I have built this circuit: Project 103 - Subwoofer Phase Controller
Now it seems as if the invert switch works as an antenna when it is open and picking up disturbance, atleast with the board still on the desk.
This is a tricky case since the "antenna" is directly on the opamp input and there is 22k+47k to ground. Do you still think it could be better with board+switch mounted in a metal case?
I am using a microphone cable with screen grounded.
Now it seems as if the invert switch works as an antenna when it is open and picking up disturbance, atleast with the board still on the desk.
This is a tricky case since the "antenna" is directly on the opamp input and there is 22k+47k to ground. Do you still think it could be better with board+switch mounted in a metal case?
I am using a microphone cable with screen grounded.
Are all loops closed up by twisting all wire pairs tightly? You have to keep both wires of each pair together, everywhere, or the the resulting loop will be an antenna.
The switch wiring should be placed very close to R1 and R2 and their connection, all the way from one end to the other. Otherwise you get enclosed loop area, and an antenna.
The switch wiring should be placed very close to R1 and R2 and their connection, all the way from one end to the other. Otherwise you get enclosed loop area, and an antenna.
Last edited:
Sorry. I did not explain clearly-enough.
Whichever two conductors go to the switch, they should be twisted tightly together, all the way from the board to the switch.
But there is a little more to it. The input and the ground ALREADY need to be twisted together, ALL the way to the top of R1, from which the ground follows the body of R1 to its other end. (Or, maybe you have a ground plane on a PCB from where the twisted input signal and input ground enter the copper, with the ground plane under at least the area where R1 is.) Also, the loop that R2 and R3 are part of will be "collapsed", compared to the way the schematic is drawn, probably with R2 and R3 against each other, and very close to the IC pins and R1.
If there is a ground plane, the ground connection for the switch can originate right next to the other switch connection, near R2. Otherwise, simply run one of the two switch wires from where it connects to the circuit, along the bodies of R2 and R1, to where the other switch wire connects to the circuit, and then twist them together from there to the switch. That way there won't be much "geometric area enclosed by a loop", so your antennas will not be as good, which is what you want.
Of course, you would also break the "ground" conductor shown in the schematic, so that the input ground reference (bottom of R1 and bottom of switch) is NOT connected to the bottom of C3, or anything to its right in the schematic, except at the star ground point, which will probably be just after the last power supply smoothing cap. [The two grounds should each have their own separate conductor that goes back to the star ground. Otherwise, the currents from the decoupling caps and output will induce voltages across the ground-return conductor that will appear at the non-ground end. If those voltages were allowed to be at the bottom of R1, they would be arithmetically summed with the input voltage, which you do not want.]
Also, keep the feedback loops as short as possible. And always connect all components as close as possible to IC pins.
l would also parallel the 0.1 uF X7R ceramic caps at the power supply pins each with a standard 10 uF electrolytic.
Whichever two conductors go to the switch, they should be twisted tightly together, all the way from the board to the switch.
But there is a little more to it. The input and the ground ALREADY need to be twisted together, ALL the way to the top of R1, from which the ground follows the body of R1 to its other end. (Or, maybe you have a ground plane on a PCB from where the twisted input signal and input ground enter the copper, with the ground plane under at least the area where R1 is.) Also, the loop that R2 and R3 are part of will be "collapsed", compared to the way the schematic is drawn, probably with R2 and R3 against each other, and very close to the IC pins and R1.
If there is a ground plane, the ground connection for the switch can originate right next to the other switch connection, near R2. Otherwise, simply run one of the two switch wires from where it connects to the circuit, along the bodies of R2 and R1, to where the other switch wire connects to the circuit, and then twist them together from there to the switch. That way there won't be much "geometric area enclosed by a loop", so your antennas will not be as good, which is what you want.
Of course, you would also break the "ground" conductor shown in the schematic, so that the input ground reference (bottom of R1 and bottom of switch) is NOT connected to the bottom of C3, or anything to its right in the schematic, except at the star ground point, which will probably be just after the last power supply smoothing cap. [The two grounds should each have their own separate conductor that goes back to the star ground. Otherwise, the currents from the decoupling caps and output will induce voltages across the ground-return conductor that will appear at the non-ground end. If those voltages were allowed to be at the bottom of R1, they would be arithmetically summed with the input voltage, which you do not want.]
Also, keep the feedback loops as short as possible. And always connect all components as close as possible to IC pins.
l would also parallel the 0.1 uF X7R ceramic caps at the power supply pins each with a standard 10 uF electrolytic.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.