Topic: Are preamps for pu$$ies?

[chris319]

looking at your card's specs for balanced input it states the maximum level is +20dBu (+18dBV), so if you had the card set to balanced, that should put the noise floor at -102dBA.  I don't know if it's possible to set the card for balanced and -10dBV nominal

+4 and -10 are the only options. My preamp has an unbalanced output.

Let's say the preamp's noise floor is -120 dBV and the sound card's noise floor is -114 dBV. That gives us a 6 dB difference, plus the 6 dB you add for summing, for 12 dB of amplification.

I honestly don't know the S/N of my preamp, but it is based on this chip:

http://www.national.com/ds/LM/LME49720.pdf

Scroll way down to the balanced input mic amp schematic. I am using just the two opamps on the left. It just occurred to me that I could give it a balanced output if I didn't ground the output of the bottom opamp.

[jerryfreak]

by my experience, internal computer soundcards are rarely capable of exceeding 90-100dB signal/noise due to the large amount of rf noise typically found in a computer case, i'm guessing you have nowhere near the dynamic range they quote with that setup. a quick peek at the emu specs for that card show that they do quote 120 dB dynamic range, but i find it hard to beleive they actually got that unless on a test bench and not inside a computer chassis.

even the ad2k, which is designed all around a philosophy of ultra-low noise and distortion, still has a slightly higher noise floor than my mics. Keep in mind that unit is solely a DA and costs over 10x what the emu card does, youre asking a lot of that emu card. i dont think preamps can be eliminated in ALL cases.

looking at your card's specs for balanced input it states the maximum level is +20dBu (+18dBV), so if you had the card set to balanced, that should put the noise floor at -102dBA.  I don't know if it's possible to set the card for balanced and -10dBV nominal

+4 and -10 are the only options. My preamp has an unbalanced output.

Let's say the preamp's noise floor is -120 dBV and the sound card's noise floor is -114 dBV. That gives us a 6 dB difference, plus the 6 dB you add for summing, for 12 dB of amplification.

I honestly don't know the S/N of my preamp, but it is based on this chip:

http://www.national.com/ds/LM/LME49720.pdf

Scroll way down to the balanced input mic amp schematic. I am using just the two opamps on the left. It just occurred to me that I could give it a balanced output if I didn't ground the output of the bottom opamp.

[chris319]

I am basing my gain calculations on human speech, which is typically 94 dB SPL at a distance of one inch (63 dB SPL at three feet). Conveniently this is equal to one Pascal. The output of a typical dynamic mic is -54 dBV at one Pascal. At two inches human speech is -60 dBV, at four inches it's -66 dBV, at eight inches -72 dBV. If my mic preamp is set to 30 dB of gain, it is delivering -42 dBV to the sound card, not -10 or +8 dBV. That's why I said you have gobs of headroom.

if you are just running the two amps on the left, you don't have a balanced circuit at all

Balanced input circuit or output circuit? With just the two on the left I'm giving the mic a balanced input, aren't I? Would the addition of the third opamp justify the attendant noise and distortion of a second stage? I'm OK with running an unbalanced output if it will result in lower noise, THD, etc.

But even at -126dBA, that's a good reason to use more preamp gain.

I'm doing listening tests as I build this out. Using only the two opamps on the left, I've got a 1.5K resistor in place of the 200 ohm resistor at R1. I haven't done an actual measurement yet but I suspect the gain is somewhat above 20 dB.

[chris319]

I would take a balanced input over the tiny difference in distortion any day of the week.

A balanced mic input, of course, or you'll pick up the nearest AM station. Did you mean balanced output? BTW this circuit replaces one with a THAT 1510 chip but has a nicer square wave response.

The usual approach is to use a pot (with a resistor in series to limit maximum gain) or rotary switch in place of R1, so you have gain control.

I can do that, but I'm deciding whether I want a small amount of fixed gain, say 20 dB, or a variable amount, say 20 - 40 dB, and normalize. The gist of this thread is to use a small amount of preamp gain, giving a large amount of headroom, and bring up the levels digitally. IOW what is the smallest amount of preamp gain I can get away with and still get good results?

[chris319]

How would this work: use a LME49740 which has four opamps on one chip, giving me two amps on the "right" side instead of one, and make a balanced output out of them? Then make R1 a pot variable from 20 to 40 dB gain? Or unity to 40 dB? If more gain is needed, normalize.

This has been very instructive, thanks.

[chris319]

use an impedance balanced output (30 ohm or so resistor from pin 3 to ground)

You're saying make R7 33 ohms.

