LM3886 without electrolytic feedback capacitor and (hardly) no DC offset voltage - Page 2

CharlieLaub · 12th September 2012, 08:23 PM

I happen to own the very nicely written book "Designing Audio Amplifiers" by Bob Cordell. Your suggested LM3886 circuit is exactly the same as Figure 8.3b on page 158. So, this is not something new.

The chapter in which the figure is located discusses DC offset, and various techniques (including the usual electrolytic cap in an RC filter to ground from the negative feedback path). His comments on this topology are:
"This arrangement reduces some of the input offset concerns without resorting to an electrolytic capacitor. However, the DC gain of the amplifiers is now equal to the closed-loop gain, so a big part of the advantage of having an electrolytic in the conventional arrangement is lost (due to input voltage offset). This approach can be used in combination with a DC servo to reduce the amount of correction that must be supplied by the servo."

So, while it seems you have removed an electrolytic cap, which is probably a good thing, it does away with the function of that cap as well. What's the overall advantage??? Not sure.

Say, while we are on the subject (hope that this is not thread hikacking), why don't people use a capacitance multiplier in place of the electro? The cap has a grounded terminal, so I believe that type of circuit can be used...

For example, substitute the terminal marked +C in the left most circuit in the figure in place of the usual grounded electrolytic, use a 1uF film for C1 and then select the resistor ratio to give you 100uF or 1000uF or whatever you need.

Thoughts?

-Charlie

Marc Vi. · 13th September 2012, 07:19 AM

Hello Charlie,

I didn't say that my method is new.

What is new in my approach is that I adjust the value of the resistor to the -input so that the output is almost exactly 0,000 V. For one LM3886 amplifier this results in a value of 27 kohm and for the other of 18 kohm. As I posed in a thread before, this value is not influencing the amplification. I then heated the amplifiers by full powering them. This results in a output offset voltage of only a few milliVolts, but when you put down the power, the output offset voltage is returning to its original point

So I dont agree with Bob Cordell. In my approach you can compensate the input offset current and voltage by adjusting just one resistor, so that the output offset voltage is nearly zero volt (In my case only 1 - 2 milliVolt). I found out in practice with the LM3886 amplifier that this output offset voltage doesn't run away (In my case max. 5 milliVolt). I was amazed by this result myself.

So, despite of the fact that de DC gain is equal to the closed loop gain, you not "reduces some of the input offset" but you can almost fully cancel it out by adjusting one resistor!! And it is not running away.

The overall advantage is a quite stable output offset voltage of (nearly) zero while not needing an electrolytic capacitor.

Marc.

zakman35 · 13th September 2012, 07:55 AM

How this translates in practice.
Is there any audible effect of it comparing with the usual ci cap cofiguration or they both sound the same?

ddd · 13th September 2012, 11:21 AM

Quote:

Originally Posted by CharlieLaub

...why don't people use a capacitance multiplier in place of the electro? The cap has a grounded terminal, so I believe that type of circuit can be used...
Thoughts?

-Charlie

Dear Charlie,
I think it's mistake to try this.
The point +C has the offset of the new OpAmp, which could be significant (with 1MOhm resistors). And you connect it to very sensitive point, multiplying its DCV by 20. Instead of benefit you'll get harm. Not mentioning several new elements in comparison with simple resistor.
These circuits are gyrators (Wiki).

CharlieLaub · 13th September 2012, 06:01 PM

Quote:

Originally Posted by ddd

Dear Charlie,
I think it's mistake to try this.
The point +C has the offset of the new OpAmp, which could be significant (with 1MOhm resistors). And you connect it to very sensitive point, multiplying its DCV by 20. Instead of benefit you'll get harm. Not mentioning several new elements in comparison with simple resistor.
These circuits are gyrators (Wiki).

I am gussing that zeroing out any DC offset is a simple thing to do in the circuit, and as long as the capacitance scaling resistor is small compared to 1M the input impedances will be relatively well balanced anyway.

-Charlie

Marc Vi. · 13th September 2012, 08:11 PM

Hallo Charlie,
Can you also answer my last post?
Marc Vi.

CharlieLaub · 13th September 2012, 08:40 PM

Quote:

Originally Posted by Marc Vi.

Hello Charlie,

I didn't say that my method is new.

