Dynamic mic pre-amp into sound card

[systemloc]

It isn't an opamp. It is a power amp. It is pretty noisy if it has a gain more than 20.

Well, the packaging used an op amp symbol within the DIP, but I understand the difference. I also snagged a 741 op amp while I was there too. If I'm going for more than 20x gain, would building a circuit around the 741 be any less noisy than the LM386? How much clean gain should I expect from the 741? Perhaps a greater amount of gain could be achieved by running the signal through both the LM386 and 741 in series, or perhaps two of the same?

For something completely different, does anyone know of a cheap local source of small tin boxes to house little projects like this? RadioShack has plastic boxes, but I'm thinking this really should be shielded by using a metal container, right?

TJ

 

[audioguru]

An old, old, old 741 opamp is one of the 1st and one of the noisiest ever made. Its max frequency is only 9kHz at full output.

A preamp needs a low noise wide bandwidth audio opamp like a TL071 or one much better.

To amplify a 1mV mic to 1V, you need a voltage gain of 1000. Even a wide bandwidth TL071 has its response drop above only 3kHz with such a high gain, so a TL072 dual opamp could be used with each opamp in series with a gain of 32 for a response that is flat to 100kHz.

 

[Hero999]

Are you sure? Wouldn't that oscillate?

Using two op-amps in series would double the phase shift as well as the gain, so you're sure to get oscillation.

 

[audioguru]

Are you sure? Wouldn't that oscillate?

Using two op-amps in series would double the phase shift as well as the gain, so you're sure to get oscillation.

They will be completely separate and won't oscillate. Each opamp will have a gain of 32.

If the negative feedback was around both of them in series then they will oscillate.

 

[systemloc]

An old, old, old 741 opamp is one of the 1st and one of the noisiest ever made. Its max frequency is only 9kHz at full output.

A preamp needs a low noise wide bandwidth audio opamp like a TL071 or one much better.

To amplify a 1mV mic to 1V, you need a voltage gain of 1000. Even a wide bandwidth TL071 has its response drop above only 3kHz with such a high gain, so a TL072 dual opamp could be used with each opamp in series with a gain of 32 for a response that is flat to 100kHz.

Ahhhh, after taking a look at the spec sheet for the 741 and the TL071, I see what you mean! The bandwidth is much wider, the THD is two orders of magnitude lower and the slew rate is incredibly better for the TL071. I'm learning a lot on this board. I'm pretty sure I can snag a TL082, which looks as if it would work the same as a TL072, so I can try building both in series.


I'm still pretty unclear on how the 0.5uF cap and 10ohm resistor on the output prevent an LM386 from oscillating. Using google, I found a lot of info on how oscillation can start and how to prevent it, but not with this particular method. I'm guessing the point is to provide a guaranteed minimum amount of AC current leaks off in the presence of a high load resistance?

TJ

 

[Nigel Goodwin]

I'm still pretty unclear on how the 0.5uF cap and 10ohm resistor on the output prevent an LM386 from oscillating. Using google, I found a lot of info on how oscillation can start and how to prevent it, but not with this particular method. I'm guessing the point is to provide a guaranteed minimum amount of AC current leaks off in the presence of a high load resistance?

Try googing for "zobel network" - which is what it's called. It's basically to help prevent oscillation under certain load conditions - it might not be needed, but it's a GOOD IDEA to always include it in amplifiers.

 

[audioguru]

Ahhhh, after taking a look at the spec sheet for the 741 and the TL071, I see what you mean!

Did you notice that the old 741 doesn't even have a spec for noise?

I'm still pretty unclear on how the 0.5uF cap and 10ohm resistor on the output prevent an LM386 from oscillating. Using google, I found a lot of info on how oscillation can start and how to prevent it, but not with this particular method. I'm guessing the point is to provide a guaranteed minimum amount of AC current leaks off in the presence of a high load resistance?

The cap should be 0.05uF (use 0.047uF), not 0.5uF.
Correct, it provides a minimum amount of load at high frequencies because the inductance of a speaker causes it to be a high impedance at high frequencies.
Look at Zobel Network in Google and look at the datasheet and application notes of the LM380 for more info.

TJ[/QUOTE]

 

[Nigel Goodwin]

Did you notice that the old 741 doesn't even have a spec for noise?

Does that mean it doesn't have any noise at all?

The cap should be 0.05uF (use 0.047uF), not 0.5uF.

Interestingly, a couple of years ago my daughter was at a "musical youth club" sort of thing, actually run by the police. She was playing bass guitar when the Carlsbro 80W bass combo she was using started smoking!. So when I went to pick her up I was asked if I would have a look at it, so I took it home.

When I got it to pieces all that was wrong was the Zobel capacitor had blown to pieces - this was a totally unexpected occurance, I've never seen it happen before. However, the Zobel capacitor was an ELECTROLYTIC - and I've never seen that before either, and it was a far higher value than I would I would ever use (perhaps 10uF or 22uF). My immediate impression was to stick a 0.1uF in, as I always use that value when I build amps - but as I only live a few miles from the Carlsbro factory, I rang them and asked to speak to technical. The amp, by the way, used a Thomson IC - can't remember the number, but one of the popular 'chip amp' types. The technical guy asked me a number of questions, as apparently the Zobel capacitor had changed value a number of times during production - but the outcome was to check the value of the resistor, and fit a 0.1uF capacitor!.

 

[audioguru]

I have also seen products that were built with a wrong part value.
A high end (really expensive) locally made speaker sounded bad, it was missing its midrange. I measured it with pink noise and a real-time-analyser to prove it. A coil in its crossover was made with the wrong value.

