Can R2 (22k) be omitted (as shown in National's schematics)? Why or why not?
I am using a 10K ohm ladder-type stepped attenuator (as a passive pre) and may get a 100k one latter. Could either attenuator work well with LM3886 without R2? Could the attenuator simply replace R2's position and set the input impedance for LM3886? Would there be any difference: 10K vs 100k, with or without R2? Thanks!
I am using a 10K ohm ladder-type stepped attenuator (as a passive pre) and may get a 100k one latter. Could either attenuator work well with LM3886 without R2? Could the attenuator simply replace R2's position and set the input impedance for LM3886? Would there be any difference: 10K vs 100k, with or without R2? Thanks!
There is no R2 of 22k in National's schematics. The application notes use Rf1 with 20k. And no, you cannot leave that out, because together with Ri it sets the gain.
Which of the attenuators is better suited depends on how your circuit looks. Attenuators may interact with other elements (capacitors) and form filters with them. There is also impedance matching between source and amplifier to take into account. Use the search function for that topic, there are already threads that deal with it.
The attenuator must not replace Rf1. The gain setting should be between 10 and 50 times. Certain attenuator settings would give you gains outside of that range.
Which of the attenuators is better suited depends on how your circuit looks. Attenuators may interact with other elements (capacitors) and form filters with them. There is also impedance matching between source and amplifier to take into account. Use the search function for that topic, there are already threads that deal with it.
The attenuator must not replace Rf1. The gain setting should be between 10 and 50 times. Certain attenuator settings would give you gains outside of that range.
a break before make stepped attenuator will leave a momentary opencircuit during the switching operation.
A make before break will have a momentary paralleling of two steps during the switching operation.
Either of these must be looked at and what effect it will have on the operation of the amplifier.
A make before break will have a momentary paralleling of two steps during the switching operation.
Either of these must be looked at and what effect it will have on the operation of the amplifier.
Sorry, I meant that the omission was in National's datasheet.
The R2 is in Chipamp.com's LM3886's design. It is indeed absent in National's datasheet. It's a 22K resistor from the input to the ground. In National's datasheet, on page 21, it's however said that, when using a 10k pot, it's recommended that a 100k resistor be put after the pot. That 100k resistor would function in the same way as the 22k R2 in chipamp's design. But are they really necessary? If I were to use a 100k pot, shouldn't I put a 1meg resistor after it, or can the 10k/100k ladder-type pot stand by itself before the chip without an input resistor following it? Would a ladder-type pot set a constant input impedance (in the same way as an input resistor) for the chip? Thanks!
The R2 is in Chipamp.com's LM3886's design. It is indeed absent in National's datasheet. It's a 22K resistor from the input to the ground. In National's datasheet, on page 21, it's however said that, when using a 10k pot, it's recommended that a 100k resistor be put after the pot. That 100k resistor would function in the same way as the 22k R2 in chipamp's design. But are they really necessary? If I were to use a 100k pot, shouldn't I put a 1meg resistor after it, or can the 10k/100k ladder-type pot stand by itself before the chip without an input resistor following it? Would a ladder-type pot set a constant input impedance (in the same way as an input resistor) for the chip? Thanks!
Yes, they are. The differential pair at the chip's input needs bias current because it is bipolar. Normally, this (very tiny) DC current flows through the wiper of the pot (or attennuator) and the amp is properly biased. But if you turn it, and the pot is a bit dirty, the wiper goes open momentarily. This leaves the + input floating if there's no alternative provision for the bias current. Then, due to input bias current, the + input will plunge towards V-, and the chip will faithfully amplify this false signal. This gives a disturbingly loud crackle through the speakers when operating the volume control.
The resistor is there to provide the alternative path should the pot (or attenuator)'s wiper momentarily go open. Don't be tempted to leave it out, because your pot/attenuator doesn't have this issue. Over time, even the most expensive switches and pots develop it.
Ah, I see, R2 is Rin. National only shows the pot as Rin in their Typical Application Circuits. In their Reference Schematics they use an additional 15k resistor for R2.
The 100k resistor on page 21 is supposed to be in series with the pot to achieve the required input impedance for the example project. R2 is not in series with the pot.
The 100k resistor on page 21 is supposed to be in series with the pot to achieve the required input impedance for the example project. R2 is not in series with the pot.
Yes, the 100k was in series with the pot, unlike R2. I didn't read it carefully.
TImpert:
Thanks for the explanation. What would be a good relationship between the values of the pot and of the Rin? I read somewhere that the value of Rin should be 10x of the pot, for the purpose of impedance matching. But it's also said somewhere else that the values should be same, for the same purpose of impedance matching.
TImpert:
Thanks for the explanation. What would be a good relationship between the values of the pot and of the Rin? I read somewhere that the value of Rin should be 10x of the pot, for the purpose of impedance matching. But it's also said somewhere else that the values should be same, for the same purpose of impedance matching.
Hi,
Impedance matching is not an issue here. However, if you make R2 about equal to the pot's value, at max volume, R2 is parallel to the pot's own resistance. This has the effect of halving the input impedance of the amp at max volume wrt the impedance at zero volume. If the source is cap coupled to the amp's input, then the lower cutoff frequency will shift when the volume control is operated. This is not desirable, since we intend to control only the volume and leave the other properties alone as much as possible. If you make R2 big compared to the pot's value, the relative impedance change will become small and the shift of the cutoff frequency will become negligible.
However, you can also make R2 too big. Input bias current is typically 0.2 uA and its limit is 1 uA (datasheet). That means that, if the DC resistance seen by the + input jumps by 100k Ohms, the voltage on it jumps anywhere between 20 and 100 mV which is amplified about 30 times, giving a spike between 0.6 and 3V (worst case) on the output. This would typically be the case with a 10k pot and 100k R2. So in a nutshell, the higher the value of R2, the less crackle-proof the amp will be. Apart from noise and frequency response, another reason to keep the impedances around the chip low.
Impedance matching is not an issue here. However, if you make R2 about equal to the pot's value, at max volume, R2 is parallel to the pot's own resistance. This has the effect of halving the input impedance of the amp at max volume wrt the impedance at zero volume. If the source is cap coupled to the amp's input, then the lower cutoff frequency will shift when the volume control is operated. This is not desirable, since we intend to control only the volume and leave the other properties alone as much as possible. If you make R2 big compared to the pot's value, the relative impedance change will become small and the shift of the cutoff frequency will become negligible.
However, you can also make R2 too big. Input bias current is typically 0.2 uA and its limit is 1 uA (datasheet). That means that, if the DC resistance seen by the + input jumps by 100k Ohms, the voltage on it jumps anywhere between 20 and 100 mV which is amplified about 30 times, giving a spike between 0.6 and 3V (worst case) on the output. This would typically be the case with a 10k pot and 100k R2. So in a nutshell, the higher the value of R2, the less crackle-proof the amp will be. Apart from noise and frequency response, another reason to keep the impedances around the chip low.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.