What is new in my approach is that I adjust the value of the resistor to the -input so that the output is almost exactly 0,000 V. For one LM3886 amplifier this results in a value of 27 kohm and for the other of 18 kohm. As I posed in a thread before, this value is not influencing the amplification. I then heated the amplifiers by full powering them. This results in a output offset voltage of only a few milliVolts, but when you put down the power, the output offset voltage is returning to its original point

So I dont agree with Bob Cordell. In my approach you can compensate the input offset current and voltage by adjusting just one resistor, so that the output offset voltage is nearly zero volt (In my case only 1 - 2 milliVolt). I found out in practice with the LM3886 amplifier that this output offset voltage doesn't run away (In my case max. 5 milliVolt). I was amazed by this result myself.

So, despite of the fact that de DC gain is equal to the closed loop gain, you not "reduces some of the input offset" but you can almost fully cancel it out by adjusting one resistor!! And it is not running away.

The overall advantage is a quite stable output offset voltage of (nearly) zero while not needing an electrolytic capacitor.

Marc.

Your scheme is essentially the same as Cordell's, except you are trimming one of the resistors that is trying to balance input impedances. I think that by "trimming" the resistor, you are trimming out the input offset.

Did you try this with the regular scheme using the SAME LM3886 IC? It would be a better test to do a comparison between the two circuits holding everything else constant. There is part-to-part variation and it could be that the particular LM3886 chip that you are using just happens not to drift with temperature in the first place. Once you make the correction at one temp, then it would be valid at other too. Not sure, just trying to think up alternate explanations...

-Charlie

Marc Vi. · 19th September 2012, 09:12 PM

Hallo Charlie,

My scheme is the same as Cordell's but the essential difference is that I trim the resistor to a zero output offset voltage.

I made two amplifiers with different LM3886's. In theory the value of the trimming resistor is 20 kohm but in practice I found that the optimal value is 18 kohm for one amplifier and 27 kohm for the other amplifier. The output offset voltage is then 1 - 2 milliVolt in both cases !! I also found that in both cases the output offset voltage is running away for max. 5 milliVolt when heating the LM3886 chips.

So I don't agree with Bob Cordell that "a big part of the advantage of having an electrolytic in the conventional arrangement is lost (due to input voltage offset). This approach can be used in combination with a DC servo to reduce the amount of correction that must be supplied by the servo."

Using this trimming resistor there is no need for a servo because the output offset voltage stays very small. The overall advantage is no electrolytic capacitor.

Even Bob Cordell is not always right. I repeat my question why this solution is not often used?

Marc.

CharlieLaub · 19th September 2012, 10:36 PM

[/QUOTE]

Quote:

Originally Posted by Marc Vi.

Hallo Charlie,

My scheme is the same as Cordell's but the essential difference is that I trim the resistor to a zero output offset voltage.

Yes, that's what I said above in my post.

Quote:

Originally Posted by Marc Vi.

I made two amplifiers with different LM3886's. In theory the value of the trimming resistor is 20 kohm but in practice I found that the optimal value is 18 kohm for one amplifier and 27 kohm for the other amplifier. The output offset voltage is then 1 - 2 milliVolt in both cases !! I also found that in both cases the output offset voltage is running away for max. 5 milliVolt when heating the LM3886 chips.

You already mentioned this. Is this so much different than other methods of trimming out offset voltage in terms of stability?

Quote:

Originally Posted by Marc Vi.

So I don't agree with Bob Cordell that "a big part of the advantage of having an electrolytic in the conventional arrangement is lost (due to input voltage offset). This approach can be used in combination with a DC servo to reduce the amount of correction that must be supplied by the servo."

Using this trimming resistor there is no need for a servo because the output offset voltage stays very small. The overall advantage is no electrolytic capacitor.

No servo needed = correct. The problem is now that the amplifier will have large DC gain (Same as AC gain, e.g. 25-30x). Even though you have a HP input filter, this is not a good idea, but this is just my opinion.

Quote:

Originally Posted by Marc Vi.

Even Bob Cordell is not always right. I repeat my question why this solution is not often used?

Marc.

I'm not sure why it would not be used more often, except that it requires trimming, the feedback electrolytic cap is just NOT that bad, and there are other ways to trim that are probably preferable and cheaper. That's just a guess on my part, however.

-Charlie

AndrewT · 20th September 2012, 08:23 AM

if the two input transistors are exactly matched and the two input sources resistances are exactly matched then the feedback ensures that the output offset is always zero.