 

[systemloc]

Ok, after looking at several different hobby websites and whatnot, as well as raiding the local RadioShack once more, I came up with a TL082 (Yay! ). I want to make a two stage amp from this with volume trimmer. The only thing I could dig up that I could fit easily was a 100K hm: stereo trimmer, R3. Since it's got two pots ganged together, I figured I could wire them both into both TL082 stages' negative feedback. I don't know if it is acceptable to control both stages' gain at the same time, or if I should only adjust a single stage. If I should only adjust a single stage, I'm guessing that it would be better to adjust the secondary stage. This way the primary stage would output a constant gain at low noise to ensure a strong clean signal for the secondary, correct?

I drew out and attached my design. Any obvious flaws? I was unsure of how to couple the two, so I borrowed this idea from a hobby site. I have a 1uF tantalum capacitor, C3, and a 10K hm: resistor, R5. Also, in order to provide both positive and negative voltage, I used two 100K hm: resistors, R1, to provide ground so that I have + and - 4.5 Volts from a single 9V battery. I'm unsure if this is sufficient voltage differential to perform my desired task. I'm guessing this sets the maximum peak to peak output possible at 9V?

Also, I found on one site the idea of sticking a 10K resistor, R7, across the output. The reason given was to provide a DC ground reference to the next circuit, since the output is decoupled. The example circuit did not have a zobel network. A zobel network doesn't look like it would provide a DC ground reference. Is it necessary to provide this ground reference? I have had previous troubles with AM reception when touching the mic, so I'm wondering if this would solve the ground problem. Also, if this is a good thing to have, should I throw in a zobel network in addition, or does this resistance provide sufficient load as well?

Lastly, the site I used as a guide to design this circuit used 10uF capacitors to decouple at both input and output, C2 and C4. In the previous designs discussed, the value for the output cap was larger. Using a reactance calculator, it seems as if either 10uF or 22uF would have much less than 1 hm: resistance at audio frequencies. This makes me think that either value would work equally well? I picked up a few tantalum 10uF capacitors I could use for C2 and C4. Would these be reasonable to use? I'm interested to know if electrolytics are better/worse, and if 10uF is acceptable.

TJ

 

[Blueteeth]

Hi,

In my experience soundcard mic inputs are awful anyway, regardless of the mic used (or any external circuit). It seems that every sound card I've used (haven't tried the really high end ones) have a poorly designed, cheap analogue input for the mic, as its generally used for voice comms (telephone quality). Even when I wacked the sampling rate up to 48Khz still lost a lot of frequency response, and that was with a relatively clean opamp circuit...gain of 20, 1Mohm input impedence, and 22kohm output impedence.

If you are recording high quality audio (16bit+ 48Khz+) then the only real solution would be to use the 'line in'. It'll need an amplifier/buffer for the mics tiny signal but the quality is far superior.

However, if this is simply for voice comms (internet games, MSN etc..) then that circuit looks fine to me, although I haven't tried it. Beware though, the voltage on the mic socket, for the bias on my soundcard is 3V, thats just not enough for a decent preamp, you'll want more headroom, an amp powered by 9v would do nicely. here's some links:

**broken link removed**
http://www.vk1od.net/FM/SoundCardPreAmp.htm
**broken link removed**
**broken link removed** same as above
http://www.discovercircuits.com/M/mircophone.htm

good luck. I know you wanted an explaination of how that circuit work, but I tihnk these guys have really helped, plus, they can explain it better than I ever could.

Blueteeth.

 

[systemloc]

After reading some more, I'm starting to think that the tantalum caps definitely require a refinement in design. I'm thinking I need to add a DC bias possibly in order to ensure they are not reversed. My other thoughts were to replace C3 and C4 with 0.1uF ceramic discs.

I'm trying to understand the selection of the values for DC blocking caps. The primary criterion is reactance across audio frequencies, correct? Well, the inputs to the first and second stages are both 10Khm: by R2 and R5. 0.1uF is 5hm: at 30 Hz, and much less than one at 30,000 Hz. This is magnitudes smaller than 10Khm:, so I'm thinking that there would be no significant difference if I simply used 0.1uF caps. For the output, I think the output impedance of the TL082 is low, less than 100hm:, correct? In this case, perhaps the reactance at low frequencies would be more of a problem?

Any links to informative sites that describe the basics of audio circuits such as DC blocking cap selection would be much appreciated.

TJ

 

[audioguru]

After reading some more, I'm starting to think that the tantalum caps definitely require a refinement in design. I'm thinking I need to add a DC bias possibly in order to ensure they are not reversed. My other thoughts were to replace C3 and C4 with 0.1uF ceramic discs.

Tantalum caps cause severe distortion when used for coupling audio, and they are not reliable. Ceramic caps also cause distortion and are "microphonic" so are not used for coupling audio but are OK for toys. Film caps are perfect. Electrolytic, even bi-polar electrolytic aren't too bad for coupling audio.

I'm trying to understand the selection of the values for DC blocking caps. The primary criterion is reactance across audio frequencies, correct?

The reactance of a capacitor is less at high frequencies and can be ignored.
When the reactance equals the source impedance and next impedance added, then the response is down 3dB which is 0.707 times the voltage which sounds a little low.

For each of your coupling caps, the load is 10k. So if the reactance of each cap equals 10k at the lowest frequency you want then their responses all add to a drop of 9dB which is too far down at nearly 1/10th.
If you want the lowest level frequency to be 100Hz, then calculate caps with a reactance of the 10k load resistance but at 10Hz. Then when their slight drops at 100Hz add, the response will be correct.
A cap of 1.6uF has a reactance of 10k at 10Hz.