That looks about right Fedde. I just tried a 30K, 75K and 91K in series with the 20K trimmer and the DC offset got lower as the R went up! At 111K it was about 13mV.
I'll see if I have some 200K's around.
BTW - I have 52K to ground on the input as I don't include a pot in my monoblocks. Does that affect the value of the resistor from in 7 to ground?
Nuuk said:
I don't have a pot too, I use my GC as a power amp.
I have 22k to ground on the input (the inverting input).
Maby that's the difference.
Nuuk, just come around, I'll give you some 200k resistors.
This is getting a little confusing...
We should ask super-heroe Kuei to give us a help.
Kuei, are you there?
You do not need our Kuei for everything. The poor man is quite occupied with all requests already, you see 
Kuei added the 22 k resistor to improve the stability of the amp under difficult conditions (no input pot present, no input connected etc.). Because the opamp has a small bias through both inputs, you get a small voltage across the 22 k resistor (and the 220 k feedback). These two resistors are parallel so the total resistance is around 20k (220k*22k/(220k+22k)). To reach 0 V at the output of the opamp, the + input needs to have the same impedance to earth, so around 20 k. Due to tolerances in the opamp, the actual impedance needed for zero DC level can vary...
Hope this clears some things up...
Fedde
Kuei added the 22 k resistor to improve the stability of the amp under difficult conditions (no input pot present, no input connected etc.). Because the opamp has a small bias through both inputs, you get a small voltage across the 22 k resistor (and the 220 k feedback). These two resistors are parallel so the total resistance is around 20k (220k*22k/(220k+22k)). To reach 0 V at the output of the opamp, the + input needs to have the same impedance to earth, so around 20 k. Due to tolerances in the opamp, the actual impedance needed for zero DC level can vary...
Hope this clears some things up...
Fedde
you're like Kuei
I was calling Kuei because it was he that first advised me to change the 18k resistor for a 50k multi-turn pot.
He recommends it.
But you answered as well as Kwei, I should say.
For me, that resistor has always to be of a smaller value than the input resistor on the inverting input.
That's why Nuuk is confusing me in saying that putting a larger resistor he gets less DC offset.
fedde said:You do not need our Kuei for everything. The poor man is quite occupied with all requests already, you see
Kuei added the 22 k resistor to improve the stability of the amp under difficult conditions (no input pot present, no input connected etc.). Because the opamp has a small bias through both inputs, you get a small voltage across the 22 k resistor (and the 220 k feedback). These two resistors are parallel so the total resistance is around 20k (220k*22k/(220k+22k)). To reach 0 V at the output of the opamp, the + input needs to have the same impedance to earth, so around 20 k. Due to tolerances in the opamp, the actual impedance needed for zero DC level can vary...
Hope this clears some things up...
Fedde
I was calling Kuei because it was he that first advised me to change the 18k resistor for a 50k multi-turn pot.
He recommends it.
But you answered as well as Kwei, I should say.
For me, that resistor has always to be of a smaller value than the input resistor on the inverting input.
That's why Nuuk is confusing me in saying that putting a larger resistor he gets less DC offset.
Well, eh thanx
The input resistor is out of play because of the DC (current and voltage) blocking cap. The main thing is that the voltages at the + and - input are equal. So because of the parasitic current flow through the inputs of the opamps, the impedances at both inputs should be equal to have equal voltages at the inputs. Because the parasitic current flow is not exactly the same, you need a pot or a resistor matched to the system. I would prefer to put a resistor parallel to the potmeter to improve the sound quality and to reduce bad noises while turning a dusty potmeter. You have to experiment a little to find the optimal configuration. I would use a resistor slighty larger than the resistor needed for zero output DC. And then use a 50 k pot or so to correct to exactly zero DC.
I think it's worthwhile to reach low output DC because you maximize the 'class-A' region of the opamp (among other things). And also speakers may perform a little better with reduced DC.
Fedde
The input resistor is out of play because of the DC (current and voltage) blocking cap. The main thing is that the voltages at the + and - input are equal. So because of the parasitic current flow through the inputs of the opamps, the impedances at both inputs should be equal to have equal voltages at the inputs. Because the parasitic current flow is not exactly the same, you need a pot or a resistor matched to the system. I would prefer to put a resistor parallel to the potmeter to improve the sound quality and to reduce bad noises while turning a dusty potmeter. You have to experiment a little to find the optimal configuration. I would use a resistor slighty larger than the resistor needed for zero output DC. And then use a 50 k pot or so to correct to exactly zero DC.
