New Poster, looking for critique of my mixer..

[JLavoie]

Hi All,

I'm looking for some help with my mixer.
I know this type of circuit has been done to death, but I have a specific need that has to fit into a specific box so nothing off the shelf will do.

Basically, it's a mic/line mixer (think SM58 and iPod) with a link input to mix other units in. so think of an installed location, with 12VDC fed down a pair of wires, and line level audio back (about 200'+) then the same thing to unit#2 (sometimes to a 3rd, but rarely)

the circuit as shown here http://www.twin-x.com/groupdiy/index.php?cat=10770 is working well on the breadboard but I'm hoping for pointers on any component values that could be optimized.
Also any tips on how to avoid ground loop hums (due to the long lengths of wire and varying grounds between different parts of buildings)
And guidelines for minimizing cellphone interference.

a lot of this is hard or impossible to test on a breadboard so hopefully you can steer me in the right direction before I do my PCB layout.
I'm hoping to go all SMT (or as much as possible) which will give me most of the bottom side of the PCB as a ground plane, which I'm hoping will go a long way to avoid interference.

Thanks in advance!

Jason

 

[JohnRoberts]

One obvious option for lowering input noise is to use a dedicated mic preamp IC.  Something like the THAT corp 1570 without all the pads and phantom power poop would not add a lot of complexity to the design, and get the mic preamp noise down a few more dB.

http://www.thatcorp.com/datashts/dn140.pdf

JR

 

[JLavoie]

A proper preamp chip is a possibility, but I wasn't sure if my 12V (after 400' of wire) would be enough to provide the +/-5V that the THAT chips spec as a minimum (and not sure if I'm comfy being that close to the minimum)
Also, at more than twice the cost per chip, and half the devices per chip that adds cost and layout space.
If a dedicated preamp chip is likely to be more resistant to cell interference it may be worth it, but something tells me that will have more to do with layout, shielding, and filtering than the IC choice.

I have some sample THAT1512's that I might try if my first prototype round doesn't live up to expectations.

Jason

 

[abbey road d enfer]

Apart from John's recommandations, I would increase the value of the 10pF caps to at least 100p and install ferrite beads on the mic input, if you're worried about cell phone interference. You may also have to investigate opamps. FET opamps are notoriously less sensitive to RF interference than bipolars.

 

[JLavoie]

Caps, done..
Ferrite beads, sounds like a good idea. any hints on a starting value?
I assume they go in series first thing after the input connector.. would I put one on pin1 as well?

is there any tradeoff to moving to FET opamps? sounds like a no-brainer, but there must be a catch

for what it's worth, the circuit I'm trying to replace picks up cell noise loud and clear even with the volume controls turned down.. so it's either coming in on the power, or some other weak circuit after the volume controls.
should I be putting ferrite beads on the power inlet?

Jason

 

[JohnRoberts]

Caps, done..
Ferrite beads, sounds like a good idea. any hints on a starting value?

FB are good for adding some impedance at RF but still being low impedance at audio frequency for low noise. In your application if not chasing the last dB of noise floor an input RC will get you there OK. When sizing a FB you can look at impedance vs frequency charts to help dial it in to reject a given source of Rf, you also want to size it to handle the path current, but for a mic input the current will be very low.

I assume they go in series first thing after the input connector.. would I put one on pin1 as well?

no, XLR pin 1 is hard bonded to chassis.

is there any tradeoff to moving to FET opamps? sounds like a no-brainer, but there must be a catch

In general FET opamps will be higher input noise voltage than bipolar opamps. You can get some pretty good modern premium JFET input opamps, but you will pay to get low noise equivalent to a decent bipolar.

for what it's worth, the circuit I'm trying to replace picks up cell noise loud and clear even with the volume controls turned down.. so it's either coming in on the power, or some other weak circuit after the volume controls.
should I be putting ferrite beads on the power inlet?

Jason

It couldn't hurt... any unshielded wire connected to the box will be an antenna.  If cellphone noise is a known problem, you can bench test the design for that.  In general you want to passively filter out RF before it hits any active stages to prevent rectification, in active stages that aren't fast enough to follow the RF signal slew rate.

JR

 

[JohnRoberts]

A proper preamp chip is a possibility, but I wasn't sure if my 12V (after 400' of wire) would be enough to provide the +/-5V that the THAT chips spec as a minimum (and not sure if I'm comfy being that close to the minimum)
Also, at more than twice the cost per chip, and half the devices per chip that adds cost and layout space.
If a dedicated preamp chip is likely to be more resistant to cell interference it may be worth it, but something tells me that will have more to do with layout, shielding, and filtering than the IC choice.

