Phantom power and audio noise

[soundguy]

Purely hypothetical idea, but wondering if any has any *real world* experience with this- how critical is the integrity of the 48v for phantom powering? I like to use the 48v rail for relay power and LED illumination, discounting perhaps a click when relays engage, is there any worry of getting any noise in the audio band from perhaps not the cleanest 48v power rail. Ive been building my boxes like this for three years now and havent noticed anything, but was thinking about it today and figured Id ask to see if anyone has ever run into problems doing that. So long as a mic gets the proper voltage for power, I figure its cool, but I occasionally hear people talking about clean 48v as being important. Assuming the grounding scheme in the rack case is good, I cant see how some noise on the 48v power would ever get into the audio band (where the 48v is only used for phantom in this example). Any thoughts? Im feeling this is more scholastic than practical, but would love to hear about a really good reason to not do this.

dave

 

[gyraf]

I've read a lot of hype about how important a clean 48V is.

But in actual tests, I've found that this is very, very unchritical when the input stage has a decent CMRR (read transformer input)

Take e.g. the G9-diy phantom - that is more or less lifted form a ISA110 - is not very low-noise (it's a raw boosted zener), yet both G9 and ISA are very well behaved indeed.

But for sand-state inputs, things may behave different because of poorer CMRR. I haven't experimented with that.

Jakob E.

 

[Scodiddly]

The other thing is that it's not a lot of power... so you could run it through a nice big choke and past some caps without spending much $.

 

[Svart]

i've wondered about this too, but a few things have made me think that it's not as important as some might want you to think.. most(if not all) mics in need of phantom are condenser type mics.. and what are the capsules of the mic in reality.. capacitors.. couple this with the fact that the phantom power is really just creating a potential on the capsule without any serious need for current draw and the "need" for clean power just doesn't justify itself. .. Unless of course you have a *bad* designed mic, in which filtering the phantom is really just masking the real problem.

just my take on the idea.. i've never run into problems with any phantom power DIY or not as long as the mic circuit is good.

 

[Dan Kennedy]

It is an issue, and I was bit early on by it.

Some mics don't have equal impedance outputs, Audio Technica comes to mind, and will convert any common mode noise to differential, meaning anything that flops the main 48v feed around will show up in the signal.

A good mic won't. Well maybe I shouldn't say "good", because I like the sound of some AT's, but I think it was the 4033 that did this. It was also about 8 years ago when I ran into this.

It was noticable enough that the noise of an LM317 without it's adjust pin bypassed increased the noise in the signal by an audible amount. Bypassing that with a 10uF dropped it well below the preamp circuits input noise.

 

[DrFrankencopter]

Dan's right, it's an issue. All you have to do is answer some simple questions, and do some simple math.

1. What's the ripple on your 48V supply?
2. How tightly matched are your 6.8k current limiting resistors?
3. What's the CMRR of your input stage?

Ideally, the phantom power ripple, once you go through these calculations should be at around -130 dBV (i.e. around, or just below the noise floor of the preamp itself).

If say you have 1% matched 6.8K resistors and a not so great input stage you might end up with an effective CMRR of around 40dB. So your ripple noise needs to be down around -90dBV, which is around 30 microvolts or so. That's pretty low. Now, obviously as the CMRR of the preamps goes up, you can get away with more phantom noise, but you need to be aware of how it can bite you in the ass.

For phantom supplies I like to see around 150-300 microvolts of ripple. You can get there with a LM317 and good bypassing/layout.

Cheers.

Kris

 

[Svart]

well again, not bypassing Vreg pins is a blatent disregard for proper design. nothing will work right no matter how loose the bounds are.

So i think what we meant with the topic was supposed to also read: Within reason...

 

[soundguy]

right, within reason. I guess Im trying to understand HOW this could be a problem. Like I said, its been SOP for me to use 48 supply for relays and LED's and Ive never had a problem, but maybe Im just lucky or just have decent mics. I think if I was offering a commercial product, an additional 5v rail for relays and LEDs woudl probably be smart, but diy allows me to be lazy I guess (not the best attitude). Perhaps I shoudl rephrase the question to address my general lack of understanding with electronics:

In what form of noise does a relay or led effect the "cleanliness" of the power rail? Is this solely in the form of additional ripple?

Assuming so, how can ripple modulate the audio from the mic input?

Can this ripple be smoothed with an input cap at the mic input (yeah, I know thats probably a poorly applied band aid) the same way caps are used in a filtering power supply?

thanks

dave

 

[Scodiddly]

Well, I'm curious where you're getting 48vdc relays and LEDs from...