I think it's worthwhile to reach low output DC because you maximize the 'class-A' region of the opamp (among other things). And also speakers may perform a little better with reduced DC.
Fedde
fedde,
Very good advices you're giving there.
I think the same.
Anyway, the pots I use are multi-turn, completely sealed and they make no noise.
But your parallel resistor idea is a good one.
And yes, no DC offset means the speaker drivers will work in a more linear way, resulting in less distorsion and finally better sound.
Very good advices you're giving there.
I think the same.
Anyway, the pots I use are multi-turn, completely sealed and they make no noise.
But your parallel resistor idea is a good one.
And yes, no DC offset means the speaker drivers will work in a more linear way, resulting in less distorsion and finally better sound.
I just got back from my cycle ride down to see Carlos. Thanks for those resistors C
Well, with a 220K resistor and the trimmer set at 132R, I get 0.5mV. Turning the trimmer either way starts increasing the offset!
At turn on the offset rises momentarily to 49mV before sinking back to the 0.5 level.
As the trimmer is only adding 132R, it may as well be left out as it only makes a tiny difference to the DC offset level.
And now time for something to eat - I'm hungry after all that cycling!
Well, with a 220K resistor and the trimmer set at 132R, I get 0.5mV. Turning the trimmer either way starts increasing the offset!
At turn on the offset rises momentarily to 49mV before sinking back to the 0.5 level.
As the trimmer is only adding 132R, it may as well be left out as it only makes a tiny difference to the DC offset level.
And now time for something to eat - I'm hungry after all that cycling!
carlosfm said:Something's wrong there, Nuuk.
220k is exactly the value Thorsten recommended first, and then he changed it to 18k, or better still, a 50k multi-turn pot.
How did you connect the pot?
The pot is in series with the 220K resistor (between pin 7 and signal star ground with nothing between pin 8 and signal star ground)). Last night I had another play around and it seems that 0v offset is somewhere just below 220K so I ended up with the pot set to zero ohms (effectively it isn't there)!
When you say Thorsten changed the resistor to 18K, isn't that with 22K from the inveting input to ground as well?
And why do you get 0v offset with only 4K? ASAIK we are all building these GC's to the same spec so what is happening here?
Fedde said he got his lowest offset around 20K. Perhaps we need to collate everybody's findings on this issue! I know that some may say it is academic and a straight wire to ground with 28mV offset is fine but we should be a little more scientific and sort out this once and for all.
I'm off to do some more fiddling!
Nuuk said:
Nuuk, there are no two equal Gainclones.
And no two equal chips, and caps...
That's why you can have 28mv on one channel and 10mv on the other.
And that's why the pot is a good thing.
Because you can pick a resistor to get around 0mv on one channel, but you may have a different value on the other channel and you'll have to pick another value resistor.
The 18k resistor I'm talking about you can see in this schematic.
That's what you can change for the pot.
I use the 22k resistor as you can see here, from the inv. to the non inv. input (actually, the star ground, the pot).
This 22k resistor is important, it helps the stability of the chip.
I made my Gainclone basically as this schematic.
On the first Thorsten Gainclone schematics you can see a 220k and a 0.1uf cap in place of the 18k resistor.
He now recommends that you forget that 220k/0.1uf and put an 18k resistor, or a 50k multi-turn pot.
I hope this helps.
I didn't say that I did use 20 kohm, that is the theoretical value if you use 22 k on the - input. I used 220 k feedback and on the + input. Since yesterday I use 330 k on both positions. DC is <0.7 mV. And the sound is IMHO much better, better defined, more dynamic. But most important the amp seems now fully stable. I had some problems recently that the sound of my amp was sometimes fantastic and sometimes tiring. Seems to be fixed
And I have more gain, that's good for the neighborhood competition
Fedde
And I have more gain, that's good for the neighborhood competition
Fedde
Hi guys, I've put on my suit of armour, got my flag and sword handy and am continuing on my crusade to help the beginners! 
Carlos, thanks for that circuit but I was using exactly the same circuit with the 22K to ground from pin 8 and the 18K to ground from pin 7 and got near enough to 0mV offset on one channel 4.5mV on the other.