The problem is that it is impossible to get the two inputs transistors exactly matched. You are using trimming of the source resistances to add a correction for the unmatched transistors.

You will find that the output offset will drift slightly. This drift may be so small that a servo is never needed.

13th September 2012, 07:55 AM	#13
zakman35 diyAudio Member Join Date: Mar 2009 Location: Athens	How this translates in practice. Is there any audible effect of it comparing with the usual ci cap cofiguration or they both sound the same? __________________ Modesty is the art to make others discover how great you are.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
DC offset and gain on GC LM3886	Xerxes75	Chip Amps	19	9th October 2012 10:02 PM
LM3886 - DC offset problem	luis.jimenez	Chip Amps	22	10th April 2010 11:34 AM
Feedback capacitor to minimize DC offset	Rafal	Chip Amps	31	15th August 2008 10:22 PM
Same old story LM3886 and DC offset	officeboy	Chip Amps	11	24th June 2004 02:51 AM

New To Site?	Need Help?
Register to Participate Search Privacy Statement Contact Us	Frequently Asked Questions Did you forget your password? Mark Forums Read


	Please consider donating to help us continue to serve you. Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving

12th September 2012, 08:23 PM	#11
CharlieLaub diyAudio Member Join Date: Mar 2007 Location: California	I happen to own the very nicely written book "Designing Audio Amplifiers" by Bob Cordell. Your suggested LM3886 circuit is exactly the same as Figure 8.3b on page 158. So, this is not something new. The chapter in which the figure is located discusses DC offset, and various techniques (including the usual electrolytic cap in an RC filter to ground from the negative feedback path). His comments on this topology are: "This arrangement reduces some of the input offset concerns without resorting to an electrolytic capacitor. However, the DC gain of the amplifiers is now equal to the closed-loop gain, so a big part of the advantage of having an electrolytic in the conventional arrangement is lost (due to input voltage offset). This approach can be used in combination with a DC servo to reduce the amount of correction that must be supplied by the servo." So, while it seems you have removed an electrolytic cap, which is probably a good thing, it does away with the function of that cap as well. What's the overall advantage??? Not sure. Say, while we are on the subject (hope that this is not thread hikacking), why don't people use a capacitance multiplier in place of the electro? The cap has a grounded terminal, so I believe that type of circuit can be used... For example, substitute the terminal marked +C in the left most circuit in the figure in place of the usual grounded electrolytic, use a 1uF film for C1 and then select the resistor ratio to give you 100uF or 1000uF or whatever you need. Thoughts? -Charlie

13th September 2012, 07:19 AM	#12
Marc Vi. diyAudio Member Join Date: Feb 2004 Location: The Netherlands	Hello Charlie, I didn't say that my method is new. What is new in my approach is that I adjust the value of the resistor to the -input so that the output is almost exactly 0,000 V. For one LM3886 amplifier this results in a value of 27 kohm and for the other of 18 kohm. As I posed in a thread before, this value is not influencing the amplification. I then heated the amplifiers by full powering them. This results in a output offset voltage of only a few milliVolts, but when you put down the power, the output offset voltage is returning to its original point So I dont agree with Bob Cordell. In my approach you can compensate the input offset current and voltage by adjusting just one resistor, so that the output offset voltage is nearly zero volt (In my case only 1 - 2 milliVolt). I found out in practice with the LM3886 amplifier that this output offset voltage doesn't run away (In my case max. 5 milliVolt). I was amazed by this result myself. So, despite of the fact that de DC gain is equal to the closed loop gain, you not "reduces some of the input offset" but you can almost fully cancel it out by adjusting one resistor!! And it is not running away. The overall advantage is a quite stable output offset voltage of (nearly) zero while not needing an electrolytic capacitor. Marc.

13th September 2012, 08:11 PM	#16
Marc Vi. diyAudio Member Join Date: Feb 2004 Location: The Netherlands	Hallo Charlie, Can you also answer my last post? Marc Vi.

20th September 2012, 08:23 AM	#20
AndrewT diyAudio Member Join Date: Jul 2004 Location: Scottish Borders	if the two input transistors are exactly matched and the two input sources resistances are exactly matched then the feedback ensures that the output offset is always zero. The problem is that it is impossible to get the two inputs transistors exactly matched. You are using trimming of the source resistances to add a correction for the unmatched transistors. You will find that the output offset will drift slightly. This drift may be so small that a servo is never needed.