But to answer the question about controlling noise, I imagine that you'd get occasional clicks from relays and that's probably not at all a good thing. What I'd do is split the output of the supply into two rails, phantom vs. control. Putting chokes on either/both would do a lot, as would some capacitance on the phantom side.

 

[soundguy]

aromat makes a 48v relay and I just use a dropping reistor for the LED's.

dave

 

[Svart]

i doubt LEDs will cause too much trouble since they rarely do much more than suckle at the teat of the current source. relays on the other hand will cause you great trouble if you let them. biggest problem you face is the induced spike of voltage and current when you make or break a field on the coil. the spikes can be orders of magnitude bigger than the rail... and the rating of the components on that same rail. sometimes the transient is so fast that you might not notice it. usually i will place a TVS or some equally FAST(no slow silicon!) diode on the coil to try to "flywheel" the current through the coil instead of allowing it to run amok through the rest of the system. while this works great for the power switching stuff i design, I don't have much experience with this and audio but i would think the same concerns apply.

i guess this follows the same rules as most everything else:

make sure you have enough spare capacitance to keep the rail from sagging under the initial load. this can't be said enough! see below...

make sure your parts are protected or rated for the possible spikes( coupling caps might take a number of hits before they die as we know tants will die immediately most of the time and 'lytics will usually die over time in one fashion or another.. Films usually do fairly well, some have a "self healing" property that allows their use in power applications..(ever wonder why the AC input caps on the line-ins of PSUs are expensive film caps..?)

I usually switch my relays using MOSFETS with the integral body diode. this will help kill the inductive kickback too and allows the use of very low current (cheap..) switches and smaller wires.


on the contact side: contact bounce and dirty contacts can make life miserable. this applies for EVERY relay no matter how expensive. you want to switch that relay ON as hard and fast as you can. fast and hard will minimize bounce times. if you listen to your "click" is it a solid "click" or more like a "clack"? this is untechnical i know but your ear is great for things like this if you don't want to sit there and try to see this on a storage scope or something. This also pertains to the voltage rail's ability. the more sag, the more time during actuation and the more bounce. I doubt starving your condenser mic of it's 48v for a couple nanoseconds will really do much to it, but watch that current/voltage spike...

I guess it also depends on what you are switching too. Switching power NEEDS a snubber no questions asked. this is usually a series RC across the output contacts. I could go into graphic parts-explosion detail about the lack of snubbers but that'll have to wait. Switching audio will likely need something similar, maybe a coupling cap similar to the interstage caps we see so often.

or...

you only switch your relays when clicking and popping are allowed, like between takes or something then none of this really matters..



:green:

 

[PRR]

> what are the capsules of the mic in reality.. capacitors.. couple this with the fact that the phantom power is really just creating a potential on the capsule without any serious need for current draw and the "need" for clean power just doesn't justify itself

The capsule is "a capacitor" but a VERY small cap. It isn't a filter cap. And the mike sensitivity varies directly with bias voltage: unfiltered bias voltage WILL screw-up the output. Fortunately the bias current is about zero, so trivial filtering works fine. (though these days some "budget" mikes can't even cover trivial filters...)

And the amplifier may suck significant current. Ideally it rejects ripple but.....

> 1. What's the ripple on your 48V supply? 2. How tightly matched are your 6.8k current limiting resistors? 3. What's the CMRR of your input stage?

AND what Dan says: what does the mike output look like? It may attenuate 68:1 against the 6K8 resistors. Or it may attenuate 68:1 on one pin, 34:1 on the other pin, and generate substantial differential output that even a hi-CMRR input won't reject.

FWIW, I have run good mikes very critically on un-regulated supplies. 24VAC, rectifier, big cap, resistor, 'nother big cap. Like 4,700uFd and 470Ω. Switching to another more sophisticated mike-amp did not change stray noise, so I don't think my low-tech supply hurt. (True, I knew the mikes were totally floating transformer-out: I paid more for that.)

LEDs and relays turned on and off will kick the "48V" down and up, send clicks or thumps toward your mikes. If you have even a little R-C between the bulk 48V and the Phantom bus, that may not be a problem, especially if you don't switch relays at critical moments. LEDs just pull-down the 48V; if it drops 48.0V to 47.9V then you have a 0.1V click on the Phantom bus and much less at the mike. Relays do that, harder, but also kick-up when turned off. Use diode-clamps even if not needed to protect the switching, and have good caps on the 48V, and an R-C section beterrn dirty 48V and Phantom bus.