Are the component values on the schematic correct?

The preamp gain seems awfully high.

Posted at 8:44 am on Jun 1st, 2009 by ac7zl

I agree !

1000x is way way too high for any application at all.

Posted at 9:30 am on Jun 1st, 2009 by ino

You’re right caleb, binaural recording refers to creating recordings that simulate the way we hear, they generally include two microphones mounted in a dummy head.

As for the project, his recording quality is terrible. I’m also leery of his preamp circuit, the component values seem a bit odd to me.

Posted at 9:34 am on Jun 1st, 2009 by dirk

also worth noting that placing the microphone capsules behind holes of smaller diameter than the microphone itself will definitely alter the frequency response. A smaller hole will add as much as several dB to the high end of a signal, this might be a source of noise as well.

Posted at 10:01 am on Jun 1st, 2009 by dirk

Just a few notes on the comments so far:
- Yes, the component values for the circuit are correct. The circuit doesn’t actually produce a gain that high due to the crappiness of the LM358 and the fact that the electret condensors are not perfect voltage sources.
- The recording quality was not meant to be perfect, just good enough for voice. The project was done on a budget, and it was simply a mod to the already average onboard sound. Had I wanted better quality audio, I’d probably have transplanted a good quality USB ADC into the Wind, and used decent, audio quality electrets and opamps.
- The binaural effect is certainly enhanced by adding a barrier between the microphones, but a large part of the effect is due to the time difference between sounds reaching each ear.

Posted at 10:19 am on Jun 1st, 2009 by Matt

@dirk i totally built one of those for holophonic recording (stereo mics inside a manikin/dummy head). Works quite well actually. I won’t spam a link though, follow sig.

.dok

Posted at 10:59 am on Jun 1st, 2009 by Addictronics

And here I thought that wind was bad for microphones.

I got a little confused, maybe I should catch up on some sleep.

Posted at 2:43 pm on Jun 1st, 2009 by nubie

he could reduce teh noise by software, almost every realtek audio driver has a noise filter (which atually sucks but works,atleast for phonecalls)

Posted at 3:10 pm on Jun 1st, 2009 by arrangemonk

Matt said:

>Yes, the component values for the circuit are
> correct. The circuit doesn’t actually produce a
> gain that high due to the crappiness of the >LM358 and the fact that the electret condensors
>are not perfect voltage sources.

Could you please elaborate on your comments?

a) I took a look at the data sheet for your opamps…I’m not sure what you mean by the “crappiness” factor and how that would effect gain. Theoretical gain should be on the order of r4/r1 or 1000x.

If it does not produce gain that high, I can think of a few reasons. First, gain will be somewhat less for lower frequencies because C1 is rather small, and its impedance rises as frequency drops. Second, you’ve biased the op amp to 2.5 volts, and the opamp will saturate as the output approachs the supply rail. With a gain of 1000, that’s easy to do, as less than 2.5 mV on the input will cause clipping. The apparent gain will drop for any input larger than that.

b) Does your comment about “crappiness” have to do with the op amp’s gain-bandwidth? I still don’t understand why you need a gain of 1000.

c) Electret mike capsules contain a fet that tries to pull their “output” to ground. This current to ground is modulated by the audio hitting the mike. This causes a corresponding voltage drop across R11, which likewise varies with the audio. What do you mean when you say your mikes are imperfect “voltage sources?”

d)… be *very* careful with the conclusions you draw from Spice. Spice simulations are not the same as living in the real world. Since Spice is only as good as the assumptions baked into the models, it is possible for a circuit to look great in simulation and fail in the real world. In fact, it is possible to build fully “functional” nonsense-circuits in Spice.

Overall, good idea, nice project, very nice circuit board and nice mechanical implementation. The circuit needs some more scrutiny, I think.

Posted at 5:48 pm on Jun 1st, 2009 by curious george

embed the mics in foam and it will kill much of the noise from the netbook itself.

Simply gluing a mic to something means you will be transmitting all the vibrations that object experiences to that microphone.

Done properly, the mic(s) will “float” in the foam, dampening surrounding vibrations.

-and this isn’t binaural, it’s just stereo.

Stereo mics in a netbook is impressive enough without needless inaccurate jargon.

I don’t know enough about circuit design to make any judgments on your design, but referencing it against known successful preamplifier designs would have been a great way to see if you are on track.

Personally I would have just lifted someone else’s preamp circuit, but i’m lazy that way.

regardless, it’s neat!

Posted at 5:07 am on Jun 2nd, 2009 by strider_mt2k

Thanks to all of those that have commented. I have a few followup comments:
- The mics are mounted on foam tape.
- The holes in the front bezel are much larger than the hole in the case of the electret microphones.
- The mod is not truly binaural, thanks to those who have pointed it out. I’ve changed the title of the post on my travel blog.
- The circuit was loosely based off the one here: http://www.reconnsworld.com/audio_singleicpreamp.html and a few others that I found on the net. The final values of the feedback and input resistors were produced from a mixture of SPICE simulations and real world testing. I tried a number of resistors to produce different gains. The 1M/1K combination was about as high as I could go before the amp started saturating the ADC (or hitting the power rails, I’m not sure which).
- Simulating the circuit with a JFET in place of the voltage source significantly lowers the peak gain.
- The LM358 is fairly low bandwidth. The high frequency rolloff is soley due to the opamp. If it is replaced with a Burr-Brown or similar, it does not have this rolloff.

Posted at 6:05 am on Jun 2nd, 2009 by Matt

Turn off your MIDI synth chip. The FM synth is like a hissy radio station inside the computer. Helps take a lot of noise out of recordings. Also I would worry more about the freq response of the mics. Storebought/repurposed elements aren’t always top notch. Anyhoo, good luck.

Posted at 7:13 am on Jun 2nd, 2009 by coldwar23