Increasing DC offset

[Tripmaster]

Hi

I have just cobbled together a P2P chipamp using the following schematic. Its working fine and sounds quite nice but I do have a small issue.



When I turn off the amp it continues to play for 5-10 seconds as the reservoir caps discharge (quite normal), the sound stops and then comes back for a second or two. I realise I can turn the source off first, but with no music playing I get a pop through the speakers. I monitored this by connecting a DMM to the speaker terminals and the the DC offset jumps from about 1mV to 200mV when I hear the pop.

The following paragraph is from Decibel Dungeon

'Why do I get strange sounds after I switch off the power?

This is either caused by poor quality decoupling capacitors or by oscillation. If you have used cheapo decoupling caps, try replacing them with something better. If the problem persists then ask for help on one of the forums.'


I'm using Panansonic FCs for decoupling and I have bypassed these using 0.1uf WIMAs

Any ideas?

[abraxalito]

What value have you ended up with for the DC offset adjusting resistor (to pin7)? I just Spiced this up using my homegrown LM3886 model (1mV offset voltage) and I found I needed a 15k. You might try bypassing this with one of your 0.1uF WIMAs to see if it helps.

[janneman]

Originally Posted by Tripmaster

Hi

I have just cobbled together a P2P chipamp using the following schematic. Its working fine and sounds quite nice but I do have a small issue.



When I turn off the amp it continues to play for 5-10 seconds as the reservoir caps discharge (quite normal), the sound stops and then comes back for a second or two. I realise I can turn the source off first, but with no music playing I get a pop through the speakers. I monitored this by connecting a DMM to the speaker terminals and the the DC offset jumps from about 1mV to 200mV when I hear the pop.

The following paragraph is from Decibel Dungeon

'Why do I get strange sounds after I switch off the power?

This is either caused by poor quality decoupling capacitors or by oscillation. If you have used cheapo decoupling caps, try replacing them with something better. If the problem persists then ask for help on one of the forums.'


I'm using Panansonic FCs for decoupling and I have bypassed these using 0.1uf WIMAs

Any ideas?

This has nothing to do with the quality of the decoupling caps. Assuming that by decoupling caps you mean the reservoir caps in the power supply? Or power supply caps at the chip? Same thing.

Lets assume you have no oscillations, then the pop etc is caused by the gradual discharge of the power supply caps. At a certain point the supply voltage gets so low that the amp can't work normally anymore and the internal balance is upset, and the offset gets out of control. Then the supply voltage decreases further and the amp stops working.

One thing that can help to minimize the pop is to make sure the supply voltage decreases faster, by placing a bleeder resistor across the supply caps. Try something between 500 ohms and 1k, but these do dissipate a few watts during operation so must be say 2W types. You can try it and install if it helps.

Another thing worth a try is to lower the mute pin current to the minimum (assuming it is with a resistor to the supply) so that when the supply starts to drop the amp goes into mute quickly.

jd

[ratza]

For low voltage offset, both inverting and non inverting inputs should be polarized in the same way, i.e. having the same resistor values.

[abraxalito]

Originally Posted by janneman

This has nothing to do with the quality of the decoupling caps. Assuming that by decoupling caps you mean the reservoir caps in the power supply? Or power supply caps at the chip? Same thing.

Yep, agreed. Whoever said this was just blowing smoke.

Lets assume you have no oscillations, then the pop etc is caused by the gradual discharge of the power supply caps. At a certain point the supply voltage gets so low that the amp can't work normally anymore and the internal balance is upset, and the offset gets out of control. Then the supply voltage decreases further and the amp stops working.

On further reflection to my previous post, I think that the relatively large noise gain at DC for this particular circuit might be the culprit. When its operating normally, the DC offset is only low because the two input bias currents (typically 0.2uA) give equal voltages at the two inputs. So if my first suggestion produces no result, my next suggestion is to place a 470uF in series with the 100R MF resistor. This ensures the noise gain at DC is unity.

One thing that can help to minimize the pop is to make sure the supply voltage decreases faster, by placing a bleeder resistor across the supply caps. Try something between 500 ohms and 1k, but these do dissipate a few watts during operation so must be say 2W types. You can try it and install if it helps.

I thought of that too but its not a particularly green solution

Another thing worth a try is to lower the mute pin current to the minimum (assuming it is with a resistor to the supply) so that when the supply starts to drop the amp goes into mute quickly.

Would be a good solution if the LM3875 had a mute pin Were you thinking of the LM3886 by any chance? I have been studying that datasheet in order to try to get a handle on how the LM3875 power down muting might work...

[Tripmaster]

Originally Posted by abraxalito

What value have you ended up with for the DC offset adjusting resistor (to pin7)? I just Spiced this up using my homegrown LM3886 model (1mV offset voltage) and I found I needed a 15k. You might try bypassing this with one of your 0.1uF WIMAs to see if it helps.

Hi

Ive adjusted the the temporary trim pot to 7K and there is no offset when I short the the speaker terminals with a 10R resistor. With the lightspeed attenuator connected its between 1-3mv. I haven't removed the trim pot as I'm still testing...

Thanks

[Tripmaster]

Originally Posted by janneman

This has nothing to do with the quality of the decoupling caps. Assuming that by decoupling caps you mean the reservoir caps in the power supply? Or power supply caps at the chip? Same thing.


One thing that can help to minimize the pop is to make sure the supply voltage decreases faster, by placing a bleeder resistor across the supply caps. Try something between 500 ohms and 1k, but these do dissipate a few watts during operation so must be say 2W types. You can try it and install if it helps.

jd

This has nothing to do with the quality of the decoupling caps. Assuming that by decoupling caps you mean the reservoir caps in the power supply? Or power supply caps at the chip? Same thing.

2x 10000uf at each regulated supply and 2x 1000uf at the chip pins


One thing that can help to minimize the pop is to make sure the supply voltage decreases faster, by placing a bleeder resistor across the supply caps. Try something between 500 ohms and 1k, but these do dissipate a few watts during operation so must be say 2W types. You can try it and install if it helps.

I currently have 2W 2K2 bleeder just after the 10000uf caps

[Tripmaster]

Here are a few photos...the wires will be re-routed

[janneman]

Originally Posted by abraxalito

[snip]On further reflection to my previous post, I think that the relatively large noise gain at DC for this particular circuit might be the culprit. When its operating normally, the DC offset is only low because the two input bias currents (typically 0.2uA) give equal voltages at the two inputs. So if my first suggestion produces no result, my next suggestion is to place a 470uF in series with the 100R MF resistor. This ensures the noise gain at DC is unity.[snip]

Yes, good point. That might help for sure.

Originally Posted by abraxalito

[snip]Would be a good solution if the LM3875 had a mute pin Were you thinking of the LM3886 by any chance? I have been studying that datasheet in order to try to get a handle on how the LM3875 power down muting might work...

I was indeed thinking about the 3886... My bad.

jd

[Tripmaster]

Originally Posted by janneman

One thing that can help to minimize the pop is to make sure the supply voltage decreases faster, by placing a bleeder resistor across the supply caps. Try something between 500 ohms and 1k, but these do dissipate a few watts during operation so must be say 2W types. You can try it and install if it helps.

Do you think 2K2 is too much for a bleeder? I just copied the value from my old chipamp.com unregulated snubbed supply. I have some 470R and I think 1K 2W resistors in stock.