woefully underbiased output stages, as those have widely varying gm.
Actually there even are some commercial headphone amps employing the LM1876, notably the Lake People G103 series and their prodecessor, the G93. They're run at a gain of 8 dB so presumably with some loop gain drained off. Inverting operation may be worth a shot, and some more attention to output DC offset probably won't hurt.
Actually there even are some commercial headphone amps employing the LM1876, notably the Lake People G103 series and their prodecessor, the G93. They're run at a gain of 8 dB so presumably with some loop gain drained off. Inverting operation may be worth a shot, and some more attention to output DC offset probably won't hurt.
Well it should be, but if you look at just about any THD+N vs power output graph, the distortion spikes below 1 watt. These chips are designed to drive speakers and are optimized for that. People think of them as a high powered op amp, but would you use one as a line level input buffer or a gyrator in an active filter?
As you pointed out, the output stages have widely varying gm. They've very nonlinear at low bias and as they approach cutoff. This is nothing unusual. The feedback mechanism has to work overtime and distortion rises rapidly. I think it was Doug Self who wrote a paper on sources of distortion and he discusses this, classifying it as a "complex distortion."
It's not hard to build a decent dedicated headphone amplifier. With devices available today it's downright easy to do with just a handful of parts.
As you pointed out, the output stages have widely varying gm. They've very nonlinear at low bias and as they approach cutoff. This is nothing unusual. The feedback mechanism has to work overtime and distortion rises rapidly. I think it was Doug Self who wrote a paper on sources of distortion and he discusses this, classifying it as a "complex distortion."
It's not hard to build a decent dedicated headphone amplifier. With devices available today it's downright easy to do with just a handful of parts.
if you need more than 2 watts with a modern headphone, then it means you do not need to listen to music, you want to ventillate your brain out via your nose.
a quite decent headphone amp i had used paralelled opamps with a verr small gain.
that is supposed to be more than enough today.
series resistor on the ouput.. well, some headphones actually sound a lot better with it, some with out it.
the reason can be found in the impedance curve 😀
nothing else.
a quite decent headphone amp i had used paralelled opamps with a verr small gain.
that is supposed to be more than enough today.
series resistor on the ouput.. well, some headphones actually sound a lot better with it, some with out it.
the reason can be found in the impedance curve 😀
nothing else.
Thanks to all for their input! I have decided to abandon the LM3886 as a headphone amp and make two, using differing chips. One using the LME49600 and another using LM1875 to compare and contrast. Since It will be used as a monitor for a bass player, in a live musical venue, my priority will be to enhance low end response. No big rigs allowed on stage. Any ideas on "simple" tweaks to accomplish this, or maybe none will be needed? Most bass players like to literally "feel" the notes coming out of their BIG rigs.
- Status
- Not open for further replies.