Gainclone LM3886 volume is abnormally weak

[kubark42]

Hi, guys, I'm trying to figure out what I've done wrong with my gainclone. Basically, at 50% volume, almost nothing's coming out. It sounds about as loud as those little headphone-jack speakers they sold back in the 90s. At full, I get lots of distortion, but nothing that would even really hurt my ears. Note, I'm not looking to hurt my ears, but an amp on full should at least make you feel uncomfortable! Again, I like to listen to music <b>quietly</b>, so I'm not some volume junky complaining that only my upstairs neighbors can hear my stereo, not my neighbors across the street. It's so weak that I'm convinced that there's a problem somewhere and I should solve it before I make the 2nd channel and commit on final values for the stepped attenuators.

I built it based on the schematic here: http://www.circuit-projects.com/audi...r-project.html

I've used the headphone output on a Dell, along with its line-out. In both cases, the volume was weak.

The input is a 25kΩ stepped attenuator. Perhaps 25kΩ is too low? Could this be the problem?

I'm powering an 8Ω speaker I had lying around. It's nothing phenomenal, but I know that it can at least get louder than "miserable".

250-->25V transformer, thus output at 35V. It's quality stuff bought from Farnell. I don't think it's the PS because the volume gets louder when I go from 0 to 100%, it's just that it doesn't get at all loud! If the PS were having problems, then I would expect that after a moment it wouldn't get louder at all.

10,000uF caps. Likewise, good stuff bought from Farnell. In any case, the problem isn't the bass, so the caps shouldn't be an issue.

Lastly, I do not have a heat sink on the chip right now. I'm just turning it on for a couple seconds. In any case, a heat sink would wick heat away too slowly for this to be the problem. I would expect thermal effects to be on the seconds level, not milliseconds. However, the chip *does* get very hot even if there is nothing being amplified. So hot that I unplug it after a few seconds anyway out of fear of melting down the chip. This seems a little abnormal to me, so I wonder if this rapid heating could signify an improperly wired chip or component value.

It reminds me of "This is Spinal Tap", only I'd be happy just to get all the way to 1.1! So, any ideas as to what this could be? Are there any components that I could have poorly chosen that would lead to this weak a volume level?

[pacificblue]

The typical quiescent current is 50 mA. With 35 V rails that gives you 1,75 W power dissipation. At 43 K/W the IC will heat up to ~75° above ambient, so it will probably be around 90-100 °C even with no music signal. Even a weak music signal will easily heat the IC above the thermal protection threshold. Attach a heatsink, before performing any further tests.

Other than that you can only check that everything is connected correctly and all resistor values are correct. First suspect is the attenuator connection, second suspect is the muting resistor value and third is the complex feedback loop, which invites for errors. You could try to replace that with the recommended design just to check, if the fault is in the feedback loop, i. e. disconnect R2 and R3 and jumper C3.

[kubark42]

Originally posted by pacificblue
The typical quiescent current is 50 mA. With 35 V rails that gives you 1,75 W power dissipation. At 43 K/W the IC will heat up to ~75° above ambient, so it will probably be around 90-100 °C even with no music signal. Even a weak music signal will easily heat the IC above the thermal protection threshold. Attach a heatsink, before performing any further tests.

It takes time for the components to heat up, and time for them to cool down. There's no way that a heat sink could have such an important role in the first few seconds. Moreover, that would show up as cutting out. Not as a full range of 0-100%, where everything from 0 to 100% is abnormally weak.

That being said, I finally dug up a heat sink in my lab and hooked it up. As expected, no dice. However, it's nice to finally be able to run the amp for a few more than 10 seconds at a time!

Other than that you can only check that everything is connected correctly and all resistor values are correct. First suspect is the attenuator connection, second suspect is the muting resistor value and third is the complex feedback loop, which invites for errors. You could try to replace that with the recommended design just to check, if the fault is in the feedback loop, i. e. disconnect R2 and R3 and jumper C3.

As near as I can tell, everything is correct, as shown on the schematic. However, I had no idea that that was a complex feedback circuit. Heck, until reading about this problem I had no idea that it was even a feedback circuit. I'll check out the reference design values and see how they work out.

