Author |
output buffer question |
br>lasesentaysiete |
br>if I use this buffer circuit by Dr. Ken Stone on an output (i.e. coming from the pcb) can I omit the 100k resistor?
br> br> |
|
br>Synthiq |
br>The 100kohm resistor is only used to pull to input ground when the input is not connected to anything so it is well defined. If the input is always connected, the 100kohm resistor can be eliminated. Not sure you need the 1kohm resistor in this case either as long as the input voltage is within the +/-15V supply voltages. br> br> |
|
br>lasesentaysiete |
br>Synthiq
br> br> |
|
br>cygmu |
br>I'm no electrical engineer, so this may be nonsense, but don't you need to ensure that there is a DC path for the input bias current?
For example, if the previous stage that is connected to the input is capacitively coupled, then without the 100k resistor, that input bias current can only come through the capacitor, which it won't want to do.
I suppose the effect will be small / slow with a FET op amp like a TL07x but it would exist.
I think.
But if the previous stage has a DC path to feed the input then I think you are fine to omit the resistor. br> br> |
|
br>mskala |
br>You do need to ensure a DC path for the input bias current. The bias current on a TL071 is very small, but it's still possible for it to be a problem. Most outputs you might patch into will provide that DC path, but you're right that not all do, and people have had problems from something similar before. See for instance https://www.muffwiggler.com/forum/viewtopic.php?t=187044
Also worth thinking about is that the TL071 requires its input voltages to be at least 4V above the negative supply. That means in this circuit with a +-15V supply, it won't work on signals below -11V. Try to drive it lower and you get phase inversion trouble. Some other op amp might be preferable in this circuit for this reason - but some other op amps may have larger bias current, making that an even more important consideration. br> br> |
|
br>Synthiq |
br>cygmu and mskala are both absolutely right that the input need a dc path. Since most synth circuits are dc coupled I didn't consider a case with a capacitively coupled input signal. My bad. br> br> |
|
br>schenkzoola |
br>It is usually a good idea to put a 100 to 240 ohm resistor between the output of the amp and the load. This helps prevent oscillations due to capacitive loads. If it is a CV output where the absolute voltage is critical, you can put the feedback after the resistor to account for any voltage drop. br> br> |
|
br>Synthiq |
br> schenkzoola wrote: | It is usually a good idea to put a 100 to 240 ohm resistor between the output of the amp and the load. This helps prevent oscillations due to capacitive loads. If it is a CV output where the absolute voltage is critical, you can put the feedback after the resistor to account for any voltage drop. |
This will compensate for any voltage drop over the resistor between the output and the load and maintain an accurate CV voltage, but will also create a 90 degrees phase shift at higher frequencies between the amplifier output and the load capacitance and will also result in oscillatons if the capacitive load is large enough. To prevent this, a capacitive feedback directly from the amplifier output to the inverting input is also included, effectively overriding the phase shifted feedback signal taken from the load capacitance at higher frequencies. br> br> |
|
br>schenkzoola |
br>Perhaps I'm being daft here, but at higher frequencies (presumably audio or higher) does the phase shift really matter for our purposes? I would also expect the capacitive load to be just a few pf due to the cable.
EDIT: Just measured a couple cables. It is about 60pf for an 18" cable, and about 260 for a 72" cable. A little more than I expected. br> br> |
|
br>Synthiq |
br>The phase shift is important because if the total phase shift in the feedback loop is more than 180 degrees when the loop gain is 1, the circuit will oscillate. Practically all amplifiers have more than 90 degrees phase shift internally at unity gain, so adding another 90 degrees from an external RC circuit will push it over the edge if it happens below the unity gain frequency.
A cable I measured had 30pF/ft so it's similar. With just 100ohm in series with the output and a 200pF load the phase shift will only be 45 degrees at 8MHz so in this case it would probably work fine. Increasing the resistance to 1kohm as is common and increasing the load capacitance could however still result in oscillation. br> br> |
|
br>cygmu |
|
br>schenkzoola |
br>Synthiq Thank you for the explanation, cygmu Thank you for the link!
I think I will explore this a little further this weekend. Time to dig out the books! br> br> |
|
br>Grumble |
br>Yeah, opams are peculiar little buggers: if you want them to oscillate they just amplify, but if you want just some amplification they oscillate br> br> |
|