So I built an amp using a kit from chipamp.com about 6 years ago and it's been working nicely ever since. The main problem was that it was always too loud, I kind of got round the problem by putting a volume pot on the input but it doesn't seem like a great solution. The amp is connected primarily to a raspberry pi and I recently upgraded to a hifiberry DAC, which sounds great but again, way too loud, more so than other devices. I believe it's output is 2v peak to peak, I think that's why it's worse than before.
So, I thought I'd swap R3/Ri for a higher rated resistor to reduce the gain. I started with some 1k trim pots but even wound all the way up, the volume was still too high. I just replaced them with some 1.5k resistors and it's still not much better, so I guess I'm missing something vital. According to everything I can find, Gain =(Ri+Rf)/Ri. Rf is 22k, with the default 680 ohm Ri my gain should have been 33db and with a 1.5k resistor it should be ~15db but it really doesn't seem any quieter.
As it stands, I have to have the volume pot turned pretty much to zero and the output volume from the pi set under 30% or i get nasty noises during loud bits which I assume is the amp clipping.
So, what have I missed? Should reducing the gain not reduce the output volume? Can the lm3886 deal with a 2v peak to peak input? Am I doing something really stupid? (Probably...)
Thanks!
edit: I think (pretty sure) I'm using an 18V+18V 160VA transformer
So, I thought I'd swap R3/Ri for a higher rated resistor to reduce the gain. I started with some 1k trim pots but even wound all the way up, the volume was still too high. I just replaced them with some 1.5k resistors and it's still not much better, so I guess I'm missing something vital. According to everything I can find, Gain =(Ri+Rf)/Ri. Rf is 22k, with the default 680 ohm Ri my gain should have been 33db and with a 1.5k resistor it should be ~15db but it really doesn't seem any quieter.
As it stands, I have to have the volume pot turned pretty much to zero and the output volume from the pi set under 30% or i get nasty noises during loud bits which I assume is the amp clipping.
So, what have I missed? Should reducing the gain not reduce the output volume? Can the lm3886 deal with a 2v peak to peak input? Am I doing something really stupid? (Probably...)
Thanks!
edit: I think (pretty sure) I'm using an 18V+18V 160VA transformer
Last edited:
An input pot should work just fine. What value are you using? I'm building one now and using an 100k pot which is probably over kill.
It's a 50k pot but I'm always working in the bottom 10% of its range, which seems foolish. I'd rather adjust the amp so I can have my output volume fairly high and use the full range rather than limiting the output and having 90% of the range useless.
From a more practical point of view, I can control the output volume from a remote whereas I have to get up to move the pot, which I have to do fairly often since I'm working in a tiny range. I'd rather eliminate the pot and control the volume from the remote, but I can't as the output is too high. Am I right to think that reducing the gain should achieve this? I know the minimum stable gain is 10db, I might try that tomorrow but I get the impression it won't be enough...
edit: i was also under the impression that a 3db drop should half the effective volume, so a 15db drop should be massive, but it doesn't seem to have been. I might try the 680 ohm resistor on one channel and 2k2 on the other to see how they compare...
From a more practical point of view, I can control the output volume from a remote whereas I have to get up to move the pot, which I have to do fairly often since I'm working in a tiny range. I'd rather eliminate the pot and control the volume from the remote, but I can't as the output is too high. Am I right to think that reducing the gain should achieve this? I know the minimum stable gain is 10db, I might try that tomorrow but I get the impression it won't be enough...
edit: i was also under the impression that a 3db drop should half the effective volume, so a 15db drop should be massive, but it doesn't seem to have been. I might try the 680 ohm resistor on one channel and 2k2 on the other to see how they compare...
Last edited:
A 1 dB change is the auditory change threshold, i.e. the amount of change that's just barely noticeable. A 3 dB change is a 2x change in power, but only a slight (but noticeable) change in perceived volume. You need a 10 dB change (10x) power for something to be perceived as twice/half as loud.
I suggest changing the gain of the LM3886 circuit. The LM3886 is stable down to a gain of 10 V/V. See attached (ripped from the data sheet).
~Tom
I suggest changing the gain of the LM3886 circuit. The LM3886 is stable down to a gain of 10 V/V. See attached (ripped from the data sheet).
~Tom
Attachments
I'll swap R3 for 2k2 resistors this afternoon and see if it helps, if dropping the gain from 33->15db hasn't achieved much, I don't see that 15->10db is going to work miracles but I'll give it a go.