Incidentally, C3 should be a capacitor, and not a jumper. If I disconnect all of those, there will be no more feedback at all! I'm guessing you mean to remove it, rather than disconnect it.

I'll run to the lab right away and check out whether the suggested modifications gives a better output.

[kubark42]

No joy. I switched to this schematic (minus the 22k resistor at pin 10) http://diyaudioprojects.com/Chip/LM3.../LM3886_CA.htm and it made no audible difference whatsoever.

However, I do notice that there is a slight difference of several hundred mV between the +35V and -35V rails. This is not present when I unplug the amp from the power-supply, so the amp must be drawing more current from the - than the +.

Moreover, I've little experience with amps, but it doesn't seem to me that the amp should heat up so much that the heat sink is very hot to the touch within a couple seconds, even when there is no input and the volume is turned down completely. Is this normal?

Lastly, all this is built on a bread-board, with uncut resistors, so is it possible that the problem is just that wires are too long?

[pacificblue]

Originally posted by kubark42
Incidentally, C3 should be a capacitor, and not a jumper. If I disconnect all of those, there will be no more feedback at all! I'm guessing you mean to remove it, rather than disconnect it.

If you put a jumper across C3 or in the place of C3 and disconnect R2 and R3, R1 should become your feedback resistor.

Originally posted by kubark42
However, I do notice that there is a slight difference of several hundred mV between the +35V and -35V rails. This is not present when I unplug the amp from the power-supply, so the amp must be drawing more current from the - than the +.

Moreover, I've little experience with amps, but it doesn't seem to me that the amp should heat up so much that the heat sink is very hot to the touch within a couple seconds, even when there is no input and the volume is turned down completely. Is this normal?

Did you measure the output offset voltage?
Can you measure the current coming out of the mute pin?
Did you retouch all solder joints to make sure they are all conducting?
Does it work better, when you skip the attenuator?
Can you post photos?

Originally posted by kubark42
Lastly, all this is built on a bread-board, with uncut resistors, so is it possible that the problem is just that wires are too long?

That should not have such an impact.

[kubark42]

Originally posted by pacificblue
[B]
If you put a jumper across C3 or in the place of C3 and disconnect R2 and R3, R1 should become your feedback resistor.

Of course, of course. I guess I was just having trouble that day making sense of, well, my native tongue.

Did you measure the output offset voltage?

The DC offset is 30-50mV, depending on the volume level. Is this normal (enough)?

Can you measure the current coming out of the mute pin?

I could measure the current, but it'd be pretty difficult because of space. However, since the voltage on the pin is -35V, the 15k resistor should pass a little over 2mA, which according to the chipdoc is more than enough to guarantee full on.

Does it work better, when you skip the attenuator?

Hmmm...

Did you retouch all solder joints to make sure they are all conducting?
Can you post photos?

The joints are soldered the entire length of the pin. However, it's not beyond reason that there be a bad connection in the breadboard. I'll give it a jiggle tomorrow or the next.

If the problem doesn't clear up, I'll take the camera to work and take some pictures.

Thanks a lot for the advice, I've read a lot about chipamps in the past months, but there's just so much more to know.

[pacificblue]

Originally posted by kubark42
The DC offset is 30-50mV, depending on the volume level. Is this normal (enough)?

As opposed to transistor amplifiers that level is accepted for chipamps. Although it seems very high, given that you have a DC blocking cap in the feedback leg. Do you have a DC blocking cap at the input as well? If not, you should definitely try one.

Originally posted by kubark42
I could measure the current, but it'd be pretty difficult because of space. However, since the voltage on the pin is -35V, the 15k resistor should pass a little over 2mA,

..if the solder joints are good, if the resistor value is correct, etc. You could measure the voltage across that resistor and calculate the corresponding current. That should be easier to do, if there is little space.[/B][/QUOTE]

[Stuey]

The headphone out would give you 'weak' sound. Are you using the line-out for all your tests? What about the sound settings on the Dell?

Maybe try another source, like a CD player or tuner, to eliminate this from the equation?

Just a suggestion.

Stuey