[chris319]

I know it's an incomplete schematic. You shouldn't make inferences about the THAT 1510 preamp from the LME49720 data sheet schematic you are looking at. The 1510 is all built and has phantom power protection, RF bypass caps, an output resistor, a split power supply (two 9-volt batteries) and an unbalanced output by design. See figure 4 of the THAT 1510 data sheet for an idea. It does not provide phantom power by design -- that has to be provided by an external supply. It works great and has served me well for a couple of years now. The new preamp will have all of those things added to it which you don't see in the incomplete schematic on the data sheet. Last night I built the output stage adding two more opamps in a balanced configuration. It is basically a carbon copy of the left half except there is a 10K resistor where R1 is. So far it tests great, but needs to have the caps and diodes added to it.

[jerryfreak]

ms, is there a tiny circuit than can take an unbalanced signal from the cmr and simply convert it to balanced?

It is basically a carbon copy of the left half except there is a 10K resistor where R1 is. So far it tests great, but needs to have the caps and diodes added to it.

That will still depend on a differential amp in your converter.  In other words, if you plug that into an unbalanced input, the entire signal chain will be unbalanced.

Copying the left buffer amps again will just add gain (6dB with 10K, I believe) and slightly degrade CMRR.

What you probably want to do is build the schemo as shown in the datasheet, but feed its output to pin 2 out AND the negative input of an opamp in inverting configuration, with that opamp's output to pin 3.

Good basic opamp page:

http://www.bcae1.com/opamp.htm

[jerryfreak]

CMR?  You lost me there

i was referring to the unbalanced schoeps cmr amplifier body, which is where this thread all began

http://www.schoeps.de/E-2004/cmr.html

[Scooter]

similar schems here kindly drawn up by mshilarious...

http://taperssection.com/index.php/topic,102355.0.html

i gathered all the parts to make it, but then I got a good deal on some PM-4's, so it never happened.  This version doesn't use transformers though, so I still might cobble it together at some point as I have a hunch it would be more transparent.

[jerryfreak]

yeah, that was me, but i dont need to use phantom, just wanted to know if there was an adapter circuit that could be built and used close to the mics to allow for longer cable runs

Oh yeah!  There was another thread on the mic board where I talked about that.  In fact I think I built such a device for somebody, but I forgot who it was, and I don't recall if they reported back on how it worked.

I think I posted this somewhere too:

[Keyd]

I think normalising is much more pu$$y.

[jerryfreak]

you mean 3 conductor cable, right?

so, bear with me, this is explaining the cabling running backwards counter to teh aactual signal flow:

from device with balanced input (ad2k/v3/etc), run normal 3 conductors up to the mic, then at the mic, terminate the (-) balanced leg from the cable to ground thru a resistor matching the output impedance of the mic.

thus the mic itself is not balanced, but the cable leading to the mic will be balanced, and reject most all of the inducted noise. again, i will see some rejection, but probably not the full cmrr due to matching.

that said,the mic is spec'd at 15 ohms at 1 khz. im sure its an approximation of sorts should i get a batch of 15 ohm resistors and try the various values with a long mic cable and do some sort of simple test to see how much noise results?

yeah, that was me, but i dont need to use phantom, just wanted to know if there was an adapter circuit that could be built and used close to the mics to allow for longer cable runs

Yes, just use 2-conductor cable, and tie the unused lead across a 30 ohm resistor (or whatever the output impedance of the mic is) to the cable shield on the mic side of the cable.

You will need a differential (balanced input) amp at the other side of the cable to take advantage of that though.

[jerryfreak]

the 100 ohm pot sound slike the way to do it.

should i use like 100+feet of mic cable and how should i 'look' at the noise, i dont have a scope

you mean 3 conductor cable, right?

2 conductor plus shield, your standard basic mic cable.

so, bear with me, this is explaining the cabling running backwards counter to teh aactual signal flow:

from device with balanced input (ad2k/v3/etc), run normal 3 conductors up to the mic, then at the mic, terminate the (-) balanced leg from the cable to ground thru a resistor matching the output impedance of the mic.

Yes.

that said,the mic is spec'd at 15 ohms at 1 khz. im sure its an approximation of sorts should i get a batch of 15 ohm resistors and try the various values with a long mic cable and do some sort of simple test to see how much noise results?

Put a 100 ohm pot in the circuit and turn it until the noise is minimal.  Then disconnect it and measure its resistance.

The other way to do it is to measure the actual output impedance.  There are a few ways to do that.  I have a variable input impedance preamp, so I just measure the signal difference at two different input impedance settings, and calculate the impedance from there.

Or just use 15 ohms . . . I always buy 1% tolerance resistors, so I don't think trying out a bunch of different 15 ohm resistors will help too much.  It's more of a issue of whether you buy 18 ohm or 15 ohm or whatever.

[chris319]

- an impedance-balanced line.  Figure out the output impedance of your device, and match it with a resistor from pin 3 to ground.

What becomes of R7? Pin 3 is the non-inverting input of the final opamp. You've already got a 10K resistor (R7) between it and ground.