I didn't realise that you were using the pot as well as the 22K resistor!
Likewise, Fedde, your last post
"22K on the - input" - I know you mean 22K from the inverting input to ground but someone less experienced may take the 22K to be the input resistor! So you are now using 330K feedback resistor and 330K from inverting input (pin 8) to ground but you don't say what you are using from non-inverting input (pin 7) to ground.
If I am getting confused here, imagine what it is like for somebody new to all this!
I'm not having a go at anyone here, just asking, on behalf of all of us if we could perhaps be a little more careful in describing our circuits. There are not many permutations so perhaps we could even standardize our terminology. Say something like:
Fb 220K
Ir 10K
7-G 18K
8-G 22K
DCO=0.4mV
I hope I am not being too pedantic here but I do feel it would prevent some misunderstandings and clear communication never did any harm
So, back to my own situation. I reckon the sound is probably better with
Fb 216K
Ir 10K
7-G Straight wire
8-G NC (No connection)
DCO=28mV
rather than
Fb 216K
Ir 10K
7-G 18K
8-G 22K
DCO=0mV
so I may go back to that and have another listen.
Carlos, thanks for that circuit but I was using exactly the same circuit with the 22K to ground from pin 8 and the 18K to ground from pin 7 and got near enough to 0mV offset on one channel 4.5mV on the other.
I didn't realise that you were using the pot as well as the 22K resistor!
Likewise, Fedde, your last post
"22K on the - input" - I know you mean 22K from the inverting input to ground but someone less experienced may take the 22K to be the input resistor! So you are now using 330K feedback resistor and 330K from inverting input (pin 8) to ground but you don't say what you are using from non-inverting input (pin 7) to ground.
If I am getting confused here, imagine what it is like for somebody new to all this!
I'm not having a go at anyone here, just asking, on behalf of all of us if we could perhaps be a little more careful in describing our circuits. There are not many permutations so perhaps we could even standardize our terminology. Say something like:
Fb 220K
Ir 10K
7-G 18K
8-G 22K
DCO=0.4mV
I hope I am not being too pedantic here but I do feel it would prevent some misunderstandings and clear communication never did any harm
So, back to my own situation. I reckon the sound is probably better with
Fb 216K
Ir 10K
7-G Straight wire
8-G NC (No connection)
DCO=28mV
rather than
Fb 216K
Ir 10K
7-G 18K
8-G 22K
DCO=0mV
so I may go back to that and have another listen.
Nuuk said:
I recommend you use the 8-G
22k resistor.More important it is if you use your Gainclone as a power amp (without input pot), as I do.
The chip may become instable without it.
And did you know that with the 8-G resistor in place the chip becomes stable even with a gain of 1 (a buffer)?
Aha!
That doesn't come in the datasheet.
Koinichiwa,
This applies if an input capacitor is used. This is classic "Op-Amp" theory. With a 10K & 4u7 Input section the DC Gain of the loop is 1 and the DC resistance is in effect Rfb.
You will get zero offset in this case if Rfb = R+ (7-G). Because the Chip has bipolar inputs they "leak" a certain amount of static base current. This base current causes a voltage across the resistor, which is amplified.
Ideally one would balance the impedances also for Audio, implying a 10K resistor in series with 4u7 across that resistor. in the original "inverting gainclone" I suggested 0.1uF as "close enough".
The later schematic with the 22k & 18K resistors are of course aimed at making the circuit "capacitor-less", so all the above can be discarded in this case.
Sayonara
Nuuk said:
This applies if an input capacitor is used. This is classic "Op-Amp" theory. With a 10K & 4u7 Input section the DC Gain of the loop is 1 and the DC resistance is in effect Rfb.
You will get zero offset in this case if Rfb = R+ (7-G). Because the Chip has bipolar inputs they "leak" a certain amount of static base current. This base current causes a voltage across the resistor, which is amplified.
Ideally one would balance the impedances also for Audio, implying a 10K resistor in series with 4u7 across that resistor. in the original "inverting gainclone" I suggested 0.1uF as "close enough".
The later schematic with the 22k & 18K resistors are of course aimed at making the circuit "capacitor-less", so all the above can be discarded in this case.
Sayonara
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.