I have some sample THAT1512's that I might try if my first prototype round doesn't live up to expectations.

Jason

yup  1510/12 is less parts.  I suspect the +/- 6V is not a stopper but this may be more performance than you need or are willing to pay for.  If you have parts laying around you can bread board one up and answer for yourself.

FWIW you can get some pretty low noise single opamps, while the topology of the THAT IC mimics better discrete or hybrid mic preamps in less space and parts insertions. So will always beat simple single opamp solution. 

My guess is you may be satisfied for your application with a decent low noise bipolar opamp, with adequate input RF filtering etc. One way to find out.

JR

 

[PRR]

> how to avoid ground loop hums (due to the long lengths of wire and varying grounds between different parts of buildings)

Don't ground the user stations. Use plastic box. Or whatever box but DON'T connect to it. (This does assume the buildings "share a ground", however noisy. Even the 4V AC I had between stage and booth is fine. If the power distribution is much more erratic, dozens of volts of ground differences, then there are safety issues.)

> hard or impossible to test on a breadboard

When RCA developed telemetry, they rented two phone lines from shop up into the hills, and cross-connected the far end. Then they could test with 7 miles of bad wire all on one bench. When I did something like what you are doing, I ran 100 feet of wire under all the fluorescent lamps in the room and back over variac, dimmer, projector etc. to my bench.

> I'm hoping to go all SMT... most of the bottom side of the PCB as a ground

That's not a cure-all. And we did a LOT of quiet work on terminal strips.

> I wasn't sure if my 12V (after 400' of wire) would be enough to provide the +/-5V that the THAT chips spec as a minimum

You could do math. Be less un-sure.

What do you have, a single 12V supply? That's awkward for this chip, but follow this thought and see where it goes.

So you can have up to 2V line drop. Two wires, 1V each way.

The 1570 draws up to 10mA idle at 30V total supply. I suspect it will be similar down near 10V. There is a dynamic load but you may not have much line-load; IAC it will be under 3mA.

Maximum 1-way line resistance is 1V/10mA= 100 ohms.

Line length is 200 feet.

Right away you can guess that the cheapest stage-speaker cable (100 feet onto 8 ohms with something coming out) will be good enough. Can we be cheaper?

Wire tables sometimes quote ohms per 1,000 feet. We want under 100 ohm in 200' so under 500 ohms in 1000'.

In US gauges, #37 fails and #36 is 432 ohms/1000' so will work.

In fact #36 is way too fine to leave the box, you must use #28 or more to be as tough as phone or network cable. 200' of #28 is 14 ohms, 0.14V 1-way drop at 10mA, 0.3V 2-way drop, six times better than requirement. 

> minimizing cellphone interference.

I think your best bet here is a TRANSFORMER.

This may also simplify getting good results from that IMHO puny 12V supply.

Since you have not asked for fidelity (and cite SM58), it can be inexpensive. Radio Shack XLR Jack-to-1/4" Plug Adapter/Transformer Catalog #274-016 has the XLR and a 1:7 step-up all for $19.69. It may be droopy by 17KHz, but that's probably what you need.

 

[PRR]

.

 

[JLavoie]

> how to avoid ground loop hums (due to the long lengths of wire and varying grounds between different parts of buildings)

Don't ground the user stations. Use plastic box. Or whatever box but DON'T connect to it. (This does assume the buildings "share a ground", however noisy. Even the 4V AC I had between stage and booth is fine. If the power distribution is much more erratic, dozens of volts of ground differences, then there are safety issues.)

grounding the stations is preferable so that when someone plugs in a laptop or other source that is locally grounded we won't have any issues.
we've gone the plastic box route in the past to avoid ground loop hums only to have the hum come back when someone plugs in a wireless mic receiver or something with a local power supply.

Jason

 

[Speedskater]

CAT-3 cable is almost free, use one pair for the input, use one pair for the output, use two pairs for the power. Make everything balanced input and output.

 

[JLavoie]

@PRR:
that's a cute little circuit, but I think I'd like to stick with balanced connections as much as possible.
I'll definitely be keeping that one in mind if I need something simpler and more local.

