various opamp questions

[traw]

have some more specific questions about passive components around opamps. learned things from snippets here and there, but haven't hit anything that is like all things opamps for audio circuits...

1. input cap followed by resistor to ground, 1/2piRC low rolloff and dc block. still need 1k resistor going to + on NI config?

2. does single 1k, for example, straight to + input just make sure there's voltage? is that same as input straight to + no inline resistor but a resistor to ground purpose? basically explanation for when that resistor is inline from signal to + and other times not.

3. input to resistor, resistor to ground. voltage divider. any similarity to resitor to ground, then input to resistor prior to going to + input (kind of how a pot out in front would look)? kind of a fuzzy question

actually think i maybe generalizing too much... mostly in reference to non inverting config.

4. capacitor parallel with feedback, 47pf for example, that to cut HF noise?

5. output side. use of 100-300 ohm to help deal with capacitive lines / loads. required if wish to use cap then resistor to ground ? what happens if just capacitor and no resistors (worst case). curious...

6. whats the cap across the + to - for again? stability?...

just try to gather up questions that don't have clear answers in my head. often resources cover basics, ideal opamp theory, jump into various applications (integrators, filters etc.).... have looked over 'opamps for everyone'.... but maybe someone can chime in with explanations for which way when/why

[janneman]

Originally posted by traw
have some more specific questions about passive components around opamps. learned things from snippets here and there, but haven't hit anything that is like all things opamps for audio circuits...

1. input cap followed by resistor to ground, 1/2piRC low rolloff and dc block. still need 1k resistor going to + on NI config?

2. does single 1k, for example, straight to + input just make sure there's voltage? is that same as input straight to + no inline resistor but a resistor to ground purpose? basically explanation for when that resistor is inline from signal to + and other times not.

3. input to resistor, resistor to ground. voltage divider. any similarity to resitor to ground, then input to resistor prior to going to + input (kind of how a pot out in front would look)? kind of a fuzzy question

actually think i maybe generalizing too much... mostly in reference to non inverting config.

4. capacitor parallel with feedback, 47pf for example, that to cut HF noise?

5. output side. use of 100-300 ohm to help deal with capacitive lines / loads. required if wish to use cap then resistor to ground ? what happens if just capacitor and no resistors (worst case). curious...

6. whats the cap across the + to - for again? stability?...

just try to gather up questions that don't have clear answers in my head. often resources cover basics, ideal opamp theory, jump into various applications (integrators, filters etc.).... have looked over 'opamps for everyone'.... but maybe someone can chime in with explanations for which way when/why

1 each input must see a path to gnd for DC, because those inputs have tiny currents, which have to go somewhere. Whether it is 1k depends on the topology. Generally, you want to have equal DC impedance from each input to ground, to balance the tiny input DC currents to avoid DC on the output (aka DC offset). But the feedback resistors also determine the DC impedance so it is more involved (but not difficult).

2 not sur I understand but if you do DC coupling on the input (can do that if your source is at 0V DC) you automatically provide a path for the tiny DC input currents and the inline resistor may just be there to limit input current in case of overload which might damage the chip.

3 I wouldn't do this, seems like a waste of input signal. have you actually seem this? May be done to decrease the input signal if there is a danger of overload, but it is better to have the circuit being able to handle the expected input level.

4 No, is sometimes necessary for stability with heavy feedback. Value is normally less than 47pF, sometimes just 10pF. Depends on amp chracteristics, often done by trail and error.

5 series R is indeed to decouple cap load (or cap cable) from output for stability. Don't know about the R and C, unless you refer to power amps where this is sometimes done (called a zobel) to maintain amp load with long speaker cables. Those speaker cables are inductive and have a climbing impedance for increasing freq and therefore make the load (speaker) at the end of that cable decrease. If the amp became unloaded that could give stability problems. But, if this zobel is used, it should be directly at the output before the series output resistor.

6 you mean + and - power supply? That's done to improve the smoothness of the supply, to prevent ripple and noise on the supply that could appear at the output as noise or distortion. Cap across + and - inputs is to improve stability by increasing what is called the noise gain. It is a bad way to improve stability and generally is bad design practise.

Hope this helps.

Jan Didden

[traw]

that helps...

3 - seems most clones go 1k then 22k to ground. then additional throw a pot in front of that. in some educational material, they often omit resistors on input. sometimes show a resistor to ground saying to ensure a voltage. also seen resistor in series for same reason (hope i'm not mixing up NI and I config). some more reading seems to indicate on inverting gain config, input impedence essentially that resistor value.

5 - more less dc blocking thing.... similar to how cap in series, then resistor to ground. seen on rod elliot p88 ~ 2uf then 100k to ground at output. although i imagine the next stages input impedence trumps that 100k but suppose 100k is have least impact when parallel with that load

4, 6 - saw this on commercial pro amp schematic and other places around . maybe brute force accomodate for the worst...

[janneman]

Originally posted by traw
that helps...

3 - seems most clones go 1k then 22k to ground. then additional throw a pot in front of that. in some educational material, they often omit resistors on input. sometimes show a resistor to ground saying to ensure a voltage. also seen resistor in series for same reason (hope i'm not mixing up NI and I config). some more reading seems to indicate on inverting gain config, input impedence essentially that resistor value.

5 - more less dc blocking thing.... similar to how cap in series, then resistor to ground. seen on rod elliot p88 ~ 2uf then 100k to ground at output. although i imagine the next stages input impedence trumps that 100k but suppose 100k is have least impact when parallel with that load

4, 6 - saw this on commercial pro amp schematic and other places around . maybe brute force accomodate for the worst...

3. The major problem here is to be able to separate the BS from the sensible. Everybody jumps on the chip amp bandwagon and some info is plain disastrously wrong. There is no clear solution to this and is a disadvantage of the way the 'net works. read the datasheet, while not iffalible, at least it doesn't tell you plain BS. The data sheet has tested designs, and READ the text. You'll learn!

5 I may have misunderstood you previous question. If you do a cap in series with the output, you may include the R to ground after the cap to make sure that whatever you will drive with it sees a path to gnd for the reasons discussed on the inputs. Also you need to provide a ground path for the series cap to charge (at switch-on and discharge at switch off. If no R to gnd you have the risk of two caps in series (the output cap and the input cap of the next equipment) and that may give rise to slicks and pops at switch-on.

4,6 yes, its possible.

Jan Didden

[traw]

another question...

on the input side. say device is said to have 10k input impedance. want to put cap for dc blocking. if i put 2.2uf, then 22k to ground... is it 22k ||10k for the 1/(2piRC) calc for cutoff? where would be 10.5 hz versus 3.28 if figuring the 22k alone? or is there a different way to look at it?

[mikee55]

http://sound.westhost.com/dwopa.htm:)