High gain pre amp noise problems.

[nigelwright7557]

I have made a microphone pre amp using a simple op - amp none inverting tl084. It works great for line level but the microphone amp with a gain of 47 is quite noisy even with the input shorted. The resistors on the op amp are 470k feedback and 10k to ground through a capacitor. Its pretty bog standard.

Does anyone have any ideas on improving the signal to noise on this ?

[Mooly]

I'm no expert on mic preamps but I would suggest looking at the impedances of both the mic and your circuit values.

Is it a dynamic mic or an electret ? Dynamics are often 600 ohm or 10k and electrets with inbuilt FET preamp can drive a lowish impedance.

Consider replacing the TL084. JFET opamps are noisy (voltage noise). Do you need a quad device ? Something like the LM4562 (NE5532) and lowering the feedback resistors to say 4K7 and 100R with an appropriately sized feedback return cap would probably be much much better noise wise. Keep the input impedance of the non inverting input low too (4K7).

[Bonsai]

Mooly is right, your feedback resistors are quite high - the parallel value of 10k and 470k is up near 10k Ohms. You could drop these to 4k7 and 100 for starters as he says (or 8.2k and 220 Ohms is you are worried about loading). What input bias resistor value are you using (i.e. on the non-inverting input)?

The other problem may be around your layout - Can you share it with us?

Is the op-amp decoupled close to the supply pins

Make sure that the junction of the feedback resistors is as close as physically possible to the inverting input.

[Original Burnedfingers]

This is probably a stupid question but are you using 1% matched resistors?

[AndrewT]

The tl084 is a fairly noisy opamp.
For line level, the noise is probably acceptable.
For mic duty, I suspect no one would accept that level of relative noise.

[Conrad Hoffman]

IMO, the resistor and bypassing is icing on the cake. If you have too much broadband "hissy" noise, you just need a better opamp, suited to the impedances you're working with. JFET for high impedance, bipolar for low impedance. Once you have that, combine it with the right resistor values and bypassing and you shouldn't hear much of anything.

[Bonsai]

Agreed - the TL084 is not a low noise amp and is best suitd to GP applications. However, if this is all nigel has at his disposal (as I assumed) then making the changes I mentioned above will bring real benefits. And the feedback resistor junction close the inverting input pin is a absolute must.

[AndrewT]

This seems to suggest that smd components for the opamp and the feedback will perform better than through hole components.
Has anyone seen comparative data for the same circuits done in both styles?
Can anyone give me/us a link to good papers on this topic?

[Bonsai]

Andrew, in a closed loop amplifer, any signal (read noise/interference) appearing between the junction of the feedback resistors and the inverting input will be amplified by the loop gain. If you are using an op-amp - and especially a high open loop gain variety - this gain factor may be 90 or 100dB at low frequencies, and still 40 or 50dB at high frequencies. By keeping the feedback node as compact as possible, you can avoid noise injection. Same issue BTW on a discrete power amp - or any feedback amplifier - but the lower loop gains may mask the issue. Using SMD components can help reduce the problem because you can make the payout more compact. That said, careful attention to layout using leaded components will in my view also deliver great results.

[Bonsai]

No need to use 1% matched resistors 6B