@Speedskater:
we have hundreds of installations out there with 2-pair shielded that we need to be compatible with, but you've got me thinking.. maybe I could make a version with RJ45 connectors.. then back at the rack I could have a breakout board to feed power to multiple zones.
I think you're confused about the physical layout of these things though..
it would go [DSP]----200'-----[station 1]----200'----[station 2]

I've never done balanced audio over cat5 without a transformer, anyone have experience with this? should be fine right?

Jason

 

[JLavoie]

This may also simplify getting good results from that IMHO puny 12V supply.

this has got me thinking..
if 12V is enough to power the chip, why is it puny?
I mean, if the target output is only 1V p-p that still leaves tons of headroom. what would be the advantage of moving up to say 24V?
the way I see it the function of my circuit wouldn't change at all. or would it?

Jason

 

[JohnRoberts]

This may also simplify getting good results from that IMHO puny 12V supply.

this has got me thinking..
if 12V is enough to power the chip, why is it puny?
I mean, if the target output is only 1V p-p that still leaves tons of headroom. what would be the advantage of moving up to say 24V?
the way I see it the function of my circuit wouldn't change at all. or would it?

Jason

Back in the ugly old days it was useful to increase the power supply voltage and peak to peak signal swing to keep the signal adequately above the noise floor. These days the noise floor is much lower, so this isn't as necessary inside the box. In fact they make respectable 20 bit A/D chips inside 5V ICs. 

But we also need to look at outside the box. Hot signal swing is useful to keep the signal well above the hum and sundry noise interference trying to corrupt the signal when sent over distance.

Doubling your power supply to 24V gives you 6 dB more signal swing "capability". This extra 6 dB will only improve S/N between your box and where the signal is going if you increase the nominal interface level. However, it also factors into headroom or clipping point.

A 1Vp-p signal from a 12V supply, after we account for the opamps not swinging all the way to the rail will still deliver close to 20dB headroom. Using differential output drivers means you actually get another 6 db of headroom, so on paper the 12V rail looks OK, if 1V p-p is as hot as you ever need to push it.

It is pretty common for professional gear to use +/-15 to +/- 18V rails so they would have approximately 10 dB more headroom than your 12V rail.

The only scenario I can imagine where you would really want the stronger rails, is a very noisy, lousy ground venue, where you push up the voltage coming from the distant mixer and pad it down at the receiver end to reduce channel noise.

JR 

 

[JLavoie]

I'm having second thoughts about my 100k pots.. I chose them because they were the only ones available in a conductive plastic log version that fit my form factor, but something just doesn't seem right about a 100k pot into 22k mixing resistors. Is there a standard ratio for that?
I've seen schematics on the web with values all over the map.

Jason

 

[JohnRoberts]

It won't cause any audio problems, but the 22k resistors will bend the taper of the pots some near the top of the range.

If concerned and you want to stay with 100k pots, you could scale up the impedance of 22k resistors (and r1 a like amount) to reduce the loading effect. Since you are dealing with near line levels at that point, the self noise of even 100k resistors will be insignificant.

JR

 

[JLavoie]

Thanks.
If I special ordered pots what would be the ideal value?

Jason

 

[JohnRoberts]

I generally standardized on 10kA for audio level pots.  It isn't very critical.

JR

 

[JLavoie]

no, XLR pin 1 is hard bonded to chassis.

What about the input bypass caps? Should they go with pin1 to chassis, or to circuit ground?
Part of me says put them to chassis to drain away rf noise without affecting the circuit, but then I'm thinking that they'll provide a path for rf noise on pin1 or the chassis back into the mic line if the chassis isn't perfectly grounded.

Jason

 

[JohnRoberts]

no, XLR pin 1 is hard bonded to chassis.

What about the input bypass caps? Should they go with pin1 to chassis, or to circuit ground?
Part of me says put them to chassis to drain away rf noise without affecting the circuit, but then I'm thinking that they'll provide a path for rf noise on pin1 or the chassis back into the mic line if the chassis isn't perfectly grounded.

Jason

If the caps are connected directly to pins 2/3 of XLR, ground them to pin 1. There already is capacitance between 2/3 to the shield from the cable, so additional caps from 2/3 to audio ground would indeed look like caps in series directly dumping shield current into the audio ground at HF.

If there is a compliance between pins 2/3 and audio circuitry, like small series resistors or ferrite beads (inductors), the caps can be connected to local audio ground to form a LPF. For very difficult RF environment perhaps a little of both.

JR