Microphones vs. ear sensitivity

[Jim Thompson]

On Tue, 07 Jun 2005 14:08:44 GMT, "Kryten"

[snip]

I made up this preamplifier
http://www.zen22142.zen.co.uk/Circuits/Audio/dyn_mic/dyn_mic.htm
which is all-discrete so should have low noise.

[snip]

What a crap "design"....

(1) Shunt feedback stage with a pot as the source impedance.

(2) CB input stage... barf. For low noise you'd be better served with
CE at low current and low VCE.

...Jim Thompson

 

[Ban]

Okay, it is pretty small so not got a great sensing area.
And for £3 I did not expect much.

I went out and bought a dynamic mic insert for £5, and this is
labelled as having 76 dB sensitivity. It is a lot bigger and heavier.
http://www.maplin.co.uk/Module.aspx?ModuleNo=29594 (type DM13)

I made up this preamplifier
http://www.zen22142.zen.co.uk/Circuits/Audio/dyn_mic/dyn_mic.htm
which is all-discrete so should have low noise.

However I didn't find it was much improvement.
There is noticeable hiss and buzz.

Hiss seems to be coming from the first transistor (BC109, common
base) and the buzz from the mic insert, because the buzz disappears
when the signal input is shorted to ground. The hiss and buzz
disappears when there is no signal to the other transistors.

Circuit constructed on a bit of veroboard - not ideal but fairly neat
and compact.

So, is this mic insert also crap, and if so how much would I expect
to pay for a good one?

Also, is this BC109 transistor just too noisy?

Or am I going to have to make a much more complex pre-amp like this
one:
http://sound.westhost.com/project66.htm

Sorry to say, but this preamp is what you said, crap. You cannot have a
dynamic mike work into low impedance inputs, it kills the output level
alters the frequency response or even oscillates, what I strongly suspect
yours too. Usually the minimum load is 1k, 4.7k-22k being usual. You also
need a balanced amplifier, so the hiss and hum cancels out. You cannot
design so easily a good mike preamp. You better use a specialized IC with
your state of knowledge, then you get at least some reference circuit and a
good result. Maybe you can get hold of some SSM2017 or the successor INA103
http://focus.ti.com/docs/prod/folders/print/ina103.html
With 1nV/rtHz you will get a much lower noise than from your circuit. Win
has made a proposition for a good Mic preamp last year, maybe you find it
with google.

 

[Kryten]

Hi all.

thanks for all the comments, very educational to this mainly digital guy.

I'd assumed the input impedance should match the source for maximum power
transfer.
Seems this does not give the best quality.

Surprised to hear batteries are noisy. I had a decent decoupler.

Never let the front-end transistor amplify flat out (grounded emitter)

crap circuit

I plugged the preamp into my good-quality hi-fi, and found far less hum than
using my TV.
No surprise there. Turning up the gain, I found I was picking up radio!
Sticking it in a crude faraday cage (okay, biscuit tin) didn't help either.

Sigh, more prototyping to do and more complex circuits.


JT> (1) Shunt feedback stage with a pot as the source impedance.

Hmm.. I thought "hello, that looks a bit odd"

JT> (2) CB input stage... barf.

I had to read up on why people used a transistor for unity gain, and found
that it was useful for buffering signals that need a low input impedance
(like 75R video) and avoiding the Miller effect.

Ban > You better use a specialized IC with your state of knowledge,
Ban > then you get at least some reference circuit and a
Ban > good result.

Yep, I'd be happy to buy a chip specifically designed for dynamic mic
preamping.
I expected they would be pretty common, but haven't seen any.

Ban > Win made a proposition for a good Mic preamp last year

Thanks, I'll look for it.

Has it been tested and the performance measured?

 

[Pooh Bear]

Hi all.

thanks for all the comments, very educational to this mainly digital guy.

I'd assumed the input impedance should match the source for maximum power
transfer.
Seems this does not give the best quality.

Indeed. Power matching was only needed when gain was expensive.

You get better results with 'voltage matching'. I.e. low source Z higher input
Z.

Graham