Switched Phantom Power

[Northy]

Hi all, I've been drawn here by me researching Phantom power on google.
I'm looking at hacking switchable phantom power onto a multi input device and have read the great threads on here about series resistors, current rush at switch on etc and it all makes sense. Typically the circuit goes:

48V supply -> Switch -> limiting resistor (100R to 200R ish) -> Bulk el cap (typ 220uF) -> 6k8 resistors

As I said, I've seen the suggestion to put the limiting resistor between the supply and switch, but I was just wondering what the cap was really there to achieve? if the 48V supply is low noise is it really needed?

Thanks,
G

 

[JohnRoberts]

Phantom power is fed directly to microphones that are connected to high gain stages. It is common sense to avoid adding noise before a high gain stage.

JR

 

[emrr]

Different people juggle the series R and the C bucket differently to achieve particular ramp up/down, among other things.

 

[Northy]

Phantom power is fed directly to microphones that are connected to high gain stages. It is common sense to avoid adding noise before a high gain stage.

JR

I agree, but if the 48V supply is 'quiet' anyway, why the big bulk cap at the output?

G

 

[Northy]

Different people juggle the series R and the C bucket differently to achieve particular ramp up/down, among other things.

Ah, ok. The typical value I've seen is 220uF. How do we spec the value/size of the cap properly? What's the trade off here?

G

 

[JohnRoberts]

I agree, but if the 48V supply is 'quiet' anyway, why the big bulk cap at the output?

G

Quiet may not still be quiet after 1,000x gain...

JR

 

[Bo Deadly]

I agree, but if the 48V supply is 'quiet' anyway, why the big bulk cap at the output?

The PCB trace (or wire) supplying the 48V has some minute resistance and inductance that electromagnetic radiation, which is all around us always, can induce tiny currents in. Said resistance / inductance will convert those currents to voltages which is a pedantic way to describe "noise". Similarly, if you have multiple devices connected to the same supply, signals on one device could bleed into the other because of said resistance / inductance being shared by all devices. That capacitor provides a low impedance path to ground to shunt these currents. Without it, I wager you would experience problems.

This can also be described in terms of minimizing loop areas. If that capacitor was not present, AC return currents would have to go through a longer path, possibly all the way through the chassis (since pin 1 is usually connected to the chassis if it's metal) to the filter capacitor at the power supply and then through the supply wire which might be physically distant. This creates a large open loop which is a great antenna for electromagnetic radiation. For this reason, the capacitor should be close to pin 1 of the input and to the junction of the two 6k8 resistors. The smaller the loop from pin 1 through the cap to the 6k8 and on to pins 2, 3 the better.

There might also be an issue with common mode rejection which is the central mechanism by which noise is suppressed in a long balanced cable like a microphone cable. A tiny difference in the two legs can have a significant negative impact on that rejection mechanism. Although the resistance / inductance should be common to both so it's not clear that CMR would be affected if that cap were left out.

At any rate, you should get a sense that there's more to E than just making the connections listed on the schematic. As a general rule, if everyone does something the same way and you don't have a real reason to believe they're doing it wrong, you should probably just do it the same way.

 

[Northy]

Thanks for the great replies.
How would you determine the size of the capacitor?

 

[Bo Deadly]

How would you determine the size of the capacitor?

Ideally it should be very large but in practice it cannot be because it needs to discharge through the resistor which is usually smallish to limit the voltage drop. Usually a DPST switch is wired so that the common terminal feeds the RC but when power is switched off, the other pole goes to ground to discharge the cap backward through R.

So for example, if you used a value like 220R and 47u, that would put 219mA through R and the switch, which is a lot, but at least it would only do it briefly. If you used like 100R and 470u, that would be bad because the switch contacts and R would be subjected to prolonged 480mA. After a while the switch contacts and / or R would probably get burned up. So you have to pick a value that has an RC that gets LF but doesn't stress the switch and resistor.

If it was a high quality NKK switch and the 48V could be adjusted upward enough to compensate for the drop in V from R, I would use maybe 180R / 100u.

 

[Northy]

Thanks for your reply.
Would there be any benefit from a mixture of electrolytic and ceramic here?

Are we really just adding an RC filter right at the output? Is that all it is?