OK folks! Here is the question that has been hanging on everybody's lip for the past 40 years! The leaders of the world awaits your verdict.
Can one use a single voltage divider (two resistors only) for several op amps + inputs or will the voltage be distributed very uneven, like when you use a single resistor for several LEDs? My initial thought was no but hey I could be wrong.
So you want to make ONE bias supply, and feed that to an opamp? Are you nuts????? ;) ha ha
Of course you can, but you didn't know that, and now you do! This is because an opamp draws no current!!!! The input impedance is so high, the voltage divider doesn't know it's there. So it doesn't get dragged down. Check the schematic for a Tube Screamer, look at Vref.
Good question, Stoner! Youtube has some great vids on opamps and how they do this 'no current' thing...EEVBlog is a good guy to search for about that, has a super tut for you. Here: https://www.youtube.com/watch?v=7FYHt5XviKc (https://www.youtube.com/watch?v=7FYHt5XviKc)
BAM! Thank you GibsonGM! Now I know! ;D
You can also feed the downstream opamp with the previous opamp's DC output which is equal to it's input bias assuming you don't need a coupling cap in between which in many cases you don't. I have a pedal with three dual opamps, the voltage divider only feeds channel 1 of the first opamp and I bias the rest of them with each subsequent output. Saves jumpers, traces and complexity when possible.
Quote from: karbomusic on March 20, 2016, 01:46:04 PM
You can also feed the downstream opamp with the previous opamp's DC output which is equal to it's input bias assuming you don't need a coupling cap in between which in many cases you don't. I have a pedal with three dual opamps, the voltage divider only feeds channel 1 of the first opamp and I bias the rest of them with each subsequent output. Saves jumpers, traces and complexity when possible.
YEP....if you haven't blocked that offset with a cap, it's still THERE! :) Tube screamer does this for stages 1 and 2, then there's a cap...so the bias voltage is brought back for the output buffer (it's there for the vol. pot, too, but i think just going to ground would work the same as it is blocked on either side by caps).
> a single voltage divider ...for several
Plagiarize, plagiarize, plagiarize!! (Only call it "research".)
Read EVERY schematic you can find.
In there you will find plans with one Vref shared by many loads.
If you don't know what you are doing (and especially if you do), you want a heavy capacitor also. First to cut crap from the power supply. Also so signal at one stage does not sneak-path into another stage.
Quote from: karbomusic on March 20, 2016, 01:46:04 PM
You can also feed the downstream opamp with the previous opamp's DC output which is equal to it's input bias
It often is, but don't be so certain about it. In several cases a gain stage will also amplify its DC offset, in which case the DC bias voltage derived from opamp output ends up being different to intented DC bias voltage.
Probably doesn't matter much if the bias shift is only few millivolts but starts to portray effects if the bias shift is in the range of few volts.
Sometimes those coupling caps are essential even though theoretically it would seem that DC offsets of the stages should be equal. Practice is often different than theory.
Quoteassuming you don't need a coupling cap in between which in many cases you don't. Saves jumpers, traces and complexity when possible.
Yes,
when possible it makes sense to eliminate those components. It's not always possible, though. If you have a gain stage with high gain then it probably will need AC coupling because high gain also means high gain to amplify any DC offsets. If you direct couple several gain stages the gain multiplies, so in the end of the chain the amplified DC offset error can be significant. Because of that it is very rare to successfully direct couple several gain stages without upsetting bias in some of the stages. Something like a unity-gain buffer stage (with unity gain also to DC offset) you can probably direct couple without any problems.
QuoteIf you don't know what you are doing (and especially if you do), you want a heavy capacitor also. First to cut crap from the power supply. Also so signal at one stage does not sneak-path into another stage.
Yes, AC voltage appearing in the DC bias supply should be "decoupled". The filtering capacitance is dependent on overall component values of the resistive divider, naturally. For example, a 1K-1K divider requires more coupling capacitance than a 1M-1M divider. But even basic electronics theory should tell you that.
Quote from: teemuk on March 21, 2016, 04:50:16 AM
For example, a 1K-1K divider requires more coupling capacitance than a 1M-1M divider. But even basic electronics theory should tell you that.
You don't need "basic electronic theory" as long as you realize that capacitor's role in this case (capacitance) is tight with current's quantity..
(which quantity is 1000 times bigger in 1k/1k divider that in 1M/1M divider..)
But you'll have to make a compromise between big capacitor (small resistors - low noise) or small capacitor (big resistors - high noise)
P.S.
The above refer only to practically negligible current consumption from the divider point (like Op-Amps Inputs). In other cases (like BJT biasing) where you need a "stiff" divider, medium to low resistor values are figurative (with all impedance issues, of course..)
You could go read http://www.geofex.com/circuits/biasnet.htm