The rest of the parts are right, the amp's been working very well for 6 years and works fine with other sources, my phone for example, it's just the new soundcard which is tricky.
Would adjusting the value of R1/Rin to increase my input impedance be any use? I don't want to just go swapping random components but I'm struggling to find what exactly I should be doing :/
The rest of the parts are right, the amp's been working very well for 6 years and works fine with other sources, my phone for example, it's just the new soundcard which is tricky.
Would adjusting the value of R1/Rin to increase my input impedance be any use? I don't want to just go swapping random components but I'm struggling to find what exactly I should be doing :/
Not sure how you are getting your dB numbers:
22K and 680R give a voltage gain of 33.35 or 30.46dB
22K and 1K5 give a voltage gain of 15.67 or 23.9dB
22K and 2K2 give a voltage gain of 11 or 20.83dB
22000
680
33.35
30.46
22000
1500
15.67
23.90
22000
2200
11
20.83
22K and 680R give a voltage gain of 33.35 or 30.46dB
22K and 1K5 give a voltage gain of 15.67 or 23.9dB
22K and 2K2 give a voltage gain of 11 or 20.83dB
22000
680
33.35
30.46
22000
1500
15.67
23.90
22000
2200
11
20.83
I thought the gain was in dB, if that gives you an idea of my level of knowledge...
My point stands though, my problem exists at a gain of 33 or 23, i 'm pretty sure it will happen at a gain of 20.
Is 2v peak to peak really unusual? Should the lm3886 be able to handle it?
My point stands though, my problem exists at a gain of 33 or 23, i 'm pretty sure it will happen at a gain of 20.
Is 2v peak to peak really unusual? Should the lm3886 be able to handle it?
I would leave the gain of the amp at design levels. (22K & 680R)
There are thousands of these amp around and I would suspect they are built that way.
I would use a pot of the lowest value that your source can drive.
Lower resistance = lower noise (all else being equal)
(There are thousands of amps with pots that seem to be OK)
If you really don't want a pot then investigate an active volume control like the LM1036:
Linear - Audio Processing | Integrated Circuits (ICs) | DigiKey
There are better ones than this but I cannot find them right now.
If there is a gross problem with the sound it is not with the gain of the amp.
🙂
There are thousands of these amp around and I would suspect they are built that way.
I would use a pot of the lowest value that your source can drive.
Lower resistance = lower noise (all else being equal)
(There are thousands of amps with pots that seem to be OK)
If you really don't want a pot then investigate an active volume control like the LM1036:
Linear - Audio Processing | Integrated Circuits (ICs) | DigiKey
There are better ones than this but I cannot find them right now.
If there is a gross problem with the sound it is not with the gain of the amp.
🙂
.
You might be interested in posts 11 and 12 here. The subject is a do-all volume control. http://www.diyaudio.com/forums/anal...ions-simple-opamp-based-low-gain-stage-2.html
If you are interested, be sure to also look at post 17 on the same page. The schematics are corrected by the addition of 100 ohm resistors.
.
You might be interested in posts 11 and 12 here. The subject is a do-all volume control. http://www.diyaudio.com/forums/anal...ions-simple-opamp-based-low-gain-stage-2.html
If you are interested, be sure to also look at post 17 on the same page. The schematics are corrected by the addition of 100 ohm resistors.
.
are the vol pots you are trying linear or log taper? A log taper pot usually has a value of around 10% of the pots rated value when the control is in the 12 o'clock position and 1% or less around the 8-9 o'clock position. A linear pot will be 50% at the 12 o'clock position and about 25% at the 9 o'clock position.
You can always add a pre-attenuator, it might need buffering though. It does seem counter intuitive to attenuate a signal and then re-amplify it though.
The datasheet says gain >= 10times which equals >=+20dB gain.
The 3886 is reputed to perform better at gains between +26dB and +30dB
Look at the stability margins in that range of gain.
1. Don't change amp's gain, LM are not stable at too low gain.
2. All volume pots - 10K, 50K, 100K regulate volume the same, no difference. Difference is between log and linear pots. Be sure to use log for volume.
2. If you have problems with too much gain, use attenuator before pot. Simplest and best way - put resistor with value 2x -4x of pot resistance in series before pot. Example - 20K-40K before 10K pot, 100K-200K before 50K pot e.t.c.
