I built this based on the cct posted in this forum. Worked nicely and sound great. Nevertheless I got about 50mV DC across the 8Ohm loudspeaker. This 50mV is rather stable over time. It does not hurt my open baffle full range driver but I would like to see it be reduced. For those who have built this what kind of offset you see? Any suggestion on how to reduce it? Thanks.
Attachments
I don't think it is unusual. The data sheet says typical 1mV, max 15mV. This is input referred of course, and with a voltage gain of 32, the output DC offset can in theory vary between 32mV and 480mV.
You can null it by inserting a very small DC into the pos or neg input (depending on polarity) but it is not something to worry about.
Jan
You can null it by inserting a very small DC into the pos or neg input (depending on polarity) but it is not something to worry about.
Jan
A few points:
1) 50 mV DC is fine, thousands of amplifiers show that and it´s simply allowable, zero damage or problems.
2) that said, your amp has 32 X gain, which is fine, but both AC (good) and DC (bad).
3) rather than correct it with a servo, a trimmer,etc. , we may simply turn DC gain to 1X , adding needed capacitor(s)
a) replace the 22k resistor connected to pin 2 (you should REALLY label all parts : R1 .. R2 ... C1 ... C2 ... etc.
) with a 100uF capacitor.
In theory a bipolar but actually a polarized one will do fine.
b) restore the DC NFB path by connecting 2 series 10k resistors from pin 2 to pin 5 , decoupling the center point joining them to ground with another 100uF capacitor, same considerations as theother one.
Presto, you now have a transconductance amplifier, with 32X AC gain, 1X DC gain; offset will drop to 1 or 2 mV .
Note:I do this all the time, for decades now, I make Guitar amplifiers where speakers love high impedance drive
EDIT: 2 extra notes:
a) amp will probably "pop" at turn on, it takes a second for the added caps to charge and stabilize.
b) if you are real cheap (or minimalistic as I am) you can connect pin 2 to pin 5 with just a 22k resistor (the one you just removed) and no decoupling.
It works because pin 2 will practically "see" the current feedback point through the 100uF cap anyway.
1) 50 mV DC is fine, thousands of amplifiers show that and it´s simply allowable, zero damage or problems.
2) that said, your amp has 32 X gain, which is fine, but both AC (good) and DC (bad).
3) rather than correct it with a servo, a trimmer,etc. , we may simply turn DC gain to 1X , adding needed capacitor(s)
a) replace the 22k resistor connected to pin 2 (you should REALLY label all parts : R1 .. R2 ... C1 ... C2 ... etc.
) with a 100uF capacitor.In theory a bipolar but actually a polarized one will do fine.
b) restore the DC NFB path by connecting 2 series 10k resistors from pin 2 to pin 5 , decoupling the center point joining them to ground with another 100uF capacitor, same considerations as theother one.
Presto, you now have a transconductance amplifier, with 32X AC gain, 1X DC gain; offset will drop to 1 or 2 mV .
Note:I do this all the time, for decades now, I make Guitar amplifiers where speakers love high impedance drive
EDIT: 2 extra notes:
a) amp will probably "pop" at turn on, it takes a second for the added caps to charge and stabilize.
b) if you are real cheap (or minimalistic as I am) you can connect pin 2 to pin 5 with just a 22k resistor (the one you just removed) and no decoupling.
It works because pin 2 will practically "see" the current feedback point through the 100uF cap anyway.
Last edited:
A few points:
1) 50 mV DC is fine, thousands of amplifiers show that and it´s simply allowable, zero damage or problems.
2) that said, your amp has 32 X gain, which is fine, but both AC (good) and DC (bad).
3) rather than correct it with a servo, a trimmer,etc. , we may simply turn DC gain to 1X , adding needed capacitor(s)
a) replace the 22k resistor connected to pin 2 (you should REALLY label all parts : R1 .. R2 ... C1 ... C2 ... etc.
In theory a bipolar but actually a polarized one will do fine.
b) restore the DC NFB path by connecting 2 series 10k resistors from pin 2 to pin 5 , decoupling the center point joining them to ground with another 100uF capacitor, same considerations as theother one.
Presto, you now have a transconductance amplifier, with 32X AC gain, 1X DC gain; offset will drop to 1 or 2 mV .
Note:I do this all the time, for decades now, I make Guitar amplifiers where speakers love high impedance drive
EDIT: 2 extra notes:
a) amp will probably "pop" at turn on, it takes a second for the added caps to charge and stabilize.
b) if you are real cheap (or minimalistic as I am) you can connect pin 2 to pin 5 with just a 22k resistor (the one you just removed) and no decoupling.
It works because pin 2 will practically "see" the current feedback point through the 100uF cap anyway.
Hi!
it uses bipolar power supply, he do not need those things from 3a,b
Why do you say so?
I am breaking the DC path in the NFB loop, what´s your point?
*Precisely* because he has a split supply, there´s only a few mV (offset) across each capacitor.
But we are trying to avoid *amplifying* those few mV by 32X
Last edited:
Mark, there's 8 ohms in parallel to that 220 ohms. Gain is around 32.
The 22k in series with the inv input is there to balance the DC impedance on both inputs to minimize offset. That said, as Padamiecki said, playing with that also allows you to tune the offset towards zero, but it isn't really necessary.
Jan
Last edited:
A few points:
1) 50 mV DC is fine, thousands of amplifiers show that and it´s simply allowable, zero damage or problems.
2) that said, your amp has 32 X gain, which is fine, but both AC (good) and DC (bad).
3) rather than correct it with a servo, a trimmer,etc. , we may simply turn DC gain to 1X , adding needed capacitor(s)
a) replace the 22k resistor connected to pin 2 (you should REALLY label all parts : R1 .. R2 ... C1 ... C2 ... etc.
In theory a bipolar but actually a polarized one will do fine.
b) restore the DC NFB path by connecting 2 series 10k resistors from pin 2 to pin 5 , decoupling the center point joining them to ground with another 100uF capacitor, same considerations as theother one.
Presto, you now have a transconductance amplifier, with 32X AC gain, 1X DC gain; offset will drop to 1 or 2 mV .
Note:I do this all the time, for decades now, I make Guitar amplifiers where speakers love high impedance drive
EDIT: 2 extra notes:
a) amp will probably "pop" at turn on, it takes a second for the added caps to charge and stabilize.
b) if you are real cheap (or minimalistic as I am) you can connect pin 2 to pin 5 with just a 22k resistor (the one you just removed) and no decoupling.
It works because pin 2 will practically "see" the current feedback point through the 100uF cap anyway.
Thanks for all the inputs from various friends. This reconfirms my initial judgement that 50mV is not going to hurt, so I have time to evaluate what is the best option moving forward. The "dirt cheap" solution sounds very interesting.
Ah, I didn't spot that - makes more sense now that its transconductance sensing speaker current in the 0.26 ohm shunt!
Then there's an issue to think about, its a lot of unnecessary noise to my mind.
Last edited:
Hi Mark,
You are correct, there is quite a bit of noise. So I am planning to drive down the resistance value. I could be in a lucky position as I use this amp in an active crossover driving an open baffle full range from 300Hz up. Driving down the 22k both at the input and feedback path can be done even I keep the input cap at 2.2u. The LF roll off does not concern me in such case.
You are correct, there is quite a bit of noise. So I am planning to drive down the resistance value. I could be in a lucky position as I use this amp in an active crossover driving an open baffle full range from 300Hz up. Driving down the 22k both at the input and feedback path can be done even I keep the input cap at 2.2u. The LF roll off does not concern me in such case.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.