I need to make a high input impedance amplifier for a mic preamp with a single supply. I was thinking of using a TL071 (non-inverting) opamp with high resistance bias resistors (to bias the opamp to half the supply voltage) so the input impedance results is aprox. 1MegOhm. But I don't know if such a scheme would work. I figured since the input bias currents in the JFET input opamps are very low (max 200pA @ 25=BAC,
7nA full range) I could bias it with two 2.2MegOhm resistors. Also I think adding high value resistors increases noise at the input...and the overall gain of the circuit would be kind of large (~1000, or perhaps larger), so it would give me a large noise at the output. The input from the mic is in the micro-volt range.
I don`t know if a discrete bipolar solution would be better...or any other clever circuit configuratios for that matter...
** The SM58 mic and its many clones have a source impedance of 200 to 300 ohms and a self noise of about 0.25uV in the audio band - which in practice equates to an SPL of about 15 dB.
A speaking voice at a range of say 1 foot will cause the mic to generate about 10mV of signal.
Yelling into the mic with it close held as possible to the lips will generate upwards of 1 volt rms.
The mic can tolerate SPLs of up to 150dB even at low audio frequencies.
A genuine Shure SM58 has a very low impedance diaphragm of about 14 ohms followed by an internal 4:1 auto-transformer (magnetically unshielded ) wired to step up the impedance to 270 ohms, once winding resistances are taken into account. The output signal floats and is connected across pins 2 and 3 of a male XLR socket at the end of the handle.
However, I have seen a Chinese made clone of the SM58 where there was no transformer, the impedance was 600 ohms and the output signal was not floating.
The SM58 datasheet says 1.85 mV output at 94 dB spl. And 300 ohms impedance.
What's a reasonable equivalent SPL noise level for a mic amp? At some point, even in a good studio, there must be some background noise. How does that compare to typical electrical noise?
And what do you think is the equivalent noise temperature of a dynamic mic? Does brownian motion matter?
At 14 nV/rtHz and 20 kHz bandwidth, the noise voltage is sqrt(20 kHz) x 14 nV/rtHz = 2 uV.
The SM58 sensitivity is quoted as 1.85 mV@94 dB SPL, so the amplifier noise would correspond to 35 dB SPL.
Assuming 0.3 uV microphone self noise (the noise of a 300 ohm resistor at room temperature), this is 2 nV/rtHz, thus amplifiers like LT1028/1115 would not degrade the system noise figure significantly.
That requires quite a lot of screaming :-).
10 mV = 109 dB, 100 mV = 129 dB, 1 V = 149 dB with this microphone.
High. You don't want to push much opamp current noise into the 300 ohms of the mike, so lots of transistors and "low noise" opamps are out. Consider a BF862 jfet maybe, or even a few in parallel, if you want extreme quiet.
There are some jfet opamps in the 4 nv/rthz noise range, but a BF862 is 0.8. You could usually get by running one at Idss, especially if you'd be willing to trim a resistor. A pseudo-inductor drain load might work too.
A thread on dynamic microphone noise, in the context of thermodynamics and conservation of energy, would be interesting. Or maybe it's been done in another group.
According to the Shure SM58 specs, the rated output impedance is 150 ohms, actual 300 ohms (whatever that means).
For best noise performance, power matching should be used i.e. a load resistance of 150-300 ohms. However, most audio transducers are not designed for power matched conditions, instead, the frequency response is typically measured with an input resistance at e.g. 10 times the source resistance (1-10 kohm).
In RF design, power matching is the accepted practice for lowest noise figures (although some low noise devices have their lowest noise figures just above or below the power matching point).
Some diode ring mixers (such as the SBL-1) require 50 ohm load resistance on the IF port. In direct conversion receivers, the IF is the final audio. The first audio amplification stage is typically one or more transistors in parallel (or a single TO-220 transistor) connected in common base configuration to create the 50 ohm load resistance.
Of course one should use all the transducer power that's available. But driving a 300 ohm mic into a 300 ohm, 0.8 nV/rthz opamp circuit just throws away half the signal, not to mention the current noise; it would be better to dump that into a 0.8 nv/rthz jfet. Twice better, actually. The real way to get low noise is to increase the transformer ratio and use the jfet, or a jfet opamp, up to the point that the high-frequency response gets hurt. Or use a few jfets in parallel, again using more of the available signal power.
At 10 pF each, and a 300 ohm source, rolling at 20 KHz, you could use
2650 BF862s in parallel, for a voltage noise density of 16 pv/rthz. Audiophools would line up for that. You would need a good, low-noise,
40 amp power supply.
The best uncooled microwave amps are probably phemts, with noise figures around 0.3 dB. The matching networks are essentially step-up transformers, giving free, almost noiseless voltage gain before the fet gate.
The mmics I've tested tend to have input impedances in the 30-40 ohm range; I've only found one that actually hits 50 ohms. I wonder if that's the optimum nf point, or whether there's some other reason to make them that low.
On a sunny day (Sun, 13 Jun 2010 11:03:49 -0700) it happened John Larkin wrote in :
Some thoughts, not saying that any of this matters, but anyways:
First frequency characteristic, a resistive terminated 300 ohm inductor (say 'generator') behaves different from a open generator. The loading of the membrane: If you load the thing with any impedance, then you sort of put the brakes on the movement of the membrane. How did the manufacturer specify the load when measuring the frequency characteristic. Dynamic range, as Phil pointed out it is microvolts to volts, when mike close the the food opening. A Simple JFET follower is not very linear over that range. Differential input, maybe you want to drive the mike into some diff amplifier. All that taken into account I think the OPs idea of a TIL071 ?? or whatever JFET opamp is not a bad idea, but have not done the math. It would certainly score better in linearity than that example circuit diagram somebody showed here driving that mike into a BJT emitter (but good impedance matching).
As for the minimum noise level, do not get carried away, these mikes pick up all sort of hum from anything near, plus the hum and noise some run of the mill microphone cables pick up. And also keep in mind the sort of usage, sort of performance the mike is used for, so I think super low noise amps with thousands of fets as you suggest is not realistic for this purpose. But indeed maybe it would sell, maybe 300 nuvistors in parallel...
Then finally there is the parametric amplifier, and related to that, I used one of these long time ago directly in the collector of a simple one transistor
100 MHz oscillator, enough signal to FM modulate it via Vce changes to hear a watch ticking in the same room via the radio in the other :-) So.... What is the SNR of FM broadcast radio? do not remember, but I remember that experiment:-)
+9V tuning | trimmer mike 100 pF | ----------- | | |/ )_____| == === )--------------------------------------------- antenna |10p | ) 7 turns air space 8 mm diameter ---100k ---- | c ---- b e Some NPN | | ///
Is this the common-base "Norton Amplifier" configuration? or something using a similar trick?
Was that a measurement at DC, or at the operating frequency? If the former, the designers might be assuming that there will be some parasitic inductance in the input lead... a bit of shunt capacitance added right at the end of the input trace could then result in a
50-ohm resistive input Z and thus a good match for the trace?
--
Dave Platt AE6EO
Friends of Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!
I think people usually ground the source and drive the gate through a matching network. CEL/NEC make nice parts, and I think they have example circuits on their data sheets.
Wideband, measured with low-level TDR. The ERA-type mmics make nice wideband pulse amplifiers. The Zin is typically 30-40 ohms, pretty much resistive, so added reactance won't help.
The SGA3586 is the only mmic I've found that hits 50 real ohms; you can tune device current to nail it exactly. I have a graph somewhere... I used them as the front ends for a 2-axis delay-line imaging microchannel plate detector (say that fast 5 times) for the ill-fated atom-probe project.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.