This resistor with pot make voltage divider. Choice right resistor value experimental, depending on your system. Normal listening volume must be between 12 - 2 o'clock.
Best way is using 5K or 10K pot with resistor. Higher value can increase noise, theoretically.
2. All volume pots - 10K, 50K, 100K regulate volume the same, no difference. Difference is between log and linear pots. Be sure to use log for volume.
2. If you have problems with too much gain, use attenuator before pot. Simplest and best way - put resistor with value 2x -4x of pot resistance in series before pot. Example - 20K-40K before 10K pot, 100K-200K before 50K pot e.t.c.
This resistor with pot make voltage divider. Choice right resistor value experimental, depending on your system. Normal listening volume must be between 12 - 2 o'clock.
Best way is using 5K or 10K pot with resistor. Higher value can increase noise, theoretically.
2 V RMS (2.8 V peak) is normal line level these days. The LM3886 is perfectly able to handle that. At a gain of +11, it'll turn that 2.8 V peak into 30.8 V peak just fine if you provide a high enough supply voltage. The problem here is that 30.8 V peak is nearly 60 W into 8 ohm. That's very loud (120 dB SPL at 1 m assuming 85 dB/W*m efficiency). Deafeningly loud, in fact.
To change this, you need to use lower gain on the LM3886. The LM3886 is stable down to a gain of 10 V/V (or 11 V/V in non-inverting configuration). This means, Rf1 = 10*Ri in attached figure.
If 10x gain is still too much, a volume control or a voltage divider is your best option. You can also use an input attenuator as I've drawn in attached figure. Set Rx to 1 kOhm for approx 20 dB of attenuation, 330 ohm for 30 dB of attenuation, or 100 ohm for 40 dB of attenuation.
~Tom
To change this, you need to use lower gain on the LM3886. The LM3886 is stable down to a gain of 10 V/V (or 11 V/V in non-inverting configuration). This means, Rf1 = 10*Ri in attached figure.
If 10x gain is still too much, a volume control or a voltage divider is your best option. You can also use an input attenuator as I've drawn in attached figure. Set Rx to 1 kOhm for approx 20 dB of attenuation, 330 ohm for 30 dB of attenuation, or 100 ohm for 40 dB of attenuation.
~Tom
Attachments
Last edited:
.
A lot of people are telling you right, but they're all telling you in different ways. Now here comes me taking my shot.
Don't fix the wrong problem. Your LM3886 amp is fine, don't change it. Why would you want to anyway, next week you might want to connect it to something else. The problem is not with the amp, the real problem is an input signal that overloads the amp.
Apparently this "hifiberry DAC" (whatever that is) has no volume control? Well then, there's really only one reasonable solution, which is also the obvious solution, which is also pretty much "the way you do it," it's been done a million times.
You add a resistor to the input (not the ground) side of your volume control. The input will be one of the side terminals of the volume potentiometer, not the center terminal. This "dropping resistor" burns off some of the input signal, allowing your volume control to function normally.
You might say, "I don't want to waste all that signal," but you're already wasting it because it's being burned off in the same way, except by the unused portion of your volume control pot.
Adding the dropping resistor in effect gives the volume control more rotation, moving the center point. For a bonus, the volume control won't be as irritatingly sensitive as it is now.
Circuit below.
.
A lot of people are telling you right, but they're all telling you in different ways. Now here comes me taking my shot.
Don't fix the wrong problem. Your LM3886 amp is fine, don't change it. Why would you want to anyway, next week you might want to connect it to something else. The problem is not with the amp, the real problem is an input signal that overloads the amp.
Apparently this "hifiberry DAC" (whatever that is) has no volume control? Well then, there's really only one reasonable solution, which is also the obvious solution, which is also pretty much "the way you do it," it's been done a million times.
You add a resistor to the input (not the ground) side of your volume control. The input will be one of the side terminals of the volume potentiometer, not the center terminal. This "dropping resistor" burns off some of the input signal, allowing your volume control to function normally.
You might say, "I don't want to waste all that signal," but you're already wasting it because it's being burned off in the same way, except by the unused portion of your volume control pot.
Adding the dropping resistor in effect gives the volume control more rotation, moving the center point. For a bonus, the volume control won't be as irritatingly sensitive as it is now.
Circuit below.
.
Attachments
Last edited:
- Status
- Not open for further replies.